**Crystallization by Antisolvent Addition and Cooling**

Marco Giulietti\* and André Bernardo

*Chemical Engineering Department Federal University of São Carlos UFSCar Brasil* 

## **1. Introduction**

378 Crystallization – Science and Technology

Schmolka, I. (1977). A review of block polymer surfactants. *Journal of the American Oil* 

Sumner, J. (1926). The isolation and crystallization of the enzyme urease. Preliminary paper.

Takáts, Z.; Vékey, K. & Hegedüs, L. (2001). Qualitative and quantitative determination of

Tardieu, A.; Le Verge, A.; Riès-Kautt, M.; Malfois, M.; Bonneté, F.; Finet, S. & Belloni, L.

Tessier, P. M.; Vandrey, S. D.; Berger, B. W.; Pazhianur, R.; Sandler, S. I. & Lenhoff, A. M. (2002).

Velev, O. D.; Kaler, E. W. & Lenhoff, A. M. (1998). Protein interactions in solution

Visuri, K.; Kaipainen, E.; Kivimaki, J.; Niemi, H.; Leisola, M. & Palosaari, S. (1990). A New

Vivares, D. & Bonneté, F. (2002). X-ray scattering studies of Aspergillus flavus urate oxidase:

Vivares, D.; Astier, J.-P.; Veesler, S. & Bonnete, F. (2006). Polymorphism of urate oxidase in PEG solutions. *Crystal Growth & Design*, Vol. 6, No. pp. 287-292, ISSN: 1528-7483 Vivares, D.; Belloni, L.; Tardieu, A. & Bonneté, F. (2002). Catching the PEG-induced attractive interaction between proteins. *Eur. Phys. J. E.*, Vol. 9, No. pp. 15-25, Weber, M.; Jones, M. J. & Ulrich, J. (2008). Crystallization as a Purification Method for Jack

Precipitants. *Cryst. Growth Des.*, Vol. 8, No. 2, pp. 711-716, ISBN 1528-7483 Wulff-Pérez, M.; Torcello-Gómez, A.; Gálvez-Ruíz, M. J. & Martín-Rodríguez, A. (2009).

Ye, X.; Narayanan, T.; Tong, P.; Huang, J. S.; Lin, M. Y.; Carvalho, B. L. & Fetters, L. J. (1996).

Zheng, B.; Gerdts, C. J. & Ismagilov, R. F. (2005). Using nanoliter plugs in microfluidics to

proteins. *Acta Cryst.*, Vol. D58, No. pp. 472-479, ISSN: 0907-4449

potentials. *J. Crystal Growth*, Vol. 196, No. pp. 193-203, ISSN: 0022-0248 ten Wolde, P. R. & Frenkel, D. (1997). Enhancement of protein crystal nucleation by critical

poloxamer surfactants by mass spectrometry. *Rapid Communications in Mass* 

(1999). Proteins in solution : from X-ray scattering intensities to interaction

Self-interaction chromatography: a novel screening method for rational protein crystallization. *Acta Cryst.*, Vol. D58, No. 10 Part 1, pp. 1531-1535, ISSN: 0907-4449 Vekilov, P. G. (2010). Nucleation. *Crystal Growth & Design*, Vol. 10, No. 12, pp. 5007-5019,

characterized by light and neutron scattering: comparison of lysozyme and chymotrypsinogen. *Biophys. J.*, Vol. 75, No. 6, pp. 2682-2697, ISSN: 0006-3495 Vérétout, F.; Delaye, M. & Tardieu, A. (1989). Molecular basis of eye lens transparency.

Osmotic pressure and X-ray analysis of alpha-crystallin solutions. *J. Mol. Biol.*, Vol.

Method for Protein Crystallization Using High Pressure. *Nature Biotechnology*, Vol.

towards a better understanding of PEG effects on the crystallization of large

Bean Urease: On the Suitability of Poly(Ethylene Glycol), Li2SO4, and NaCl as

Stability of emulsions for parenteral feeding: Preparation and characterization of o/w nanoemulsions with natural oils and Pluronic f68 as surfactant. *Food* 

Depletion interactions in colloid-polymer mixtures. *Phys. Rev. E*, Vol. 54, No. 6, pp.

facilitate and understand protein crystallization. *Current Opinion in Structural* 

*Chemists' Society*, Vol. 54, No. 3, pp. 110-116,

*Spectrometry*, Vol. 15, No. 10, pp. 805-810,

205, No. 4, pp. 713-728, ISSN: 0022-2836

*Hydrocolloids*, Vol. 23, No. 4, pp. 1096-1102,

*Biology*, Vol. 15, No. pp. 548–555, ISSN: 0959-440X

6500-6510, ISSN: 1063-651X

WOS:000284675100001

8, No. 6, pp. 547-549,

*Journal of Biological Chemistry*, Vol. 69, No. 2, pp. 435-441,

density fluctuations. *Science*, Vol. 277, No. pp. 1975-1978,

Crystallization is the second most important separation process in chemical industry after distillation. Crystallization consists of a solid disperse phase formation into a continuous medium, which usually is a liquid solution in industrial systems. This solid phase formation occurs in two main steps: the appearance of transition structures between solid and fluid phase, or nucleation; and the growth of these structures into solid particles, crystals. The solution concentration must be higher than the equilibrium concentration at that temperature (solubility) in order to nucleation and crystal growth occur. The difference between actual concentration and equilibrium concentration is called supersaturation and is the driving force of crystallization. Supersaturation can be generated in the system by cooling, solvent evaporation, or changing of medium – addition of an antisolvent which reduces the solute solubility in the resultant system, or changing the solute by chemical reaction producing another substance with much lower solubility. Frequently other secondary processes occur, as agglomeration and breakage of those particles, which affect the final product (crystal) size distribution.

Fig. 1. Mechanisms of nucleation

<sup>\*</sup> Corresponding Author

Crystallization by Antisolvent Addition and Cooling 381

RT r R m Rn kTr T r Tr VM N

the chemical potential (4), the critical Gibbs energy for homogeneous nucleation is:

between equilibrium concentration and actual solution concentration, was reached.

Fig. 3. Surface energies in heterogeneous nucleation. cs, sl, cl are, respectively surface energies between cluster and surface, surface and liquid, and cluster and liquid.

express the affinity between the cluster and the surface (Mullin, 2001):

Therefore, in heterogeneous nucleation there is a contact angle and a factor which

 sl cs cl

cos (6)

In primary heterogeneous nucleation, an alien surface (dust particle, reactor wall etc.)

2M 2 2 2 kT ln S ln S

where is the molecular volume, NA is the Avogadro number, n is the number of mols, and

Considering the homogeneous nucleus radius definition (2) and the obtained expression for

crit 2 <sup>16</sup> <sup>G</sup>

The critical Gibbs energy may be interpreted as an energy barrier to be transposed to the appearance of the nucleus, as primary nucleation depends on ordered clustering of solute until a critical size (or critical radius if cluster and nucleus are considered spherical). The practical consequences of this barrier are the metastability and the induction time. Metaestability is the state of a clear supersaturated solution, i.e. there is driving force for crystallization but there is no crystal in the system yet. The metastable state may persist for different induction times – time interval between the supersaturation imposition and the appearance of first nucleus – depending on supersaturation degree. As this supersaturation degree increases, smaller induction times are observed until the limit of instantaneous nucleation. At this point, the metastable zone width (abbreviated as MZW), i.e. a region

3 2

3 kT ln S (5)

 A

(4)

Manipulating equation (3) and considering equation (2):

k is the constant of Boltzmann.

facilitates nucleation (Figure 3):

Nucleation may occur by different mechanisms. It may be primary, when is independent of crystal presence in solution, or secondary, when occurs as result of crystal presence in solution. When nucleation is primary, it may occur as a homogeneous mechanism, i.e. spontaneous depending only of supersaturation degree, or heterogeneous when is facilitated by alien surfaces (reactor or agitator wall, for instance). These different mechanisms are shown in Figure 1.

Primary homogeneous nucleation is the result of successive equilibrium of solute clustering (Figure 2). When a stable cluster is formed, it remains in solution as a transition structure from which the solid dispersed phase may appear. This stable cluster is called nucleus (Mullin, 2001).

$$\begin{aligned} \mathbf{A} + \mathbf{A} &\leftrightarrow \mathbf{A}\_2 \\ \mathbf{A}\_2 + \mathbf{A} &\leftrightarrow \mathbf{A}\_3 \\ \vdots \\ \mathbf{A}\_{n-1} + \mathbf{A} &\leftrightarrow \mathbf{A}\_n \end{aligned}$$

Fig. 2. Solute clustering

The excess Gibbs energy is result of a resistance for the formation of a new surface, or the product of the surface tension and the cluster area, and a tendency to reach equilibrium, product of chemical potential and cluster volume:

$$
\Delta \mathbf{G} = \Delta \mathbf{G}\_{\rm S} + \Delta \mathbf{G}\_{\rm V} = \boldsymbol{\gamma} \cdot \mathbf{A}\_{\rm cluster} - \Delta \boldsymbol{\mu} \cdot \mathbf{V}\_{\rm cluster} = \boldsymbol{\gamma} \cdot 4 \pi \mathbf{r}^2 - \Delta \boldsymbol{\mu} \cdot \frac{4}{3} \pi \mathbf{r}^3 \tag{1}
$$

where is the surface energy of the system solute-solvent, is the chemical potential, Acluster is the cluster surface area, Vcluster is the cluster volume, and r is the cluster radius (therefore, there is an implicit hypothesis considering cluster spherical)

Because the clusters start to grow from very small sizes (theoretically from the solute volume), it is possible to infer that initially the surface term of the excess Gibbs energy is larger than the volume term. As clusters continue to grow, they reach a critical size which corresponds to the maximum value of the excess Gibbs energy (the critical Gibbs energy). From this point on, Gibbs energy starts to decrease, and crystals grow from the existent stables clusters (nuclei). Mathematically, the critical size is calculated deriving equation (1) and equating to zero the value:

$$\mathbf{r}\_{\rm crit} = \frac{2\gamma}{\Delta\mu} \tag{2}$$

The above equations as well as Gibbs-Thomson equation for non-electrolytes (Mullin, 2001) may be manipulated to obtain an expression for the chemical potential:

$$\ln \frac{\mathbf{c}(\mathbf{r})}{\mathbf{c}^\*} = \ln \mathbf{S} = \frac{2 \mathbf{M} \mathbf{y}}{\mathbf{R} \mathbf{T} \rho \mathbf{r}} \tag{3}$$

where M is the molecular weight of solute, S is the supersaturation ratio defined as c/c\*, R is the gas constant, T the absolute temperature, the crystal density and r the particle radius.

Nucleation may occur by different mechanisms. It may be primary, when is independent of crystal presence in solution, or secondary, when occurs as result of crystal presence in solution. When nucleation is primary, it may occur as a homogeneous mechanism, i.e. spontaneous depending only of supersaturation degree, or heterogeneous when is facilitated by alien surfaces (reactor or agitator wall, for instance). These different

Primary homogeneous nucleation is the result of successive equilibrium of solute clustering (Figure 2). When a stable cluster is formed, it remains in solution as a transition structure from which the solid dispersed phase may appear. This stable cluster is called nucleus

S V cluster cluster

(therefore, there is an implicit hypothesis considering cluster spherical)

may be manipulated to obtain an expression for the chemical potential:

 

AA A AAA

A AA

n 1 n

The excess Gibbs energy is result of a resistance for the formation of a new surface, or the product of the surface tension and the cluster area, and a tendency to reach equilibrium,

2 3

<sup>4</sup> G G G A V 4r r

where is the surface energy of the system solute-solvent, is the chemical potential, Acluster is the cluster surface area, Vcluster is the cluster volume, and r is the cluster radius

Because the clusters start to grow from very small sizes (theoretically from the solute volume), it is possible to infer that initially the surface term of the excess Gibbs energy is larger than the volume term. As clusters continue to grow, they reach a critical size which corresponds to the maximum value of the excess Gibbs energy (the critical Gibbs energy). From this point on, Gibbs energy starts to decrease, and crystals grow from the existent stables clusters (nuclei). Mathematically, the critical size is calculated deriving equation (1)

> crit

The above equations as well as Gibbs-Thomson equation for non-electrolytes (Mullin, 2001)

 c r 2M ln ln S

where M is the molecular weight of solute, S is the supersaturation ratio defined as c/c\*, R is the gas constant, T the absolute temperature, the crystal density and r the particle radius.

2

r (2)

c \* RT r (3)

3 (1)

2 3

2

mechanisms are shown in Figure 1.

(Mullin, 2001).

Fig. 2. Solute clustering

and equating to zero the value:

product of chemical potential and cluster volume:

Manipulating equation (3) and considering equation (2):

$$\ln\text{S} = \frac{2\text{M}\text{y}}{\text{RT}\text{/}\text{r}} = \frac{2\gamma\text{v}}{\upsilon\frac{\text{R}}{\text{V}}\text{T}\frac{\text{m}}{\text{M}}\text{r}} = \frac{2\gamma\text{v}}{\text{R}\text{n}} = \frac{2\gamma\text{v}}{\text{kTr}} \Rightarrow \Delta\mu = \frac{\text{kT}\ln\text{S}}{\upsilon} \tag{4}$$

where is the molecular volume, NA is the Avogadro number, n is the number of mols, and k is the constant of Boltzmann.

Considering the homogeneous nucleus radius definition (2) and the obtained expression for the chemical potential (4), the critical Gibbs energy for homogeneous nucleation is:

$$
\Delta \mathbf{G}\_{\rm crit} = \frac{16\pi \mathbf{\hat{v}}^3 \mathbf{\hat{v}}^2}{\Re \left( \mathbf{k} \mathbf{T} \ln \mathbf{S} \right)^2} \tag{5}
$$

The critical Gibbs energy may be interpreted as an energy barrier to be transposed to the appearance of the nucleus, as primary nucleation depends on ordered clustering of solute until a critical size (or critical radius if cluster and nucleus are considered spherical). The practical consequences of this barrier are the metastability and the induction time. Metaestability is the state of a clear supersaturated solution, i.e. there is driving force for crystallization but there is no crystal in the system yet. The metastable state may persist for different induction times – time interval between the supersaturation imposition and the appearance of first nucleus – depending on supersaturation degree. As this supersaturation degree increases, smaller induction times are observed until the limit of instantaneous nucleation. At this point, the metastable zone width (abbreviated as MZW), i.e. a region between equilibrium concentration and actual solution concentration, was reached.

In primary heterogeneous nucleation, an alien surface (dust particle, reactor wall etc.) facilitates nucleation (Figure 3):

Fig. 3. Surface energies in heterogeneous nucleation. cs, sl, cl are, respectively surface energies between cluster and surface, surface and liquid, and cluster and liquid.

Therefore, in heterogeneous nucleation there is a contact angle and a factor which express the affinity between the cluster and the surface (Mullin, 2001):

$$\cos \Theta = \frac{\chi\_{\rm cl} - \chi\_{\rm cs}}{\chi\_{\rm cl}} \tag{6}$$

Crystallization by Antisolvent Addition and Cooling 383

as supersaturation C (the difference between C and C\*), impeller rotation W, and

Seeding is a common practice in industry. It consists of adding a small quantity of crystal in the supersaturated solution that will facilitate the crystal growth by the existence of a surface. If the process of crystallization is seeded, it is expected that secondary nucleation be

The crystal growth may be defined as the variation in time of the characteristic size of the

dL <sup>G</sup>

There are many theories to explain and model crystal growth. Briefly explaining, these theories may be grouped in three sets: surface energy theories, that postulate that shape of growing crystals search the minimum energy condition, are limited to molecular modeling studies; diffusion theories, which states that crystal growth is limited by the diffusion of the solute to the crystal surface; adsorption theories, which states that the integration of the solute molecule to the crystal surface is the rate-determining step. For the crystallization of melts, the crystal growth may be limited by heat release. It seems quite obvious that crystal growth may be limited by diffusion of solute to the surface, integration of the solute in the surface, or by the heat release depending on the system composition or the process conditions (Mulin, 2001). In the engineering practice, it is common to describe crystal

where kG is a constant which may vary with temperature according to an Arrhenius-like

As already mentioned, supersaturation may be generated by changing the solubility of the system by the addition of an antisolvent – a liquid miscible with the solvent which reduces solute solubility in this new mixed solvent. An advantage of the antisolvent crystallization is that the process can be carried out at temperatures near the ambient temperature. It is quite convenient for heat-sensitive substances. Also, the process would demand less energy than a solvent evaporation process. However, the solvent-antisolvent mixture must be separated in order to recover and recycle one or both solvents. Another advantage of antisolvent crystallization is that the change in solvent composition may favor one crystalline structure in those cases where the solute may crystallize in two or more crystalline phases (what is called polymorphism), and only one of them is desired for product application. Because of these characteristics, antisolvent crystallization has been widely used to crystallize

equation, and g is the crystal growth order, generally a number between 1 and 2.

where G is the crystal growth rate and L is the characteristic size of the crystal.

growth rate as a power law (Myerson and Ginde, 2002):

**1.1 Antisolvent crystallization** 

0 in <sup>j</sup> B k WM C <sup>N</sup> T (12)

dt (13)

<sup>g</sup> G k C (14) G

concentration of solids MT in a power law equation (Myerson and Ginde, 2002):

the dominant mechanism of nucleation.

crystal:

$$\Phi = \frac{\left(2 + \cos \theta\right) \left(1 - \cos \theta\right)^2}{4} \tag{7}$$

The factor varies from 0 (total affinity between cluster and surface) to 1 (no affinity). The primary nucleation rate, J, is the rate of appearance of nucleus in a given volume:

$$\mathbf{J} = \frac{\mathbf{dN}}{\mathbf{Vdt}} \tag{8}$$

where N is the number of nucleus, t is time and V is the system volume.

For its characteristics, primary nucleation may be modeled as an Arrhenius thermally activated rate:

$$\mathbf{J} = \mathbf{A} \exp\left(-\frac{\mathbf{A}\mathbf{G}}{\mathbf{k}\mathbf{T}}\right) \tag{9}$$

Equations 5, 7 and 9 allow determining a primary nucleation rate:

$$\mathbf{J} = \mathbf{A} \exp\left(-\frac{16\mathbf{f}\gamma^3 \mathbf{u}^2}{3\mathbf{k}^3 \mathbf{T}^3 \left(\ln \mathbf{S}\right)^2}\right) \tag{10}$$

Where f is a factor that equals 1 for homogeneous nucleation or is lower than 1 for heterogeneous nucleation (Bernardo et al, 2004). Consequently, it should be expected that in industrial systems heterogeneous nucleation occurs before homogeneous, and the metastability or the induction times be smaller than in homogeneous nucleation.

It is reported that melts frequently demonstrate abnormal nucleation characteristics – nucleation rate follows the expected exponential curve as supersaturation is imposed to the system, but reaches a maximum and decreases for higher supersaturation values. It was suggested that the viscosity of melts increase starkly with cooling and restricts molecular movement inhibiting the formation of ordered structures. It was modeled by a modification in equation 10, including a 'viscosity' term (Mullin, 2001):

$$\mathbf{J} = \mathbf{A} \exp\left(-\frac{16\mathbf{f}\gamma^3 \boldsymbol{\upsilon}^2}{3\mathbf{k}^3 \mathbf{T}^3 \left(\ln \mathbf{S}\right)^2} + \frac{\Delta \mathbf{G}^\prime}{\mathbf{k} \mathbf{T}}\right) \tag{11}$$

where G' is the activation energy for molecular motion across the system, and is very large for viscous liquids and glasses.

When there are crystals suspended in solution, they may be shorn by relative movement of liquid or broken by collision with other crystals or with crystallizer or impellers surfaces. The consequence of these mechanical processes is the appearance in suspension of small embryos which allow the growth of new crystals, i.e. secondary nucleation. As it exists because of physical interactions between crystals and the system, it cannot be modeled by thermodynamic equations, as primary nucleation. The wide-spread solution is to relate secondary nucleation rate B0 with process variables which may cause secondary nucleation,

The factor varies from 0 (total affinity between cluster and surface) to 1 (no affinity). The

For its characteristics, primary nucleation may be modeled as an Arrhenius thermally

 

Where f is a factor that equals 1 for homogeneous nucleation or is lower than 1 for heterogeneous nucleation (Bernardo et al, 2004). Consequently, it should be expected that in industrial systems heterogeneous nucleation occurs before homogeneous, and the

It is reported that melts frequently demonstrate abnormal nucleation characteristics – nucleation rate follows the expected exponential curve as supersaturation is imposed to the system, but reaches a maximum and decreases for higher supersaturation values. It was suggested that the viscosity of melts increase starkly with cooling and restricts molecular movement inhibiting the formation of ordered structures. It was modeled by a modification

metastability or the induction times be smaller than in homogeneous nucleation.

3 2 3 3 2

3 2 3 3 2

where G' is the activation energy for molecular motion across the system, and is very large

When there are crystals suspended in solution, they may be shorn by relative movement of liquid or broken by collision with other crystals or with crystallizer or impellers surfaces. The consequence of these mechanical processes is the appearance in suspension of small embryos which allow the growth of new crystals, i.e. secondary nucleation. As it exists because of physical interactions between crystals and the system, it cannot be modeled by thermodynamic equations, as primary nucleation. The wide-spread solution is to relate secondary nucleation rate B0 with process variables which may cause secondary nucleation,

primary nucleation rate, J, is the rate of appearance of nucleus in a given volume:

<sup>2</sup> 2 cos 1 cos

4 (7)

dN <sup>J</sup> Vdt (8)

<sup>G</sup> J Aexp kT (9)

16f J Aexp 3k T ln S (10)

16f G' J Aexp 3k T ln S kT (11)

where N is the number of nucleus, t is time and V is the system volume.

Equations 5, 7 and 9 allow determining a primary nucleation rate:

in equation 10, including a 'viscosity' term (Mullin, 2001):

for viscous liquids and glasses.

activated rate:

as supersaturation C (the difference between C and C\*), impeller rotation W, and concentration of solids MT in a power law equation (Myerson and Ginde, 2002):

$$\mathbf{B}^0 = \mathbf{k}\_{\rm N} \mathbf{W}^i \mathbf{M}\_{\rm I}^l \Delta \mathbf{C}^n \tag{12}$$

Seeding is a common practice in industry. It consists of adding a small quantity of crystal in the supersaturated solution that will facilitate the crystal growth by the existence of a surface. If the process of crystallization is seeded, it is expected that secondary nucleation be the dominant mechanism of nucleation.

The crystal growth may be defined as the variation in time of the characteristic size of the crystal:

$$\mathbf{G} = \frac{\mathbf{d}\mathbf{L}}{\mathbf{d}\mathbf{t}}\tag{13}$$

where G is the crystal growth rate and L is the characteristic size of the crystal.

There are many theories to explain and model crystal growth. Briefly explaining, these theories may be grouped in three sets: surface energy theories, that postulate that shape of growing crystals search the minimum energy condition, are limited to molecular modeling studies; diffusion theories, which states that crystal growth is limited by the diffusion of the solute to the crystal surface; adsorption theories, which states that the integration of the solute molecule to the crystal surface is the rate-determining step. For the crystallization of melts, the crystal growth may be limited by heat release. It seems quite obvious that crystal growth may be limited by diffusion of solute to the surface, integration of the solute in the surface, or by the heat release depending on the system composition or the process conditions (Mulin, 2001). In the engineering practice, it is common to describe crystal growth rate as a power law (Myerson and Ginde, 2002):

$$\mathbf{G} = \mathbf{k}\_{\mathbf{G}} \boldsymbol{\Delta} \mathbf{C}^{\mathbf{g}} \tag{14}$$

where kG is a constant which may vary with temperature according to an Arrhenius-like equation, and g is the crystal growth order, generally a number between 1 and 2.

#### **1.1 Antisolvent crystallization**

As already mentioned, supersaturation may be generated by changing the solubility of the system by the addition of an antisolvent – a liquid miscible with the solvent which reduces solute solubility in this new mixed solvent. An advantage of the antisolvent crystallization is that the process can be carried out at temperatures near the ambient temperature. It is quite convenient for heat-sensitive substances. Also, the process would demand less energy than a solvent evaporation process. However, the solvent-antisolvent mixture must be separated in order to recover and recycle one or both solvents. Another advantage of antisolvent crystallization is that the change in solvent composition may favor one crystalline structure in those cases where the solute may crystallize in two or more crystalline phases (what is called polymorphism), and only one of them is desired for product application. Because of these characteristics, antisolvent crystallization has been widely used to crystallize

Crystallization by Antisolvent Addition and Cooling 385

The major amount of crystallization studies for sugars, sucrose is almost an exception, were made concerning the undesirable crystallization or the need to impose the crystallization of these substances in food formulation (Hartel, 2001; Hartel et al., 2011). Very little research concerns industrial crystallization processes, and even for sucrose, industrial processes remain based on practical operational practice. Fundamental studies on sugars solutions behavior have been being done, almost always concerning the

In fact, the most used sugars in industrial formulations – sucrose, glucose, fructose, lactose – seem to have specific characteristics that become the study of their crystallization quite complex. These substances are all very soluble in water, which implies that their industrial processes of crystallization will have very high initial (feed) concentration. Figure 4 shows

Fig. 4. Solubility in water of fructose (Silva, 2010), sucrose (Ouiazzane et al., 2008),

It may be observed in Figure 1 that despite of lactose, all the other sugar have solubilities larger than 40 % weight. Even lactose have its solubility larger than 30 % weight at 60°C,

Other important physical properties to be considered are solution viscosity and glass transition temperature Tg. Chirife and Buera (1997) presented a simple equation to predict the viscosity of sugar solutions at 20°C. Table 1 utilized this equation to calculate viscosities for selected sugars in a 60% weight solution. Hartel et al. (2011) compiled Tg values for different sugars (Table 2). For temperature values around or below Tg, the solution will be

**2. Crystallization of sugars** 

behavior of food formulation.

the solubility of sucrose, glucose, fructose, lactose in water.

glucose (Alves et al., 2007), lactose (Brito, 2007).

expected to exhibit a very low mobility.

where the others haves solubilities larger than 60 % weight.

pharmaceutical products, which are generally sensitive to degradation by heating and frequently have polymorphism occurrence.

Takiyama et al. (2010) utilized the antisolvent crystallization to control the formation of two possible polymorphs of indomethacin (alpha and gamma forms).To obtain only the desired stable polymorph, it is required to avoid the precipitation of meta-stable polymorph crystals; they postulate that in the antisolvent crystallization, the solubility profiles are essential data for crystallization operation design to selectively isolate the target polymorph. In antisolvent crystallization, indomethacin was dissolved in acetone and heptane as used as antisolvent. Agitation speed and rate of addition control, as well as seeding gamma crystals were done to obtain the desired gamma form. The experiments were done at 288 and 313 K.

Granberg et al. (1999) investigated the influence of solvent composition on the antisolvent crystallization of paracetamol in acetone-water mixtures where extra water was added as antisolvent and concluded that supersaturation degree and not solvent composition defines induction time. They noticed increasing nucleation and agglomeration rate with increasing initial supersaturation, but at a given initial supersaturation, the solvent composition has no clear influence on the crystalline product characteristics. Crystal growth rate showed good relationship with solubility. Their work concluded that antisolvent modifies solubility and crystal shape, but has low influence in crystallization kinetics, governed by supersaturation degree.

Analyzing data of benzoic acid crystallization in water-ethanol solution by the water addition, Kubota (2008) concluded that solvent composition has no effect on induction time or primary nucleation rate, which could be modeled considering only the supersaturation imposed to the system as water as added.

Antisolvent crystallization may be combined with cooling strategies to enhance crystallization. Sheikhzadeh et al. (2008) implemented an adaptive MIMO neuro-fuzzy logic control for crystallization of paracetamol in isopropanol-water system in which water was added as antisolvent and temperature was varied from 40 to 10°C. When seeds were added, product yield reached 99%, while unseeded experiments reached 95% product yield. Seeding allowed to significantly reduce batch time without reduction in crystal mean size.

In combined cooling antisolvent crystallization, it seems that when antisolvent is added before cooling, the results are better than the opposite. Studying crystallization of paracetamol in isopropanol-water system in which water was added as antisolvent, Knox et al. (2009) increased the yield from 78.4% to 93.5% when antisolvent was added before cooling.

Nagy et al. (2006, 2008) utilized the method of moments to model the combined cooling and antisolvent crystallization in order to obtain the optimum recipe for crystallizing lovastatin (a hypolipidemic agent in drugs) in acetone/water mixture and achieve a maximized crystal size. Compared to cooling-only strategy, antisolvent-only strategy improved the product mean size in 15%, while combined strategy improved the mean size in 22%. The width of particle size distribution was lowered in 17 and 23% when only antisolvent and combined cooling antisolvent was used, respectively, compared to cooling-only strategy.

## **2. Crystallization of sugars**

384 Crystallization – Science and Technology

pharmaceutical products, which are generally sensitive to degradation by heating and

Takiyama et al. (2010) utilized the antisolvent crystallization to control the formation of two possible polymorphs of indomethacin (alpha and gamma forms).To obtain only the desired stable polymorph, it is required to avoid the precipitation of meta-stable polymorph crystals; they postulate that in the antisolvent crystallization, the solubility profiles are essential data for crystallization operation design to selectively isolate the target polymorph. In antisolvent crystallization, indomethacin was dissolved in acetone and heptane as used as antisolvent. Agitation speed and rate of addition control, as well as seeding gamma crystals were done to obtain the desired gamma form. The experiments

Granberg et al. (1999) investigated the influence of solvent composition on the antisolvent crystallization of paracetamol in acetone-water mixtures where extra water was added as antisolvent and concluded that supersaturation degree and not solvent composition defines induction time. They noticed increasing nucleation and agglomeration rate with increasing initial supersaturation, but at a given initial supersaturation, the solvent composition has no clear influence on the crystalline product characteristics. Crystal growth rate showed good relationship with solubility. Their work concluded that antisolvent modifies solubility and crystal shape, but has low influence in crystallization

Analyzing data of benzoic acid crystallization in water-ethanol solution by the water addition, Kubota (2008) concluded that solvent composition has no effect on induction time or primary nucleation rate, which could be modeled considering only the supersaturation

Antisolvent crystallization may be combined with cooling strategies to enhance crystallization. Sheikhzadeh et al. (2008) implemented an adaptive MIMO neuro-fuzzy logic control for crystallization of paracetamol in isopropanol-water system in which water was added as antisolvent and temperature was varied from 40 to 10°C. When seeds were added, product yield reached 99%, while unseeded experiments reached 95% product yield. Seeding allowed to significantly reduce batch time without reduction in

In combined cooling antisolvent crystallization, it seems that when antisolvent is added before cooling, the results are better than the opposite. Studying crystallization of paracetamol in isopropanol-water system in which water was added as antisolvent, Knox et al. (2009)

Nagy et al. (2006, 2008) utilized the method of moments to model the combined cooling and antisolvent crystallization in order to obtain the optimum recipe for crystallizing lovastatin (a hypolipidemic agent in drugs) in acetone/water mixture and achieve a maximized crystal size. Compared to cooling-only strategy, antisolvent-only strategy improved the product mean size in 15%, while combined strategy improved the mean size in 22%. The width of particle size distribution was lowered in 17 and 23% when only antisolvent and combined

increased the yield from 78.4% to 93.5% when antisolvent was added before cooling.

cooling antisolvent was used, respectively, compared to cooling-only strategy.

frequently have polymorphism occurrence.

kinetics, governed by supersaturation degree.

imposed to the system as water as added.

crystal mean size.

were done at 288 and 313 K.

The major amount of crystallization studies for sugars, sucrose is almost an exception, were made concerning the undesirable crystallization or the need to impose the crystallization of these substances in food formulation (Hartel, 2001; Hartel et al., 2011). Very little research concerns industrial crystallization processes, and even for sucrose, industrial processes remain based on practical operational practice. Fundamental studies on sugars solutions behavior have been being done, almost always concerning the behavior of food formulation.

In fact, the most used sugars in industrial formulations – sucrose, glucose, fructose, lactose – seem to have specific characteristics that become the study of their crystallization quite complex. These substances are all very soluble in water, which implies that their industrial processes of crystallization will have very high initial (feed) concentration. Figure 4 shows the solubility of sucrose, glucose, fructose, lactose in water.

Fig. 4. Solubility in water of fructose (Silva, 2010), sucrose (Ouiazzane et al., 2008), glucose (Alves et al., 2007), lactose (Brito, 2007).

It may be observed in Figure 1 that despite of lactose, all the other sugar have solubilities larger than 40 % weight. Even lactose have its solubility larger than 30 % weight at 60°C, where the others haves solubilities larger than 60 % weight.

Other important physical properties to be considered are solution viscosity and glass transition temperature Tg. Chirife and Buera (1997) presented a simple equation to predict the viscosity of sugar solutions at 20°C. Table 1 utilized this equation to calculate viscosities for selected sugars in a 60% weight solution. Hartel et al. (2011) compiled Tg values for different sugars (Table 2). For temperature values around or below Tg, the solution will be expected to exhibit a very low mobility.

Crystallization by Antisolvent Addition and Cooling 387

water be removed and evacuated from crystal integration surface to the bulk solution to allow the growth of crystals. It seems to occur because for disaccharides in dilute or concentrated aqueous solutions, folding around the glycosidic linkage and hydrogen bonding influences very much the solution behavior. Specifically in the case of properties such as solubility, viscosity and molecular arrangements that take place before crystallization. As a general rule, high Tg sugars exhibit a greater degree of freedom to

Molinero et al. (2004) utilizing atomistic simulations investigated the nature of combination of water and glucose in supercooled solutions and concluded that there is a concentration

Bensoussi et al. (2010) compared the metastable zone width of aqueous solutions of sucrose, maltitol, mannitol and xylitol, and attributed the observed differences to the interactions between water and solute molecules, as well as the conformation of solute molecule in solution. Further, they concluded that these factors are at the origin of solution properties like viscosity, diffusivity and surface tension, which interfere in nucleation and crystal growth. They concluded that nucleation of sugars is affected by their solubility, as it affects viscosity and the consequent solution diffusivity. Besides, the stability of bonds established with water may also affect nucleation. Sucrose and xylitol, for instance, have high potential of forming stable hydrogen bonds with water, as well as more favourable water-sugar interactions than to sugar-sugar interaction, which implies in large MZW, low capacity to form spontaneous nuclei, and high hydrophilic behavior. Mannitol and maltitol have similar MZW despite the difference in their solubility in water, because of, in the case of maltitol, its high viscosity at saturation and flexibility of glucitol moiety which decreases the stability of water-maltitol interactions. On the other hand, mannitol has low affinity for water and low viscosity, which favors the conditions to form spontaneous nuclei; its rigid conformation

Shortly, as general rule, sugars have high affinity with water, which frequently implies in highly viscous, highly soluble solutions with large MZW, low ease of nucleation and small

Sugars are very polar compounds, which explain their affinity with water. The dielectric constant of water is 78.54 at 25°C. A 50 weight percent of sucrose aqueous solution has its dielectric constant equals to 60.19; a similar dextrose solution has a dielectric constant of 63.39 (Malmberg and Maryott, 1950). Dielectric constants of ethanol and acetone at 25°C are 24.55 and 20.7, respectively. As dielectric constant provides a good measure of a system polarity, it is obvious that aqueous solutions of sugars are much more polar that the common organic solvents. The solubility of a solute in aqueous solution should be decreased by the addition of an organic solvent with a dielectric constant lower than that of water. Another factor which contributes to precipitation by organic solvents is the redistribution of water and the organic solvent around solute molecule (Arakawa and Timasheff, 1985). In fact, water-organic solvent mixture cannot be regarded as a continuous medium in the vicinity of a sugar molecule, since the sugar surface may be a mosaic of regions with different polarities and different affinities for the solvent components. Furthermore, large

rearrange hydrogen bonds during changes in temperature than low Tg sugars.

limit not to have water freezing and keep a glassy state for all system.

explains the ease of nucleation and the narrow metastable zone.

organic molecules as sugars may also be excluded by steric hindrance.

**2.2 Antisolvent cooling crystallization of sugars** 

crystal growth rates.


Table 1. Viscosity at 20°C for 60% weight solution of selected sugars


Table 2. Glass transition temperatures of selected sugars

Considering a sugar crystallization process, if no seed is added, primary nucleation rate will occurs as predict equation (11), i. e. nucleation rate will grow exponentially until a maximum and decrease for higher supersaturation. Crystal growth rate will be limited by diffusion of solute to crystal face, which implies that even seeding policies has limited efficiency.

Further, metastable zone width (MZW) may have reach 30°C as reported in several references (Gharsallaoui et al. (2008); Brito (2007); Silva (2010)). The combination of large MZW, flat solubility curves (Figure 1) and high viscosity makes simple cooling crystallization practically unfeasible. Mathlouthi and Genotelle (1998) compare sucrose crystallization to a 'hurdle race', where viscosity seems to be a minor hurdle and the disassociation of hydration water a major one.

## **2.1 The role of water affinity in sugar crystallization**

Mathlouthi and Genotelle (1998) considered two steps in sucrose crystallization: diffusion of sucrose molecules from the bulk solution to the interface crystal/solution and the incorporation of these molecules to the crystal after releasing their hydration water. Utilizing X-ray diffraction and laser-Raman spectroscopy, they concluded that hydrogen bonds between sucrose molecules in concentrated solutions is so strong that it hinders completely the free diffusion of molecules, and that diffusion in concentrated solutions is not due to viscous flow, but to the transfer of water molecules from one sucrose molecule to another by rotation of these sugar molecules. Consequently, water would diffuse in the concentrated solution and sucrose molecules would remain immobile, becoming the migration of hydration water from the crystal surface to the bulk solution very likely to be the controlling step in sucrose crystal growth.

Gharsallaoui et al. (2008) studied the interactions between water and disaccharides (sucrose, maltitol, and trehalose) in saturated solution and in crystallization conditions. According to them, narrowest metastable zone width was observed for maltitol and the largest for trehalose, because of the higher affinity of trehalose for water. They conclude that the crystallization of anhydrous disaccharides in aqueous solution necessitates that hydration

Sugar Tg (°C) Glucose 31 Fructose 5-10 Sucrose 62-70 Lactose 101

Considering a sugar crystallization process, if no seed is added, primary nucleation rate will occurs as predict equation (11), i. e. nucleation rate will grow exponentially until a maximum and decrease for higher supersaturation. Crystal growth rate will be limited by diffusion of

Further, metastable zone width (MZW) may have reach 30°C as reported in several references (Gharsallaoui et al. (2008); Brito (2007); Silva (2010)). The combination of large MZW, flat solubility curves (Figure 1) and high viscosity makes simple cooling crystallization practically unfeasible. Mathlouthi and Genotelle (1998) compare sucrose crystallization to a 'hurdle race', where viscosity seems to be a minor hurdle and the

Mathlouthi and Genotelle (1998) considered two steps in sucrose crystallization: diffusion of sucrose molecules from the bulk solution to the interface crystal/solution and the incorporation of these molecules to the crystal after releasing their hydration water. Utilizing X-ray diffraction and laser-Raman spectroscopy, they concluded that hydrogen bonds between sucrose molecules in concentrated solutions is so strong that it hinders completely the free diffusion of molecules, and that diffusion in concentrated solutions is not due to viscous flow, but to the transfer of water molecules from one sucrose molecule to another by rotation of these sugar molecules. Consequently, water would diffuse in the concentrated solution and sucrose molecules would remain immobile, becoming the migration of hydration water from the crystal surface to the bulk solution very likely to be

Gharsallaoui et al. (2008) studied the interactions between water and disaccharides (sucrose, maltitol, and trehalose) in saturated solution and in crystallization conditions. According to them, narrowest metastable zone width was observed for maltitol and the largest for trehalose, because of the higher affinity of trehalose for water. They conclude that the crystallization of anhydrous disaccharides in aqueous solution necessitates that hydration

solute to crystal face, which implies that even seeding policies has limited efficiency.

Table 1. Viscosity at 20°C for 60% weight solution of selected sugars

Table 2. Glass transition temperatures of selected sugars

disassociation of hydration water a major one.

the controlling step in sucrose crystal growth.

**2.1 The role of water affinity in sugar crystallization** 

Sugar Viscosity (cP) Glucose 36,4 Fructose 32,9 Sucrose 59,4 Lactose 65,7

water be removed and evacuated from crystal integration surface to the bulk solution to allow the growth of crystals. It seems to occur because for disaccharides in dilute or concentrated aqueous solutions, folding around the glycosidic linkage and hydrogen bonding influences very much the solution behavior. Specifically in the case of properties such as solubility, viscosity and molecular arrangements that take place before crystallization. As a general rule, high Tg sugars exhibit a greater degree of freedom to rearrange hydrogen bonds during changes in temperature than low Tg sugars.

Molinero et al. (2004) utilizing atomistic simulations investigated the nature of combination of water and glucose in supercooled solutions and concluded that there is a concentration limit not to have water freezing and keep a glassy state for all system.

Bensoussi et al. (2010) compared the metastable zone width of aqueous solutions of sucrose, maltitol, mannitol and xylitol, and attributed the observed differences to the interactions between water and solute molecules, as well as the conformation of solute molecule in solution. Further, they concluded that these factors are at the origin of solution properties like viscosity, diffusivity and surface tension, which interfere in nucleation and crystal growth. They concluded that nucleation of sugars is affected by their solubility, as it affects viscosity and the consequent solution diffusivity. Besides, the stability of bonds established with water may also affect nucleation. Sucrose and xylitol, for instance, have high potential of forming stable hydrogen bonds with water, as well as more favourable water-sugar interactions than to sugar-sugar interaction, which implies in large MZW, low capacity to form spontaneous nuclei, and high hydrophilic behavior. Mannitol and maltitol have similar MZW despite the difference in their solubility in water, because of, in the case of maltitol, its high viscosity at saturation and flexibility of glucitol moiety which decreases the stability of water-maltitol interactions. On the other hand, mannitol has low affinity for water and low viscosity, which favors the conditions to form spontaneous nuclei; its rigid conformation explains the ease of nucleation and the narrow metastable zone.

Shortly, as general rule, sugars have high affinity with water, which frequently implies in highly viscous, highly soluble solutions with large MZW, low ease of nucleation and small crystal growth rates.

### **2.2 Antisolvent cooling crystallization of sugars**

Sugars are very polar compounds, which explain their affinity with water. The dielectric constant of water is 78.54 at 25°C. A 50 weight percent of sucrose aqueous solution has its dielectric constant equals to 60.19; a similar dextrose solution has a dielectric constant of 63.39 (Malmberg and Maryott, 1950). Dielectric constants of ethanol and acetone at 25°C are 24.55 and 20.7, respectively. As dielectric constant provides a good measure of a system polarity, it is obvious that aqueous solutions of sugars are much more polar that the common organic solvents. The solubility of a solute in aqueous solution should be decreased by the addition of an organic solvent with a dielectric constant lower than that of water. Another factor which contributes to precipitation by organic solvents is the redistribution of water and the organic solvent around solute molecule (Arakawa and Timasheff, 1985). In fact, water-organic solvent mixture cannot be regarded as a continuous medium in the vicinity of a sugar molecule, since the sugar surface may be a mosaic of regions with different polarities and different affinities for the solvent components. Furthermore, large organic molecules as sugars may also be excluded by steric hindrance.

Crystallization by Antisolvent Addition and Cooling 389

 T

2k f z L L

 N

Silva (2010) studied the antisolvent cooling crystallization of fructose utilizing ethanol as antisolvent. She varied the initial concentration of the aqueous solution of fructose, the quantity of added ethanol expressed as ratio ethanol/water (E/S) and the cooling rate. The agitation rate was 500 rpm and the final temperature was 30°C for all experiments. The

As it was expected, MZW decreases with added ethanol quantity. The crystals yield, not shown in table, was more than 93% of available fructose quantity for all experiments. The obtained crystals had cubic habit, and agglomeration occurred in all experiments. The crystal mean size, and the crystallization kinetics, calculated by Nývlt's method had no

> Saturation temperature (°C)

Table 3. MZW of antisolvent cooling crystallization of fructose with ethanol as antisolvent

Flood et al. (2000) studying the same system also concluded that ethanol quantity and temperature did not affect significantly the crystal growth rate. They cited other studies that

86,88 1.5 50.5 30 after 40 min 0.60 86,88 4.0 50.5 30 after 30 min 0.55 86,88 6.0 50.5 42 0.55 86,88 9.0 50.5 47 0.58 88,10 1.5 55 30 after 130 min 0.50 88,10 4.0 55 30 after 20 min 0.55 88,10 6.0 55 38.5 0.60 88,10 9.0 55 42 0.55 89,36 4.0 60 38 0.22 89,36 6.0 60 40.5 0.65 89,36 9.0 60 46.5 0.65

 4(m 1) <sup>n</sup> g3 1 n

<sup>k</sup> <sup>n</sup> ln N ln cln M ln G k g

n/g T

27M G <sup>N</sup>

g

A multi-linear regression of N as function of G and MT, gives:

**2.3 Antisolvent cooling crystallization of fructose** 

significant difference with the quantity of ethanol added (Table 4).

Ethanol (E/S)

Nucleation order may be calculated:

results of MZW are shown in table 3.

Initial concentration (% weight)

concluded the same.

vC n m n

4

(20)

(21)

(22)

Nucleation temperature (°C) Cooling rate (°C/min)

If an organic solvent is mixed to aqueous solution of sugar as antisolvent, it is possible to suppose that it would surround the hydrophobic moieties of sugar molecule surface, dehydrating sugar molecule by a steric hindrance mechanism. This dehydration of sugar molecule, exposed to a less polar medium in which it has little affinity to solvent, would decrease viscosity (increasing solute mobility) and facilitate sugar-sugar interaction. The result would be higher ease of nucleation and higher crystal growth rate.

Following, it is presented results of antisolvent cooling crystallization of fructose utilizing ethanol as antisolvent, and for lactose utilizing acetone, ethanol, and iso-propyl alcohol. Crystallization was evaluated utilizing Nýlvlt's method to calculate crystallization kinetics (Nývlt et al., 2001).

Nýlvlt's method utilizes a set of at least nine experiments of no-seeded batch cooling crystallization, with three different cooling rates and three initial concentrations, to determine the apparent average crystal growth rate, expressed as a power-law equation similar to equation (13), and the apparent average nucleation rate N , expressed as a powerlaw equation:

$$\dot{\mathbf{N}} = \mathbf{k}\_{\text{N}} \mathbf{M}\_{\text{T}}^{\text{c}} \mathbf{A} \mathbf{C}^{\text{n}} \tag{15}$$

where c value allows to comprehend the nucleation mechanism – c = 0 means true primary or secondary nucleation; c = 1 means that crystal-crystal interaction provokes nucleation; c = 2 means that friction between crystals provokes nucleation.

For all experiments, the metastable zone width Tmax is measured, and from equation:

$$\log \Delta \mathbf{T}\_{\text{max}} = \frac{1-\mathbf{m}}{\mathbf{m}} \log \frac{\mathbf{d} \mathbf{C}^\*}{\mathbf{d} \mathbf{T}} - \frac{1}{\mathbf{m}} \log \mathbf{k}\_{\text{N}} + \frac{1}{\mathbf{m}} \log \frac{\mathbf{d} \mathbf{T}}{\mathbf{d} \mathbf{t}} \tag{16}$$

where m is the apparent nucleation order and C\* is the solubility, it is possible to obtain the values of kN and m by a multiple linear regression.

The cumulative mass distributions of crystals M(L) may be described as the function:

$$\mathbf{M(L)} = 100 \left( 1 + \mathbf{z} + \frac{\mathbf{z}^2}{2} + \frac{\mathbf{z}^3}{6} \right) \exp(-\mathbf{z}) \tag{17}$$

where z = L/(Gtbatch), is the crystal dimensionless size. It is possible to calculate the z values iteratively, and the average crystal growth rate for that experiment may be calculated from the relation between z and L. Crystal mean size Lm is the L value for z = 3, and

$$\mathbf{G} = \frac{\mathbf{L}\_{\mathbf{m}}}{\Re \mathbf{t}\_{\text{batch}}} \tag{18}$$

The linear coefficient of z-L relation is called zn and

$$\text{f}\left(\mathbf{z}\_{\text{n}}\right) = 100 \left(\mathbf{1} + \mathbf{z}\_{\text{n}} + \frac{\mathbf{z}\_{\text{n}}\,^2}{2} + \frac{\mathbf{z}\_{\text{n}}\,^3}{6}\right) \exp\left(-\mathbf{z}\_{\text{n}}\right) \tag{19}$$

The nucleation rate be calculated by the following equation

If an organic solvent is mixed to aqueous solution of sugar as antisolvent, it is possible to suppose that it would surround the hydrophobic moieties of sugar molecule surface, dehydrating sugar molecule by a steric hindrance mechanism. This dehydration of sugar molecule, exposed to a less polar medium in which it has little affinity to solvent, would decrease viscosity (increasing solute mobility) and facilitate sugar-sugar interaction. The

Following, it is presented results of antisolvent cooling crystallization of fructose utilizing ethanol as antisolvent, and for lactose utilizing acetone, ethanol, and iso-propyl alcohol. Crystallization was evaluated utilizing Nýlvlt's method to calculate crystallization kinetics

Nýlvlt's method utilizes a set of at least nine experiments of no-seeded batch cooling crystallization, with three different cooling rates and three initial concentrations, to determine the apparent average crystal growth rate, expressed as a power-law equation similar to equation (13), and the apparent average nucleation rate N , expressed as a power-

where c value allows to comprehend the nucleation mechanism – c = 0 means true primary or secondary nucleation; c = 1 means that crystal-crystal interaction provokes nucleation; c =

where m is the apparent nucleation order and C\* is the solubility, it is possible to obtain the

 

where z = L/(Gtbatch), is the crystal dimensionless size. It is possible to calculate the z values iteratively, and the average crystal growth rate for that experiment may be calculated from

> <sup>m</sup> batch

 n n nn n

<sup>L</sup> <sup>G</sup>

For all experiments, the metastable zone width Tmax is measured, and from equation:

max <sup>N</sup>

The cumulative mass distributions of crystals M(L) may be described as the function:

the relation between z and L. Crystal mean size Lm is the L value for z = 3, and

c n N kM C N T (15)

1 m dC \* 1 1 dT log T log log k log m dT m m dt (16)

z² z³ M L 100 1 z exp z 2 6 (17)

z² z³ f z 100 1 z exp z 2 6 (19)

3t (18)

result would be higher ease of nucleation and higher crystal growth rate.

2 means that friction between crystals provokes nucleation.

values of kN and m by a multiple linear regression.

The linear coefficient of z-L relation is called zn and

The nucleation rate be calculated by the following equation

(Nývlt et al., 2001).

law equation:

$$\dot{\mathbf{N}} = \frac{27 \mathbf{M}\_{\rm T} \mathbf{G}}{2 \mathbf{k}\_{\rm v} \rho\_{\rm C} \mathbf{f} \left(\mathbf{z}\_{\rm n}\right) \left(\mathbf{L}\_{\rm m} - \mathbf{L}\_{\rm n}\right)^{4}} \tag{20}$$

A multi-linear regression of N as function of G and MT, gives:

$$\text{l}\ln\dot{\text{N}} = \ln\left(\frac{\text{k}\_{\text{N}}}{\text{k}\_{\text{g}}^{\text{n/g}}}\right) + \text{cln}\left(\text{M}\_{\text{T}}\right) + \frac{\text{n}}{\text{g}}\ln\text{G} \tag{21}$$

Nucleation order may be calculated:

$$\mathbf{m} = \frac{4(\mathbf{m} - 1)}{3\frac{\mathbf{g}}{\mathbf{n}} + 1} \tag{22}$$

#### **2.3 Antisolvent cooling crystallization of fructose**

Silva (2010) studied the antisolvent cooling crystallization of fructose utilizing ethanol as antisolvent. She varied the initial concentration of the aqueous solution of fructose, the quantity of added ethanol expressed as ratio ethanol/water (E/S) and the cooling rate. The agitation rate was 500 rpm and the final temperature was 30°C for all experiments. The results of MZW are shown in table 3.

As it was expected, MZW decreases with added ethanol quantity. The crystals yield, not shown in table, was more than 93% of available fructose quantity for all experiments. The obtained crystals had cubic habit, and agglomeration occurred in all experiments. The crystal mean size, and the crystallization kinetics, calculated by Nývlt's method had no significant difference with the quantity of ethanol added (Table 4).


Table 3. MZW of antisolvent cooling crystallization of fructose with ethanol as antisolvent

Flood et al. (2000) studying the same system also concluded that ethanol quantity and temperature did not affect significantly the crystal growth rate. They cited other studies that concluded the same.

Crystallization by Antisolvent Addition and Cooling 391

the same temperature variation, higher initial concentration (that means higher average

Ci (% weight) Saturation (°C) Final (°C) Nucleation (°C) Cooling rate (°C/min) 42.86 70 20 68 1.00 42.86 70 30 69 0.34 42.86 70 40 68.5 0.53 35.48 60 20 57 1.10 35.48 60 30 57 0.59 35.48 60 40 58 0.33 31.03 52 10 50 0.33 31.03 52 20 49 0.30 31.03 52 30 51.5 0.42

Table 6. MZW of antisolvent cooling crystallization of lactose with isopropanol as

Ci (% weight) Saturation (°C) Final (°C) Nucleation (°C) Cooling rate (°C/min) 35.06 50 20 50 0.15 35.00 50 20 50 0.17 35.07 50 20 48 0.21 35.07 50 20 50 0.27 35.00 50 20 50 0.51 35.07 50 25 50 0.52 35.07 50 35 50 0.45 33.32 45 20 43 0.61 33.32 45 25 43 1.00 33.32 45 30 43.5 0.44 29.88 40 10 39 0.66 30.00 40 20 40 1.19 29.98 40 25 40 0.50 Table 7. MZW of antisolvent cooling crystallization of lactose with acetone as antisolvent

For acetone, larger concentration experiments had a yield of about 90%, increasing cooling rate (which means to increase average supersaturation) caused nucleation rate to increase and mean crystal size to decrease. For lower initial concentration, increasing cooling rate seems to decrease yield and mean size. Intermediate concentration experiments had an

The presented results corroborated the expectation that the addition of an organic antisolvent eases the crystallization of sugar, as the antisolvent would decrease sugar-water interaction increasing solute mobility (Miranda et al., 2009). An indirect measurement of this effect is the sugar solubility in the solvent mixture. Figure 5 presents lactose solubility in different pH values and in a mixture of 50 percent weight of water and ethanol. Figure 6

presents solubility of lactose in water and in mixture of water with different solvents.

supersaturation) implies in larger crystal sizes.

increase of mean size with cooling rate.

antisolvent


¹As in these experiments nucleation occurred after cooling cessation, calculated kinetic parameters G and N must be considered cautiously.

Table 4. Crystallization kinetics of fructose calculated by Nývlt's method

## **2.4 Antisolvent cooling crystallization of lactose**

Brito (2007) studied the antisolvent cooling crystallization of lactose utilizing ethanol (at different pH), isopropanol, and acetone as antisolvents. She varied the initial concentration of the aqueous solution of lactose, the final temperature and the cooling rate; the quantity of added antisolvent was always the same quantity of water in solution (E/S = 1), and the agitation rate was 350 rpm for all experiments. The results of MZW are shown in tables 5, 6 and 7.


Table 5. MZW of antisolvent cooling crystallization of lactose with ethanol as antisolvent

Tables 5, 6 and 7 show that adding the same quantity of antisolvent of water in solution MZW almost disappear for all studied conditions. Tables 8, 9 and 10 present the calculated kinetic parameters (Nývlt's method).

Results presented in tables 8, 9 and 10 allow to conclude that pH has an important role in ethanol cooling crystallization of lactose – crystal growth rate and yield increase with pH, and nucleation rate decreases. For isopropanol, an increase in batch time almost always implies in higher yields and larger mean sizes due to the increase in growth rate. Also, for

86,88 1.5 42.50 1.28¹ 202.1¹ 86,88 4.0 109.7 3.14¹ 12.36¹ 86,88 6.0 44.07 1.01 133.5 86,88 9.0 56.75 1.27 49.54 88,10 1.5 51.64 2.82¹ 192.5¹ 88,10 4.0 42.06 1.15¹ 224.1¹ 88,10 6.0 44.02 1.04 156.1 88,10 9.0 63.82 1.46 35.43 89,36 4.0 55.87 1.22 62.32 89,36 6.0 43.95 1.03 177.0 89,36 9.0 60.07 1.36 56.62

¹As in these experiments nucleation occurred after cooling cessation, calculated kinetic parameters G

Brito (2007) studied the antisolvent cooling crystallization of lactose utilizing ethanol (at different pH), isopropanol, and acetone as antisolvents. She varied the initial concentration of the aqueous solution of lactose, the final temperature and the cooling rate; the quantity of added antisolvent was always the same quantity of water in solution (E/S = 1), and the agitation rate was 350 rpm for all experiments. The results of MZW are shown in tables 5, 6

Ci (% weight) pH Saturation (°C) Final (°C) Nucleation (°C) Cooling rate (°C/min) 25.54 4.00 60 25 52 0.35 36.95 7.00 60 25 55 0.52 33.24 12.41 60 25 53 0.58 Table 5. MZW of antisolvent cooling crystallization of lactose with ethanol as antisolvent

Tables 5, 6 and 7 show that adding the same quantity of antisolvent of water in solution MZW almost disappear for all studied conditions. Tables 8, 9 and 10 present the calculated

Results presented in tables 8, 9 and 10 allow to conclude that pH has an important role in ethanol cooling crystallization of lactose – crystal growth rate and yield increase with pH, and nucleation rate decreases. For isopropanol, an increase in batch time almost always implies in higher yields and larger mean sizes due to the increase in growth rate. Also, for

Table 4. Crystallization kinetics of fructose calculated by Nývlt's method

(m) G (106m/s) <sup>N</sup> (10-11#/m³s)

(% weight) Ethanol (E/S) Product mean size

Initial concentration

and N must be considered cautiously.

kinetic parameters (Nývlt's method).

and 7.

**2.4 Antisolvent cooling crystallization of lactose** 


the same temperature variation, higher initial concentration (that means higher average supersaturation) implies in larger crystal sizes.

Table 6. MZW of antisolvent cooling crystallization of lactose with isopropanol as antisolvent


Table 7. MZW of antisolvent cooling crystallization of lactose with acetone as antisolvent

For acetone, larger concentration experiments had a yield of about 90%, increasing cooling rate (which means to increase average supersaturation) caused nucleation rate to increase and mean crystal size to decrease. For lower initial concentration, increasing cooling rate seems to decrease yield and mean size. Intermediate concentration experiments had an increase of mean size with cooling rate.

The presented results corroborated the expectation that the addition of an organic antisolvent eases the crystallization of sugar, as the antisolvent would decrease sugar-water interaction increasing solute mobility (Miranda et al., 2009). An indirect measurement of this effect is the sugar solubility in the solvent mixture. Figure 5 presents lactose solubility in different pH values and in a mixture of 50 percent weight of water and ethanol. Figure 6 presents solubility of lactose in water and in mixture of water with different solvents.

Crystallization by Antisolvent Addition and Cooling 393

and below 93.5°C -form is the constituent of stable crystals. It is known that mutarotation is affected by temperature, pH and solution impurities. Further, -form crystallization rate may be faster than mutarotation rate, causing mutarotation to be the rate-determining step

Fig. 5. Solubility of lactose in water at pH 4.0 (), water pH 7.0 (), water pH 12.41 (),

Fig. 6. Solubility of lactose in different solvents: water (), solution 50% weight acetone-water (), solution 50% weight ethanol-water (), solution 50% weight isopropanol-water ().

for crystallization (McLeod, 2007).

and a mixture 50% weight ethanol water ().


Table 8. Crystallization kinetics of lactose (antisolvent ethanol) calculated by Nývlt's method




Table 10. Crystallization kinetics of lactose (antisolvent acetone) calculated by Nývlt's method

Figure 5 explains the variation in the obtained yields in experiments of table 8, as solubility of lactose has a maximum value in neutral pH and also decreases with ethanol addition. However solubility curves do not explain the yields themselves. In systems where lactose has very low solubility, it would be expected that lactose has low interaction with solvent system and, therefore, high mobility of lactose molecule. A probable reason why lactose crystal yields are about 40% even in system with low solubility could be mutarotation (Miranda et al., 2009). Lactose molecule has two conformations (anomers), and forms,

Lm (m)

25.54 4.00 0.35 121.18 0.734 5.683 62.4 36.95 7.00 0.52 115.77 3.473 5.308 46.04 33.24 12.41 0.58 168.37 5.051 2.313 82.46 Table 8. Crystallization kinetics of lactose (antisolvent ethanol) calculated by Nývlt's method

Ci (% weight) Cooling rate (°C/min) Lm (m) G (105m/s) N (10-11#/m³s) Yield (%) 42.86 1 103.5 0.6029 3.6751 86.23 42.86 0.34 91.74 5.5047 7.4673 85.72 42.86 0.53 68.09 7.1675 13.769 80.42 35.48 1.1 81.44 6.1077 11.027 49.60 35.48 0.59 53.27 0.2948 18.195 75.65 35.48 0.33 89.98 6.8336 6.2648 49.20 31.03 0.33 47.00 3.1688 28.797 61.01 31.03 0.3 56.47 5.2936 21.703 73.74 31.03 0.42 80.70 5.6968 8.3657 43.85

Table 9. Crystallization kinetics of lactose (antisolvent isopropanol) calculated by Nývlt's

Ci (% weight) Cooling rate (°C/min) Lm (m) G (105m/s) N (10-11#/m³s) Yield (%) 35.06 0.15 118.71 0.3458 0.7797 88.74 35.00 0.17 99.04 0.3396 1.6071 92.63 35.07 0.21 53.98 0.2722 21.347 89.81 35.07 0.27 77.31 0.4179 4.0916 91.55 35.00 0.51 70.11 0.7130 11.599 86.60 35.07 0.52 93.24 1.1655 6.2562 47.45 35.07 0.45 106.43 1.9351 5.2757 41.55 33.32 0.61 113.84 1.3940 3.3583 73.74 33.32 1.00 72.96 1.7510 17.515 43.85 33.32 0.44 73.94 0.9859 10.476 82.40 29.88 0.66 114.01 1.2906 2.0553 55.72 30.00 1.19 82.07 2.3449 1.2720 40.91 29.98 0.50 149.77 2.2466 1.5606 58.08 Table 10. Crystallization kinetics of lactose (antisolvent acetone) calculated by Nývlt's method

Figure 5 explains the variation in the obtained yields in experiments of table 8, as solubility of lactose has a maximum value in neutral pH and also decreases with ethanol addition. However solubility curves do not explain the yields themselves. In systems where lactose has very low solubility, it would be expected that lactose has low interaction with solvent system and, therefore, high mobility of lactose molecule. A probable reason why lactose crystal yields are about 40% even in system with low solubility could be mutarotation (Miranda et al., 2009). Lactose molecule has two conformations (anomers), and forms,

G (105m/s)

N (10-11#/m³s) Yield (%)

Ci (%weight)

method

pH Cooling rate

(°C/min)

and below 93.5°C -form is the constituent of stable crystals. It is known that mutarotation is affected by temperature, pH and solution impurities. Further, -form crystallization rate may be faster than mutarotation rate, causing mutarotation to be the rate-determining step for crystallization (McLeod, 2007).

Fig. 5. Solubility of lactose in water at pH 4.0 (), water pH 7.0 (), water pH 12.41 (), and a mixture 50% weight ethanol water ().

Fig. 6. Solubility of lactose in different solvents: water (), solution 50% weight acetone-water (), solution 50% weight ethanol-water (), solution 50% weight isopropanol-water ().

Crystallization by Antisolvent Addition and Cooling 395

Crystallization of sugars may be improved by adding an organic liquid antisolvent (as alcohol or ketone) and cooling the system. This addition shuts nucleation hindrance off, as it decreases system viscosity. Simultaneously, the antisolvent competes with solute for water of hydration, throwing solute out of the solution: promoting crystallization. As the solubility of sugars in the mixture water-organic solvent is much lower than in water only, antisolvent addition increases

Sugars may present complex structures which interconvert in solution by mutarotation. As generally only one anomer crystallizes, mutarotation may decrease crystals yield. Mutarotaton may be affected by pH, temperature, and by solvent composition. However, for the studied cases of fructose and lactose, antisolvent cooling crystallization showed to be

Fructose was studied utilizing ethanol, and lactose was studied utilizing ethanol, acetone and isopropanol. It is possible to vary antisolvent addition rate and cooling rate simultaneously (Nagy et al., 2008), allowing to optimize crystal quality in industrial operations. The presented kinetic data as well as shape and size distribution measurement for antisolvent cooling crystallization corroborate its utilization in industrial operation.

Combined antisolvent and cooling crystallization is an important technique for obtaining products that are difficult to crystalize due to inherent solution properties like high viscosities, large metastable zone width, low kinetic of nucleation and growth, like sugars and others materials. The good choice of antisolvent must be done carefully with preliminary experiments that can allow to obtain high yields and easiness of solvent recovery. Optimal path to combine the two techniques, antisolvent and cooling, to obtain good crystal size distribution must be evaluated for each particular system, taking into account the couples solvent-antisolvent, solute-mixed solvent. The phase diagram of this

Alves, L. A., Almeida e Silva, J. B., Giulietti, M. (2007). Solubility of D-Glucose in Water and

Arakawa, T., Timasheff, S. N. (1985). Theory of Protein Solubility, in: *Methods of Enzymology 114*, edited by: Wyckoff, H. W., Hirs, C. H. W., Timasheff, S. N., Academic Press. Bensouissi, A., Roge, B., Mathlouthi, M. (2010). Effect of conformation and water

Bernardo, A., Calmanovici, C. E., Miranda, E. A. (2004). Induction Time as an Instrument to

Brito, A. B. N. (2007). *Study of lactose crystallization in different solvents*, PhD Thesis (in

Chirife, J., Buera, M. P. (1997). A Simple Model for Predicting The Viscosity of Sugar and

Flood, A. E., Johns, M. R., White, E. T., Crystal Growth Rates and Dispersion for D-Fructose from Aqueous Ethanol, *AIChE Journal*, Vol. 46, No. 2, 239-246 (2000).

interactions of sucrose, maltitol, mannitol and xylitol on their metastable zone

Enhance Comprehension of Protein Crystallization, *Crystal Growth and Design*, 4

the crystallization rate. Cooling the system maximizes the drowning-out effect.

ternary system is very important to evaluate that path and possible yield.

Ethanol/Water Mixtures, *J. Chem. Eng. Data*, 52, 2166-2170.

width and ease of nucleation, *Food Chemistry* 122 443–446

portuguese), Federal University of São Carlos, São Carlos.

Oligosaccharide Solutions, *Journal of Food Engineering* 33 221-226.

advantageous even considering mutarotation occurrence.

**3. Conclusion** 

**4. References** 

799-805.

Flood at al. (2000) describes fructose mutarotation issue. According to them, fructose interconverts naturally in solution in five tautomeric forms by mutarotation, but only the - D-fructopyranose form crystallizes. In aqueous solutions the mutarotation rates would be higher than the crystallization kinetics, but in aqueous ethanolic solutions mutarotation would be sufficiently slow to move the tautomeric equilibrium away from the equilibrium -D-fructopyranose. So, it would be important express supersaturation in terms of the tautomer, -D-fructopyranose. In experiments driven by Silva (2010), crystal yield was always higher than 93%, but lower than 100%. In lactose crystallization experiments of Brito (2007), higher yield were about 90%. Therefore, it seems that mutarotation may reduce crystallization yields for lactose and fructose.

From an industrial perspective, in which the maximum crystal yield is an aim, it is important to emphasize that cooling and antisolvent addition must be combined. Figure 7 shows fructose solubility in weight percentage as function of water content in solvent (a mixture of ethanol and water) for the temperatures of 20°C and 60°C.

Fig. 7. Fructose solubility in a mixture of ethanol and water at 20°C (continuous line) and at 60°C (dashed line).

The data showed in Figure 7 is based on Silva (2010) work. Figure 7 allows to understand that despite the addition of antisolvent eases crystallization, crystal yield is strongly dependent on final temperature. For 10% of water in solvent (thus 90% ethanol), fructose solubility varies from 53,13% at 60°C to 9,75% at 20°C. For instance, 1000 kg of fructose aqueous solution at 60°C (point A in Figure 7) would contain 106,9 kg of water and 893,1 kg of fructose. Adding 962,1 kg of ethanol to this system, total solvent content would be 1069 kg, with 10% weight of water. In this solvent system, more than 1200 kg of fructose could be dissolved at 60°C (point B in Figure 7) – more than initial quantity with no crystal produced – but only 115,5 kg of fructose would be soluble at 20°C (point C in Figure 7) – giving a theoretical crystal yield of 87% of total fructose, desconsidering possible mutarotation effects, or 81% of total fructose, considering that in Silva's experiments, yield was always more than 93% of available fructose.

## **3. Conclusion**

394 Crystallization – Science and Technology

Flood at al. (2000) describes fructose mutarotation issue. According to them, fructose interconverts naturally in solution in five tautomeric forms by mutarotation, but only the - D-fructopyranose form crystallizes. In aqueous solutions the mutarotation rates would be higher than the crystallization kinetics, but in aqueous ethanolic solutions mutarotation would be sufficiently slow to move the tautomeric equilibrium away from the equilibrium -D-fructopyranose. So, it would be important express supersaturation in terms of the tautomer, -D-fructopyranose. In experiments driven by Silva (2010), crystal yield was always higher than 93%, but lower than 100%. In lactose crystallization experiments of Brito (2007), higher yield were about 90%. Therefore, it seems that mutarotation may reduce

From an industrial perspective, in which the maximum crystal yield is an aim, it is important to emphasize that cooling and antisolvent addition must be combined. Figure 7 shows fructose solubility in weight percentage as function of water content in solvent (a

Fig. 7. Fructose solubility in a mixture of ethanol and water at 20°C (continuous line) and at

The data showed in Figure 7 is based on Silva (2010) work. Figure 7 allows to understand that despite the addition of antisolvent eases crystallization, crystal yield is strongly dependent on final temperature. For 10% of water in solvent (thus 90% ethanol), fructose solubility varies from 53,13% at 60°C to 9,75% at 20°C. For instance, 1000 kg of fructose aqueous solution at 60°C (point A in Figure 7) would contain 106,9 kg of water and 893,1 kg of fructose. Adding 962,1 kg of ethanol to this system, total solvent content would be 1069 kg, with 10% weight of water. In this solvent system, more than 1200 kg of fructose could be dissolved at 60°C (point B in Figure 7) – more than initial quantity with no crystal produced – but only 115,5 kg of fructose would be soluble at 20°C (point C in Figure 7) – giving a theoretical crystal yield of 87% of total fructose, desconsidering possible mutarotation effects, or 81% of total fructose, considering that in Silva's experiments, yield was always

mixture of ethanol and water) for the temperatures of 20°C and 60°C.

crystallization yields for lactose and fructose.

60°C (dashed line).

more than 93% of available fructose.

Crystallization of sugars may be improved by adding an organic liquid antisolvent (as alcohol or ketone) and cooling the system. This addition shuts nucleation hindrance off, as it decreases system viscosity. Simultaneously, the antisolvent competes with solute for water of hydration, throwing solute out of the solution: promoting crystallization. As the solubility of sugars in the mixture water-organic solvent is much lower than in water only, antisolvent addition increases the crystallization rate. Cooling the system maximizes the drowning-out effect.

Sugars may present complex structures which interconvert in solution by mutarotation. As generally only one anomer crystallizes, mutarotation may decrease crystals yield. Mutarotaton may be affected by pH, temperature, and by solvent composition. However, for the studied cases of fructose and lactose, antisolvent cooling crystallization showed to be advantageous even considering mutarotation occurrence.

Fructose was studied utilizing ethanol, and lactose was studied utilizing ethanol, acetone and isopropanol. It is possible to vary antisolvent addition rate and cooling rate simultaneously (Nagy et al., 2008), allowing to optimize crystal quality in industrial operations. The presented kinetic data as well as shape and size distribution measurement for antisolvent cooling crystallization corroborate its utilization in industrial operation.

Combined antisolvent and cooling crystallization is an important technique for obtaining products that are difficult to crystalize due to inherent solution properties like high viscosities, large metastable zone width, low kinetic of nucleation and growth, like sugars and others materials. The good choice of antisolvent must be done carefully with preliminary experiments that can allow to obtain high yields and easiness of solvent recovery. Optimal path to combine the two techniques, antisolvent and cooling, to obtain good crystal size distribution must be evaluated for each particular system, taking into account the couples solvent-antisolvent, solute-mixed solvent. The phase diagram of this ternary system is very important to evaluate that path and possible yield.

## **4. References**


**15** 

*Colombia* 

**Thin Film Growth Through Sputtering** 

During the last decade the dc and rf sputtering techniques have been used extensively in their two configurations — balanced and unbalanced magnetron. The main applications have been in the fields of industry and research. Examples of industrial applications are: decorative thin films (Raymond & Baham, 1999), hard wear-resistant thin films (Rodil & Olaya, 2006), low-friction thin films (Heimberg *et al.*, 2001) corrosion-resistant thin films (Flores *et al.*, 2006), and thin films used as a protective optical system (Stefan *et al.*, 2008), as well as maybe the most interesting applications, thin films used in the electronic industry (Monroy *et al.*, 2011). In the research field, the investigation has been oriented toward understanding the main physical mechanisms, such as: interaction between charged particles and the surface of the target material, adherence between the substrate and the deposited material, and chemical reactions near the substrate, as well as the influence of the deposit parameters (substrate temperature, working pressure, density power applied to the target). This research has produced thin films with a high degree of crystallinity and with

Moreover, researchers have made an effort to improve the system of operation. These efforts have been initiated through the so-called conventional or balanced magnetron sputtering in the early 1970s (Waits R, 1978), followed by the development of unbalanced systems in the late 1980s (Window, 1986) and its incorporation into multi-source "closed-field" systems in the early 1990s (Teer, 1989). Finally, the sputtering technique can increase the rate of deposition and ion energy by applying a unipolar high power pulse of low frequency and low duty cycle to the cathode target, referred to as high-power impulse magnetron sputtering (HiPIMS) or high-power pulsed magnetron sputtering (HPPMS). Common to all highly ionized techniques is very high density plasma. Implementing these discharges in sputter deposition technology modifies the surface of components, bringing improvements in mechanical, chemical, optical, electronic, and many other properties of the material. Highcurrent glows are transient discharges operating at simultaneously high voltage (*>* 300 V) and high current density (*>* 100mAcm*−*2). They have recently proven successful for the deposition of thin-film materials. These developments have made it possible to have an exceptionally versatile technique, suitable for the deposition of high-quality, well-adhered films of a wide range of materials with high rates of deposition. Table 1 show the main applications obtained in the last decade with the magnetron sputtering (balanced and

**1. Introduction** 

the possibility of various industrial applications.

unbalanced) rf and dc versions.

**Technique and Its Applications** 

Edgar Alfonso, Jairo Olaya and Gloria Cubillos

*Universidad Nacional de Colombia* 


## **Thin Film Growth Through Sputtering Technique and Its Applications**

Edgar Alfonso, Jairo Olaya and Gloria Cubillos *Universidad Nacional de Colombia Colombia* 

## **1. Introduction**

396 Crystallization – Science and Technology

Gharsallaoui, A., Roge, B., Mathlouthi, M. (2008). Water–disaccharides interactions in saturated solution and the crystallisation conditions, *Food Chemist*ry 106 1329–1339 Granberg, R. A., Bloch, D. G., Rasmuson, A. C. (1999). Crystallization of paracetamol in acetone-water mixtures, *Journal of Crystal Growth* 198/199 1287-1293 Hartel, R. W., Crystallization in foods, in: *Handbook of Industrial Crystallization*, edited by

Hartel, R. W., Ergun, R., Vogel, S., Phase/State Transitions of Confectionery Sweeteners:

Knox, M., Trifkovic, M., Rohani, S., Combining antisolvent and cooling crystallization: Effect

Kubota, N., An interpretation of the metastable zone width concerning primary nucleation in antisolvent crystallization, *Journal of Crystal Growth* 310 (2008) 4647–4651. Malmberg, C. G., Maryott, A. A., Dielectric Constants of Aqueous Solutions of Dextrose and

Mathlouthi, M., Genotelle, J., Role of water in sucrose crystallization, *Carbohydrate Polymers* 

McLeod, J., *Nucleation and Growth of Alpha Lactose Monohydrate*, PhD Thesis, Massey

Miranda, E. A., Bernardo, A., Hirata, G.A.M., Giulietti, M,. Crystallization of lactose and

Molinero, V., Çagn,T., Goddard, III, W.A., Mechanisms of Nonexponential Relaxation in

Myerson, A. S.; Ginde, R., Crystals, crystal growth and nucleation, in: *Handbook of Industrial Crystallization*, edited by Allan S. Myerson, 2nd edition, Butterworth, Woburn, 2002. Nagy, Z.K., Fujiwara, M., Braatz, R.D. Optimal control of combined cooling and antisolvent

Nagy, Z.K., Fujiwara, M., Braatz, R.D., Modelling and control of combined cooling and antisolvent crystallization processes, *Journal of Process Control*, 18, 856–864, 2008. Nývlt, J., Hostomský, J., Giulietti, M., *Cristalização*, EdUFSCar, São Carlos, 2001 (in portuguese). Ouiazzane, S., Messnaoui, B. Abderafi, S. Wouters, J., Bounahmidi, T., Estimation of sucrose

Sheikhzadeh, M., Trifkovic, M., Rohani, S. (2008). *Chemical Engineering Science* 63 1261 – 1272 Silva, A. T. C. R. (2010). *Study of fructose crystallization in different media*, MSc. Dissertassion

Takiyama, H., Minamisono, T., Osada, Y., Matsuoka, M. (2010). Operation design for

Mullin, J. W., Crystallization, 4th Edition, Butterworth-Heineman, London, 2001.

*Workshop in Industrial Crystallization*, Delft, The Netherlands, 2006.

(in portuguese), Federal University of São Carlos, São Carlos.

diagram, *Chemical Engineering Research and Design*, 88, 1242–1247.

Thermodynamic and Kinetic Aspects, *Comprehensive Reviews in Food Science and* 

of solvent composition on yield and metastable zone width, *Chemical Engineering* 

Sucrose, *Journal of Research of the National Bureau of Standards* Vol. 45, No. 4, October

whey protein. In: Jane Célia dos Reis Coimbra, Jane S. R.; Teixeira, J. A.. (Org.). *Engineering aspects of milk and dairy products*. 1 ed. Boca Raton, FL, EUA: CRC Press

Supercooled Glucose Solutions: the Role of Water Facilitation, *J. Phys. Chem. A* 2004,

pharmaceutical crystallization, *Proceedings of BIWIC 2006 13th International* 

crystallization kinetics from batch crystallizer data, *Journal of Crystal Growth* 310

controlling polymorphism in the antisolvent crystallization by using ternary phase

Myerson, A. S., Elsevier, 2001.

*Food Safety*, 10, 17-32 (2011)

*Science* 64 (2009) 3555 – 3563

Taylor & Francis Group, 2009, v. 1, p. 121-154.

1950, 299-303.

37 (1998) 335–342.

University, 2007.

108, 3699-3712

(2008) 798–803.

During the last decade the dc and rf sputtering techniques have been used extensively in their two configurations — balanced and unbalanced magnetron. The main applications have been in the fields of industry and research. Examples of industrial applications are: decorative thin films (Raymond & Baham, 1999), hard wear-resistant thin films (Rodil & Olaya, 2006), low-friction thin films (Heimberg *et al.*, 2001) corrosion-resistant thin films (Flores *et al.*, 2006), and thin films used as a protective optical system (Stefan *et al.*, 2008), as well as maybe the most interesting applications, thin films used in the electronic industry (Monroy *et al.*, 2011). In the research field, the investigation has been oriented toward understanding the main physical mechanisms, such as: interaction between charged particles and the surface of the target material, adherence between the substrate and the deposited material, and chemical reactions near the substrate, as well as the influence of the deposit parameters (substrate temperature, working pressure, density power applied to the target). This research has produced thin films with a high degree of crystallinity and with the possibility of various industrial applications.

Moreover, researchers have made an effort to improve the system of operation. These efforts have been initiated through the so-called conventional or balanced magnetron sputtering in the early 1970s (Waits R, 1978), followed by the development of unbalanced systems in the late 1980s (Window, 1986) and its incorporation into multi-source "closed-field" systems in the early 1990s (Teer, 1989). Finally, the sputtering technique can increase the rate of deposition and ion energy by applying a unipolar high power pulse of low frequency and low duty cycle to the cathode target, referred to as high-power impulse magnetron sputtering (HiPIMS) or high-power pulsed magnetron sputtering (HPPMS). Common to all highly ionized techniques is very high density plasma. Implementing these discharges in sputter deposition technology modifies the surface of components, bringing improvements in mechanical, chemical, optical, electronic, and many other properties of the material. Highcurrent glows are transient discharges operating at simultaneously high voltage (*>* 300 V) and high current density (*>* 100mAcm*−*2). They have recently proven successful for the deposition of thin-film materials. These developments have made it possible to have an exceptionally versatile technique, suitable for the deposition of high-quality, well-adhered films of a wide range of materials with high rates of deposition. Table 1 show the main applications obtained in the last decade with the magnetron sputtering (balanced and unbalanced) rf and dc versions.

Thin Film Growth Through Sputtering Technique and Its Applications 399

electrons can be made to circulate on a closed path on the target surface. This high current of electrons creates high-density plasma, from which ions can be extracted to sputter the target material, producing a magnetron sputter configuration (Penfold, 1995). A disadvantage of the magnetron sputtering configuration is that the plasma is confined near the cathode and is not available to active reactive gases in the plasma near the substrate for reactive sputter deposition. This difficulty can be overcome using an unbalanced magnetron configuration (see Fig. 1), where the magnetic field is such that some electrons can escape from the cathode region (Windows & Savvides, 1986). A disadvantage of the unbalanced magnetron is that the current of escaping electrons is not uniform, and the plasma generated is not

In ac sputtering, working at frequencies below about 50 kHz, the potential on the target is periodically reversed, and the ions have enough mobility so that a dc diode-like discharge, where the total potential drop is near the cathode, can be formed alternately on each electrode. The substrate chamber walls can be used as the counterelectrode. At frequencies above 50 kHz, the ions do not have enough mobility to allow establishing a dc-diode-like discharge and the applied potential is felt throughout the space between electrodes. The electrons acquire sufficient energy to cause ionizing collisions in the space between the electrodes. When an rf potential with a large peak-to-peak voltage is capacitively coupled to an electrode, an alternating positive-negative potential appears on the surface. During part of each half-cycle, the potential is such that ions are accelerated to the surface with enough energy to cause sputtering, while in alternate half-cycles, electrons reach the surface and prevent any charge buildup. Rf sputtering can be used to sputter insulating material, although the sputtering rate is low. A major disadvantage of rf sputtering of dielectric targets is that most insulating materials have poor thermal conductivity and high coefficients of thermal expansion, and are usually brittle materials. Since most of the bombarding energy produces heat, this means that large thermal gradients can be generated

that result in fracturing the target if high power levels are used.

Fig. 1. Schematic configuration of the a- balanced magnetron (intermediate) and bunbalanced magnetron (type 1and type 2 ( B. Window & N. Savvides, 1986)

uniform.


Table 1. Technological applications of thin films obtained with magnetron sputtering

In sputtering there are two means of operation: dc (diode and triode) and ac (radiofrequency), which also function in two configurations: magnetron dc (balanced and unbalanced) and magnetron ac (balanced and unbalanced).

In dc (diode) discharge, the cathode electrode is the sputtering target and the substrate is placed on the anode, which is often at ground potential (Vossen &Cuomo, 1978). The applied potential appears across a region very near the cathode, and the plasma generation region is near the cathode surface. The cathode in dc discharge must be an electrical conductor, since an insulating surface will develop a surface charge that will prevent ion bombardment of the surface. This condition implies that dc sputtering must be used to sputter simple electrically conductive materials such as metals, although the process is rather slow and expensive compared to vacuum deposition. An advantage of dc sputtering is that the plasma can be established uniformly over a large area, so that a solid large-area vaporization source can be established.

On the other hand, in dc sputtering the electrons that are ejected from the cathode are accelerated away from the cathode and are not efficiently used for sustaining the discharge. To avoid this effect, a magnetic field is added to the dc sputtering system that can deflect the electrons to near the target surface, and with appropriate arrangement of the magnets, the

TiC wear applications (Pavel, *et al.*, 2011)

nano-composites (Liu, *et al.*, 2009)

applications (Liu, *et al.*, 20007)

applications ( Olivares, *et al.*,2011)

applications ( Ramirez, *et al.*, 2011)

NbN films. ( Olaya, *et al.*, 2008,)

engineering (Bewilogua, *et al.*, 2009)

tribological applications (Kamath, *et al.*, 2011)

applications (Raymond & Baham, 1999)

(Rodil & Olaya, 2006).

(Lingxia & Xu, 2000).

Technique Coating Reference

Balanced magnetron Optical applications. (Stefan, *et al.*, 2008)

and wear applications

Diamond-like carbon films for infrared transmission

Thin films for automotive

TiAlCN/VCN films for

Table 1. Technological applications of thin films obtained with magnetron sputtering

In sputtering there are two means of operation: dc (diode and triode) and ac (radiofrequency), which also function in two configurations: magnetron dc (balanced and

In dc (diode) discharge, the cathode electrode is the sputtering target and the substrate is placed on the anode, which is often at ground potential (Vossen &Cuomo, 1978). The applied potential appears across a region very near the cathode, and the plasma generation region is near the cathode surface. The cathode in dc discharge must be an electrical conductor, since an insulating surface will develop a surface charge that will prevent ion bombardment of the surface. This condition implies that dc sputtering must be used to sputter simple electrically conductive materials such as metals, although the process is rather slow and expensive compared to vacuum deposition. An advantage of dc sputtering is that the plasma can be established uniformly over a large area, so that a solid large-area

On the other hand, in dc sputtering the electrons that are ejected from the cathode are accelerated away from the cathode and are not efficiently used for sustaining the discharge. To avoid this effect, a magnetic field is added to the dc sputtering system that can deflect the electrons to near the target surface, and with appropriate arrangement of the magnets, the

enhancement

Balanced magnetron Nano-composites of NC-

Balanced magnetron Optical properties of AlSiN

Balanced magnetron Hard coatings to decorative

Balanced magnetron Nd-Fe-B Film for magnetic

Unbalanced magnetron Hard films for corrosion

Unbalanced magnetron Nb Films for biological

Unbalanced magnetron NbO films for biological

Unbalanced magnetron Electrical applications of

unbalanced) and magnetron ac (balanced and unbalanced).

vaporization source can be established.

Unbalanced magnetron

High power pulsed magnetron sputtering

High power pulsed magnetron sputtering electrons can be made to circulate on a closed path on the target surface. This high current of electrons creates high-density plasma, from which ions can be extracted to sputter the target material, producing a magnetron sputter configuration (Penfold, 1995). A disadvantage of the magnetron sputtering configuration is that the plasma is confined near the cathode and is not available to active reactive gases in the plasma near the substrate for reactive sputter deposition. This difficulty can be overcome using an unbalanced magnetron configuration (see Fig. 1), where the magnetic field is such that some electrons can escape from the cathode region (Windows & Savvides, 1986). A disadvantage of the unbalanced magnetron is that the current of escaping electrons is not uniform, and the plasma generated is not uniform.

In ac sputtering, working at frequencies below about 50 kHz, the potential on the target is periodically reversed, and the ions have enough mobility so that a dc diode-like discharge, where the total potential drop is near the cathode, can be formed alternately on each electrode. The substrate chamber walls can be used as the counterelectrode. At frequencies above 50 kHz, the ions do not have enough mobility to allow establishing a dc-diode-like discharge and the applied potential is felt throughout the space between electrodes. The electrons acquire sufficient energy to cause ionizing collisions in the space between the electrodes. When an rf potential with a large peak-to-peak voltage is capacitively coupled to an electrode, an alternating positive-negative potential appears on the surface. During part of each half-cycle, the potential is such that ions are accelerated to the surface with enough energy to cause sputtering, while in alternate half-cycles, electrons reach the surface and prevent any charge buildup. Rf sputtering can be used to sputter insulating material, although the sputtering rate is low. A major disadvantage of rf sputtering of dielectric targets is that most insulating materials have poor thermal conductivity and high coefficients of thermal expansion, and are usually brittle materials. Since most of the bombarding energy produces heat, this means that large thermal gradients can be generated that result in fracturing the target if high power levels are used.

Fig. 1. Schematic configuration of the a- balanced magnetron (intermediate) and bunbalanced magnetron (type 1and type 2 ( B. Window & N. Savvides, 1986)

Thin Film Growth Through Sputtering Technique and Its Applications 401

In this chapter we will present the physical parameters involved in the growth of thin films; also discussed will be the influence that the growth parameters have on the degree crystallinity of the films, the chemical characterization, and the optical characterization of the films; and finally, we will discuss the residual stress, hardness, and corrosion and wear

The main physical phenomenon involved in the sputtering technique is the momentum transfer between energetic atomic-sized particles (usually ions of noble gases) and the atoms of the surface of the material (target). During the interchange of momentum, many effects can be produced on the elastic and inelastic collisions; in the first kind of collision, mainly reflected particles can be found (neutrals, ions of the target and the gas). In the second kind, the collisions can present secondary electrons, UV/visible photons, X-ray and implanted particles; schematically, Fig. 3 shows different processes that may occur during the

Fig. 3. The main physical process produced in sputtering technique (Weissmantel, 1983).

by the momentum transfer theory. These effects can be summarized as:

The momentum-transfer theory for physical sputtering was proposed early on, but was replaced by the "hot-spot" theory, in which the process of thermal vaporization is involved. The confusion about the physical process present in sputtering has only been overcome thanks to the work of Gunthersshulze in the 1920´s and 30´s and Wehner *et al.* in the 1950´s and 60´s, who demonstrated that the effects produced in sputtering could only be explained

1. The sputtering yield (ratio of atoms sputtered to the number of high-energy incident particles) depends on the mass of the bombarding particle as well as their energy. For

> 1 2 2 1

*t t* (1)

*m m U*

ion energies from 100 eV to 1000 eV the sputtering yield can be calculated as:

3 4 4( )

*m m <sup>E</sup> <sup>Y</sup>*

interaction between charged particles and the surface of the material.

resistance of thin films.

**2. Physical sputtering** 

Rf sputtering can be used with a magnetic field in balanced and unbalanced configurations to obtain a result similar to dc-like diode discharge. In fig 2 the different configurations of dc and rf sputtering are shown.

Fig. 2. Sputtering configurations a) dc sputtering, b) rf sputtering

It is important to state that in all the cases discussed above, the target and the substrate were facing (on- axis sputtering). In this configuration, the highly energetic electrons irradiate the substrates and/or the growing surface of the thin films during deposition. Off-axis sputtering reduces the effects of the irradiation of the high-energy particles. In off-axis sputtering, the substrates are settled at the outside of the discharge plasma. The thickness distribution of thin films deposited by off-axis sputtering will be larger than that for on-axis sputtering. A rotating substrate holder with a metal shadow mask is used for the reduction of the thickness distribution of the off-axis sputtering. Under a suitable design, the thickness distribution is less than 2% for substrates of 100 × 100 mm in an rf sputtering system using a 5-inch target (Shibahara *et al.*, 1987).

In this chapter we will present the physical parameters involved in the growth of thin films; also discussed will be the influence that the growth parameters have on the degree crystallinity of the films, the chemical characterization, and the optical characterization of the films; and finally, we will discuss the residual stress, hardness, and corrosion and wear resistance of thin films.

## **2. Physical sputtering**

400 Crystallization – Science and Technology

Rf sputtering can be used with a magnetic field in balanced and unbalanced configurations to obtain a result similar to dc-like diode discharge. In fig 2 the different configurations of dc

a)

b)

It is important to state that in all the cases discussed above, the target and the substrate were facing (on- axis sputtering). In this configuration, the highly energetic electrons irradiate the substrates and/or the growing surface of the thin films during deposition. Off-axis sputtering reduces the effects of the irradiation of the high-energy particles. In off-axis sputtering, the substrates are settled at the outside of the discharge plasma. The thickness distribution of thin films deposited by off-axis sputtering will be larger than that for on-axis sputtering. A rotating substrate holder with a metal shadow mask is used for the reduction of the thickness distribution of the off-axis sputtering. Under a suitable design, the thickness distribution is less than 2% for substrates of 100 × 100 mm in an rf sputtering system using a

Fig. 2. Sputtering configurations a) dc sputtering, b) rf sputtering

5-inch target (Shibahara *et al.*, 1987).

and rf sputtering are shown.

The main physical phenomenon involved in the sputtering technique is the momentum transfer between energetic atomic-sized particles (usually ions of noble gases) and the atoms of the surface of the material (target). During the interchange of momentum, many effects can be produced on the elastic and inelastic collisions; in the first kind of collision, mainly reflected particles can be found (neutrals, ions of the target and the gas). In the second kind, the collisions can present secondary electrons, UV/visible photons, X-ray and implanted particles; schematically, Fig. 3 shows different processes that may occur during the interaction between charged particles and the surface of the material.

Fig. 3. The main physical process produced in sputtering technique (Weissmantel, 1983).

The momentum-transfer theory for physical sputtering was proposed early on, but was replaced by the "hot-spot" theory, in which the process of thermal vaporization is involved. The confusion about the physical process present in sputtering has only been overcome thanks to the work of Gunthersshulze in the 1920´s and 30´s and Wehner *et al.* in the 1950´s and 60´s, who demonstrated that the effects produced in sputtering could only be explained by the momentum transfer theory. These effects can be summarized as:

1. The sputtering yield (ratio of atoms sputtered to the number of high-energy incident particles) depends on the mass of the bombarding particle as well as their energy. For ion energies from 100 eV to 1000 eV the sputtering yield can be calculated as:

$$Y = \frac{3\alpha}{4\pi^2} \frac{4m\_1m\_t}{\left(m\_1 + m\_t\right)^2} \frac{E}{U} \tag{1}$$

Thin Film Growth Through Sputtering Technique and Its Applications 403

the growth the columns and the various incidence angles at which the atoms arrive at the surface of the substrate. In the second zone, 0.3≤ Ts/Tm≤0.45, the substrate temperature increasing homogeneous which leads to a higher diffusion of the adatoms, which produce a dense structure with a higher degree of binding among the columns and the borders between columns, with borders of the grain beginning to form. In this zone, the size of the grain can be increased and the grains extended in equiaxed form, from interface substratefilm to film surface. In the third zone, Ts/Tm>0.45, the volumetric diffusion size has a great influence on the morphology of the film, due to the increase in the diffusion into the grains, which produces growth of the grains, formation of the equiaxed grain and re-crystallization.

Thorton (Thorton, 1974) elaborated the zone classification, considering the final working pressure, because this growth parameter can change both the kinetic energy of the ions that arrive at the substrate and the mean free path of the particles, which allows an increase or decrease in the bombardment of the surface of the substrate, which in turn determines the mobility the adatoms in that surface. In the Thorton model, the T zone as a transitional zone between first and second zone discussed above was added. The T zone is formed by grains defined by the limits of the low porosity. The surfaces of the T zone are denser and less

Moreover, Messier (Messier & Giri, 1984) found that in thin films of TiB2, BN and SiC there is a non-linear limit between the first zone and T zone, which is a function of the bias

These effects produce a greater crystalline structure.

rough than the two surfaces around them (see fig 4).

Fig. 4. Thorton zone model (Thonton, 1974)

where mi is the atomic mass of the bombarding incident ion, mt the atomic mass of the target, U the binding energy of the surface atom of the target, E the energy of the incident ion, and depends on the ratio of the masses of the target atom and the incident ion (monotonically increased with mi/mt ; for a ratio 0.1 = 0.17 for a ratio 10 =1.4) ( Ochiati, 1986).

The sputtering yield is sensitive to the angle-of-incidence of the bombarding particle.There is a "threshold energy" below which sputtering does not occur no matter how high the bombarding flow.


Effects 1 through 7 above are important for the growth of films by sputter deposition. This is particularly true for low pressure (<5 m Torr).

## **3. Physical models that explain the microstructure of thin film growth through sputtering**

The microstructure of thin films is related to the mobility of the adatoms during growth. The energy supply to the atoms is provided by the following mechanism: a- thermal effect, bionic bombarding and c- chemical reactions at the substrate. The effects that are produced by these mechanisms in the growth of thin films can be explained by the structure zone model (SZM). The SZM model can determine the morphology and microstructure of the films as a function of the adatoms, regardless of the kind of material. The parameters that the SZM model includes for determining the microstructure of the films are basically the substrate temperature, the final working pressure, the bias voltage applied to the substrate, and the thermal characteristics of the target. For example, in the research of Movchan and Demchishin (Movchan & Demchisshim, 1969), it has been established that the microstructure of the thin films of Ti, Ni, ZrO2 y Al2O3 is related to the normalized temperature, *i. e.* Ts/Tm (Ts is the temperature of the substrate and Tm is the melting temperature of the target). Movchan *et al.* have shown that in metallic films there are three well-defined zones.

The first zone is Ts/Tm <0.3. This zone is formed by small and elongated grains that form a columnar structure with porous morphology and weakly binding grains. The columnar structure is produced by a low diffusion, a low mobility of the atoms adsorbed by the substrate surface, and the atomic shadow effects, which are produced by varying velocity in

where mi is the atomic mass of the bombarding incident ion, mt the atomic mass of the target, U the binding energy of the surface atom of the target, E the energy of the incident ion, and depends on the ratio of the masses of the target atom and the incident ion (monotonically

The sputtering yield is sensitive to the angle-of-incidence of the bombarding particle.There is a "threshold energy" below which sputtering does not occur no matter how high the

2. Many sputtering atoms have kinetic energies much higher than those of thermally

3. Atoms ejected from single crystals tend to be ejected along the directions of the close-

4. In a polycrystalline material, some crystallographic planes are sputtered faster than are

5. Atoms sputtered from the alloy's surface are deposited in the ratio of the bulk composition, not their relative vapor pressure, as in the case of thermal vaporization. 6. Sputtering yields decrease at very high energies because the ions lose much of their

9. The secondary electron emission by ion bombardment is low, whereas high rates from thermo electron emission would be expected if high temperatures were present.

Effects 1 through 7 above are important for the growth of films by sputter deposition. This is

The microstructure of thin films is related to the mobility of the adatoms during growth. The energy supply to the atoms is provided by the following mechanism: a- thermal effect, bionic bombarding and c- chemical reactions at the substrate. The effects that are produced by these mechanisms in the growth of thin films can be explained by the structure zone model (SZM). The SZM model can determine the morphology and microstructure of the films as a function of the adatoms, regardless of the kind of material. The parameters that the SZM model includes for determining the microstructure of the films are basically the substrate temperature, the final working pressure, the bias voltage applied to the substrate, and the thermal characteristics of the target. For example, in the research of Movchan and Demchishin (Movchan & Demchisshim, 1969), it has been established that the microstructure of the thin films of Ti, Ni, ZrO2 y Al2O3 is related to the normalized temperature, *i. e.* Ts/Tm (Ts is the temperature of the substrate and Tm is the melting temperature of the target). Movchan *et al.* have shown that in metallic films there are three

The first zone is Ts/Tm <0.3. This zone is formed by small and elongated grains that form a columnar structure with porous morphology and weakly binding grains. The columnar structure is produced by a low diffusion, a low mobility of the atoms adsorbed by the substrate surface, and the atomic shadow effects, which are produced by varying velocity in

7. The sputtering yield is rather intensive to the temperature of the sputtering target.

**3. Physical models that explain the microstructure of thin film growth** 

8. There is no sputtering by electrons even at very high temperature.

increased with mi/mt ; for a ratio 0.1 = 0.17 for a ratio 10 =1.4) ( Ochiati, 1986).

bombarding flow.

others.

**through sputtering** 

well-defined zones.

evaporated atoms.

packed planes in the crystal.

energy far below the surface.

particularly true for low pressure (<5 m Torr).

the growth the columns and the various incidence angles at which the atoms arrive at the surface of the substrate. In the second zone, 0.3≤ Ts/Tm≤0.45, the substrate temperature increasing homogeneous which leads to a higher diffusion of the adatoms, which produce a dense structure with a higher degree of binding among the columns and the borders between columns, with borders of the grain beginning to form. In this zone, the size of the grain can be increased and the grains extended in equiaxed form, from interface substratefilm to film surface. In the third zone, Ts/Tm>0.45, the volumetric diffusion size has a great influence on the morphology of the film, due to the increase in the diffusion into the grains, which produces growth of the grains, formation of the equiaxed grain and re-crystallization. These effects produce a greater crystalline structure.

Thorton (Thorton, 1974) elaborated the zone classification, considering the final working pressure, because this growth parameter can change both the kinetic energy of the ions that arrive at the substrate and the mean free path of the particles, which allows an increase or decrease in the bombardment of the surface of the substrate, which in turn determines the mobility the adatoms in that surface. In the Thorton model, the T zone as a transitional zone between first and second zone discussed above was added. The T zone is formed by grains defined by the limits of the low porosity. The surfaces of the T zone are denser and less rough than the two surfaces around them (see fig 4).

Fig. 4. Thorton zone model (Thonton, 1974)

Moreover, Messier (Messier & Giri, 1984) found that in thin films of TiB2, BN and SiC there is a non-linear limit between the first zone and T zone, which is a function of the bias

Thin Film Growth Through Sputtering Technique and Its Applications 405

Material Parameter Change of orientation References TiN Ji/Ja (111) → (200) (Green *et al.*, 1995)

AlN P axis c → axis a (Lee & Y. Lee, 1994)

TaN N2/Ar (111) → (200) → (111) (Nie *et al.*, 2001)

Table 2. Relationship between the deposition parameters and film preferential orientation. Ji/Ja: ion–atom flow ratio, N2/Ar: flow ratios, t h: thickness, Vb: bias voltage, W: power supply, Ts: substrate temperature, P: pressure, ds-t: target-substrate distance y λ: free mean

In the thermodynamic model, it has been established that the growth orientation in thin films is produced at the thermodynamic equilibrium, which it reaches when the total energy (Whkl) of the system substrate-film is at a minimum. In this case, Whkl is formed by the addition of surface energy ( Shkl) and deformation energy (Uhkl), Shk passivation; energy is

2

(2)

*hkl*

*Z m*

*HN <sup>J</sup> <sup>S</sup>*

where H is the sublimation energy, Nhkl is the number of unsaturated bondings per atom at the plane (hkl), and Z is the amount of coordination among neighbors. Equation 2 does

On the other hand, deformation energy is related to intrinsic efforts in the film. Considering

<sup>2</sup> (1 ) *hk hk l ll hk EU*

where hkl is the deformation along the plane (hkl) of the film, Ehkl is the elastic module of the plane (hkl), and μ is the Poisson ratio. Research has established that the values of hkl and Ehkl are different for (111), (220) and (200) planes (Ma *et al.*, 2004), and using equations (2) and (3) an order relation has been obtained: S111>S220>S200 and U200>U220>U111. These

produced by unsaturated bonding at the surface. Shkl can be calculated by:

*hkl*

path.

not include impurities.

only two dimensions, Uhkl can be calculated by:

N2/Ar (111) → (200) (Noda *et al.,* 2002) Th Mixed → (200) → (111) (Pelleg *et al.*, 1991) Vb (200) → (111) → (220) (Kobayashi & Doy,

W (200) → (111) (Je *et al.*, 1997). T (111) → (200) (Cheng *et al.*, 2002)

ds-t axis c → axis a (Ishihara *et al.*, 1998) λ/ds-t axis a → axis c (Ishihara *et al.*, 1998)

N2/Ar axis a → axis c (Lee & Y. Lee, 1994) Vb axis a→ axis c→ axis a (Wang & Zhao,

W axis a → axis c (Lee & Y. Lee, 1994)

Vb (111) → (220) (Lin & Lee, 2000) Ji/Ja (111) → (200) (Shi *et al.*, 2002)

1984)

1997).

(3)

voltage applied to the substrate. The bias voltage of the substrate has the same effect on the mobility and adsorption of the atoms as an increase in the temperature, so when it is increased, the T zone increases and the first zone decreases, resulting in denser thin films and with a high degree of crystallinity.

The bias voltage also influences the mechanical properties of thin films; for example, with a voltage (≥100V) applied to the substrate, the deformation of the lattice increases, causing high residual efforts and low adherence between the substrate and the film.

A new three-dimensional SZM model has been produced in order to explain the simultaneous influence of the normalized temperature, the bias voltage, and the relation between ion density current (Ji) and density deposited atoms (Ja). In Fig. 5, it can observed that it is possible to obtain the third zone with high density using a combination highdensity ionic current, intermediate values of normalized temperature, and low bias voltage (Kelly & Arnell, 1998).

Fig. 5. The zone model including the relation between current density of ions and current density of atoms (Kelly & Arnell, 1999).

Other parameters of growth that affect the microstructure are the power supply to the target and gas flows. For example, changing the nitrogen flow during the growth of TiN and NbN films changes the preferred orientation of the films from the [111] direction to the [200] direction (Alfonso *et al.*, 2010). Explaining the change in the direction of growth of metallic nitride films is a very complex procedure, as is shown in table 2, where it can be observed that any parameter change can influence the direction of growth of the film. There are three models that explain the preferential growth direction: a) the thermodynamic model (Pelleg *et. al.*, 1991); b) the kinetic model (Greene *et al.*, 1995), and c) the atomic model (Petrov *et al.*, 2003).

voltage applied to the substrate. The bias voltage of the substrate has the same effect on the mobility and adsorption of the atoms as an increase in the temperature, so when it is increased, the T zone increases and the first zone decreases, resulting in denser thin films

The bias voltage also influences the mechanical properties of thin films; for example, with a voltage (≥100V) applied to the substrate, the deformation of the lattice increases, causing

A new three-dimensional SZM model has been produced in order to explain the simultaneous influence of the normalized temperature, the bias voltage, and the relation between ion density current (Ji) and density deposited atoms (Ja). In Fig. 5, it can observed that it is possible to obtain the third zone with high density using a combination highdensity ionic current, intermediate values of normalized temperature, and low bias voltage

Fig. 5. The zone model including the relation between current density of ions and current

Other parameters of growth that affect the microstructure are the power supply to the target and gas flows. For example, changing the nitrogen flow during the growth of TiN and NbN films changes the preferred orientation of the films from the [111] direction to the [200] direction (Alfonso *et al.*, 2010). Explaining the change in the direction of growth of metallic nitride films is a very complex procedure, as is shown in table 2, where it can be observed that any parameter change can influence the direction of growth of the film. There are three models that explain the preferential growth direction: a) the thermodynamic model (Pelleg *et. al.*, 1991); b) the kinetic model (Greene *et al.*, 1995), and

high residual efforts and low adherence between the substrate and the film.

and with a high degree of crystallinity.

density of atoms (Kelly & Arnell, 1999).

c) the atomic model (Petrov *et al.*, 2003).

(Kelly & Arnell, 1998).


Table 2. Relationship between the deposition parameters and film preferential orientation. Ji/Ja: ion–atom flow ratio, N2/Ar: flow ratios, t h: thickness, Vb: bias voltage, W: power supply, Ts: substrate temperature, P: pressure, ds-t: target-substrate distance y λ: free mean path.

In the thermodynamic model, it has been established that the growth orientation in thin films is produced at the thermodynamic equilibrium, which it reaches when the total energy (Whkl) of the system substrate-film is at a minimum. In this case, Whkl is formed by the addition of surface energy ( Shkl) and deformation energy (Uhkl), Shk passivation; energy is produced by unsaturated bonding at the surface. Shkl can be calculated by:

$$S\_{hkl} = \frac{\Delta H N\_{hkl}}{Z} \left\lfloor \frac{J}{m^2} \right\rfloor \tag{2}$$

where H is the sublimation energy, Nhkl is the number of unsaturated bondings per atom at the plane (hkl), and Z is the amount of coordination among neighbors. Equation 2 does not include impurities.

On the other hand, deformation energy is related to intrinsic efforts in the film. Considering only two dimensions, Uhkl can be calculated by:

$$
\hbar L\_{\text{hkl}} = \epsilon\_{\text{hkl}}^2 E\_{\text{hkl}} (1 - \mu) \tag{3}
$$

where hkl is the deformation along the plane (hkl) of the film, Ehkl is the elastic module of the plane (hkl), and μ is the Poisson ratio. Research has established that the values of hkl and Ehkl are different for (111), (220) and (200) planes (Ma *et al.*, 2004), and using equations (2) and (3) an order relation has been obtained: S111>S220>S200 and U200>U220>U111. These

Thin Film Growth Through Sputtering Technique and Its Applications 407

& Kadlec, 1990), Ziemann and Kay for Pd coatings (Ziemann and Kay, 1993) and passivation for CBN films (Kulisch, *et al.*,1999). This represents the energy deposited per dense particle. This bombardment-induced mobility parameter, known as the energy parameter Ep, is

> *i p i a <sup>J</sup> E E*

where Ei is incident energy of the ions and is obtained from the plasma potential (Vp) and

So far, we have studied the influence of the growth parameters on the structure and microstructure of thin films; now we discuss the main physical mechanisms involved in the nucleation and crystallization of the films over the substrate. Studies done through Xray diffraction, optical diffraction and mainly through electron diffraction have allowed establishing that there are three mechanisms of the nucleation and growth of thin films, which depend on the thermodynamic parameters of the deposit and the substrate surface interaction between the adatoms and the substrate material (Green, 1994). The three basic modes are: (i) Volmer–Weber model, (ii) Frank–Van der Merwe model and (iii) Stranski– Krastanov model. A schematic illustration of each of these modes of growth is shown in

Fig. 7. Modes of growth of film: a- Volmer–Weber island growth b- Frank–vander Merwe

layer growth and c- Stranski-Krastanov layer plus island growth (Harsha, 2005).

substrate bias (Vs) and the elementary charge by the expression ( ) *E eV V i s <sup>p</sup>* .

**3.1 Physical mechanisms of thin film growth** 

*<sup>J</sup>* (4)

defined as:

Fig. 7.

relations imply that the preferential orientation is determined by a configuration of minimum total energy, which results in competition between the (111), with minimum deformation energy, and (200) planes, with minimum surface energy.

Moreover, Pelleg has researched the variation of the total energy (Whkl) for the (111), (220) and (200) planes as a function of the thickness of the TiN film deposited at 373K through the rf sputtering technique. These results have shown that the surface energy does not change with the thickness, the deformation energy increases proportionally with the thickness, and the slope of the straight line represents the deformation per volume.

In the kinetic model, it is proposed that the kinetic process of the ions is involved in the orientation of the growth, which influences the surface substrate, the energy and flow of the ions being the main parameters. A thin film grows in the [111] direction because the adatoms have limited mobility, caused by the low temperature of the substrate and the restricted flow of the ions. Increasing the mobility, either through increasing the substrate temperature or raising the relation Ji/Ja above five, a preferential orientation in the [200] direction is obtained.

Fig. 6. Surface growth on a- (200) planes and b- (111) planes. The growth on the (200) plane has 1 binding, while the surfaces grown on (111) planes have 3 bandings (M. Marlo & V. Milman, 2000).

Finally, the atomistic model considers that in thin films deposited at high temperatures the thermodynamic parameters control the orientation of the growth, favoring planes with low energy, *i.e.* the (200) planes. This behavior can be understood if adatoms with low diffusion, as in the case of the cation Cl in NaCl, which have three bonds in the (111) surface, and have high diffusion in surface (200), are considered to have only one bond (see Fig. 6), leading to the conclusion that if NaCl thin films are grown at high temperatures, they probably will grow along the [200] direction.

On the other hand, the same model establishes that the degree of ion bombardment is influenced by the flow ratio between the density current of the ions and the density current of the atoms, Ji/Ja, and the ion energy, Ei, and both depend on the pressure, substrate target distance, and substrate bias (Ensinger, 1998; Losbichle & Mitterer, 1997). One parameter that combines both the ion energy and the flow has been suggested by Musil and Kadlec (Musil

relations imply that the preferential orientation is determined by a configuration of minimum total energy, which results in competition between the (111), with minimum

Moreover, Pelleg has researched the variation of the total energy (Whkl) for the (111), (220) and (200) planes as a function of the thickness of the TiN film deposited at 373K through the rf sputtering technique. These results have shown that the surface energy does not change with the thickness, the deformation energy increases proportionally with the thickness, and

In the kinetic model, it is proposed that the kinetic process of the ions is involved in the orientation of the growth, which influences the surface substrate, the energy and flow of the ions being the main parameters. A thin film grows in the [111] direction because the adatoms have limited mobility, caused by the low temperature of the substrate and the restricted flow of the ions. Increasing the mobility, either through increasing the substrate temperature or raising the relation Ji/Ja above five, a preferential orientation in the [200]

Fig. 6. Surface growth on a- (200) planes and b- (111) planes. The growth on the (200) plane has 1 binding, while the surfaces grown on (111) planes have 3 bandings (M. Marlo & V.

Finally, the atomistic model considers that in thin films deposited at high temperatures the thermodynamic parameters control the orientation of the growth, favoring planes with low energy, *i.e.* the (200) planes. This behavior can be understood if adatoms with low diffusion, as in the case of the cation Cl in NaCl, which have three bonds in the (111) surface, and have high diffusion in surface (200), are considered to have only one bond (see Fig. 6), leading to the conclusion that if NaCl thin films are grown at high temperatures, they probably will

On the other hand, the same model establishes that the degree of ion bombardment is influenced by the flow ratio between the density current of the ions and the density current of the atoms, Ji/Ja, and the ion energy, Ei, and both depend on the pressure, substrate target distance, and substrate bias (Ensinger, 1998; Losbichle & Mitterer, 1997). One parameter that combines both the ion energy and the flow has been suggested by Musil and Kadlec (Musil

deformation energy, and (200) planes, with minimum surface energy.

the slope of the straight line represents the deformation per volume.

direction is obtained.

Milman, 2000).

grow along the [200] direction.

$$E\_p = \frac{J\_i}{J\_a} \times E\_i \tag{4}$$

where Ei is incident energy of the ions and is obtained from the plasma potential (Vp) and substrate bias (Vs) and the elementary charge by the expression ( ) *E eV V i s <sup>p</sup>* .

#### **3.1 Physical mechanisms of thin film growth**

So far, we have studied the influence of the growth parameters on the structure and microstructure of thin films; now we discuss the main physical mechanisms involved in the nucleation and crystallization of the films over the substrate. Studies done through Xray diffraction, optical diffraction and mainly through electron diffraction have allowed establishing that there are three mechanisms of the nucleation and growth of thin films, which depend on the thermodynamic parameters of the deposit and the substrate surface interaction between the adatoms and the substrate material (Green, 1994). The three basic modes are: (i) Volmer–Weber model, (ii) Frank–Van der Merwe model and (iii) Stranski– Krastanov model. A schematic illustration of each of these modes of growth is shown in Fig. 7.

Fig. 7. Modes of growth of film: a- Volmer–Weber island growth b- Frank–vander Merwe layer growth and c- Stranski-Krastanov layer plus island growth (Harsha, 2005).

Thin Film Growth Through Sputtering Technique and Its Applications 409

intersection lines of the crystal side faces and the substrate present a specific structural precondition for the growth of these crystals. The intersection lines can be active or passive in the monolayer nucleation on the side crystal faces. In the first case, the movement of the monolayer growth steps proceeds from the intersection line to the top of the crystal, while in the second case, the movement of the growth steps proceeds in the direction of the intersection line. In the presence of impurities, the direction of the movement of the growth steps will be

Fig. 8. Types of crystal growth in polycrystalline thin films: a- and b- growth of dispersed individual crystals on the substrate surface, the intersection line between the substrate and the side crystal faces are active (a) and passive (b) in the monolayer nucleation; c- and drole of grain boundaries in the growth of crystals as parts of a polycrystalline matrix, pure grain boundary active in the monolayer nucleation (a), contaminated grain boundary

**4. Influence of the normalized temperature (Ts/ Tm) on the structure and** 

The structural behavior, as a function of the normalized temperature of Ts/Tm<0.3 is shown in Fig. 9. In general, the patterns recorded from the various films grown present the same reflections as the target (δ-NbN cubic phase), but show important differences in the relative intensities, in particular those corresponding to planes (111) and (200), which increase with an increase of the absolute temperature. Regarding this, it is important to point out that contrary to what occurs in the target and the rest of the film patterns, the XRD pattern of the film grown at 553 K shows a greater intensity for the reflection from plane (111) than from that corresponding to the (200) plane. While the (111) plane in the target pattern has a relative intensity of 100% compared to 86% for the (200) plane, in the 553 K sample the relative intensity of the (200) plane is approximately 6% of the corresponding (111) plane. This result indicates that the supply of additional 22.4 meV to the substrate (which is the energy difference between room temperature and 553 K) does not bring about changes in the polycrystalline character of the materials, but gives place to the growth of films with

passive in the monolayer nucleation (b). (Barna & Adamimik)

**microstructure of thin films** 

preferred orientation along the (111) plane.

important in determining the location of the developing second phase, *e.g.*, SCL.

In the Volmer–Weber model, equilibrium exists in a three-dimensional crystal of the film in contact with the substrate, while the rest of the substrate is devoid of any condensed phase. Nucleation of film occurs in the form of discrete three-dimensional nuclei on the surface of the substrate, for example lead on graphite. Both the number of nuclei and the size of a given nucleus increase. Finally, the nuclei grow in size until they intersperse with each other to form a continuous film.

In the Frank and Vander Merwe model, nucleation occurs in the form of a monolayer island of the deposit. Eventually the monolayer's grow together to form a complete continuous monolayer of the deposit. The process repeats itself so that the deposit grows in a layer-bylayer manner, for example rare gases on graphite. In this growth, the interaction between the substrate and the layer atoms is stronger than that between neighboring layer atoms. Layer-by-layer growth is hindered by elastic constraints at the solid–solid interface.

The Stransky and Krastanov (S-K) model combines the features of layer-by-layer growth and discrete three-dimensional nucleation The S-K nucleation is common with metal-onmetal deposition and at low temperatures where the surface mobility is low (Greene, 1987). The conditions for these types of growth are generally described in terms of thermodynamics and surface energy considerations. In this model, nucleation and growth occurs as in the layer-by-layer mode, so that a finite number of monolayer's is produced. Subsequent formation of film occurs by formation of discrete nuclei. The lattice mismatch between the substrate and the deposit cannot be accommodated when the layer thickness increases, so the three-dimensional growth follows the layer-by-layer growth. Alternatively, symmetry or orientation of the overlayers with respect to the substrate might be responsible for the production of this growth mode.

The foregoing models were summarized by Barna and Adamik (Barna & Adamimik, 1998), who established that the growth of the films has the following evolution: nucleation, island growth, coalescence of islands, formation of polycrystalline islands and channels, development of continuous structure, and thickness growth.

According to Barna, the nucleation starting the growth of individual islands takes place on the substrate surface at the very first stage of the condensation (primary nucleation*)* or later on the bare substrate surface area developing upon liquid-like coalescence ( secondary nucleation*).* A peculiar case of nucleation shows up on the surface of a growing crystal when its growth is blocked by a surface covering layer (SCL) of an impurity phase. This is the repeated nucleation*.* The primary nucleation starts the condensation and the film growth on the whole substrate surface simultaneously, while the secondary and the repeated nucleation initiates the start of the growth locally in later stages of film formation. It is important to note that on amorphous substrates the nuclei are randomly oriented.

Crystal growth is the fundamental structure-forming phenomenon which incorporates the depositing material into the condensed phase. Two main cases of crystal growth should be considered in the case of polycrystalline thin films: a- the growth of discrete crystals dispersed on the substrate surface, (Fig. 8a and b), and b- the growth of crystals which are parts of a polycrystalline structure (Fig. 8c and d). Crystals growing from the nuclei are randomly oriented due to the random orientation of the nuclei. The complete coalescence of the crystals touching each other produces a grain coarsening, resulting also in the development of discrete single crystals and is connected to some changes in the orientation controlled mainly by the minimization of the substrate–crystal interface energy. The

In the Volmer–Weber model, equilibrium exists in a three-dimensional crystal of the film in contact with the substrate, while the rest of the substrate is devoid of any condensed phase. Nucleation of film occurs in the form of discrete three-dimensional nuclei on the surface of the substrate, for example lead on graphite. Both the number of nuclei and the size of a given nucleus increase. Finally, the nuclei grow in size until they intersperse with each other

In the Frank and Vander Merwe model, nucleation occurs in the form of a monolayer island of the deposit. Eventually the monolayer's grow together to form a complete continuous monolayer of the deposit. The process repeats itself so that the deposit grows in a layer-bylayer manner, for example rare gases on graphite. In this growth, the interaction between the substrate and the layer atoms is stronger than that between neighboring layer atoms.

The Stransky and Krastanov (S-K) model combines the features of layer-by-layer growth and discrete three-dimensional nucleation The S-K nucleation is common with metal-onmetal deposition and at low temperatures where the surface mobility is low (Greene, 1987). The conditions for these types of growth are generally described in terms of thermodynamics and surface energy considerations. In this model, nucleation and growth occurs as in the layer-by-layer mode, so that a finite number of monolayer's is produced. Subsequent formation of film occurs by formation of discrete nuclei. The lattice mismatch between the substrate and the deposit cannot be accommodated when the layer thickness increases, so the three-dimensional growth follows the layer-by-layer growth. Alternatively, symmetry or orientation of the overlayers with respect to the substrate might be responsible

The foregoing models were summarized by Barna and Adamik (Barna & Adamimik, 1998), who established that the growth of the films has the following evolution: nucleation, island growth, coalescence of islands, formation of polycrystalline islands and channels,

According to Barna, the nucleation starting the growth of individual islands takes place on the substrate surface at the very first stage of the condensation (primary nucleation*)* or later on the bare substrate surface area developing upon liquid-like coalescence ( secondary nucleation*).* A peculiar case of nucleation shows up on the surface of a growing crystal when its growth is blocked by a surface covering layer (SCL) of an impurity phase. This is the repeated nucleation*.* The primary nucleation starts the condensation and the film growth on the whole substrate surface simultaneously, while the secondary and the repeated nucleation initiates the start of the growth locally in later stages of film formation. It is

Crystal growth is the fundamental structure-forming phenomenon which incorporates the depositing material into the condensed phase. Two main cases of crystal growth should be considered in the case of polycrystalline thin films: a- the growth of discrete crystals dispersed on the substrate surface, (Fig. 8a and b), and b- the growth of crystals which are parts of a polycrystalline structure (Fig. 8c and d). Crystals growing from the nuclei are randomly oriented due to the random orientation of the nuclei. The complete coalescence of the crystals touching each other produces a grain coarsening, resulting also in the development of discrete single crystals and is connected to some changes in the orientation controlled mainly by the minimization of the substrate–crystal interface energy. The

important to note that on amorphous substrates the nuclei are randomly oriented.

Layer-by-layer growth is hindered by elastic constraints at the solid–solid interface.

to form a continuous film.

for the production of this growth mode.

development of continuous structure, and thickness growth.

intersection lines of the crystal side faces and the substrate present a specific structural precondition for the growth of these crystals. The intersection lines can be active or passive in the monolayer nucleation on the side crystal faces. In the first case, the movement of the monolayer growth steps proceeds from the intersection line to the top of the crystal, while in the second case, the movement of the growth steps proceeds in the direction of the intersection line. In the presence of impurities, the direction of the movement of the growth steps will be important in determining the location of the developing second phase, *e.g.*, SCL.

Fig. 8. Types of crystal growth in polycrystalline thin films: a- and b- growth of dispersed individual crystals on the substrate surface, the intersection line between the substrate and the side crystal faces are active (a) and passive (b) in the monolayer nucleation; c- and drole of grain boundaries in the growth of crystals as parts of a polycrystalline matrix, pure grain boundary active in the monolayer nucleation (a), contaminated grain boundary passive in the monolayer nucleation (b). (Barna & Adamimik)

## **4. Influence of the normalized temperature (Ts/ Tm) on the structure and microstructure of thin films**

The structural behavior, as a function of the normalized temperature of Ts/Tm<0.3 is shown in Fig. 9. In general, the patterns recorded from the various films grown present the same reflections as the target (δ-NbN cubic phase), but show important differences in the relative intensities, in particular those corresponding to planes (111) and (200), which increase with an increase of the absolute temperature. Regarding this, it is important to point out that contrary to what occurs in the target and the rest of the film patterns, the XRD pattern of the film grown at 553 K shows a greater intensity for the reflection from plane (111) than from that corresponding to the (200) plane. While the (111) plane in the target pattern has a relative intensity of 100% compared to 86% for the (200) plane, in the 553 K sample the relative intensity of the (200) plane is approximately 6% of the corresponding (111) plane. This result indicates that the supply of additional 22.4 meV to the substrate (which is the energy difference between room temperature and 553 K) does not bring about changes in the polycrystalline character of the materials, but gives place to the growth of films with preferred orientation along the (111) plane.

Thin Film Growth Through Sputtering Technique and Its Applications 411

Fig. 10 shows the micro-structural analysis carried out on the NbN films through transmission ion electron microcopy (TEM). Micrographs 10a and 10b correspond to the image obtained in multibeam configuration, and in them one can observe an amorphous matrix with crystalline nanoparticles with spherical morphology that are homogenously distributed along the growth plane of the film. Using Gatan software, it was determined that the average grain size of the nanoparticles was 2.7±0.6 nm with normal distribution (fig. 10c). When doing magnification of nanoparticles of Fig. 10a, we can observe an interferential pattern produced by the atoms belonging to the NbN film (fig. 9d). Higher magnifications of the nanoparticles visualized in Fig. 10a let us observe atomic planes as a product of the interferential pattern produced by the arrangement of atoms belonging to the NbN film (Fig.10d). Image processing of Fig. 10d, using the Fourier transform, lets us visualize the reciprocal space of one NbN nanoparticle (Fig. 10e). Applying a mask (Fig. 10f) over the corresponding diffracting spots and directly measuring the distance between them (0.224 nm), we can confirm that they correspond to distances between (200) planes of δ-NbN. These results confirm the analysis of XRD, which established that NbN grows preferentially

To produce thin films with normalized temperature larger than 0.3 it is necessary to have ultrahigh-vacuum equipment. This condition limits the production the thin films through magnetron sputtering. However, there are studies such as that of Frederick (Frederic & Gall, 2005), who grew CrN thin films on MgO substrate through magnetron sputtering with Ts/tm between 0 .43 to 0.48. These authors found that the films grown at 993 K present complete single-crystal structure with smooth surfaces. The root-mean-square surface roughness for 230-nm-thick layers decreases from 18.8 to 9.3 to 1.1 nm as Ts is raised from

**4.1 Influence of the gas flow on the structure and microstructure of thin films** 

In a different set of experiments, nitrogen gas was introduced into the deposition chamber (maintaining the final working pressure constant) in order to study the influence that the addition of this gas during deposition exerts on the structural and micro structural properties of the NbN films. Fig. 11 shows the diffraction patterns recorded from the films grown at 300 W, 553 K and different nitrogen flows. The results obtained make clear that in all cases a preferential growth appears along the (200) plane (Fig.11a). The relative intensity of this diffraction peak is so high that it makes it impossible to distinguish the diffraction peaks corresponding to other planes. To determine the polycrystalline character of the film, we carried out X-ray diffraction experiments at grazing incidence. Fig. 11b shows a representative example. It is clear from Fig. 11 that the films show the same diffraction peaks as the target, confirming their polycrystalline character, although having a preferential orientation along the (200) plane (texture index, 0.65). The grain sizes deduced by Scherer's equation (Cullity, 2001) for the different films grown along the (200) plane vary from 35 nm (Φ= 2 sscm) to 42 nm (Φ = 6 sscm). These results indicate that incorporation of nitrogen during the fabrication process favors the preferential growth of the d-NbN phase along the (200) plane. This effect has also been produced during the growth of TiN films, due to the change of the preferred orientation of the films from the [111] direction to the

The incorporation of nitrogen during the deposition process implies changes in the dynamics of the plasma, since the increase in the number of nitrogen molecules increases the

along the (200) plane.

873 to 973 to 1073 K.

[200] direction when the gas flow is increased.

Fig. 9. XRD pattern of NbN thin films deposited at a normalized temperature (Ts/Tm<0.3) at 300W of power.

Fig. 10.a) and 10.b) TEM micrographs using multibeam configuration of NbN nanoparticles distributed homogeneously; c) average grain size of NbN nanoparticles; d) HRTEM micrographs shows atomic planes; e) reciprocal space of a NbN nanoparticle; and f) distance between (200) planes.

**Ts**

**/Tm<0.3**

**(311)**

**(222)**

**Intensity (a.u)**

300W of power.

between (200) planes.

**(111)**

**(200)**

30 40 50 60 70 80 90

**(220)**

**2**

Fig. 9. XRD pattern of NbN thin films deposited at a normalized temperature (Ts/Tm<0.3) at

a) b) c)

d) e) f)

Fig. 10.a) and 10.b) TEM micrographs using multibeam configuration of NbN nanoparticles

micrographs shows atomic planes; e) reciprocal space of a NbN nanoparticle; and f) distance

distributed homogeneously; c) average grain size of NbN nanoparticles; d) HRTEM

Target

**393K**

**513K**

**553K**

Fig. 10 shows the micro-structural analysis carried out on the NbN films through transmission ion electron microcopy (TEM). Micrographs 10a and 10b correspond to the image obtained in multibeam configuration, and in them one can observe an amorphous matrix with crystalline nanoparticles with spherical morphology that are homogenously distributed along the growth plane of the film. Using Gatan software, it was determined that the average grain size of the nanoparticles was 2.7±0.6 nm with normal distribution (fig. 10c). When doing magnification of nanoparticles of Fig. 10a, we can observe an interferential pattern produced by the atoms belonging to the NbN film (fig. 9d). Higher magnifications of the nanoparticles visualized in Fig. 10a let us observe atomic planes as a product of the interferential pattern produced by the arrangement of atoms belonging to the NbN film (Fig.10d). Image processing of Fig. 10d, using the Fourier transform, lets us visualize the reciprocal space of one NbN nanoparticle (Fig. 10e). Applying a mask (Fig. 10f) over the corresponding diffracting spots and directly measuring the distance between them (0.224 nm), we can confirm that they correspond to distances between (200) planes of δ-NbN. These results confirm the analysis of XRD, which established that NbN grows preferentially along the (200) plane.

To produce thin films with normalized temperature larger than 0.3 it is necessary to have ultrahigh-vacuum equipment. This condition limits the production the thin films through magnetron sputtering. However, there are studies such as that of Frederick (Frederic & Gall, 2005), who grew CrN thin films on MgO substrate through magnetron sputtering with Ts/tm between 0 .43 to 0.48. These authors found that the films grown at 993 K present complete single-crystal structure with smooth surfaces. The root-mean-square surface roughness for 230-nm-thick layers decreases from 18.8 to 9.3 to 1.1 nm as Ts is raised from 873 to 973 to 1073 K.

## **4.1 Influence of the gas flow on the structure and microstructure of thin films**

In a different set of experiments, nitrogen gas was introduced into the deposition chamber (maintaining the final working pressure constant) in order to study the influence that the addition of this gas during deposition exerts on the structural and micro structural properties of the NbN films. Fig. 11 shows the diffraction patterns recorded from the films grown at 300 W, 553 K and different nitrogen flows. The results obtained make clear that in all cases a preferential growth appears along the (200) plane (Fig.11a). The relative intensity of this diffraction peak is so high that it makes it impossible to distinguish the diffraction peaks corresponding to other planes. To determine the polycrystalline character of the film, we carried out X-ray diffraction experiments at grazing incidence. Fig. 11b shows a representative example. It is clear from Fig. 11 that the films show the same diffraction peaks as the target, confirming their polycrystalline character, although having a preferential orientation along the (200) plane (texture index, 0.65). The grain sizes deduced by Scherer's equation (Cullity, 2001) for the different films grown along the (200) plane vary from 35 nm (Φ= 2 sscm) to 42 nm (Φ = 6 sscm). These results indicate that incorporation of nitrogen during the fabrication process favors the preferential growth of the d-NbN phase along the (200) plane. This effect has also been produced during the growth of TiN films, due to the change of the preferred orientation of the films from the [111] direction to the [200] direction when the gas flow is increased.

The incorporation of nitrogen during the deposition process implies changes in the dynamics of the plasma, since the increase in the number of nitrogen molecules increases the

Thin Film Growth Through Sputtering Technique and Its Applications 413

(i = 1–4) admolecule or islands of adatoms (Petrov & Barna, 2003). This can be considered as causing an additional decrease in the (200) surface energy relative to that of the (111) plane. Consequently, the presence of the nitrogen atoms reduces the flow of cations from the (200) to the (111) planes, resulting in the orientation of the growth along the [200] direction. In this example, the growth of the films depends on the final working pressure as well as the

The SEM study (Fig. 12a and 12b) of film growth as a function of the gas flow indicates that the NbN films present a compact granular structure, with a columnar growth of the type described by Movchan and Demchisshim (Movchan & Demchisshim, 1969), having an

average thickness of 0.7 μm, which implies a deposition rate of 20 nm/min.

(a) (b)

(2.0 sccm) and the nitrogen flow was varied.

2001).

Fig. 12. a) Micrograph of the morphology of NbN thin film. b) Micrograph of the crosssection of NbN film, grown at 300 W and 553 K, 20 sccm of Ar and 6 sccm of N2.

An interesting example of the growth of thin films in a reactive phase through the rf sputtering technique is ZrNxOy thin films, which grow at different flow ratios (N2/O2), but with the final working pressure constant (7.4 X10-1 Pa). The results of the XRD studies are shown in figures 13a and 13b, where it is possible to observe the influence that the flows have on the crystallographic of the films. Fig. 13a shows the XRD pattern of the films where the nitrogen flow was maintained constant (2.5 sccm) and the oxygen flow was varied, and Fig. 13b shows the XRD pattern of the films where the oxygen flow was maintained constant

The results obtained allowed establishing that there is an optimum flow ratio of 1.25 in which growth of a film with a high degree of crystallinity is reached. This behavior is very similar to the NbN films discussed above; therefore, the physicochemical mechanisms involved in the growth of thin films are the same for nitrogen molecules as for oxygen molecules, due to the higher reactivity of oxygen as compared to nitrogen (Martin, *et al.*,

The SEM study (Fig. 14a) of ZrNxOy films that were grown at a flow ratio of 1.25 present a highly homogenous and maybe very compact surface in which it is not possible to find contrast, and therefore the growth mechanism is not well defined. The microstructure of the

bombardment energy of the ions.

probability of collisions, promoting a larger number of chemical reactions on the substrate surface. These reactions can be explained using a model of low energy (<20 eV) ion bombardment during film growth at a Ts/Tm ratio ranging between 0.1 and 0.3 (a condition which is satisfied in this case) (Petrov & Barna, 2003). According to this model, 25 eV are sufficient to cause collision dissociation of the N2 ions, providing a continuous source of atomic nitrogen. The nitrogen readily chemisorbs on the (200) planes but not on the Nterminated (111) planes. This in turn reduces the mean free path of the metal cation on the (200) plane due to capture by the nitrogen atoms, and promotes the formation of a NbNi

Fig. 11. a) XRD patterns -2 configuration and b) Grazing angle (2) XRD pattern recorder from the NbN film, grown at 300 W, 553 K, 20 sccm of Ar and 6 sccm of N2.

probability of collisions, promoting a larger number of chemical reactions on the substrate surface. These reactions can be explained using a model of low energy (<20 eV) ion bombardment during film growth at a Ts/Tm ratio ranging between 0.1 and 0.3 (a condition which is satisfied in this case) (Petrov & Barna, 2003). According to this model, 25 eV are sufficient to cause collision dissociation of the N2 ions, providing a continuous source of atomic nitrogen. The nitrogen readily chemisorbs on the (200) planes but not on the Nterminated (111) planes. This in turn reduces the mean free path of the metal cation on the (200) plane due to capture by the nitrogen atoms, and promotes the formation of a NbNi

**a**

=2 sccm

**b**

Target

(222)

(311)

Target

=4 sccm

=6 sccm

30 40 50 60 70 80 90

30 40 50 60 70 80 90

**2** Fig. 11. a) XRD patterns -2 configuration and b) Grazing angle (2) XRD pattern recorder

from the NbN film, grown at 300 W, 553 K, 20 sccm of Ar and 6 sccm of N2.

(220)

**2**

= 6 sccm

**(220)**

**(111)**

(111)

(200)

**(200)**

**Intensity (a.u)**

**Intensity(a.u)**

**(222)**

**(311)**

(i = 1–4) admolecule or islands of adatoms (Petrov & Barna, 2003). This can be considered as causing an additional decrease in the (200) surface energy relative to that of the (111) plane. Consequently, the presence of the nitrogen atoms reduces the flow of cations from the (200) to the (111) planes, resulting in the orientation of the growth along the [200] direction. In this example, the growth of the films depends on the final working pressure as well as the bombardment energy of the ions.

The SEM study (Fig. 12a and 12b) of film growth as a function of the gas flow indicates that the NbN films present a compact granular structure, with a columnar growth of the type described by Movchan and Demchisshim (Movchan & Demchisshim, 1969), having an average thickness of 0.7 μm, which implies a deposition rate of 20 nm/min.

Fig. 12. a) Micrograph of the morphology of NbN thin film. b) Micrograph of the crosssection of NbN film, grown at 300 W and 553 K, 20 sccm of Ar and 6 sccm of N2.

An interesting example of the growth of thin films in a reactive phase through the rf sputtering technique is ZrNxOy thin films, which grow at different flow ratios (N2/O2), but with the final working pressure constant (7.4 X10-1 Pa). The results of the XRD studies are shown in figures 13a and 13b, where it is possible to observe the influence that the flows have on the crystallographic of the films. Fig. 13a shows the XRD pattern of the films where the nitrogen flow was maintained constant (2.5 sccm) and the oxygen flow was varied, and Fig. 13b shows the XRD pattern of the films where the oxygen flow was maintained constant (2.0 sccm) and the nitrogen flow was varied.

The results obtained allowed establishing that there is an optimum flow ratio of 1.25 in which growth of a film with a high degree of crystallinity is reached. This behavior is very similar to the NbN films discussed above; therefore, the physicochemical mechanisms involved in the growth of thin films are the same for nitrogen molecules as for oxygen molecules, due to the higher reactivity of oxygen as compared to nitrogen (Martin, *et al.*, 2001).

The SEM study (Fig. 14a) of ZrNxOy films that were grown at a flow ratio of 1.25 present a highly homogenous and maybe very compact surface in which it is not possible to find contrast, and therefore the growth mechanism is not well defined. The microstructure of the

Thin Film Growth Through Sputtering Technique and Its Applications 415

a)

b) Fig. 14. a- Micrograph of the morphology of ZrNxOy thin film. b- AFM micrograph of the

The XRD patterns of MgO films (Fig. 15a and 15b) recorded from films grown at room temperature shows the influence which the power supply applied to the target has on the structural behavior of thin films. The films grown within the range from 150 to 200W showed amorphous behavior (not shown in the XRD pattern). On the other hand, starting at 250W, all films showed the (200) plane corresponding to magnesium oxide in the FCC phase. In the XRD pattern it can also be observed that the intensity of the (200) plane is so high that it makes it impossible to distinguish the diffraction peaks corresponding to other planes. To determine the polycrystalline character of the film, we carried out X-ray diffraction experiments at grazing incidence. From Fig. 15b it is clear that the films present two main diffraction peaks, (200) and (220), which belong to the target, confirming their polycrystalline character, although

**4.2 Influence of the power on the structure and microstructure of thin films** 

showing a preferential orientation along the (200) plane (texture index, 0.85).

ZrNxOy film, deposited at 350 W, 623 K and ΦN2/ΦO2= 1.25

films was evaluated through AFM studies (Fig. 14b). The micrograph reveals that the average size grain was 150 nm with an average rugosity of 5.9 nm.

Fig. 13 . XRD Patterns recorded from ZrNxOy thin films deposited on common glass at 623K and different ratios flow (ΦN2/ΦO2 ). a- ΦN2 constant at 2.5 sccm and b- ΦO2 constant at 2.0 sccm.

films was evaluated through AFM studies (Fig. 14b). The micrograph reveals that the

**a**

**b**

10 20 30 40 50 60

10 20 30 40 50 60

Fig. 13 . XRD Patterns recorded from ZrNxOy thin films deposited on common glass at 623K and different ratios flow (ΦN2/ΦO2 ). a- ΦN2 constant at 2.5 sccm and b- ΦO2 constant at 2.0

2

**(111 )**

**(-111)**

**2**

**(111)**

**(-111)**

**N2 /O2**

> **N2 /O2 =1.5**

**N2 /O2**

**N2 /O2 =1.0**

**=1.25**

**N2 /O2 =1.0**

**N2 /O2**

**=0.83**

**=1.25**

average size grain was 150 nm with an average rugosity of 5.9 nm.

**(100)**

**Intensity (a.u.)**

**Intensity (a.u.)**

sccm.

Fig. 14. a- Micrograph of the morphology of ZrNxOy thin film. b- AFM micrograph of the ZrNxOy film, deposited at 350 W, 623 K and ΦN2/ΦO2= 1.25

### **4.2 Influence of the power on the structure and microstructure of thin films**

The XRD patterns of MgO films (Fig. 15a and 15b) recorded from films grown at room temperature shows the influence which the power supply applied to the target has on the structural behavior of thin films. The films grown within the range from 150 to 200W showed amorphous behavior (not shown in the XRD pattern). On the other hand, starting at 250W, all films showed the (200) plane corresponding to magnesium oxide in the FCC phase. In the XRD pattern it can also be observed that the intensity of the (200) plane is so high that it makes it impossible to distinguish the diffraction peaks corresponding to other planes. To determine the polycrystalline character of the film, we carried out X-ray diffraction experiments at grazing incidence. From Fig. 15b it is clear that the films present two main diffraction peaks, (200) and (220), which belong to the target, confirming their polycrystalline character, although showing a preferential orientation along the (200) plane (texture index, 0.85).

Thin Film Growth Through Sputtering Technique and Its Applications 417

**(111)**

**(101)´**

**Intensity (a.u.)**

at different powers applied to the target.

substrate.

**thin films** 

10 20 30 40 50

**2** Fig. 16. XRD Patterns recorded from ZrNxOy thin films deposited on common glass at 623K

Fig. 17. Atomic force micrographs the MgO thin films deposited at 400 W.

The study of the microstructure of thin films was carried out using atomic force microscopy (AFM). The micrograph of Fig. 17 shows that in the area swept by the cantilever point (16μm2) the film has an average roughness of 20 nm and an average grain size on the order of 110 nm. These results established that there is a threshold of power for the growth of nano-structured MgO thin films with high texture without intentional heating of the

**4.3 Influence of the substrate bias voltage (Vs) on the structure and microstructure of** 

Fig. 18 shows the influence that the substrate bias voltage has on the growth of the Ti alloy thin films deposited on steel and glass substrates. The films grown on steel at –100V showed, in addition to the diffraction peaks of the substrate material (note that in all the XRD

**350 W**

**300 W 250 W**

**200 W**

Another example that shows the influence of the power supply applied to the target on the crystallization of the films is the growth of ZrNxOy thin films; the films were grown from the Zr target in the reactive phase in an atmosphere of nitrogen and oxygen (ΦN= 2.5 sccm, ΦO=2.0 sccm). In the XRD pattern of Fig. 16, it is possible to determine that there is a threshold power (250W) for producing the crystallization of the film on the substrate. At this power, the film grows with a high degree of crystallization along the (111) plane. This behavior can be explained by the energy model, since increasing the power on the target implies increasing the energy of the ions that are bombarding the substrate and therefore improving the mobility of the adatoms, which produce chemical reactions and atomic grouping along the planes with the higher surface energy, which generates films with a high degree of crystallinity.

Fig. 15. a- XRD patterns recorded from MgO films deposited at 293 K and different power supplies applied to the target b- Grazing angle (3) XRD pattern recorded from a MgO film grown at 400 W. The target XRD pattern is included as reference.

Another example that shows the influence of the power supply applied to the target on the crystallization of the films is the growth of ZrNxOy thin films; the films were grown from the Zr target in the reactive phase in an atmosphere of nitrogen and oxygen (ΦN= 2.5 sccm, ΦO=2.0 sccm). In the XRD pattern of Fig. 16, it is possible to determine that there is a threshold power (250W) for producing the crystallization of the film on the substrate. At this power, the film grows with a high degree of crystallization along the (111) plane. This behavior can be explained by the energy model, since increasing the power on the target implies increasing the energy of the ions that are bombarding the substrate and therefore improving the mobility of the adatoms, which produce chemical reactions and atomic grouping along the planes with the

30 40 50 60 70 80

30 40 50 60 70 80

**2** Fig. 15. a- XRD patterns recorded from MgO films deposited at 293 K and different power supplies applied to the target b- Grazing angle (3) XRD pattern recorded from a MgO film

**2**

**(200)**

**(111)**

grown at 400 W. The target XRD pattern is included as reference.

**(220)**

**(220)**

**a**

**400W 350W 300W 250W MgO**

**b**

400W

MgO Target

higher surface energy, which generates films with a high degree of crystallinity.

**(200)**

**Intensity (a.u.)**

**Intensity (u.a.)**

Fig. 16. XRD Patterns recorded from ZrNxOy thin films deposited on common glass at 623K at different powers applied to the target.

Fig. 17. Atomic force micrographs the MgO thin films deposited at 400 W.

The study of the microstructure of thin films was carried out using atomic force microscopy (AFM). The micrograph of Fig. 17 shows that in the area swept by the cantilever point (16μm2) the film has an average roughness of 20 nm and an average grain size on the order of 110 nm. These results established that there is a threshold of power for the growth of nano-structured MgO thin films with high texture without intentional heating of the substrate.

#### **4.3 Influence of the substrate bias voltage (Vs) on the structure and microstructure of thin films**

Fig. 18 shows the influence that the substrate bias voltage has on the growth of the Ti alloy thin films deposited on steel and glass substrates. The films grown on steel at –100V showed, in addition to the diffraction peaks of the substrate material (note that in all the XRD

Thin Film Growth Through Sputtering Technique and Its Applications 419

films. The films grown on glass at 120V and -200V show a similar behavior, although there are two differences: the two peaks appearing in the XRD pattern of the film grown at -120 V show a different intensity ratio from those of the film grown on steel, and the XRD pattern of the film grown at –200 V shows an additional low-intensity peak at 82.3, which we

In summary, it follows from the XRD data that the substrate bias voltage (Vs) has a great influence on the structure of the deposited films: an increase of the bias voltage promotes the growth of different phases of titanium having well-defined different preferential crystallographic orientations. The increase of Vs is also reflected in a better crystallinity of the deposited films. The data also show that the crystalline phases formed at different Vs are the same in both substrates (see fig 18b). These results are in accord with that established in the foregoing section, where it was indicated that the increasing V*s* improves the density of

the film, which in turn allows obtaining films with a high degree of crystallinity.

Fig. 19. a) Micrograph of a Ti6A14V film grown on steel with bias voltage of – 200 V. (b) Atomic force Micrograph recorded from Ti6Al4V grown on steel at bias voltage of –200V.

a) b)

3

μm

2

0

 0 1

1

2 μm

3

Ǻ

607 1174

The analysis of the microstructure of the Ti6Al4V was made through scenic electron microscopy (SEM) and an atomic force microscope (AFM). Fig. 19a shows the SEM results; the recorded micrograph of the film grown at –200V reveals a film with excellent texture, high homogeneity and denser microstructure with grain refinement produced under the enhanced plasma bombardment, which is induced by the substrate bias voltage. Through Scherer's equation (Cullity, 2001) and using the broadening of the (110) and (110) planes of the Ti -phase, it was found that the average grain size was 13 nm for the (100) plane and 16.5 nm for the (110) plane (Alfonso *et al.*, 2005). Fig. 19b shows the results of the AFM, which indicated that the films grown at -200 V possess a roughness of 20 nm for a scan area

**4.4 Influence of the energy parameter (Ep) on the structure and microstructure of thin** 

The X-ray diffraction patterns of the CrNx films deposited at different Ep values are shown in Fig. 20. It was necessary to use two different scales, because as the Ep increased, the intensity of the CrN (200) peak increased by two orders of magnitude and became narrower. At Ep values lower than 30 eV/atom, we observed that the films mainly showed a [200] orientation with traces of the CrN[111] and CrN[220] orientations, while at higher Ep values only the CrN[200]

associate with the (220) reflection of the -Ti phase.

of 3μmx3μm.

**films** 

patterns recorded from the films deposited onto steel diffraction peaks from the substrate are still visible), a quite broad peak, which corresponds to the (002) plane of the α- phase of Ti. The XRD pattern corresponding to the films deposited at –120 V present two welldefined peaks at 35.1 (100) and 38.4 (110), which we associate with the (α/)-alloy phase. The film deposited at a bias voltage of -160 V shows a peak at 38.4, which can be associated with the (110) plane of the - phase of Ti. When the bias voltage increases to -200 V, the XRD pattern shows only one high-intensity peak at 38.4(110), which can be associated with the - phase of Ti. It can also be observed that the width of the diffraction peaks is narrower for the films deposited at higher bias voltages, especially for that deposited at -200 V, which suggests that increasing the bias voltage increases the grain size of the deposited

Fig. 18. (a). XRD pattern of the Ti6A14V films deposited by rf magnetron on steel substrates. The substrate appears as a reference. (b) XRD pattern of the films growth on glass and XRD pattern of theTi6A14V in bulk.

patterns recorded from the films deposited onto steel diffraction peaks from the substrate are still visible), a quite broad peak, which corresponds to the (002) plane of the α- phase of Ti. The XRD pattern corresponding to the films deposited at –120 V present two well-

The film deposited at a bias voltage of -160 V shows a peak at 38.4, which can be associated with the (110) plane of the - phase of Ti. When the bias voltage increases to -200 V, the XRD pattern shows only one high-intensity peak at 38.4(110), which can be associated with the - phase of Ti. It can also be observed that the width of the diffraction peaks is narrower for the films deposited at higher bias voltages, especially for that deposited at -200 V, which suggests that increasing the bias voltage increases the grain size of the deposited

30 40 50 60 70 80 90

**2**

40 60 80 100

**2** Fig. 18. (a). XRD pattern of the Ti6A14V films deposited by rf magnetron on steel substrates. The substrate appears as a reference. (b) XRD pattern of the films growth on glass and XRD

**Ti6Al4V Target**

**Bias -200 V**

**Bias -120 V**

(110), which we associate with the (α/)-alloy phase.

a

 **AISI 420 Steel**

**b**

**Bias -100 V**

**Bias -120 V**

**Bias -200 V**

**Bias -160 V**

defined peaks at 35.1 (100) and 38.4

**(002)**

**(110)**

**(100)**

**35,21**

**Intensity (a.u.)**

pattern of theTi6A14V in bulk.

**38,42**

**Intensity (a.u.)**

films. The films grown on glass at 120V and -200V show a similar behavior, although there are two differences: the two peaks appearing in the XRD pattern of the film grown at -120 V show a different intensity ratio from those of the film grown on steel, and the XRD pattern of the film grown at –200 V shows an additional low-intensity peak at 82.3, which we associate with the (220) reflection of the -Ti phase.

In summary, it follows from the XRD data that the substrate bias voltage (Vs) has a great influence on the structure of the deposited films: an increase of the bias voltage promotes the growth of different phases of titanium having well-defined different preferential crystallographic orientations. The increase of Vs is also reflected in a better crystallinity of the deposited films. The data also show that the crystalline phases formed at different Vs are the same in both substrates (see fig 18b). These results are in accord with that established in the foregoing section, where it was indicated that the increasing V*s* improves the density of the film, which in turn allows obtaining films with a high degree of crystallinity.

Fig. 19. a) Micrograph of a Ti6A14V film grown on steel with bias voltage of – 200 V. (b) Atomic force Micrograph recorded from Ti6Al4V grown on steel at bias voltage of –200V.

The analysis of the microstructure of the Ti6Al4V was made through scenic electron microscopy (SEM) and an atomic force microscope (AFM). Fig. 19a shows the SEM results; the recorded micrograph of the film grown at –200V reveals a film with excellent texture, high homogeneity and denser microstructure with grain refinement produced under the enhanced plasma bombardment, which is induced by the substrate bias voltage. Through Scherer's equation (Cullity, 2001) and using the broadening of the (110) and (110) planes of the Ti -phase, it was found that the average grain size was 13 nm for the (100) plane and 16.5 nm for the (110) plane (Alfonso *et al.*, 2005). Fig. 19b shows the results of the AFM, which indicated that the films grown at -200 V possess a roughness of 20 nm for a scan area of 3μmx3μm.

#### **4.4 Influence of the energy parameter (Ep) on the structure and microstructure of thin films**

The X-ray diffraction patterns of the CrNx films deposited at different Ep values are shown in Fig. 20. It was necessary to use two different scales, because as the Ep increased, the intensity of the CrN (200) peak increased by two orders of magnitude and became narrower. At Ep values lower than 30 eV/atom, we observed that the films mainly showed a [200] orientation with traces of the CrN[111] and CrN[220] orientations, while at higher Ep values only the CrN[200]

Thin Film Growth Through Sputtering Technique and Its Applications 421

the residual stress remained low, but the texture of the films changed abruptly and completely from the [111] to the [002] direction. According to the Petrov Model, the key factor was that the energy of the N2 ions should be near 20 eV to promote the dissociation of the ions through collision with the film surface, in this way providing a supply of atomic nitrogen which can chemisorb on [002] oriented grains and later capture metal atoms, resulting in the development of the 〈002〉 texture. In CrN films, it was observed that at the lowest Ep values (Ep<50 eV/atom) with f(200) ~80%, the film texture was equivalent to the transition regime in Petrov *et al.*'s research. The ion energy and ion–atom flows were not sufficiently low to produce pure 〈111〉 texture, as explained above, but were also not sufficiently high to induce 100% 〈002〉. In this transition regime, a competitive growth between the high-trapping (111) plane and the low surface energy (002) plane was established, since as the ds–t increased, the ion energy of the N2 ions approached 20 eV (see Fig. 20). As the Ep value increased, the conditions favored the formation of the 〈002〉 texture. Thus we considered Petrov's theory to be a rather good model to explain the microstructure evolution of the texture in metal nitride films. The competition of the different orientations during growth also affected the microstructure, as observed in Fig. 21a–d. As the ion bombardment increased, the microstructure changed from a non-ordered columnar growth to well-oriented grains that look like fibers coming out from the substrate, *i.e*. equiaxed grains.

Fig. 21. Cross-section SEM images of samples deposited at increasing Ep values (a) 13, (b)

**1 m 1 m** 

**1 m** 

**1 m** 

46, (c) 143 (d) 316 eV/atom, respectively.

**a b**

**c d**

orientation was detected. An estimation of the 〈200〉 texture was obtained comparing the (200) peak intensity (I200) relative to the intensity of all the orientations (Ii) appearing in the XRD pattern. Table 3 summarizes some experimental conditions of the deposition of the CrN films.

Fig. 20. X-ray diffraction patterns of CrNx coatings: (a) low Ep values, (b) high Ep values.


Table 3. Summary of experimental growth conditions of CrN thin films and some of the results: deposition distance ds–t, r.f. bias voltage Vs, deposition rate R, ion energy Ei, ion– atom flow Ji/Ja, energy parameter Ep and texture factor f<200>.

The results of the microstructure behavior of the CrN thin films are in good agreement with that of the theory of Petrov *et al.,* which was explained above. In the case of CrN thin films, increasing the ion energy by applying a higher substrate bias and keeping low ion–atom flows resulted in a film densification and a change in the preferred orientation from [111] to [002] with a subsequent increase in the levels of stress and the incorporation of argon ions. Moreover, by increasing the energy to around 20 eV and working at higher ion–atom ratios,

orientation was detected. An estimation of the 〈200〉 texture was obtained comparing the (200) peak intensity (I200) relative to the intensity of all the orientations (Ii) appearing in the XRD pattern. Table 3 summarizes some experimental conditions of the deposition of the CrN films.

s

CrN (200)

40 60 65

Fig. 20. X-ray diffraction patterns of CrNx coatings: (a) low Ep values, (b) high Ep values.

CrN-3 3 0 0.10 10 1.3 13 0.80 CrN-4 4 0 0.064 16 2 32 0.89 CrN-5 5 0 0.051 12 2.2 26 0.90 CrN-6 6 0 0.035 13.5 2.6 35 0.92 CrN-7 7 0 0.026 14.5 2.7 39 0.93 CrN-8 8 0 0.019 17 2.7 46 0.92 CrN-3-100 3 -100 0.10 110 1.3 143 0.99 CrN-4-100 4 -100 0.064 115.6 2.0 231 0.99 CrN-5-100 5 -100 0.05 112.3 2.2 247 1.0 CrN-6-100 6 -100 0.034 113.5 2.6 295 1.0 CrN-7-100 7 -100 0.026 114.5 2.7 310 1.0 CrN-8-100 8 -100 0.019 117 2.7 316 1.0 Table 3. Summary of experimental growth conditions of CrN thin films and some of the results: deposition distance ds–t, r.f. bias voltage Vs, deposition rate R, ion energy Ei, ion–

The results of the microstructure behavior of the CrN thin films are in good agreement with that of the theory of Petrov *et al.,* which was explained above. In the case of CrN thin films, increasing the ion energy by applying a higher substrate bias and keeping low ion–atom flows resulted in a film densification and a change in the preferred orientation from [111] to [002] with a subsequent increase in the levels of stress and the incorporation of argon ions. Moreover, by increasing the energy to around 20 eV and working at higher ion–atom ratios,

40 60 65

CrN (220)

Ei (eV) Ji/Ja Ep

2

**b**

(eV/atom)

F<200>

**a**

 Ep (eV/atom) 13 39 27

 Ep(eV/atom) 146 243 308

0

Vs (V) R

atom flow Ji/Ja, energy parameter Ep and texture factor f<200>.

(μm/min)

1x10<sup>4</sup>

4x10<sup>4</sup>

5,0x10<sup>2</sup> 1,0x10<sup>3</sup> 1,5x10<sup>3</sup>

0,0

CrN (111)

Intensity (a.u.)

Sample d s-t

(cm)

the residual stress remained low, but the texture of the films changed abruptly and completely from the [111] to the [002] direction. According to the Petrov Model, the key factor was that the energy of the N2 ions should be near 20 eV to promote the dissociation of the ions through collision with the film surface, in this way providing a supply of atomic nitrogen which can chemisorb on [002] oriented grains and later capture metal atoms, resulting in the development of the 〈002〉 texture. In CrN films, it was observed that at the lowest Ep values (Ep<50 eV/atom) with f(200) ~80%, the film texture was equivalent to the transition regime in Petrov *et al.*'s research. The ion energy and ion–atom flows were not sufficiently low to produce pure 〈111〉 texture, as explained above, but were also not sufficiently high to induce 100% 〈002〉. In this transition regime, a competitive growth between the high-trapping (111) plane and the low surface energy (002) plane was established, since as the ds–t increased, the ion energy of the N2 ions approached 20 eV (see Fig. 20). As the Ep value increased, the conditions favored the formation of the 〈002〉 texture. Thus we considered Petrov's theory to be a rather good model to explain the microstructure evolution of the texture in metal nitride films. The competition of the different orientations during growth also affected the microstructure, as observed in Fig. 21a–d. As the ion bombardment increased, the microstructure changed from a non-ordered columnar growth to well-oriented grains that look like fibers coming out from the substrate, *i.e*. equiaxed grains.

Fig. 21. Cross-section SEM images of samples deposited at increasing Ep values (a) 13, (b) 46, (c) 143 (d) 316 eV/atom, respectively.

Thin Film Growth Through Sputtering Technique and Its Applications 423

observed for the passive oxide layer on a Ti6Al4V alloy. The O 1s peak shows three different contributions: a main one at 530.2 eV that can be associated with metal-oxygen bonds, and two much less intense contributions at 531.7 eV and 532.8 eV, which correspond to the presence of OH- groups and chemisorbed water, respectively. The fit of the V 2p spectrum is complicated by the fact that the V peaks overlap strongly with the O 1s K3,4 X-ray satellite peaks. Therefore, the whole spectrum was refined considering the presence of these satellites and the presence of vanadium peaks. The results show the presence of an intense peak at 515.4 eV, which we associate with the presence of VO2, and a less intense vanadium contribution at 516.7 eV, which corresponds to V2O5. This finding contrasts with those of Milosev (Milosev *et al.*, 2000), who did not find oxidized vanadium in the passive layer

These results show that the films that have been grown through rf magnetron sputtering reproduced the stiochoimetry of the target, although on the surface of the thin films

The elemental composition of thin films is determined through energy dispersive X-ray analysis (EDX), although is important to indicate that elements such as oxygen and nitrogen are not possible to assess because the binding energies are very near to the binding energy of the carbon. As an example, the EDX results for NbN nanoparticle are shown. Fig. 23 shows the elements that constitute the nanoparticles of NbN (N and Nb) that formed the NbN studied in section 4, and the elements of the substrate (common glass) on which the

chemical reactions will be produced, forming hydroxides and passivation layers.

Fig. 23. EDX spectrum of NbN nanoparticle that integrates the NbN thin films.

formed on bulk Ti6Al4V (Alfonso *et al.*, 2006).

**5.2 Energy dispersive X-ray (EDX) analysis** 

thin films were grown. The Cu belongs to the sampling.

## **5. Chemical characterization of thin films**

#### **5.1 X-ray photoelectron spectroscopy (XPS) analysis**

The last 10 nm of the surface of thin films can be analyzed using X-ray photoelectron spectroscopy (XPS). As an example, we show the results of XPS on Ti6Al4V thin films. The spontaneously passive film formed on the deposited Ti6Al4V films upon exposure to the air was studied through XPS. Fig. 4 shows high resolution narrow-scan spectra recorded from the Ti 2p, Al 2p, V 2p and O 1s spectra recorded from one of the samples. All the narrow-scan spectra recorded from the rest of the samples are almost identical to those presented in Fig. 22.

Fig. 22. 2p, Al 2p, V 2p and O 1s XPS narrow-scan spectra recorded from a film deposited at a bias voltage of -140 V.

The Ti 2p spectrum shows several contributions, and is similar to that reported previously by other authors for the spontaneously-formed passive oxide layer on bulk Ti6Al4V alloy. The spectrum is dominated by a major TiO2 contribution (BE Ti 2p3/2=458.4 eV, 82%), and shows smaller Ti (BE Ti 2p3/2=453.3 eV, 5%), TiO (BE Ti 2p3/2=454.8 eV, 9%) and Ti2O3 (BE Ti 2p3/2=456.5 eV, 4%) contributions. The Al 2p spectrum shows a major Al2O3 contribution at 73.6 eV and a minor Al0 contribution at 71.0 eV. Again this spectrum is similar to that

The last 10 nm of the surface of thin films can be analyzed using X-ray photoelectron spectroscopy (XPS). As an example, we show the results of XPS on Ti6Al4V thin films. The spontaneously passive film formed on the deposited Ti6Al4V films upon exposure to the air was studied through XPS. Fig. 4 shows high resolution narrow-scan spectra recorded from the Ti 2p, Al 2p, V 2p and O 1s spectra recorded from one of the samples. All the narrow-scan spectra recorded from the rest of the samples are almost identical to

Fig. 22. 2p, Al 2p, V 2p and O 1s XPS narrow-scan spectra recorded from a film deposited at

The Ti 2p spectrum shows several contributions, and is similar to that reported previously by other authors for the spontaneously-formed passive oxide layer on bulk Ti6Al4V alloy. The spectrum is dominated by a major TiO2 contribution (BE Ti 2p3/2=458.4 eV, 82%), and shows smaller Ti (BE Ti 2p3/2=453.3 eV, 5%), TiO (BE Ti 2p3/2=454.8 eV, 9%) and Ti2O3 (BE Ti 2p3/2=456.5 eV, 4%) contributions. The Al 2p spectrum shows a major Al2O3 contribution at 73.6 eV and a minor Al0 contribution at 71.0 eV. Again this spectrum is similar to that

**5. Chemical characterization of thin films** 

those presented in Fig. 22.

a bias voltage of -140 V.

**5.1 X-ray photoelectron spectroscopy (XPS) analysis** 

observed for the passive oxide layer on a Ti6Al4V alloy. The O 1s peak shows three different contributions: a main one at 530.2 eV that can be associated with metal-oxygen bonds, and two much less intense contributions at 531.7 eV and 532.8 eV, which correspond to the presence of OH- groups and chemisorbed water, respectively. The fit of the V 2p spectrum is complicated by the fact that the V peaks overlap strongly with the O 1s K3,4 X-ray satellite peaks. Therefore, the whole spectrum was refined considering the presence of these satellites and the presence of vanadium peaks. The results show the presence of an intense peak at 515.4 eV, which we associate with the presence of VO2, and a less intense vanadium contribution at 516.7 eV, which corresponds to V2O5. This finding contrasts with those of Milosev (Milosev *et al.*, 2000), who did not find oxidized vanadium in the passive layer formed on bulk Ti6Al4V (Alfonso *et al.*, 2006).

These results show that the films that have been grown through rf magnetron sputtering reproduced the stiochoimetry of the target, although on the surface of the thin films chemical reactions will be produced, forming hydroxides and passivation layers.

#### **5.2 Energy dispersive X-ray (EDX) analysis**

The elemental composition of thin films is determined through energy dispersive X-ray analysis (EDX), although is important to indicate that elements such as oxygen and nitrogen are not possible to assess because the binding energies are very near to the binding energy of the carbon. As an example, the EDX results for NbN nanoparticle are shown. Fig. 23 shows the elements that constitute the nanoparticles of NbN (N and Nb) that formed the NbN studied in section 4, and the elements of the substrate (common glass) on which the thin films were grown. The Cu belongs to the sampling.

Fig. 23. EDX spectrum of NbN nanoparticle that integrates the NbN thin films.

Thin Film Growth Through Sputtering Technique and Its Applications 425

Films of CrN, TiN, ZrN, TaN and NbN were deposited using an unbalanced magnetron sputtering system with different energy parameters to investigate its effect on some film properties. Fig. 23 shows the main results obtained in these films. In general, it may be observed that the residual compressive stresses, determined using the curvature method and Stoney's equation (Stoney, 1909), and increased with the energy parameter, although for

Fig. 25a shows the microhardness of the substrate–film system. Hardness measurements were made on samples deposited on AISI M2 tool steel. The hardness of the substrate was 800 kg mm−2. The hardness of the films was 2–3 times higher than the substrate hardness. Group IV nitrides generally are harder than those of group V (Hofmann, 1990). Moreover, the data showed that the hardness increased as the energy parameter increased. This could be attributed to the increase in ion bombardment on the substrate surface, increasing adatom mobility and producing denser films, and the effect on hardness of compressive

Fig. 25b presents the wear coefficient, Kwear, which was evaluated using a ball cratering system, which is a micro scale abrasion test. Two-body grooving abrasion was the wear mechanism observed for all coatings (not shown), probably due to the micro-cutting action of abrasive particles that were dragged across the ball, basically remaining fixed to the ball surface during the test (Adachi, K. & Hutchings, 2003). The wear coefficients were all in the 10−6 mm3 N−1m−1 range, and the lowest values were obtained for NbN and TaN films. It may be seen that there was a slight decrease in k wear as the energy parameter increased, but the variation was too small to be considered important. In addition, no clear trend was observed as the unbalance coefficient was increased. This might be a consequence of variations in other film properties, such as the coefficient of friction or the roughness, parameters known

Ceramic films like metallic nitrides on a metal substrate are commonly believed to be immune to corrosion. Fig. 25 c-d shows the potentiodynamic polarization curves for the films deposited on AISI 304 substrate at the two energy parameters and the AISI 304 substrate in an electrolyte of 0.5 M H2SO4 + 0.05 M KSCN. The corrosion resistance of a material in the polarization curve is determined by its ability to retain low current densities as the electric potential increases. Quantitatively, Tafel analysis was used to determine the corrosion potential, Ecorr, while the critical passivation current density, Icrit, was estimated from the maximum anodic dissolution current before passivation. The corrosion behavior of PVD ceramic-coated steels in aqueous solutions has been increasingly investigated in recent years (Vyas *et al*., 2010; Yang *et al*., 2008; Wesley, 2001; Li *et al.*, 2009; Kumar, A. & Kaur, 2009). One of the major drawbacks is the presence of defects, which are associated with the growth process in PVD ceramic coatings. These coating defects (*e.g.* pores) are particularly deleterious for corrosion protection, since they provide direct paths through which the electrolyte can reach the coating/substrate interface, where localized galvanic corrosion occurs due to the difference in the corrosion potential between the coating and the steel. A lot of research has been undertaken to deal with this problem of coating defects (Stansbury& Buchnan, 2000; Kaciulis *et al.*, 2000; Lang & Yu, 2001; Lee *et al.*, 2009; Chou & Huang, 2003) particularly for transition metal nitride coatings, which in general have excellent wear and

some films the variation was very small.

stresses present in these films (Olaya *et al*., 2009).

to exert a strong influence on the wear performance of the surfaces.

corrosion resistance and therefore are widely used in industry.

## **6. Optical characterization of thin films**

An important characteristic of thin films is the optical behavior, since this determines possible industrial applications, which range from transparent coatings to optical filters; optical studies are carried out through measurements of transmittance, absorbance and reflectance. Fig. 24 shows the transmittance percent as a function of the wavelength on MgO thin films that have been grown through rf sputtering with different power supplies applied to the target.

Fig. 24. Transmission spectrum vs wavelength the films the MgO with various strengths of power supply to the target.

The results show (see Fig. 24) that the films have percentages of transparency that range from 84% to 95% in films grown at 150 and 400 W, respectively. These results indicate that films that were grown at a higher power have a high value of transmittance. This behavior is due to the fact that the films are denser and possibly present lower diffraction.

## **7. Applications of thin films**

Surface modification by means of thin film deposition is an important industrial process used to protect basic materials against wear, fatigue, corrosion and many other surfacerelated damage phenomena (Vyas *et al.*, 2010; Yang *et al*., 2008; Wesley, 2001; Li *et al*., 2009; Kumar, A. & Kaur, 2009). The modern methods of plasma-assisted physical vapor deposition techniques provide great flexibility for designing films with specific chemistry and microstructure, leading to coatings with unique properties. Among these, ceramic coatings deposited on metallic substrates have shown excellent improvement of the surface properties, such as a low friction coefficient and a high degree of hardness with associated good wear resistance and also corrosion resistance to aggressive environments; (Yang, *et al*., 2008; Wesley, 2012).

An important characteristic of thin films is the optical behavior, since this determines possible industrial applications, which range from transparent coatings to optical filters; optical studies are carried out through measurements of transmittance, absorbance and reflectance. Fig. 24 shows the transmittance percent as a function of the wavelength on MgO thin films that have been grown through rf sputtering with different power supplies applied

400 500 600 700 800

**Wavelength (nm)**

Fig. 24. Transmission spectrum vs wavelength the films the MgO with various strengths of

The results show (see Fig. 24) that the films have percentages of transparency that range from 84% to 95% in films grown at 150 and 400 W, respectively. These results indicate that films that were grown at a higher power have a high value of transmittance. This behavior is

Surface modification by means of thin film deposition is an important industrial process used to protect basic materials against wear, fatigue, corrosion and many other surfacerelated damage phenomena (Vyas *et al.*, 2010; Yang *et al*., 2008; Wesley, 2001; Li *et al*., 2009; Kumar, A. & Kaur, 2009). The modern methods of plasma-assisted physical vapor deposition techniques provide great flexibility for designing films with specific chemistry and microstructure, leading to coatings with unique properties. Among these, ceramic coatings deposited on metallic substrates have shown excellent improvement of the surface properties, such as a low friction coefficient and a high degree of hardness with associated good wear resistance and also corrosion resistance to aggressive environments; (Yang, *et al*.,

due to the fact that the films are denser and possibly present lower diffraction.

 150 W 200 W 250 W 300 W 350 W 400 W

**6. Optical characterization of thin films** 

20

power supply to the target.

**7. Applications of thin films** 

2008; Wesley, 2012).

40

60

**Tranmittance (%)**

80

100

to the target.

Films of CrN, TiN, ZrN, TaN and NbN were deposited using an unbalanced magnetron sputtering system with different energy parameters to investigate its effect on some film properties. Fig. 23 shows the main results obtained in these films. In general, it may be observed that the residual compressive stresses, determined using the curvature method and Stoney's equation (Stoney, 1909), and increased with the energy parameter, although for some films the variation was very small.

Fig. 25a shows the microhardness of the substrate–film system. Hardness measurements were made on samples deposited on AISI M2 tool steel. The hardness of the substrate was 800 kg mm−2. The hardness of the films was 2–3 times higher than the substrate hardness. Group IV nitrides generally are harder than those of group V (Hofmann, 1990). Moreover, the data showed that the hardness increased as the energy parameter increased. This could be attributed to the increase in ion bombardment on the substrate surface, increasing adatom mobility and producing denser films, and the effect on hardness of compressive stresses present in these films (Olaya *et al*., 2009).

Fig. 25b presents the wear coefficient, Kwear, which was evaluated using a ball cratering system, which is a micro scale abrasion test. Two-body grooving abrasion was the wear mechanism observed for all coatings (not shown), probably due to the micro-cutting action of abrasive particles that were dragged across the ball, basically remaining fixed to the ball surface during the test (Adachi, K. & Hutchings, 2003). The wear coefficients were all in the 10−6 mm3 N−1m−1 range, and the lowest values were obtained for NbN and TaN films. It may be seen that there was a slight decrease in k wear as the energy parameter increased, but the variation was too small to be considered important. In addition, no clear trend was observed as the unbalance coefficient was increased. This might be a consequence of variations in other film properties, such as the coefficient of friction or the roughness, parameters known to exert a strong influence on the wear performance of the surfaces.

Ceramic films like metallic nitrides on a metal substrate are commonly believed to be immune to corrosion. Fig. 25 c-d shows the potentiodynamic polarization curves for the films deposited on AISI 304 substrate at the two energy parameters and the AISI 304 substrate in an electrolyte of 0.5 M H2SO4 + 0.05 M KSCN. The corrosion resistance of a material in the polarization curve is determined by its ability to retain low current densities as the electric potential increases. Quantitatively, Tafel analysis was used to determine the corrosion potential, Ecorr, while the critical passivation current density, Icrit, was estimated from the maximum anodic dissolution current before passivation. The corrosion behavior of PVD ceramic-coated steels in aqueous solutions has been increasingly investigated in recent years (Vyas *et al*., 2010; Yang *et al*., 2008; Wesley, 2001; Li *et al.*, 2009; Kumar, A. & Kaur, 2009). One of the major drawbacks is the presence of defects, which are associated with the growth process in PVD ceramic coatings. These coating defects (*e.g.* pores) are particularly deleterious for corrosion protection, since they provide direct paths through which the electrolyte can reach the coating/substrate interface, where localized galvanic corrosion occurs due to the difference in the corrosion potential between the coating and the steel. A lot of research has been undertaken to deal with this problem of coating defects (Stansbury& Buchnan, 2000; Kaciulis *et al.*, 2000; Lang & Yu, 2001; Lee *et al.*, 2009; Chou & Huang, 2003) particularly for transition metal nitride coatings, which in general have excellent wear and corrosion resistance and therefore are widely used in industry.

Thin Film Growth Through Sputtering Technique and Its Applications 427

**b**

 TiN CrN ZrN TaN NbN

**c**

 TiN CrN ZrN TaN NbN

0 15 30 45 60 75

Ep (eV/at)

0 15 30 45 60 75

(eV/at)

E p

1600

3,0x10-6

4,0x10-6

5,0x10-6

6,0x10-6

K

wear

7,0x10-6

8,0x10-6

9,0x10-6

2000

HV0.025

2400

The critical passivation current density shows that in contrast to the other properties of the films, there was a decrease in the corrosion resistance of the films as the energy parameter increased with Ep. The critical passivation current density is proportional to the exposed area of the substrate due to the existence of pores or pinholes in the film ( Olaya *et al.*, 2005). The value of Icrit depends on the grain limits that join the columns due to the fact that they can contain vacancies, micropores, pinholes, and possibly microcracks, allowing the diffusion of electrolytes of the corrosive solution toward the substrate and increasing degradation of the coatings.

Therefore, the results reported in Fig. 25 c-d suggest that as the energy parameter increased there was more substrate area exposed at the bottom of the pinholes. These pinholes or defects are usually localized at the grain boundaries, which are defined by the crystal growth process, which consequently models the final film structure. The rise in the corrosion current with Ep was very small for the NbN and TiN films, but significantly higher for TaN, ZrN and CrN. However, the higher ion bombardment induced by the degree of magnetic field created more defects, increasing the residual stresses and therefore deteriorating the film–substrate adhesion and consequently the response of the coatings to the corrosion products, as was visually observed at the end of the corrosion test for the TaN film. This phenomenon can be explained by galvanic coupling produced by a difference in the corrosion potential between coated and uncoated specimens. The potential difference is characterized by anodic dissolution of the substrate material with a high anodic current density at the defect site, leading to an adhesion failure of the coating.

The critical passivation current density shows that in contrast to the other properties of the films, there was a decrease in the corrosion resistance of the films as the energy parameter increased with Ep. The critical passivation current density is proportional to the exposed area of the substrate due to the existence of pores or pinholes in the film ( Olaya *et al.*, 2005). The value of Icrit depends on the grain limits that join the columns due to the fact that they can contain vacancies, micropores, pinholes, and possibly microcracks, allowing the diffusion of electrolytes of the corrosive solution toward the substrate and increasing

Therefore, the results reported in Fig. 25 c-d suggest that as the energy parameter increased there was more substrate area exposed at the bottom of the pinholes. These pinholes or defects are usually localized at the grain boundaries, which are defined by the crystal growth process, which consequently models the final film structure. The rise in the corrosion current with Ep was very small for the NbN and TiN films, but significantly higher for TaN, ZrN and CrN. However, the higher ion bombardment induced by the degree of magnetic field created more defects, increasing the residual stresses and therefore deteriorating the film–substrate adhesion and consequently the response of the coatings to the corrosion products, as was visually observed at the end of the corrosion test for the TaN film. This phenomenon can be explained by galvanic coupling produced by a difference in the corrosion potential between coated and uncoated specimens. The potential difference is characterized by anodic dissolution of the substrate material with a high anodic current density at the defect site, leading to an

0 15 30 45 60 75 90

(eV/at)

E p **a**

 TiN CrN ZrN TaN NbN

degradation of the coatings.

adhesion failure of the coating.





residual Stress (GPa)




Thin Film Growth Through Sputtering Technique and Its Applications 429

growth parameters have on the crystallinity and the micro-structure of the thin films has been discussed, and based on the described models, examples have been provided of the thin films' growth under each of the growth parameters. The fundamental idea of presenting these basic theories is to introduce the science of materials to young researchers in the world, helping them to understand by means of examples the basic concepts that apply to the growth of thin films, with the hope that a better understanding of these theories

Adachi, K. & Hutchings, I., Wear-mode mapping for the micro-scale abrasion test. *Wear*, Vol

B. A. Movchan and A. V. Demchishin, Study of the Structure and Properties of Dioxide thin,

B. Rauschenbach, J. W. Gerlach, Texture Development in Titanium Nitride Films Grown by Low-Energy Ion Assisted Deposition, Cryst. Res. Technol. Vol. 35, (2000) 6-7. B. Wang, Y.N. Zhao, Z. He, The effects of deposition parameters on the crystallographic

B. Window & N. Savvides, Charged particle fluxes from planar magnetron sputtering

Barna, P. , Adamik, M., Fundamental structure forming phenomena of polycrystalline films and the structure zone models. Thin Solid Films, Vol. 317, (April 1998) 27-33. C. Lin, C. Lee, Growth of Tantalum Nitride Films on Si by Radio Frequency Reactive Sputtering

C. S. Shin, D. Gall, Y. W. Kim, N. Hellgren, I. Petrov, J.E. Greene, Development of preferred

C.R. Weissmantel, Deposition of Metastable Films by Ion Beam and Plasma Techniques,

Chou, W., Yu, G. & Huang, J. Corrosion resistance of ZrN films on AISI 304 stainless steel substrate. *Surface and Coatings Technology*. Vol. 167, issue 1, (april 2003) 59-67. Emery, C., Chourasia,A., Yashar, P., A study of CrNx thin films by X-ray photoelectron

Flores, J. F.; Olaya, J. J.; Colás, R.; Rodil, S. E.; Valdez, B. S.; Fuente, I. G. Corrosion behaviour

G. G. Stoney, The measurement of stress in layered structures, in: Proc. Roy. Soc. London

G. Kamath , A.P. Ehiasarian, Y. Purandare, P.Eh. Hovsepian, Tribological and oxidation

Goldfarb, I., Pelleg, J., Zevin, L. and Croitoru, N. (1991). Lattice distortion in thin films of

orientation of AIN films prepared by RF reactive sputtering, Vacuum. Vol. 48, (1997)

of Ta in N2/Ar Gas Mixtures: Effect of Bias, J. Electrochem. Soc. Vol. 147, (2000) 713-718.

orientation in polycrystalline NaCl-structure δ-TaN layers grown by reactive magnetron sputtering: Role of low-energy ion surface interactions , J. Appl. Phys.

*Proc. 9th Int. Vacuum Congress and 5th Int. Conf. on Solid Surfaces* (Madrid), 26 Sept-1

spectroscopy J. Elect. Spectros. Relat. Phenom. Vol. 104, issue 1-3, (July 1999) 91-97.

of TaN thin PVD films on steels. Corrosion Engineering, Science and Tecjnology,

behaviour of TiAlCN/VCN nanoscale multilayer coating deposited by the combined HIPIMS/(HIPIMS-UBM) technique, Surface & Coatings Technology 205

IVB metal (Ti, Zr, Hr) Nitrides, *Thin Solid Film*, Vol. 200, issue 1, (May 1991) 117-

will produce new and better applications.

255, issues 1-6, (august-september 2003) 23-29.

sources, J. Vac. Sci. Technol. A4 (1986) 196–202

B. D. Cullity, s: R. Stock, Elements of X-ray diffraction, Prentice Hall, 2001.

Fiz. Met. Metalloved. Vol. 28, (1969) 83-90.

**9. References** 

427-429.

92, (2002) 5084.

Oct 1983, p 229-308

Vol. A82, (1909) 172.

(2011) 2823–2829

127., (May 1991), 117-127.

Vol. 4, issue 2, (June 2006) 168-176.

Fig. 25. Summary of mechanical and chemical properties of films deposited through sputtering. a) residual stress b) hardness c) wear coefficient d) critical passivation current and d) corrosion potential

#### **8. Conclusion**

In this chapter, the main physical-chemical models utilized to explain the growth of thin films through the rf sputtering technique have been presented, and the influence that the growth parameters have on the crystallinity and the micro-structure of the thin films has been discussed, and based on the described models, examples have been provided of the thin films' growth under each of the growth parameters. The fundamental idea of presenting these basic theories is to introduce the science of materials to young researchers in the world, helping them to understand by means of examples the basic concepts that apply to the growth of thin films, with the hope that a better understanding of these theories will produce new and better applications.

#### **9. References**

428 Crystallization – Science and Technology

**d**

 TiN CrN ZrN TaN NbN

**e**

 TiN CrN ZrN TaN NbN

0 15 30 45 60 75

0 15 30 45 60 75

(eV/at)

E p

In this chapter, the main physical-chemical models utilized to explain the growth of thin films through the rf sputtering technique have been presented, and the influence that the

Fig. 25. Summary of mechanical and chemical properties of films deposited through sputtering. a) residual stress b) hardness c) wear coefficient d) critical passivation current

(eV/at)

E p

0,1



E

and d) corrosion potential

**8. Conclusion** 

corr (mV)


Icrit (mA/cm2

)

1


Thin Film Growth Through Sputtering Technique and Its Applications 431

K. Shibahara, S. Nishino, and H. Matsunam, Antiphase‐domain‐free growth of cubic SiC on

Kaciulis, S., Mezzi, A., Montesperelli, G. Lamastra, F., Rapone, M. Casadei, F., Valente, T. &

Kelly,P., Arnell, R., Magnetron sputtering: a review of recent developments and

Kelly,P., Arnell, R., The influence of magnetron configuration on ion current density and

Kumar,A. & Kaur, D., Nanoindentation and corrosion studies of TiN/NiTi thin films for

L. Monroy, J. Olaya, M. Rivera2, A. Ortiz3, G. Santana. Growth study of Y-Ba-Cu-O on

Lang, F. & Yu, Z. Surf. Coat. Technolology, The corrosion resistance and wear resistance of

Lee, K., Kim, W., Young, J., Won, S. & CHOE, H. Effects of TiN film coating on

M. Ishihara, S. J. Li, H. Yumoto, K. Akashi, Y. Ide, Control of preferential orientation of AlN films prepared by the reactive sputtering method Thin Solid Films, Vol. 316 (1998) 152. M. Marlo, V. Milman, Density- functional study of bulk and surface properties of titanium

Martin, O. Banakh, A.M.E. Santo, S. Springer, R. Sanjines, J. Takadoum, F. Levy, Appl. Surf.

Mayrhofer,P., Tischler, G., Mitterer, C., Microstructure and mechanical/thermal properties

Messier, R., Giri, A.P., Roy, R.A., Revised structure zone model for thin film physical

Mitsunori Kobayashi, Yoshihiko Doi, TiN and TiC coating on cemented carbides by ion

Olaya, J., Rodil, S., Muhl, S. & Sánchez, E. Comparative study of chromium nitride coatings

Olaya, J., Rodil,S., Marulanda, D & B. Bhushan, Propiedades mecánicas de nitruros

limpia. *Revista Mexicana de física*. , Vol.55, issue 6 (november 2009), 425-431. P. Losbichler, C. Mitterer, Non-reactively sputtered TiN and TiB2 films: influence of activation energy on film growth, Surf. Coat. Technol. Vol. 97, (1997) 567.

plating, Thin Solid Films, Vol. 54, Issue 1, 2, (October 1978) 67-74

*Nonferrous Metals Society of China*, Vol. 19, issue 4, (august 2009) 857-861. Lingxia Hang , Y. Yin, Junqi Xu, Optimisation of diamond-like carbon films by unbalanced

applications. *Vacuum*, Vol. 56, (September 1999) 159-152.

*Coatings Technology,* Vol. 108–109 (1998) 317–322

LatinAm. Metal. Mat., Vol. 32 (October 2011) 21-29

Volume 515, Issue 1, (25 September 2006) 357-361.

Coat. Technol. Vol.142-144, (July 2001), 78-84.

Films, Vol. 474, issues 1-2, (march 2005), 119-126.

Vol. 145, issue 1-3, (august 2001) 80-87.

Num. 4 ( January 2000), pp. 2899-2907

Sci. Vol. 185 (2001) 123–133.

2(2, Pt. 1), 500-3 .

Gusmano, G. Multi-technique study of corrosion resistant CrN/Cr/CrN and CrN : C coatings. *Surface and Coatings Technology*, Vol. 201, issue 1-2, 12, (6

deposition rate in a dual unbalanced magnetron sputtering system. *Surface and* 

biomedical applications. *Surface and Coatings Technology*, Vol. 204, issue 6-7, (14

buffer layers and different substrates made by ultrasonic spray pyrolysis. Rev.

thick TiN coatings deposited by arc ion plating. *Surface and Coatings Technology,*

electrochemical behaviors of nanotube formed Ti-xHf alloys. *Transactions of* 

magnetron sputtering for infrared transmission enhancement , Thin Solid Films,

nitride using different exchange-correlation functional, Physical Review B, Vol. 63,

of Cr-N coatings deposited by reactive unbalanced magnetron sputtering, Surf.

structure. *Journal of Vacuum Science and Technology, A: Vacuum, Surfaces, and Films*,

deposited by unbalanced and balanced magnetron sputtering. Thin Solid

metálicos depositados con UBM: tecnología eficiente y ambientalmente

Si(100), Appl. Phys. Lett., 50, (1987) 1888.

september 2006) 313-319.

december 2009), 1132-1136.


Greene, J. E., Low Energy Ion Bombardment during Film Deposition from the Vapor Phase:

Greene, J. E., Nucleation, Film Growth, and Microstructural Evolution, *Deposition Processes for Films and Coating*, 2nd edition, (R. Bunshah,ed.), Ch. 13, Noyes Publications (1994). H. Liu, W. Tang, D. Hui, L. Hei, F. Lu. Characterization of (Al, Si)N films deposited by

H.B. Nie, S.Y. Xu, S.J. Wang, L.P. You, Z. Yang, C.K. Ong, J. Li and T.Y.F. Liew, Structural

Hofmann S., Target and substrate surface reaction kinetics in magnetron sputtering of

Hwan-Chul Leea, Guen-Hong Kimb, Soon-Ku Hongc, Ki-Young Leea, Yoon-Joong Yonga,

I. Milosev, M. Metikos-Hukovic and H.-H Strehblow, Passive film on orthopaedic TiAlV by

J. E. Alfonso, Fernando Pacheco, Alvaro Castro P, J. Torres, Influence of the substrate bias

J. E. Greene, J.-E. Sundgren, L. Hultman, I. Petrov, and D. B. Bergstrom, Development of

J. Musil, S. Kadlec, Reactive sputtering of TiN films at large substrate to target distances,

J. Pelleg, L. Z. Zevin, S. Lungo, Lattice distortion in thin films of IV B metal (Ti, Zr, Hr)

J. R. Frederick, D. Gall, Surface morphological evolution of epitaxial CrN(001) layers, J.

J.A. Heimberg, K.J. Wahl, & I. L. Singer, Superlow friction behavior of diamond like carbon

J.J. Olaya, L. Huerta, S.E. Rodil, R. Escamilla, Superconducting niobium nitride films

K. Aso, M. Hayakawa, H. Matsuda, K. Hayashi, W. Ishikawa, On specimen preparation by radio

K. Bewilogua a, G. Bra¨uer a, A. Dietz a, J. Ga¨bler a, G. Goch (1)b, B. Karpuschewski (1)c, B.

K. S. Harsha, Principles of Physical Vapor Deposition of Thin Films, Elsevier, 2005.

coatings: Time and speed effects, Applied Physics Letters, Vol. 79, issue 17, (april

deposited by unbalanced magnetron sputtering , Thin Solid Films, Volume Vol.

frequency planar magnetron sputtering , J. Vac. Sci. technol. Vol. A4, issue 1, (1986) 19-25.

Szyszka, Surface technology for automotive engineering, Manufacturing

Reactive Magnetron Sputtering, Appl. Phys. Letters, Vol. 67, (1995) 2928. J. H. Je, D. Y. Noh, H.K. Kim, K.S.Liang, Preferred orientation of TiN films studied by a real

time synchrotron x-ray scatteringJ. Appl. Phys. 81, (1997) 6126.

nitride coatings, Thin Solid Films, Vol. 191, (1990) 335-348.

Solid Films, Vol. 261, issue 1-2, (1 June 1995) 148-153.

X-ray photoelectron, Biomaterials, Vol. 21, (2000), 2103.

2009) 5988-5993.

(2001) 229- 236.

2006) 994-996.

2001) 2449-2451.

Vacuum, Vol. 40 (1990) 435.

Appl. Phys., Vol. 98, (2005) 054906.

516, issue 23, (1 october 2008) 8768-8773

Technology, Vol. 58, (2009) 608–627

nitrides Thin Solid Film, Vol. 197, (1991) 117.

Effects on Microstructure and Microchemistry, *Solid State Technol*., 30,(4), (1987)115.

balanced magnetron sputtering, Thin Solid Films, Vol. 517, issue 21, 1 (September

and electrical properties of tantalum nitride thin films fabricated by using reactive radio-frequency magnetron sputtering, Appl. Phys. A. Mater. Sci. process, Vol. 73

Chang-Hwan Chunb, Jai-Young Leea, Influence of sputtering pressure on the microstructure evolution of AlN thin films prepared by reactive sputtering, Thin

voltaje on the crystallographic structure and mechanical properties of Ti6Al4V coatings, deposited by rf magnetro, Phys. Stat. sol. Vol. 10, (2005) 3786-3789. J. E. Alfonso, J. Torres, and J. F. Marco,. Influence of the Substrate Bias Voltage on the

Crystallographic Structure and Surface Composition of Ti6A14V Thin Films Deposited by rf Magnetron Sputtering, Brazilian Journal of Physics, Vol. 36, issue 3, (September

Preferred Orientation in Polycrystalline TiN Layers Grown by Ultra-High-Vacuum


**16** 

 *Mexico* 

**Crystallization of Ge:Sb:Te Thin Films for** 

J. J. Gervacio Arciniega, E. Prokhorov, F. J. Espinoza Beltran and G. Trapaga

Chalcogenide glasses are a chemical compound consisting of at least one chalcogen element, sulphur, selenium, or tellurium, in combination with other elements. These glasses obtained great attention after discovery between 1962 and 1969 by Kolomiets, Eaton, Ovshinsky and Pearson of the S-shape current-voltage characteristic in chalcogenide glasses and the switching phenomenon from high to low resistivity states (Popescu, 2005). In 1968 S. R. Ovshinsky demonstrated very short (about of 10-10 seconds) reversible electrical switching phenomena in Te81Ge15Sb2S2 thin films due to amorphous-crystalline phase transition (Ovshinsky, 1968). This work opened a new area of phase change technology, which now is one of the most important technologies for memory devices and applications in computers,

The first electrical phase-change memory devices used various binary, ternary and quaternary Te-based films with compositions made up of Ge:Te, Si:As:Te, Ge:As:Si:Te, etc., systems (Stand, 2005). The early phase change materials used in optical storage comprised simple alloys based primarily on compositions in the vicinity of the tellurium-germanium eutectic. Antimony was primarily used, although other elements including selenium, arsenic and bismuth were all shown to have beneficial effects. Based on the results obtained from films with Ge15Sb4Te81 composition (Ovshinsky, 1971) the first application of Ge:Sb:Te alloys

In the early 1990s, a second generation of high speed phase change materials based on Ge:Sb:Te alloys was reported by several optical memory research groups (Ohta et al 1989, Yamada et al, 1991, Gonzalez-Hernandez et al, 1992). These alloys have stoichiometric compositions along the GeTe–Sb2Te3 pseudobinary line of phase diagram such as Ge2Sb2Te5,

Ge:Sb:Te stoichiometric alloys have three phases: one amorphous and two crystal structures. The first crystalline phase is the rock salt NaCl-like and the second more stable phase is hexagonal. When the amorphous films are heated, the transition from amorphous to rock salt-like structure occurs at around 120-1700C; subsequent heating transforms this phase into a stable hexagonal structure at temperatures around 200-2500C. The exact transition temperature depends on the composition of the film. The hexagonal structure remains stable over a wide temperature range from 200-2500C to around the melting point (593-630 0C,

CD, DVD, phase-change random access memories, etc.

was reported for phase change rewritable optical disks.

**1. Introduction** 

Ge1Sb2Te4 and Ge1Sb4Te7.

**Phase Change Memory Application** 

*CINVESTAV, Unidad Queretaro, Juriquilla, Querétaro,* 


## **Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application**

J. J. Gervacio Arciniega, E. Prokhorov, F. J. Espinoza Beltran and G. Trapaga *CINVESTAV, Unidad Queretaro, Juriquilla, Querétaro, Mexico* 

## **1. Introduction**

432 Crystallization – Science and Technology

P. Ziemann, E. Kay, Correlation between the ion bombardment during film growth of films

P.B. Barna, M. Adamik, in: F.C. Matacotta, G. Ottaviani Eds., Science and Technology of

P.B. Barna, M. Adamik,Fundamental structure forming phenomena of polycrystalline films

Pavel Souček, Tereza Schmidtová, Evaluation of composition, mechanical properties and

Petrov,I, Barna, P, Hultman,L, Greene,J., Microstructural evolution during film growth*. J.* 

Raymond Constantin, Performance of hard coatings, made by balanced and unbalanced

René Olivares-Navarrete , Jhon Jairo Olaya, Claudia Ramirez and Sandra Elizabeth Rodil,

S.E. Rodil and J. J Olaya, Unbalanced magnetic field configuration: plasma and film properties, Journal of Physics: Condensed Matter, Vol. 18,. 1703-1720, 2006. Siegwart, R. (2001), Indirect Manipulation of a Sphere on a Flat Disk Using Force

Stansbury E. and Buchnan R. *Fundamentals of Electrochemical Corrosion*. First edition. ASM

Stefan Wrehde, Marion Quaas, Robert Bogdanowicz, Hartmut Steffen, Harm Wulff, Rainer

T. Q. Li, S. Noda, Y. Tsuji, T. Osawa, and H. Komiyama, J. Vac. Sci. Technol. Vol. A 20 (3),

W.F. Liu, S. Suzuki, K. Machida, Magnetic properties of Nd–Fe–B film magnets prepared by

Waits R.K. J Planar magnetron sputtering, Vac. Sci. Technol. Vol. 15, Issue 2, (1978) 179-87. Window B, Savvides N, Unbalanced dc magnetron as source high ion flux, J Vac. Sci.

Y. Ochiai, K. Aso, M. Hayakawa, H. Matsuda, K. Hayashi, W. Ishikawa, On specimen

Y.H. Cheng, B.K. Tay, S.P. Lan, Influence of deposition temperature on the structure and

RF sputtering, Journal of Magnetism and Magnetic Materials, Vol. 308, Issue 1,

preparation by radio frequency planar magnetron sputtering , J. Vac. Sci. technol.

internal stress of TiN films deposited by filtered cathodic vacuum arc, J. Vac. Sci.

Teer D.G. A , magnetron sputter ion-plating system, Surf Coat Technol, Vol. 39-40, (1989) 565. Thornton, J., Influence of apparatus geometry and deposition conditions on the structure and topography of thick sputtered coating, J. Vac. Sci. Technol., Vol. 11, (1974) 666-670. W. Ensinger, Tribological behavior of duplex coating improved by ion implantation, Surf.

International). ISBN 0-87170-676-8, United States of America(2000).

magnetron sputtering, Vacuum, Vol. 82, (2008) 1115–1119

and the structure zone models., Thin Solid Films, 317, (208) 27-33.

*Vac. Sci. Technol. A,* Vol 21, No 5, (September 2003), 117-128.

Coatings Technology, Vol. 120-121, (November 1999) 728-733.

Thin Films World Scientific Publ., Singapore, 1995.

1 (1983) 512.

16 September 2011.

2555-2559

(December 2009),12-16.

Coat. Technol. Vol. 99, (1998) 1

Technol. Vol. A 4, Issue 2 (1986) 453-456.

(January 2007) 126-130.

Vol. A4, Issue 1, (1986) 19-25

Technol. Vol. A 20, (2002) 1270.

(2002) 583-588.

and their structure and electrical properties J. Vac. Sci. Technol., A, Vac. Surf. Films,

structure of NC-TiC/a-C:H coatings prepared by balanced magnetron sputtering, Surface and Coatings Technology, In Press, Accepted Manuscript, Available online

magnetron sputtering, for decorative applications Baham Miremad, Surface and

Coatings 1 (2011) 72-87 G. Ramírez, S.E. Rodil, H. Arzate, S. Muhl, J.J. Olaya Biocompatibility of Niobium coatings, Applied Surface Science, Vol. 257, (2011)

Information. *International Journal of Advanced Robotic Systems,* Vol.6, No.4,

Hippler, Optical and chemical characterization of thin TiNx films deposited by DC-

Chalcogenide glasses are a chemical compound consisting of at least one chalcogen element, sulphur, selenium, or tellurium, in combination with other elements. These glasses obtained great attention after discovery between 1962 and 1969 by Kolomiets, Eaton, Ovshinsky and Pearson of the S-shape current-voltage characteristic in chalcogenide glasses and the switching phenomenon from high to low resistivity states (Popescu, 2005). In 1968 S. R. Ovshinsky demonstrated very short (about of 10-10 seconds) reversible electrical switching phenomena in Te81Ge15Sb2S2 thin films due to amorphous-crystalline phase transition (Ovshinsky, 1968). This work opened a new area of phase change technology, which now is one of the most important technologies for memory devices and applications in computers, CD, DVD, phase-change random access memories, etc.

The first electrical phase-change memory devices used various binary, ternary and quaternary Te-based films with compositions made up of Ge:Te, Si:As:Te, Ge:As:Si:Te, etc., systems (Stand, 2005). The early phase change materials used in optical storage comprised simple alloys based primarily on compositions in the vicinity of the tellurium-germanium eutectic. Antimony was primarily used, although other elements including selenium, arsenic and bismuth were all shown to have beneficial effects. Based on the results obtained from films with Ge15Sb4Te81 composition (Ovshinsky, 1971) the first application of Ge:Sb:Te alloys was reported for phase change rewritable optical disks.

In the early 1990s, a second generation of high speed phase change materials based on Ge:Sb:Te alloys was reported by several optical memory research groups (Ohta et al 1989, Yamada et al, 1991, Gonzalez-Hernandez et al, 1992). These alloys have stoichiometric compositions along the GeTe–Sb2Te3 pseudobinary line of phase diagram such as Ge2Sb2Te5, Ge1Sb2Te4 and Ge1Sb4Te7.

Ge:Sb:Te stoichiometric alloys have three phases: one amorphous and two crystal structures. The first crystalline phase is the rock salt NaCl-like and the second more stable phase is hexagonal. When the amorphous films are heated, the transition from amorphous to rock salt-like structure occurs at around 120-1700C; subsequent heating transforms this phase into a stable hexagonal structure at temperatures around 200-2500C. The exact transition temperature depends on the composition of the film. The hexagonal structure remains stable over a wide temperature range from 200-2500C to around the melting point (593-630 0C,

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 435

Sb2Te3 pseudobinary line (Ge2Sb2Te5 Ge1Sb2Te4 and Ge1Sb4Te7) including GeTe and Sb2Te3 (which lie at the end of GeTe-Sb2Te3 pseudobinary line), taking into account the processes during the incubation time. It is necessary to note that GeTe and Sb2Te3, as well as Ge:Sb:Te, can be used in phase-change data storage. Additionally, the crystallization kinetics of Ge4Sb1Te5 films will be present in this chapter. The structure of this alloy is unclear. In contrast to other Ge:Sb:Te ternary compounds, it does not belong to the GeTe – Sb2Te3 homologous series, although it lies on (or very close to) the GeTe-Sb2Te3 pseudobinary line in the Ge:Sb:Te phase diagram between GeTe and Ge2Sb2Te5. It is suggested (Coombs et al., 1995) that this alloy is a solid solution between GeTe and a compound with a composition close to Ge4Sb1Te5. This material, in contrast to stoichiometric alloys, demonstrates only one amorphous-NaCl-type phase transition, but it has the largest optical (Kato et al., 1999) and electrical (Morales-Sanchez et al., 2005) contrast between the crystalline and amorphous

Necessary to note that as-prepared and melt-quenched Ge:Sb:Te amorphous materials show different crystallization kinetics (Nobukuni et al., 1999, Park et al., 1999, Khulbe et al., 2000, Wei et al., 2003, Kalb et al., 2004, Raoux et al., 2008). It has been proposed in the literature that for melt-quenched amorphous materials may exist: embryos following the condensation and evaporation of embryos to form crystalline clusters (Khulbe et al., 2000), preexisting clusters (Park et al., 1999), crystal nuclei (Wei et al., 2003), sinks and voids after repeated overwriting (Nobukuni et al., 1999), locally ordered regions with structure similar to that of crystalline Sb (Naito et al., 2004) or a crystalline amorphous border (Raoux et al., 2008). In spite of different models, crystallization in the melt-quenched amorphous material can start from these nucleation centers, which decrease incubation time and increase crystallization speed. But in

Studies of crystallization kinetics of phase change materials are mostly analyzed using the Johnson–Mehl–Avrami-Kolmogorov (JMAK) model for isothermal annealing (see for example: (Weidenhof et al., 2001, Ruitenberg et al., 2002, Trappe et al., 2000, Morales-Sanchez et al., 2010)), which permitted to determine the activation energy for the crystallization process. This method will be used for investigation of the crystallization

Figure 1 shows in-situ optical reflection (using a laser diode emitting at 650 nm) as a function of temperature with a heating rate of 50C/min for all investigated films with composition indicated on the graph. An abrupt change in reflection is associated with the onset of phase crystallization. Stoichiometric Ge2Sb2Te5, Ge1Sb2Te4 and Ge1Sb4Te7 materials demonstrate two changes: the first corresponds to amorphous-NaCl-type transition and the second to NaCl-type-hexagonal transformation. In contrast, GeTe, Ge4Sb1Te5 and Sb2Te3 show only one amorphous-crystalline transition. The highest crystallization temperature (Tc) has been observed in GeTe, the lowest in Sb2Te3. The crystallization kinetics of the GeTe-Sb2Te3 pseudobinary line will be analyzed according to this Figure from the highest to the

An important aspect to develop thin films of phase change materials is the deposition method. It is well known from the literature that deposition technology strongly affects

this chapter only crystallization of as-prepared materials will be discussed.

states, compared with other Ge:Sb:Te ternary alloys.

kinetics in the compositions mentioned above.

lowest crystallization temperature.

**2. Deposition of phase change films** 

depending on composition) (Gonzalez-Hernandez et al, 1992). The principle of phasechange memory operation is based on a reversible phase-transformation from the amorphous (high resistance/low reflectivity state) to crystalline NaCl-type (low resistance/high reflectivity state) under short laser or electrical pulses. It is necessary to note that for phase-change memory applications only the amorphous-NaCl-type transition has been used, probably because the heating time produced by the laser or electrical pulse is too short to form the stable hexagonal structure (Yamada, 1991). This transformation is always accompanied by abrupt changes in reflection (about 20-30 %) and resistivity (about 3 orders of magnitude) in chalcogenide films. The reversible transformation from crystalline to the amorphous structurally disordered state can be obtained by increasing the local temperature of the Ge:Sb:Te layer above its melting point by short intense laser, or electrical pulses and the subsequent quenching with a cooling rate about 1010 deg/s (Yamada et al, 1991).

The phase transition between the amorphous and NaCl-type crystalline phase in stoichiometric materials is fast because atoms in the amorphous state do not need to travel long distances to take their position in the crystal lattice. On the other hand, nonstoichiometric compounds require long-range diffusion when they crystallize from an amorphous state (Yamada et al, 1991, Yamada 1996), which in general slows down the crystallization process. For example, 30 at. % of the deviation from Ge2Sb2Te5 stoichiometric composition increased the crystallization time from 220 to 500 ns (Kyrsta et al., 2001).

For several years, Ge:Sb:Te alloys along the pseudobinary GeTe–Sb2Te3 line have been used as the main material for optical phase-change data storage devices and are currently being investigated for nonvolatile electronic storage purposes (which utilize the difference in the electrical resistance of the two phases) due to the high reflectivity and resistivity contrast between the amorphous and crystalline phases. Many investigations have been carried out on this type of phase-change materials with the purpose of improving their properties and to increase its storage capacity, stability, speed and versatility. These investigations have shown that the crystallization of the phase-change (or amorphous-NaCl-type transition) can be considered as a rate-limiting process to obtain a fast data transfer. That is why there are a considerable number of experimental and theoretical studies investigating the amorphousto-crystalline NaCl-type phase transformation. In spite of the large number of publications, the activation energy for crystallization of Ge:Sb:Te materials reported in the literature shows a large discrepancy; the activation energy of crystallization for Ge2Sb2Te5 reported in the literature varied between 0.8 and 2.9 eV, for example. Such dispersion in the kinetic parameters might depend on the differences in the deposition methods, type of substrate, dielectric cover layer, film thickness and/or parameters of deposition processes, which lead to differences in the crystallization temperature, activation energy, and the like. In addition, experimental data show relatively long incubation times for crystallization of Ge:Sb:Te films and a relatively large amount of crystallized material during this period of time (Morales-Sanchez et al 2010). Incubation time manifests itself as the time necessary to reach critical values of nucleus for the crystallization to occur (Senkader et al, 2004). The existence of a non-negligible amount of crystallite during the incubation time could also be responsible for the dispersion in the kinetic parameters reported in the literature.

Based on such antecedents, we aimed in this chapter at investigating the crystallization kinetics of thin films on a glass substrate obtained by the same deposition process (DC sputtering), with the same thickness (around 200 nm), and composition along the GeTe–

depending on composition) (Gonzalez-Hernandez et al, 1992). The principle of phasechange memory operation is based on a reversible phase-transformation from the amorphous (high resistance/low reflectivity state) to crystalline NaCl-type (low resistance/high reflectivity state) under short laser or electrical pulses. It is necessary to note that for phase-change memory applications only the amorphous-NaCl-type transition has been used, probably because the heating time produced by the laser or electrical pulse is too short to form the stable hexagonal structure (Yamada, 1991). This transformation is always accompanied by abrupt changes in reflection (about 20-30 %) and resistivity (about 3 orders of magnitude) in chalcogenide films. The reversible transformation from crystalline to the amorphous structurally disordered state can be obtained by increasing the local temperature of the Ge:Sb:Te layer above its melting point by short intense laser, or electrical pulses and

the subsequent quenching with a cooling rate about 1010 deg/s (Yamada et al, 1991).

composition increased the crystallization time from 220 to 500 ns (Kyrsta et al., 2001).

the dispersion in the kinetic parameters reported in the literature.

The phase transition between the amorphous and NaCl-type crystalline phase in stoichiometric materials is fast because atoms in the amorphous state do not need to travel long distances to take their position in the crystal lattice. On the other hand, nonstoichiometric compounds require long-range diffusion when they crystallize from an amorphous state (Yamada et al, 1991, Yamada 1996), which in general slows down the crystallization process. For example, 30 at. % of the deviation from Ge2Sb2Te5 stoichiometric

For several years, Ge:Sb:Te alloys along the pseudobinary GeTe–Sb2Te3 line have been used as the main material for optical phase-change data storage devices and are currently being investigated for nonvolatile electronic storage purposes (which utilize the difference in the electrical resistance of the two phases) due to the high reflectivity and resistivity contrast between the amorphous and crystalline phases. Many investigations have been carried out on this type of phase-change materials with the purpose of improving their properties and to increase its storage capacity, stability, speed and versatility. These investigations have shown that the crystallization of the phase-change (or amorphous-NaCl-type transition) can be considered as a rate-limiting process to obtain a fast data transfer. That is why there are a considerable number of experimental and theoretical studies investigating the amorphousto-crystalline NaCl-type phase transformation. In spite of the large number of publications, the activation energy for crystallization of Ge:Sb:Te materials reported in the literature shows a large discrepancy; the activation energy of crystallization for Ge2Sb2Te5 reported in the literature varied between 0.8 and 2.9 eV, for example. Such dispersion in the kinetic parameters might depend on the differences in the deposition methods, type of substrate, dielectric cover layer, film thickness and/or parameters of deposition processes, which lead to differences in the crystallization temperature, activation energy, and the like. In addition, experimental data show relatively long incubation times for crystallization of Ge:Sb:Te films and a relatively large amount of crystallized material during this period of time (Morales-Sanchez et al 2010). Incubation time manifests itself as the time necessary to reach critical values of nucleus for the crystallization to occur (Senkader et al, 2004). The existence of a non-negligible amount of crystallite during the incubation time could also be responsible for

Based on such antecedents, we aimed in this chapter at investigating the crystallization kinetics of thin films on a glass substrate obtained by the same deposition process (DC sputtering), with the same thickness (around 200 nm), and composition along the GeTe–

Sb2Te3 pseudobinary line (Ge2Sb2Te5 Ge1Sb2Te4 and Ge1Sb4Te7) including GeTe and Sb2Te3 (which lie at the end of GeTe-Sb2Te3 pseudobinary line), taking into account the processes during the incubation time. It is necessary to note that GeTe and Sb2Te3, as well as Ge:Sb:Te, can be used in phase-change data storage. Additionally, the crystallization kinetics of Ge4Sb1Te5 films will be present in this chapter. The structure of this alloy is unclear. In contrast to other Ge:Sb:Te ternary compounds, it does not belong to the GeTe – Sb2Te3 homologous series, although it lies on (or very close to) the GeTe-Sb2Te3 pseudobinary line in the Ge:Sb:Te phase diagram between GeTe and Ge2Sb2Te5. It is suggested (Coombs et al., 1995) that this alloy is a solid solution between GeTe and a compound with a composition close to Ge4Sb1Te5. This material, in contrast to stoichiometric alloys, demonstrates only one amorphous-NaCl-type phase transition, but it has the largest optical (Kato et al., 1999) and electrical (Morales-Sanchez et al., 2005) contrast between the crystalline and amorphous states, compared with other Ge:Sb:Te ternary alloys.

Necessary to note that as-prepared and melt-quenched Ge:Sb:Te amorphous materials show different crystallization kinetics (Nobukuni et al., 1999, Park et al., 1999, Khulbe et al., 2000, Wei et al., 2003, Kalb et al., 2004, Raoux et al., 2008). It has been proposed in the literature that for melt-quenched amorphous materials may exist: embryos following the condensation and evaporation of embryos to form crystalline clusters (Khulbe et al., 2000), preexisting clusters (Park et al., 1999), crystal nuclei (Wei et al., 2003), sinks and voids after repeated overwriting (Nobukuni et al., 1999), locally ordered regions with structure similar to that of crystalline Sb (Naito et al., 2004) or a crystalline amorphous border (Raoux et al., 2008). In spite of different models, crystallization in the melt-quenched amorphous material can start from these nucleation centers, which decrease incubation time and increase crystallization speed. But in this chapter only crystallization of as-prepared materials will be discussed.

Studies of crystallization kinetics of phase change materials are mostly analyzed using the Johnson–Mehl–Avrami-Kolmogorov (JMAK) model for isothermal annealing (see for example: (Weidenhof et al., 2001, Ruitenberg et al., 2002, Trappe et al., 2000, Morales-Sanchez et al., 2010)), which permitted to determine the activation energy for the crystallization process. This method will be used for investigation of the crystallization kinetics in the compositions mentioned above.

Figure 1 shows in-situ optical reflection (using a laser diode emitting at 650 nm) as a function of temperature with a heating rate of 50C/min for all investigated films with composition indicated on the graph. An abrupt change in reflection is associated with the onset of phase crystallization. Stoichiometric Ge2Sb2Te5, Ge1Sb2Te4 and Ge1Sb4Te7 materials demonstrate two changes: the first corresponds to amorphous-NaCl-type transition and the second to NaCl-type-hexagonal transformation. In contrast, GeTe, Ge4Sb1Te5 and Sb2Te3 show only one amorphous-crystalline transition. The highest crystallization temperature (Tc) has been observed in GeTe, the lowest in Sb2Te3. The crystallization kinetics of the GeTe-Sb2Te3 pseudobinary line will be analyzed according to this Figure from the highest to the lowest crystallization temperature.

## **2. Deposition of phase change films**

An important aspect to develop thin films of phase change materials is the deposition method. It is well known from the literature that deposition technology strongly affects

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 437

The thin films have been prepared on glass and Si substrates by DC magnetron sputtering from one composite target. The deposition conditions were: 4.5x10-6 mbar base pressure, 2.0x10-5 mbar work pressure, 180 cm3/min Ar gas input flow, 0.49 W/cm2 DC power density, and 10 min deposition time. Before films sputtering, the targets were cleaned for 10 minutes by sputtering to remove the oxide from the surface. According to XRD measurements all films were obtained in amorphous state. Energy dispersive spectroscopy has shown that deviations of films compositions from the targets were approximately 2%. Figure 2a shows typical surface topography obtained from AFM measurements of asprepared amorphous thin films. All amorphous films demonstrated vary smooth surface with the surface roughness between 1 and 3 nm. In contrast, in the crystalline films (annealed to the temperature of 2000C) the topography image shows grains with average size between 20-120 nm (Fig. 2b). The dimensions of grains are dependent on the films

GeTe is a promising candidate for the application of in phase-change technology due to its higher crystallization temperature (onset of crystallization temperature at about 1840C) when compared to other alloys, since it offers a significant improvement in data-retention at high temperature (Perniola et al., 2010, Fantini et al., 2010). Amorphous films, in contrast to Ge:Sb:Te ternary alloys, crystallize in the rhombohedral phase. This phase can be viewed as a rock salt structure distorted along the [111]-direction (Caravati et al., 2010). Additionally, the central atom is displaced along the [111]-direction from the center of the rhombohedron. Upon crystallization, in a rhombohedral (distorted rock salt) structure appeared about 10% of vacancies occurring on Ge sites (Kolobov 2004). GeTe is a classical ferroelectric material that shows displacive type ferroelectric-paraelectric transition from a rhombohedral ferroelectric phase to a rock salt type structure with paraelectric properties. Two sublattices form this rock salt structure: Ge atoms compose one of the lattices and tellurium atoms the

The NaCl type crystalline structure of GeTe is an unstable phase. This crystalline structure exists for temperatures above 300°C. The ferroelectric-paraelectric transition occurs in the interval of 327-4270C depending on various factors, such as exact composition, carrier density, etc. For example, changes in Te composition from 50 to 50.7 at. % shift the transition

Activation energy of amorphous to rhombohedral phase crystallization reported in the literature demonstrates large dispersion: 1.7 eV (Libera et al., 1993), 1.77 eV (Lu et al., 1995), 1.96 eV (Fantini et al., 2010), 2.5 eV (Fan et al., 2004), 3.9 eV (Matsushita et al., 1989). Such dispersion can be dependent on the differences in the deposition methods, type of substrate, method of results interpretation, etc. According to the literature and to our measurements (Fig. 1), GeTe has the highest crystallization temperature for materials on the GeTe-Sb2Te3 pseudobinary line and must demonstrate the highest activation energy of crystallization.

Figure 3 shows X-ray patterns of GeTe film measured at the temperature indicated on the graph. At temperatures below 1800C, the material shows only wide bands, which are characteristic of amorphous materials. At higher temperatures, a rhombohedral GeTe phase appears. At a temperature of 3500C two phases have been observed: rhombohedral and

composition.

**3. Crystallization of GeTe** 

other (Rabe et al., 1987).

temperature from 4270C to 3560C (Okura, 1992).

the microstructure and as a results the optical, electrical and crystallization (especially the crystallization temperature) properties of phase-change materials. For example, the crystallization temperature of the films with the same composition but obtained by different deposition methods can differ for more than 200C, and no suitable explanation has yet been proposed (Morales-Sanchez et al., 2005). That is why, to compare crystallization properties all studied films have to be obtained with the same deposition process (DC sputtering), with the same thickness (around 200 nm), and composition along the GeTe–Sb2Te3 pseudobinary line. This method has been chosen because sputtering can produce films with the desired stoichiometry. Compared, for example, with thermal evaporation, in sputtering the thickness and chemical composition of thin films are easier to control.

Fig. 1. Dependencies of optical reflection on temperature for films with compositions indicated on the graph.

Fig. 2. Surface topography of as-prepare amorphous (a) and crystalline (b) Ge2Sb2Te5 films.

The thin films have been prepared on glass and Si substrates by DC magnetron sputtering from one composite target. The deposition conditions were: 4.5x10-6 mbar base pressure, 2.0x10-5 mbar work pressure, 180 cm3/min Ar gas input flow, 0.49 W/cm2 DC power density, and 10 min deposition time. Before films sputtering, the targets were cleaned for 10 minutes by sputtering to remove the oxide from the surface. According to XRD measurements all films were obtained in amorphous state. Energy dispersive spectroscopy has shown that deviations of films compositions from the targets were approximately 2%.

Figure 2a shows typical surface topography obtained from AFM measurements of asprepared amorphous thin films. All amorphous films demonstrated vary smooth surface with the surface roughness between 1 and 3 nm. In contrast, in the crystalline films (annealed to the temperature of 2000C) the topography image shows grains with average size between 20-120 nm (Fig. 2b). The dimensions of grains are dependent on the films composition.

## **3. Crystallization of GeTe**

436 Crystallization – Science and Technology

the microstructure and as a results the optical, electrical and crystallization (especially the crystallization temperature) properties of phase-change materials. For example, the crystallization temperature of the films with the same composition but obtained by different deposition methods can differ for more than 200C, and no suitable explanation has yet been proposed (Morales-Sanchez et al., 2005). That is why, to compare crystallization properties all studied films have to be obtained with the same deposition process (DC sputtering), with the same thickness (around 200 nm), and composition along the GeTe–Sb2Te3 pseudobinary line. This method has been chosen because sputtering can produce films with the desired stoichiometry. Compared, for example, with thermal evaporation, in sputtering the thickness and chemical composition of thin films are easier

Fig. 1. Dependencies of optical reflection on temperature for films with compositions

a b

Fig. 2. Surface topography of as-prepare amorphous (a) and crystalline (b) Ge2Sb2Te5 films.

to control.

indicated on the graph.

GeTe is a promising candidate for the application of in phase-change technology due to its higher crystallization temperature (onset of crystallization temperature at about 1840C) when compared to other alloys, since it offers a significant improvement in data-retention at high temperature (Perniola et al., 2010, Fantini et al., 2010). Amorphous films, in contrast to Ge:Sb:Te ternary alloys, crystallize in the rhombohedral phase. This phase can be viewed as a rock salt structure distorted along the [111]-direction (Caravati et al., 2010). Additionally, the central atom is displaced along the [111]-direction from the center of the rhombohedron. Upon crystallization, in a rhombohedral (distorted rock salt) structure appeared about 10% of vacancies occurring on Ge sites (Kolobov 2004). GeTe is a classical ferroelectric material that shows displacive type ferroelectric-paraelectric transition from a rhombohedral ferroelectric phase to a rock salt type structure with paraelectric properties. Two sublattices form this rock salt structure: Ge atoms compose one of the lattices and tellurium atoms the other (Rabe et al., 1987).

The NaCl type crystalline structure of GeTe is an unstable phase. This crystalline structure exists for temperatures above 300°C. The ferroelectric-paraelectric transition occurs in the interval of 327-4270C depending on various factors, such as exact composition, carrier density, etc. For example, changes in Te composition from 50 to 50.7 at. % shift the transition temperature from 4270C to 3560C (Okura, 1992).

Activation energy of amorphous to rhombohedral phase crystallization reported in the literature demonstrates large dispersion: 1.7 eV (Libera et al., 1993), 1.77 eV (Lu et al., 1995), 1.96 eV (Fantini et al., 2010), 2.5 eV (Fan et al., 2004), 3.9 eV (Matsushita et al., 1989). Such dispersion can be dependent on the differences in the deposition methods, type of substrate, method of results interpretation, etc. According to the literature and to our measurements (Fig. 1), GeTe has the highest crystallization temperature for materials on the GeTe-Sb2Te3 pseudobinary line and must demonstrate the highest activation energy of crystallization.

Figure 3 shows X-ray patterns of GeTe film measured at the temperature indicated on the graph. At temperatures below 1800C, the material shows only wide bands, which are characteristic of amorphous materials. At higher temperatures, a rhombohedral GeTe phase appears. At a temperature of 3500C two phases have been observed: rhombohedral and

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 439

homogeneity of the system, random and uniform nucleation, and nucleation rate taking place at the very beginning of the transformation is time independent and applied for

Fig. 4. Dependencies of the volume fraction *x* versus time obtained for GeTe samples from

The value of activation energy for GeTe obtained using Equation (2) was very high: close to 12 eV. Such high value of activation energy has been obtained because GeTe, as other Ge:Sb:Te materials, shows long incubation times *τ* for crystallization, namely, the annealing time required to reach a critical nuclei size or to observe an abrupt increase in the crystalline volume fraction. In this case, the nucleation rate cannot be considered to be time independent for the entire crystallization process as is considered in classical JMAK models

In the case of GeTe, the volume fraction of crystalline phase approximately equals zero during all the incubation time (see Fig. 4). In such cases than the volume fraction of crystalline phase does not demonstrate a substantial increase during the incubation time it is possible to define the beginning of the transformation after the incubation time (Weidenhof

where *τ* is the incubation time. According to Equation (3) the plot *ln[-ln(1- x)]* versus *ln(t-τ)* must be a straight line with slope *n*. Figure 5 shows such modification of an Avrami plot for

The calculated activation energy for GeTe as determined from the modification of the Avrami Equation (3) using the Arrhenius type relation for *K* was 3.98±0.12 eV. The value of the Avrami exponent was about 1.6, which corresponds to diffusion controlled growth from

GeTe. The insert shows the rate constant as a function of the reciprocal temperature.

small dimension grains with decreasing nucleation rate (Christian, 1975).

, (3)

reflection measurements at the temperature indicated on the graph.

et al., 2001). The JMAK equation can now be expressed as:

( ) 1 exp[ ( ) ] *<sup>n</sup> xt Kt*

isotropic growth at a constant rate.

(Senkader et al., 2004).

NaCl-type, and at higher temperatures the material transforms into a NaCl-type phase. This phase exists only at temperatures above 4000C and in the process of cooling, again transforms into the stable rhombohedral phase.

Crystallization kinetics in GeTe, as in other materials, has been investigated using optical reflection during isothermal measurements. Reflection measurements were made with a 650 nm wavelength laser diode and a PIN diode as detector. In the reflectance measurements, the generally employed assumption has been used; the reflection signals are linearly related to the total transformed crystalline volume fractions *x* (Weidenhof et al., 2001):

$$\mathbf{x} = \left[ R(t) - Ra \right] / \left[ Rc - Ra \right] \,, \tag{1}$$

where R(t), the experimental measurement value of reflectance, Ra and Rc are reflectance of amorphous and fully crystalline phases, respectively.

Fig. 3. XRD patterns of GeTe film measured at the temperature indicated on the graph. Blue patterns correspond to rhombohedral phase. Red patterns correspond to NaCl-type phase.

Figure 4 shows the evolution of the total crystalline volume fraction *x*, for GeTe samples, calculated from reflection measurements for films isothermally annealed at the temperatures indicated on the graph.

In this chapter, crystallization kinetics will be analyzed using the Johnson–Mehl–Avrami-Kolmogorov (JMAK) model for isothermal annealing. According to the classical JMAK model, the transformed volume fraction *x* can be determined by the following expression:

$$\mathbf{x}(t) = 1 - \exp(-Kt^{\eta}) \, , \tag{2}$$

where *K=γexp(-E/kT), γ* is the frequency factor, *E* is the effective activation energy, *t* is the annealing time, and *n* the Avrami exponent, which provides information about the mechanisms of crystallization. The value of *n* can be evaluated from the slope of the *ln[-ln(1- x)]* versus *ln(t)* plot, which in materials with random nucleation and isotropic growth should be linear. The JMAK model is based on several assumptions such as:

NaCl-type, and at higher temperatures the material transforms into a NaCl-type phase. This phase exists only at temperatures above 4000C and in the process of cooling, again

Crystallization kinetics in GeTe, as in other materials, has been investigated using optical reflection during isothermal measurements. Reflection measurements were made with a 650 nm wavelength laser diode and a PIN diode as detector. In the reflectance measurements, the generally employed assumption has been used; the reflection signals are linearly related

where R(t), the experimental measurement value of reflectance, Ra and Rc are reflectance of

Fig. 3. XRD patterns of GeTe film measured at the temperature indicated on the graph. Blue patterns correspond to rhombohedral phase. Red patterns correspond to NaCl-type phase.

Figure 4 shows the evolution of the total crystalline volume fraction *x*, for GeTe samples, calculated from reflection measurements for films isothermally annealed at the

In this chapter, crystallization kinetics will be analyzed using the Johnson–Mehl–Avrami-Kolmogorov (JMAK) model for isothermal annealing. According to the classical JMAK model, the transformed volume fraction *x* can be determined by the following expression:

where *K=γexp(-E/kT), γ* is the frequency factor, *E* is the effective activation energy, *t* is the annealing time, and *n* the Avrami exponent, which provides information about the mechanisms of crystallization. The value of *n* can be evaluated from the slope of the *ln[-ln(1- x)]* versus *ln(t)* plot, which in materials with random nucleation and isotropic growth should be linear. The JMAK model is based on several assumptions such as:

*x R t Ra Rc Ra* [ ( ) ]/[ ] , (1)

( ) 1 exp( ) *<sup>n</sup> x t Kt* , (2)

to the total transformed crystalline volume fractions *x* (Weidenhof et al., 2001):

transforms into the stable rhombohedral phase.

amorphous and fully crystalline phases, respectively.

temperatures indicated on the graph.

homogeneity of the system, random and uniform nucleation, and nucleation rate taking place at the very beginning of the transformation is time independent and applied for isotropic growth at a constant rate.

Fig. 4. Dependencies of the volume fraction *x* versus time obtained for GeTe samples from reflection measurements at the temperature indicated on the graph.

The value of activation energy for GeTe obtained using Equation (2) was very high: close to 12 eV. Such high value of activation energy has been obtained because GeTe, as other Ge:Sb:Te materials, shows long incubation times *τ* for crystallization, namely, the annealing time required to reach a critical nuclei size or to observe an abrupt increase in the crystalline volume fraction. In this case, the nucleation rate cannot be considered to be time independent for the entire crystallization process as is considered in classical JMAK models (Senkader et al., 2004).

In the case of GeTe, the volume fraction of crystalline phase approximately equals zero during all the incubation time (see Fig. 4). In such cases than the volume fraction of crystalline phase does not demonstrate a substantial increase during the incubation time it is possible to define the beginning of the transformation after the incubation time (Weidenhof et al., 2001). The JMAK equation can now be expressed as:

$$\mathbf{x}(t) = \mathbf{1} - \exp\left[-\mathbf{K}(t-\tau)^{\eta}\right],\tag{3}$$

where *τ* is the incubation time. According to Equation (3) the plot *ln[-ln(1- x)]* versus *ln(t-τ)* must be a straight line with slope *n*. Figure 5 shows such modification of an Avrami plot for GeTe. The insert shows the rate constant as a function of the reciprocal temperature.

The calculated activation energy for GeTe as determined from the modification of the Avrami Equation (3) using the Arrhenius type relation for *K* was 3.98±0.12 eV. The value of the Avrami exponent was about 1.6, which corresponds to diffusion controlled growth from small dimension grains with decreasing nucleation rate (Christian, 1975).

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 441

As mentioned above, the structure of Ge4Sb1Te5 is not known. It lies on or is close to the GeTe-Sb2Te3 pseudobinary line in the Ge:Sb:Te phase diagram between GeTe and Ge2Sb2Te5. Because of such position, the onset of crystallization temperature is about 1600C, less than in GeTe but higher than in another Ge:Sb:Te ternary alloys. This alloy crystallizes in NaCl-type structure (Gonzalez-Hernandez et al., 1992, Wamwangi et al., 2002, Ruiz Santos et al., 2010). For Ge4Sb1Te5 films the reported values of the activation energy varied in a wide range: 1.13 eV (Kato et al., 1999), 3.09 eV (Morales-Sanchez et al.,

In addition, similarly to GeTe, bulk material and thin Ge4Sb1Te5 films demonstrate at a temperature of approximately 3270C ferroelectric-paraelectric transition (Bahgat et al., 2004, Ruiz Santos et al., 2010). But in contrast to GeTe, in which ferroelectric-paraelectric transition relates to a transformation from a rhombohedral phase to a rock salt type structure, in Ge4Sb1Te5 the ferroelectric-paraelectric transition relates to a transformation

Figure 7 shows DSC measurements on a crystalline Ge4Sb1Te5 bulk alloy and on the amorphous material obtained by removing the as-deposited films from the glass substrates. In the bulk alloys, a nonsymmetrical endothermic peak at a temperature of 3290C has been observed. In contrast, DSC measurements on amorphous material showed an exothermic peak at 1760C, which is associated with the crystallization of the sample and a broad endothermic peak, with a maximum at approximately 3270C, and that corresponds to the same temperature as in the bulk alloy. The observed endothermic peak can be associated

Fig. 7. DSC measurements on a bulk crystalline alloy and on the amorphous material,

obtained by removing the as-deposited films from the glass substrates.

**4. Crystallization of Ge4Sb1Te5** 

2005), 3.48 eV (Wamwangi et al., 2002).

between two different rock salt type structures.

with the ferroelectric-paraelectric transition.

In addition, X-ray measurements have shown that during isothermal annealing from amorphous phase only one rhombohedral phase appeared. As time increases, the volume fraction of amorphous phase decreases (Figure 6); in the same time, the volume fraction of crystalline phase increases. No shift in the position of diffraction peaks was observed. This observation will be important in the analysis of crystallization processes of Ge2Sb2Te5 and Ge1Sb2Te4 materials.

Fig. 5. Avrami plot of *ln[−ln(1−x)]* vs *ln(t−τ)* for GeTe. Points – experiment, lines – result of fitting using JMAK equation (3). The insert shows the rate constant *K* as function of the reciprocal temperature.

Fig. 6. X-ray diffraction spectra for a GeTe film obtained in the process of isothermal annealing at a temperature of 1640C during the time indicated on the plot. The upper pattern corresponds to film annealing at 2000C.

## **4. Crystallization of Ge4Sb1Te5**

440 Crystallization – Science and Technology

In addition, X-ray measurements have shown that during isothermal annealing from amorphous phase only one rhombohedral phase appeared. As time increases, the volume fraction of amorphous phase decreases (Figure 6); in the same time, the volume fraction of crystalline phase increases. No shift in the position of diffraction peaks was observed. This observation will be important in the analysis of crystallization processes of Ge2Sb2Te5 and

Fig. 5. Avrami plot of *ln[−ln(1−x)]* vs *ln(t−τ)* for GeTe. Points – experiment, lines – result of fitting using JMAK equation (3). The insert shows the rate constant *K* as function of the

Fig. 6. X-ray diffraction spectra for a GeTe film obtained in the process of isothermal annealing at a temperature of 1640C during the time indicated on the plot. The upper

pattern corresponds to film annealing at 2000C.

Ge1Sb2Te4 materials.

reciprocal temperature.

As mentioned above, the structure of Ge4Sb1Te5 is not known. It lies on or is close to the GeTe-Sb2Te3 pseudobinary line in the Ge:Sb:Te phase diagram between GeTe and Ge2Sb2Te5. Because of such position, the onset of crystallization temperature is about 1600C, less than in GeTe but higher than in another Ge:Sb:Te ternary alloys. This alloy crystallizes in NaCl-type structure (Gonzalez-Hernandez et al., 1992, Wamwangi et al., 2002, Ruiz Santos et al., 2010). For Ge4Sb1Te5 films the reported values of the activation energy varied in a wide range: 1.13 eV (Kato et al., 1999), 3.09 eV (Morales-Sanchez et al., 2005), 3.48 eV (Wamwangi et al., 2002).

In addition, similarly to GeTe, bulk material and thin Ge4Sb1Te5 films demonstrate at a temperature of approximately 3270C ferroelectric-paraelectric transition (Bahgat et al., 2004, Ruiz Santos et al., 2010). But in contrast to GeTe, in which ferroelectric-paraelectric transition relates to a transformation from a rhombohedral phase to a rock salt type structure, in Ge4Sb1Te5 the ferroelectric-paraelectric transition relates to a transformation between two different rock salt type structures.

Figure 7 shows DSC measurements on a crystalline Ge4Sb1Te5 bulk alloy and on the amorphous material obtained by removing the as-deposited films from the glass substrates. In the bulk alloys, a nonsymmetrical endothermic peak at a temperature of 3290C has been observed. In contrast, DSC measurements on amorphous material showed an exothermic peak at 1760C, which is associated with the crystallization of the sample and a broad endothermic peak, with a maximum at approximately 3270C, and that corresponds to the same temperature as in the bulk alloy. The observed endothermic peak can be associated with the ferroelectric-paraelectric transition.

Fig. 7. DSC measurements on a bulk crystalline alloy and on the amorphous material, obtained by removing the as-deposited films from the glass substrates.

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 443

relation for *K*, was 3.46±0.22 eV (insert in Fig. 10). This value is in good agreement with other values reported in the literature (3.48 eV±0.12 eV) obtained using the Kissinger analysis (Wamwangi et al., 2002). The Avrami exponent for Ge4Sb1Te5 was close to 1.8, which corresponds to a diffusion controlled growth from small dimension grains with

Additional X-ray measurements have shown that during isothermal annealing from

Fig. 9. Dependencies of the volume fraction *x* versus time obtained for Ge4Sb1Te5 samples

Fig. 10. Avrami plot of *ln[−ln(1−x)]* vs *ln(t−τ)* for Ge4Sb1Te5 films. Points – experiment, lines – results of fitting using JMAK equation (3). The insert shows the rate constant *K* as function

of the reciprocal temperature.

from reflection measurements at the temperature indicated on the graph.

decreasing nucleation rate, an exponent similar to that of GeTe (Christian 1975).

amorphous phase only one NaCl-type phase appeared.

Additional confirmation of ferroelectric-paraelectric transition in Ge4Sb1Te5 films has been obtained by temperature capacitance measurements (Ruiz Santos et al., 2010). In this work, it has been shown that the reciprocal capacitances (which are proportional to the reciprocal dielectric constant) as a function of temperature show a typical Curie–Weiss behavior for temperatures above 3270C.

Specific data of crystallization process and ferroelectric-paraelectric transition in Ge4Sb1Te5 films can be obtained by in situ XRD measurements at different temperatures. Figure 8 shows that below the crystallization temperature the film is in an amorphous state (pattern at 1400C). As the temperature is increased, the material crystallizes in the NaCl type structure. As the temperature continue increasing, the position of all Ge4Sb1Te5 peaks shift to lower 2θ values, reaching saturation at a temperature of approximately 3270C at positions close to the NaCl type of GeTe structure (ICSD cart #602124), which demonstrate paraelectric properties. The position of the NaCl type of GeTe is marked on the graph with the horizontal line. This shift can be clearly seen on the insert of Figure 8, which shows the temperature dependence of the (200) diffraction peak in Ge4Sb1Te5. The horizontal line on the insert indicates the position of the (200) peak in crystalline GeTe.

Figure 9 shows the evolution of the total crystalline volume fraction *x*, for Ge4Sb1Te5 samples, calculated from reflection measurements for films isothermally annealed at temperatures indicated on the graph. As in the case of GeTe, the volume fraction of crystalline phase of Ge4Sb1Te5 is approximately equal to zero during all the incubation time. The same as for GeTe, kinetics parameters for Ge4Sb1Te5 can be calculated using Eq. (3).

Fig. 8. XRD patterns of Ge4Sb1Te5 film measured at the temperature indicated on the graph. The insert shows temperature dependence of (200) diffraction line in Ge4Sb1Te5.

Figure 10 shows the modification of an Avrami plot for Ge4Sb1Te5. The insert shows the rate constant as a function of the reciprocal temperature. The effective activation energy, determined from the Avrami plot for different temperatures using the Arrhenius type

Additional confirmation of ferroelectric-paraelectric transition in Ge4Sb1Te5 films has been obtained by temperature capacitance measurements (Ruiz Santos et al., 2010). In this work, it has been shown that the reciprocal capacitances (which are proportional to the reciprocal dielectric constant) as a function of temperature show a typical Curie–Weiss behavior for

Specific data of crystallization process and ferroelectric-paraelectric transition in Ge4Sb1Te5 films can be obtained by in situ XRD measurements at different temperatures. Figure 8 shows that below the crystallization temperature the film is in an amorphous state (pattern at 1400C). As the temperature is increased, the material crystallizes in the NaCl type structure. As the temperature continue increasing, the position of all Ge4Sb1Te5 peaks shift to lower 2θ values, reaching saturation at a temperature of approximately 3270C at positions close to the NaCl type of GeTe structure (ICSD cart #602124), which demonstrate paraelectric properties. The position of the NaCl type of GeTe is marked on the graph with the horizontal line. This shift can be clearly seen on the insert of Figure 8, which shows the temperature dependence of the (200) diffraction peak in Ge4Sb1Te5. The horizontal line on

Figure 9 shows the evolution of the total crystalline volume fraction *x*, for Ge4Sb1Te5 samples, calculated from reflection measurements for films isothermally annealed at temperatures indicated on the graph. As in the case of GeTe, the volume fraction of crystalline phase of Ge4Sb1Te5 is approximately equal to zero during all the incubation time. The same as for GeTe, kinetics parameters for Ge4Sb1Te5 can be calculated using Eq. (3).

Fig. 8. XRD patterns of Ge4Sb1Te5 film measured at the temperature indicated on the graph.

Figure 10 shows the modification of an Avrami plot for Ge4Sb1Te5. The insert shows the rate constant as a function of the reciprocal temperature. The effective activation energy, determined from the Avrami plot for different temperatures using the Arrhenius type

The insert shows temperature dependence of (200) diffraction line in Ge4Sb1Te5.

the insert indicates the position of the (200) peak in crystalline GeTe.

temperatures above 3270C.

relation for *K*, was 3.46±0.22 eV (insert in Fig. 10). This value is in good agreement with other values reported in the literature (3.48 eV±0.12 eV) obtained using the Kissinger analysis (Wamwangi et al., 2002). The Avrami exponent for Ge4Sb1Te5 was close to 1.8, which corresponds to a diffusion controlled growth from small dimension grains with decreasing nucleation rate, an exponent similar to that of GeTe (Christian 1975).

Additional X-ray measurements have shown that during isothermal annealing from amorphous phase only one NaCl-type phase appeared.

Fig. 9. Dependencies of the volume fraction *x* versus time obtained for Ge4Sb1Te5 samples from reflection measurements at the temperature indicated on the graph.

Fig. 10. Avrami plot of *ln[−ln(1−x)]* vs *ln(t−τ)* for Ge4Sb1Te5 films. Points – experiment, lines – results of fitting using JMAK equation (3). The insert shows the rate constant *K* as function of the reciprocal temperature.

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 445

crystallite during the incubation time, as will be shown below, is the most important factor

Figure 11 shows the evolution of the total crystalline volume fraction *x*, for Ge2Sb2Te5 samples, calculated from reflection measurements for films isothermally annealed at the temperatures indicated on the graph. In contrast to GeTe and Ge4Sb1Te5, the volume fraction

Fig. 11. Dependencies of the volume fraction *x* versus time obtained for Ge2Sb2Te5 samples from reflection measurements at the temperature indicated on the graph. Black curves

A typical JMAK plot for Ge2Sb2Te5 films is shown in Figure 12. If we evaluate the slope only on a linear behavior, which is observed after incubation time τ, using a classical equation (2), the effective activation energy and the Avrami exponent will be equal to 5.61 eV and 3.3, respectively. The evaluation using a modification of a JMAK equation (3), which takes into account the incubation time, renders values for effective activation energy and an Avrami exponent equal to 2.07 eV and 1.5, respectively. This value of activation energy correlates well with what is reported in the literature (2.0 eV) and is obtained from the same interpretation of the reflection measurements (Weidenhof et al., 2001). Thus what comes into

As has been mentioned above, the JMAK model is based on the following assumptions: the system is homogeneous, nucleation is random and uniform, the nucleation rate is time independent and takes place at the very beginning of the transformation, the isotropic growth maintains a constant rate, etc. (Henderson, 1979). However, nucleation in Ge2Sb2Te5 films is not random nor uniform (Senkader et al., 2004) and due to partial crystallization during incubation time the nucleation rate cannot be considered to be time independent for the entire crystallization process as is considered in the classical model (equation 2). To overcome this limitation it has been proposed to avoid using a modification of the JMAK model (equation 3): after incubation time the rate constant can be considered as independent in time (Weidenhof et al., 2001). But in the case of Ge2Sb2Te5 a large amount of crystallized

results from fittings using equation (7).

question is the pertinence of using these two approximations.

responsible for the dispersion in the kinetic parameters reported in the literature.

of crystalline phase of Ge2Sb2Te5 does not equal zero during incubation time.

## **5. Crystallization of Ge2Sb2Te5**

Ge2Sb2Te5 thin films are the most commonly employed materials for phase-change memory technology application due to its high crystallization speed and relatively high crystallization temperature (but less than in GeTe and Ge4Sb1Te5), which lead to high thermal stability. Because of this, many extensive experimental and theoretical studies have been conducted to understand the structure and crystallization phenomena in this material (see for example (Yamada et al, 1991, Weidenhof et al., 2001, Matsunaga et al., 2004, Matsunaga et al., 2006, Paesler et al., 2007, Im et al., 2008, Claudio et al 2006)).

As mentioned above, Ge2Sb2Te5 material demonstrates two phase changes: first at a temperature close to 1450C there is an amorphous-NaCl-type transition and second (at about 2400C) a NaCl-type-hexagonal transformation.

Under heating, the amorphous Ge2Sb2Te5 films crystallize at around 130-1700C, depending on the preparation method and heating rate, into a phase with a NaCl-type structure (*Fm* 3 *m*). In this structure, the 4(a) site is fully occupied by Te atoms, whereas the 4(b) site is randomly occupied by Ge and Sb atoms and vacancies. The composition of Ge:Sb:Te ternary system, which lie on the GeTe-Sb2Te3 pseudobinary line, can be described as (GeTe)*x+*  (Sb2Te3)1-*<sup>x</sup>* (0<*x*<1). In such case the site occupancy of the vacancy varies continuously according to *(1-x)/(3-2x)* (Matsunaga et al., 2004, Matsunaga et al., 2006). But Ge and Sb atoms deviate from the ideal rock-salt positions in Ge2Sb2Te5 not in a random way but in a strongly correlated manner with respect to the neighboring Te atoms (Kolobov et al., 2004, Kolobov et al., 2006). The off-center location of Ge atoms means that there is a net dipole moment and suggests that a NaCl-type phase of the Ge2Sb2Te5 is a ferroelectric material (Tominaga et al., 2004). The ferroelectric properties in Ge2Sb2Te5 NaCl-type phase have been observed using capacitance-temperature measurements. The temperature dependence of the capacitance shows an abrupt change with a maximum at the temperature that corresponds to the end from a NaCl-type to a hexagonal transition. In addition, the reciprocal capacitance for temperatures above this transition shows the Curie–Weiss dependence, which is typical of ferroelectric materials (Gervacio Arciniega et al., 2010).

In spite of the large number of publications about crystallization phenomena in Ge2Sb2Te5, the activation energy of crystallization reported in the literature shows a large discrepancy in the range between 0.8 and 2.9 eV and between 1.2 and 4.4 for the Avrami exponent (Morales-Sanchez et al., 2010, Liu et al., 2009). Such parameter dispersion can be related not only to the difference in the preparation methods but to the specific crystallization process.

Experimental data in Ge2Sb2Te5 films, as in all materials along the pseudobinary GeTe-Sb2Te3 line, show relatively long incubation times during isothermal crystallization (Weidenhof et al., 2001, Laine et al., 2004, Zhang et al., 2008, Morales-Sanchez et al., 2010). But in GeTe and Ge4Sb1Te5 the amount of crystallized material is close to zero during the incubation. In contrast, in Ge2Sb2Te5 films large amount of crystallized material has been observed during this period of time (ranging from 50% in Ref. (Zhang et al., 2008), 10% in Ref. (Sian et al., 2008), 9% in Ref. (Laine et al., 2004) and 8% in Ref. (Weidenhof et al., 2001)). It is necessary to note that the amount of crystalline phase material during incubation time depends on the annealing temperature. The existence of a non-negligible amount of

Ge2Sb2Te5 thin films are the most commonly employed materials for phase-change memory technology application due to its high crystallization speed and relatively high crystallization temperature (but less than in GeTe and Ge4Sb1Te5), which lead to high thermal stability. Because of this, many extensive experimental and theoretical studies have been conducted to understand the structure and crystallization phenomena in this material (see for example (Yamada et al, 1991, Weidenhof et al., 2001, Matsunaga et al., 2004,

As mentioned above, Ge2Sb2Te5 material demonstrates two phase changes: first at a temperature close to 1450C there is an amorphous-NaCl-type transition and second (at about

Under heating, the amorphous Ge2Sb2Te5 films crystallize at around 130-1700C, depending on the preparation method and heating rate, into a phase with a NaCl-type structure (*Fm* 3 *m*). In this structure, the 4(a) site is fully occupied by Te atoms, whereas the 4(b) site is randomly occupied by Ge and Sb atoms and vacancies. The composition of Ge:Sb:Te ternary system, which lie on the GeTe-Sb2Te3 pseudobinary line, can be described as (GeTe)*x+*  (Sb2Te3)1-*<sup>x</sup>* (0<*x*<1). In such case the site occupancy of the vacancy varies continuously according to *(1-x)/(3-2x)* (Matsunaga et al., 2004, Matsunaga et al., 2006). But Ge and Sb atoms deviate from the ideal rock-salt positions in Ge2Sb2Te5 not in a random way but in a strongly correlated manner with respect to the neighboring Te atoms (Kolobov et al., 2004, Kolobov et al., 2006). The off-center location of Ge atoms means that there is a net dipole moment and suggests that a NaCl-type phase of the Ge2Sb2Te5 is a ferroelectric material (Tominaga et al., 2004). The ferroelectric properties in Ge2Sb2Te5 NaCl-type phase have been observed using capacitance-temperature measurements. The temperature dependence of the capacitance shows an abrupt change with a maximum at the temperature that corresponds to the end from a NaCl-type to a hexagonal transition. In addition, the reciprocal capacitance for temperatures above this transition shows the Curie–Weiss dependence,

Matsunaga et al., 2006, Paesler et al., 2007, Im et al., 2008, Claudio et al 2006)).

which is typical of ferroelectric materials (Gervacio Arciniega et al., 2010).

In spite of the large number of publications about crystallization phenomena in Ge2Sb2Te5, the activation energy of crystallization reported in the literature shows a large discrepancy in the range between 0.8 and 2.9 eV and between 1.2 and 4.4 for the Avrami exponent (Morales-Sanchez et al., 2010, Liu et al., 2009). Such parameter dispersion can be related not only to the difference in the preparation methods but to the specific

Experimental data in Ge2Sb2Te5 films, as in all materials along the pseudobinary GeTe-Sb2Te3 line, show relatively long incubation times during isothermal crystallization (Weidenhof et al., 2001, Laine et al., 2004, Zhang et al., 2008, Morales-Sanchez et al., 2010). But in GeTe and Ge4Sb1Te5 the amount of crystallized material is close to zero during the incubation. In contrast, in Ge2Sb2Te5 films large amount of crystallized material has been observed during this period of time (ranging from 50% in Ref. (Zhang et al., 2008), 10% in Ref. (Sian et al., 2008), 9% in Ref. (Laine et al., 2004) and 8% in Ref. (Weidenhof et al., 2001)). It is necessary to note that the amount of crystalline phase material during incubation time depends on the annealing temperature. The existence of a non-negligible amount of

**5. Crystallization of Ge2Sb2Te5** 

2400C) a NaCl-type-hexagonal transformation.

crystallization process.

crystallite during the incubation time, as will be shown below, is the most important factor responsible for the dispersion in the kinetic parameters reported in the literature.

Figure 11 shows the evolution of the total crystalline volume fraction *x*, for Ge2Sb2Te5 samples, calculated from reflection measurements for films isothermally annealed at the temperatures indicated on the graph. In contrast to GeTe and Ge4Sb1Te5, the volume fraction of crystalline phase of Ge2Sb2Te5 does not equal zero during incubation time.

Fig. 11. Dependencies of the volume fraction *x* versus time obtained for Ge2Sb2Te5 samples from reflection measurements at the temperature indicated on the graph. Black curves results from fittings using equation (7).

A typical JMAK plot for Ge2Sb2Te5 films is shown in Figure 12. If we evaluate the slope only on a linear behavior, which is observed after incubation time τ, using a classical equation (2), the effective activation energy and the Avrami exponent will be equal to 5.61 eV and 3.3, respectively. The evaluation using a modification of a JMAK equation (3), which takes into account the incubation time, renders values for effective activation energy and an Avrami exponent equal to 2.07 eV and 1.5, respectively. This value of activation energy correlates well with what is reported in the literature (2.0 eV) and is obtained from the same interpretation of the reflection measurements (Weidenhof et al., 2001). Thus what comes into question is the pertinence of using these two approximations.

As has been mentioned above, the JMAK model is based on the following assumptions: the system is homogeneous, nucleation is random and uniform, the nucleation rate is time independent and takes place at the very beginning of the transformation, the isotropic growth maintains a constant rate, etc. (Henderson, 1979). However, nucleation in Ge2Sb2Te5 films is not random nor uniform (Senkader et al., 2004) and due to partial crystallization during incubation time the nucleation rate cannot be considered to be time independent for the entire crystallization process as is considered in the classical model (equation 2). To overcome this limitation it has been proposed to avoid using a modification of the JMAK model (equation 3): after incubation time the rate constant can be considered as independent in time (Weidenhof et al., 2001). But in the case of Ge2Sb2Te5 a large amount of crystallized

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 447

corresponding to the NaCl-type phase of Ge2Sb2Te5 composition. In this figure the blue pattern corresponds to material annealed at 1800C; the upper pattern corresponds to Ge1Sb4Te7 film annealed at 1800C. Thus, in Ge2Sb2Te5 films the formation of a stable rock salt

On the basis of these results, the crystallization of Ge2Sb2Te5 material during isothermal annealing can be considered as a process which takes place in two stages: in the first stage nuclei of a metastable Ge1Sb4Te7 phase appear; in the second stage, the nuclei transform into

In materials in which phase transformation starts with the formation of metastable phases the isothermal crystallization process cannot be simply described by JMAK theory and the plots of *ln(-ln(1 - x))* versus *ln(t)* are not linear, which implies that the Avrami exponent *n*

In ref. (Claudio et al., 2006) an analytical model that can describe the isothermal crystallization process of materials is proposed. In this process, the stable crystalline phase is preceded by the formation of a metastable phase. The model assumes that the volume fraction of metastable phase *fm* grows up to a maximum value *fmax* and then stops growing when the stable phase that has nucleated into it overpasses the metastable grain boundaries. The kinetics of the metastable fraction can be represented by a JMAK-type equation

The kinetic behavior of the stable phase into a metastable phase can be represented by the modification of classical JMAK formula, which takes into account the incubation time *tsm* of

> ( ) (1 exp( ( ) )) for t t and ( ) 0 for t t

where *Ksm* and *nsm* are the crystallization rate constant and the Avrami exponent, respectively, which represent the kinetic parameters of the transformation from stable phase

Furthermore, it is possible to propose that the stable phase can grow into the amorphous phase. The stable phase transformation kinetics in amorphous phase can be described by

> ( ) (1 )(1 exp( ( ) )) for t t and ( ) 0 for t t

where *fsa* is the fraction of the stable phase in amorphous phase (from 0 to *1-fmax*), *Ksa* and *nsa* are the JMAK parameters and *tsa* is the incubation time of stable phase in amorphous phase.

*sa*

*sa sa sa*

*ft f Ktt*

The total fraction of transformed material is given by the equation:

max sm

max sa

max ( ) (1 exp( )) ( ) *nm fm <sup>t</sup> <sup>f</sup> K t m smf <sup>t</sup>* , (4)

sm

sa

*nsa*

*f t* (5)

*f t* (6)

*nsm*

subtracting the fraction of stable phase *fsm* which grows inside the metastable phase:

where *Km* is the crystallization rate constant and *nm* the Avrami exponent.

*sm sm sm*

*f t f K tt*

*sm*

type crystalline phase is preceded by the formation of a metastable Ge1Sb4Te7 phase.

the equilibrium NaCl-type stoichiometric Ge2Sb2Te5 structures.

does not remain constant during the crystallization process.

the stable phase into the metastable phase:

JMAK-type equation with its specific parameters:

into metastable phase.

material has been observed during incubation time. Moreover, experimental investigation has shown that nuclei, which appeared in Ge2Sb2Te5 films during incubation time, have a composition corresponding to Ge1Sb4Te7 which differs from the nominal value for the amorphous matrix (Laine et al., 2004, Claudio et al., 2006). Figure 13 shows X-ray diffraction patterns for Ge2Sb2Te5 film isothermal annealing at the temperature 1180C and times indicated on the graph. The patterns of samples annealed during 50, 100, and 200 minutes show weak peaks, with positions corresponding to Ge1Sb4Te7 NaCl-type phase. After annealing for more than 200 min, the positions of the peaks begin to change to those

Fig. 12. Avrami plot of *ln[−ln(1−x)]* vs *ln(t)* for Ge2Sb2Te5 films. Points – experiment, black line – results of fitting using equation (7).

Fig. 13. X-ray diffraction spectra for a Ge2Sb2Te5 film obtained in the process of isothermal annealing at a temperature of 1180C during the time indicated in the plot. Upper patterns correspond to Ge1Sb4Te7 (red curve) and Ge2Sb2Te5 (blue curve) taken at 1800C. The positions of the Ge1Sb4Te7 peaks are indicated with the vertical lines.

material has been observed during incubation time. Moreover, experimental investigation has shown that nuclei, which appeared in Ge2Sb2Te5 films during incubation time, have a composition corresponding to Ge1Sb4Te7 which differs from the nominal value for the amorphous matrix (Laine et al., 2004, Claudio et al., 2006). Figure 13 shows X-ray diffraction patterns for Ge2Sb2Te5 film isothermal annealing at the temperature 1180C and times indicated on the graph. The patterns of samples annealed during 50, 100, and 200 minutes show weak peaks, with positions corresponding to Ge1Sb4Te7 NaCl-type phase. After annealing for more than 200 min, the positions of the peaks begin to change to those

Fig. 12. Avrami plot of *ln[−ln(1−x)]* vs *ln(t)* for Ge2Sb2Te5 films. Points – experiment, black

Fig. 13. X-ray diffraction spectra for a Ge2Sb2Te5 film obtained in the process of isothermal annealing at a temperature of 1180C during the time indicated in the plot. Upper patterns correspond to Ge1Sb4Te7 (red curve) and Ge2Sb2Te5 (blue curve) taken at 1800C. The

positions of the Ge1Sb4Te7 peaks are indicated with the vertical lines.

line – results of fitting using equation (7).

corresponding to the NaCl-type phase of Ge2Sb2Te5 composition. In this figure the blue pattern corresponds to material annealed at 1800C; the upper pattern corresponds to Ge1Sb4Te7 film annealed at 1800C. Thus, in Ge2Sb2Te5 films the formation of a stable rock salt type crystalline phase is preceded by the formation of a metastable Ge1Sb4Te7 phase.

On the basis of these results, the crystallization of Ge2Sb2Te5 material during isothermal annealing can be considered as a process which takes place in two stages: in the first stage nuclei of a metastable Ge1Sb4Te7 phase appear; in the second stage, the nuclei transform into the equilibrium NaCl-type stoichiometric Ge2Sb2Te5 structures.

In materials in which phase transformation starts with the formation of metastable phases the isothermal crystallization process cannot be simply described by JMAK theory and the plots of *ln(-ln(1 - x))* versus *ln(t)* are not linear, which implies that the Avrami exponent *n* does not remain constant during the crystallization process.

In ref. (Claudio et al., 2006) an analytical model that can describe the isothermal crystallization process of materials is proposed. In this process, the stable crystalline phase is preceded by the formation of a metastable phase. The model assumes that the volume fraction of metastable phase *fm* grows up to a maximum value *fmax* and then stops growing when the stable phase that has nucleated into it overpasses the metastable grain boundaries. The kinetics of the metastable fraction can be represented by a JMAK-type equation subtracting the fraction of stable phase *fsm* which grows inside the metastable phase:

$$f\_m(t) = f\_{\max}(1 - \exp(-K\_m t^{\eta\_m})) - f\_{\sup}(t) \,\,\,\,\,\tag{4}$$

where *Km* is the crystallization rate constant and *nm* the Avrami exponent.

The kinetic behavior of the stable phase into a metastable phase can be represented by the modification of classical JMAK formula, which takes into account the incubation time *tsm* of the stable phase into the metastable phase:

$$\begin{aligned} f\_{sm}(t) &= f\_{\text{max}} \left( 1 - \exp(-K\_{\text{sm}}(t - t\_{\text{sm}})^{n\_{\text{sm}}}) \right) \quad \text{for } t \ge t\_{\text{sm}}\\ \text{and} \quad f\_{sm}(t) &= 0 \text{ for } t < t\_{\text{sm}} \end{aligned} \tag{5}$$

where *Ksm* and *nsm* are the crystallization rate constant and the Avrami exponent, respectively, which represent the kinetic parameters of the transformation from stable phase into metastable phase.

Furthermore, it is possible to propose that the stable phase can grow into the amorphous phase. The stable phase transformation kinetics in amorphous phase can be described by JMAK-type equation with its specific parameters:

$$f\_{sa}(t) = (1 - f\_{\text{max}}) (1 - \exp(-K\_{sa}(t - t\_{sa})^{\eta\_{sa}})) \text{ for } t \ge t\_{sa} \\ \tag{6}$$
 
$$\text{and} \quad f\_{sa}(t) = 0 \text{ for } t < t\_{sa}$$

where *fsa* is the fraction of the stable phase in amorphous phase (from 0 to *1-fmax*), *Ksa* and *nsa* are the JMAK parameters and *tsa* is the incubation time of stable phase in amorphous phase.

The total fraction of transformed material is given by the equation:

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 449

The results obtained have shown that in Ge1Sb2Te4, the same as in Ge2Sb2Te5 film, during the isothermal process of amorphous to crystalline phase transformation, a metastable phase appeared with composition of Ge1Sb4Te7. Thus, for the interpretation of the crystallization process in Ge1Sb2Te4 an analytical model can be used in which the formation during

Fig. 14. Dependencies of the volume fraction *x* versus time obtained for Ge1Sb2Te4 samples from reflection measurements at the temperature indicated on the graph. Black curves –

Fig. 15. Avrami plot of *ln[−ln(1−x)]* vs *ln(t*) for Ge1Sb2Te4 films. Points – experiment, black

crystallization of metastable phase is taken into account (Equation 7).

results from fittings using equation (7).

curve – results from fitting using equation (7).

$$f\_{\text{total}}(t) = f\_m(t) + f\_{sm}(t) + f\_{sa}(t) \tag{7}$$

This model allows fitting the experimental transformation data obtained in Ge2Sb2Te5 films using a genetic algorithm. The continuous lines on Fig. 11 show the evolution on time of the volume fractions of metastable Ge1Sb4Te7 and stable Ge2Sb2Te5 phase obtained from fitting using Equation (7). Results of simulations have shown that the volume fraction of metastable Ge1Sb4Te7 phase grows up to a maximum value and then decreases and disappears. After some incubation time, the Ge2Sb2Te5 phase began to grow and all material is transformed into a stable crystalline phase.

In addition, the model is capable of predicting the three slopes clearly shown in the JMAK plot (Fig. 12) corresponding to three distinguishable stages in the crystallization process observed at dependencies obtained at low annealing temperature: the first one related to the metastable transformation, the second one (low value) with the step between metastable and stable transformation, and the last one with the stable phase nucleation and growth. It is quite evident from these simulated results that the first slope of the JMAK plot can be related to the kinetic behavior of the metastable phase.

The results of the investigation have shown that during the isothermal process of amorphous to crystalline phase transformation in Ge2Sb2Te5 film a metastable phase with composition Ge1Sb4Te7 and a stable NaCl-type Ge2Sb2Te5 phase coexist within a certain time range. The appearance of nuclei of the Ge1Sb4Te7 composition could be related to the local fluctuation in the film composition and to the fact that the crystallization temperature of Ge1Sb4Te7 is lower than in Ge2Sb2Te5 material.

## **6. Crystallization of Ge1Sb2Te4**

Ge1Sb2Te4, as with Ge2Sb2Te5 film, during annealing crystallize at a temperature of approximately 1330C (lower than the crystallization temperature of Ge2Sb2Te5) into a NaCltype structure (*Fm* 3 *m*) (Matsunaga et al., 2004) and at a higher temperature (about 2350C) into a hexagonal phase. Because of the lower crystallization temperature, if compared with Ge2Sb2Te5, this material may possess a low programming current in random access memory.

Crystallization kinetics, the same as in Ge2Sb2Te5 material, is sufficiently complicated. Figure 14 shows the dependence of the total crystalline volume fraction *x* versus time for Ge1Sb2Te4 samples calculated from reflection measurements for films isothermally annealed at the temperatures indicated on the graph and Figure 15 shows a typical JMAK plot. If we evaluate the slope only on a linear behavior, which is observed after incubation time τ, using classical equation (2), the effective activation energy will be equal to 5.45 eV. The evaluation using modification of JMAK equation (3), which takes into account incubation time, renders values of effective activation energy equal to 1.77 eV. But in Ge1Sb2Te4, the same as in Ge2Sb2Te5 films, a large amount of crystallized material has been observed during the incubation time. Moreover, XRD measurements (Figure 16) have shown that nuclei, which appeared in Ge1Sb2Te4 films during the incubation period, have a composition corresponding to the Ge1Sb4Te7 structure. The upper XRD pattern in this Figure corresponds to a fully crystallized film with the Ge1Sb4Te7 composition; the peak positions of this phase are marked with the vertical lines. The pattern mark as Ge1Sb2Te4 shows diffraction lines of Ge1Sb2Te4 film annealing to 1800C.

This model allows fitting the experimental transformation data obtained in Ge2Sb2Te5 films using a genetic algorithm. The continuous lines on Fig. 11 show the evolution on time of the volume fractions of metastable Ge1Sb4Te7 and stable Ge2Sb2Te5 phase obtained from fitting using Equation (7). Results of simulations have shown that the volume fraction of metastable Ge1Sb4Te7 phase grows up to a maximum value and then decreases and disappears. After some incubation time, the Ge2Sb2Te5 phase began to grow and all material

In addition, the model is capable of predicting the three slopes clearly shown in the JMAK plot (Fig. 12) corresponding to three distinguishable stages in the crystallization process observed at dependencies obtained at low annealing temperature: the first one related to the metastable transformation, the second one (low value) with the step between metastable and stable transformation, and the last one with the stable phase nucleation and growth. It is quite evident from these simulated results that the first slope of the JMAK plot can be

The results of the investigation have shown that during the isothermal process of amorphous to crystalline phase transformation in Ge2Sb2Te5 film a metastable phase with composition Ge1Sb4Te7 and a stable NaCl-type Ge2Sb2Te5 phase coexist within a certain time range. The appearance of nuclei of the Ge1Sb4Te7 composition could be related to the local fluctuation in the film composition and to the fact that the crystallization temperature of

Ge1Sb2Te4, as with Ge2Sb2Te5 film, during annealing crystallize at a temperature of approximately 1330C (lower than the crystallization temperature of Ge2Sb2Te5) into a NaCltype structure (*Fm* 3 *m*) (Matsunaga et al., 2004) and at a higher temperature (about 2350C) into a hexagonal phase. Because of the lower crystallization temperature, if compared with Ge2Sb2Te5, this material may possess a low programming current in random access memory. Crystallization kinetics, the same as in Ge2Sb2Te5 material, is sufficiently complicated. Figure 14 shows the dependence of the total crystalline volume fraction *x* versus time for Ge1Sb2Te4 samples calculated from reflection measurements for films isothermally annealed at the temperatures indicated on the graph and Figure 15 shows a typical JMAK plot. If we evaluate the slope only on a linear behavior, which is observed after incubation time τ, using classical equation (2), the effective activation energy will be equal to 5.45 eV. The evaluation using modification of JMAK equation (3), which takes into account incubation time, renders values of effective activation energy equal to 1.77 eV. But in Ge1Sb2Te4, the same as in Ge2Sb2Te5 films, a large amount of crystallized material has been observed during the incubation time. Moreover, XRD measurements (Figure 16) have shown that nuclei, which appeared in Ge1Sb2Te4 films during the incubation period, have a composition corresponding to the Ge1Sb4Te7 structure. The upper XRD pattern in this Figure corresponds to a fully crystallized film with the Ge1Sb4Te7 composition; the peak positions of this phase are marked with the vertical lines. The pattern mark as Ge1Sb2Te4 shows diffraction lines of

is transformed into a stable crystalline phase.

related to the kinetic behavior of the metastable phase.

Ge1Sb4Te7 is lower than in Ge2Sb2Te5 material.

**6. Crystallization of Ge1Sb2Te4** 

Ge1Sb2Te4 film annealing to 1800C.

() () () () *total m sm sa f t ft f t ft* (7)

The results obtained have shown that in Ge1Sb2Te4, the same as in Ge2Sb2Te5 film, during the isothermal process of amorphous to crystalline phase transformation, a metastable phase appeared with composition of Ge1Sb4Te7. Thus, for the interpretation of the crystallization process in Ge1Sb2Te4 an analytical model can be used in which the formation during crystallization of metastable phase is taken into account (Equation 7).

Fig. 14. Dependencies of the volume fraction *x* versus time obtained for Ge1Sb2Te4 samples from reflection measurements at the temperature indicated on the graph. Black curves – results from fittings using equation (7).

Fig. 15. Avrami plot of *ln[−ln(1−x)]* vs *ln(t*) for Ge1Sb2Te4 films. Points – experiment, black curve – results from fitting using equation (7).

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 451

case, as is for GeTe and Ge4Sb1Te5, it is also possible to neglect the amount of crystallized material during τ, and describe the transformation using the modified JMAK equation (3). Figure 19 shows a plot of *ln[-ln(1-x)]* versus *ln(t-τ)* for a Ge1Sb4Te7 film, which demonstrates a linear dependence. This means that the crystallization process in Ge1Sb4Te7 material displays random nucleation and isotropic growth with effective activation energy of 1.7±0.27 eV and an Avrami exponent *n* close to 1.94. According to (Christian, 1975), the values of *n* in the range 1.5 < *n* < 2.5 correspond to a crystallization process dominated by all

Fig. 17. X-ray diffraction spectra for a Ge1Sb4Te7 film obtained in the process of isothermal

Fig. 18. Dependencies of the volume fraction *x* versus time obtained for Ge1Sb4Te7 samples

from reflection measurements at the temperature indicated on the graph.

annealing at a temperature of 900C during the time indicated on the plot.

shapes growing from small dimensions with decreasing nucleation rate.

Continuous black curves on Fig. 14 show the calculated the volume fractions of metastable Ge1Sb4Te7 and stable Ge1Sb2Te4 phase obtained from fittings using Equation (7). Results of simulation have shown that the same as in Ge2Sb2Te5, the volume fraction of metastable Ge1Sb4Te7 phase grows up to a maximum value and then decreases and disappears. After some incubation time, Ge1Sb2Te4 phase began to growth and all material was transformed into a stable crystalline phase.

Additionally, the model is capable to simulate the JMAK plot of Fig. 15, in which the first slope of the JMAK plot can be related to the kinetic behavior of the metastable Ge1Sb4Te7 phase.

Fig. 16. X-ray diffraction spectra for a Ge1Sb2Te4 film obtained in the process of isothermal annealing at a temperature of 1200C during the time indicated on the plot. Upper patterns correspond to Ge1Sb4Te7 (red curve) and Ge1Sb2Te4 (blue curve) taken at 1800C. The position of the Ge1Sb4Te7 peaks is indicated with the vertical lines.

## **7. Crystallization of Ge1Sb4Te7**

Similarly, as with Ge2Sb2Te5 and Ge1Sb2Te4, Ge1Sb4Te7 demonstrates a two phase transition: amorphous to the rock salt-like structure at a temperature of approximately 1100C and a second transition to the hexagonal phase at the temperature close to 1680C. This material demonstrates the fastest phase transformation among Ge:Sb:Te ternary alloys and phasereversible transformations can be observed in the femtoseconds range (Huang et al., 2006). But because of the low crystallization temperature, Ge1Sb4Te7 has lower thermal stability when compared with other ternary alloys (Miao et al., 2006).

Crystallization transformation from amorphous to NaCl-type structure in Ge1Sb4Te7 differs from the transformations observed in Ge2Sb2Te5 and Ge1Sb2Te4 films. In contrast with Ge2Sb2Te5 and Ge1Sb2Te4, in-situ X-ray diffraction measurements show that amorphous films with the Ge1Sb4Te7 composition crystallize in the NaCl-type phase with the same composition as at the beginning of the transformation (Figure 17). Moreover, isothermal reflection measurements (Figure 18) have shown practically zero value of crystalline phase during the much shorter incubation time compared with Ge2Sb2Te5 films (Figure 11). In this

Continuous black curves on Fig. 14 show the calculated the volume fractions of metastable Ge1Sb4Te7 and stable Ge1Sb2Te4 phase obtained from fittings using Equation (7). Results of simulation have shown that the same as in Ge2Sb2Te5, the volume fraction of metastable Ge1Sb4Te7 phase grows up to a maximum value and then decreases and disappears. After some incubation time, Ge1Sb2Te4 phase began to growth and all material was transformed

Additionally, the model is capable to simulate the JMAK plot of Fig. 15, in which the first slope of the JMAK plot can be related to the kinetic behavior of the metastable Ge1Sb4Te7

Fig. 16. X-ray diffraction spectra for a Ge1Sb2Te4 film obtained in the process of isothermal annealing at a temperature of 1200C during the time indicated on the plot. Upper patterns correspond to Ge1Sb4Te7 (red curve) and Ge1Sb2Te4 (blue curve) taken at 1800C. The position

Similarly, as with Ge2Sb2Te5 and Ge1Sb2Te4, Ge1Sb4Te7 demonstrates a two phase transition: amorphous to the rock salt-like structure at a temperature of approximately 1100C and a second transition to the hexagonal phase at the temperature close to 1680C. This material demonstrates the fastest phase transformation among Ge:Sb:Te ternary alloys and phasereversible transformations can be observed in the femtoseconds range (Huang et al., 2006). But because of the low crystallization temperature, Ge1Sb4Te7 has lower thermal stability

Crystallization transformation from amorphous to NaCl-type structure in Ge1Sb4Te7 differs from the transformations observed in Ge2Sb2Te5 and Ge1Sb2Te4 films. In contrast with Ge2Sb2Te5 and Ge1Sb2Te4, in-situ X-ray diffraction measurements show that amorphous films with the Ge1Sb4Te7 composition crystallize in the NaCl-type phase with the same composition as at the beginning of the transformation (Figure 17). Moreover, isothermal reflection measurements (Figure 18) have shown practically zero value of crystalline phase during the much shorter incubation time compared with Ge2Sb2Te5 films (Figure 11). In this

of the Ge1Sb4Te7 peaks is indicated with the vertical lines.

when compared with other ternary alloys (Miao et al., 2006).

**7. Crystallization of Ge1Sb4Te7** 

into a stable crystalline phase.

phase.

case, as is for GeTe and Ge4Sb1Te5, it is also possible to neglect the amount of crystallized material during τ, and describe the transformation using the modified JMAK equation (3).

Figure 19 shows a plot of *ln[-ln(1-x)]* versus *ln(t-τ)* for a Ge1Sb4Te7 film, which demonstrates a linear dependence. This means that the crystallization process in Ge1Sb4Te7 material displays random nucleation and isotropic growth with effective activation energy of 1.7±0.27 eV and an Avrami exponent *n* close to 1.94. According to (Christian, 1975), the values of *n* in the range 1.5 < *n* < 2.5 correspond to a crystallization process dominated by all shapes growing from small dimensions with decreasing nucleation rate.

Fig. 17. X-ray diffraction spectra for a Ge1Sb4Te7 film obtained in the process of isothermal annealing at a temperature of 900C during the time indicated on the plot.

Fig. 18. Dependencies of the volume fraction *x* versus time obtained for Ge1Sb4Te7 samples from reflection measurements at the temperature indicated on the graph.

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 453

the Kissinger analysis in the films deposited on the glass substrates by the same DC

Fig. 20. Dependencies of the volume fraction *x* versus time obtained for Sb2Te3 samples from

Fig. 21. Avrami plot of *ln[−ln(1−x)]* vs *ln(t−τ)* for Sb2Te3 films. Points – experiment, lines – results from fittings using JMAK equation (3). Insert shows the rate constant *K* as function of

This chapter presents some of the results concerning the crystallization properties of the most commonly employed materials on the GeTe–Sb2Te3 pseudo-binary line for phase

reflection measurements at the temperatures indicated on the graph.

magnetron sputtering (Zhai et al., 2009).

the reciprocal temperature.

**9. Conclusions** 

## **8. Crystallization of Sb2Te3**

Sb2Te3 lies at the end of the GeTe-Sb2Te3 pseudobinary line and has the lowest crystallization temperature but the highest crystallization speed. This material is an attractive candidate for phase change random access memory due to its rapid crystallization speed, but the crystallization temperature of Sb2Te3 (about 94°C in material investigated in this chapter and between 90-100°C reported in the literature (Yin et al., 2007, Kim et al., 2010)) is too low for it to be of practical use.

Upon annealing, amorphous Sb2Te3 films crystallize in rhombohedral Sb2Te3 (Kim et al., 2008, Lv et al., 2010) or fcc structure, which at a temperature above 2000C transforms into an hexagonal phase (Yin et al., 2007, Zhu et al., 2011). In the investigated Sb2Te3 films within the temperature range comprised between 95-2000C, only the rhombohedral phase (JCPDS data #15-0874) has been observed.

Fig. 19. Avrami plot of *ln[−ln(1−x)]* vs *ln(t−τ)* for Ge1Sb4Te7 films. Points – experiment, lines – results from fittings using JMAK equation (3). Insert shows the rate constant *K* as a function of the reciprocal temperature.

Isothermal reflection measurements (Figure 20) have shown the same as for Ge1Sb4Te7: practically zero value of crystalline phase during the incubation time. In this case, the same as for GeTe, Ge4Sb1Te5 and Ge1Sb4Te7 it is possible to neglect the amount of crystallized material during τ, and describe the transformation using the modified JMAK equation (3).

Figure 21 shows a plot of *ln[-ln(1-x)]* versus *ln(t-τ)* for a Sb2Te3 film, which demonstrates a linear dependence. This means that the crystallization process in Sb2Te3 films, the same as in Ge1Sb4Te7 material, display random nucleation and isotropic growth with an effective activation energy of 1.54±0.15 eV and an Avrami exponent *n* close to 1.1, which according to ref. (Christian, 1975), corresponds to a crystallization growth of particles of appreciable initial volume. It is necessary to note that the values obtained for crystallization activation energy is in good agreement with values reported in the literature (1.51 eV) obtained using

Sb2Te3 lies at the end of the GeTe-Sb2Te3 pseudobinary line and has the lowest crystallization temperature but the highest crystallization speed. This material is an attractive candidate for phase change random access memory due to its rapid crystallization speed, but the crystallization temperature of Sb2Te3 (about 94°C in material investigated in this chapter and between 90-100°C reported in the literature (Yin et al., 2007, Kim et al.,

Upon annealing, amorphous Sb2Te3 films crystallize in rhombohedral Sb2Te3 (Kim et al., 2008, Lv et al., 2010) or fcc structure, which at a temperature above 2000C transforms into an hexagonal phase (Yin et al., 2007, Zhu et al., 2011). In the investigated Sb2Te3 films within the temperature range comprised between 95-2000C, only the rhombohedral phase (JCPDS data

Fig. 19. Avrami plot of *ln[−ln(1−x)]* vs *ln(t−τ)* for Ge1Sb4Te7 films. Points – experiment, lines – results from fittings using JMAK equation (3). Insert shows the rate constant *K* as a

Isothermal reflection measurements (Figure 20) have shown the same as for Ge1Sb4Te7: practically zero value of crystalline phase during the incubation time. In this case, the same as for GeTe, Ge4Sb1Te5 and Ge1Sb4Te7 it is possible to neglect the amount of crystallized material during τ, and describe the transformation using the modified JMAK equation (3). Figure 21 shows a plot of *ln[-ln(1-x)]* versus *ln(t-τ)* for a Sb2Te3 film, which demonstrates a linear dependence. This means that the crystallization process in Sb2Te3 films, the same as in Ge1Sb4Te7 material, display random nucleation and isotropic growth with an effective activation energy of 1.54±0.15 eV and an Avrami exponent *n* close to 1.1, which according to ref. (Christian, 1975), corresponds to a crystallization growth of particles of appreciable initial volume. It is necessary to note that the values obtained for crystallization activation energy is in good agreement with values reported in the literature (1.51 eV) obtained using

**8. Crystallization of Sb2Te3** 

#15-0874) has been observed.

function of the reciprocal temperature.

2010)) is too low for it to be of practical use.

the Kissinger analysis in the films deposited on the glass substrates by the same DC magnetron sputtering (Zhai et al., 2009).

Fig. 20. Dependencies of the volume fraction *x* versus time obtained for Sb2Te3 samples from reflection measurements at the temperatures indicated on the graph.

Fig. 21. Avrami plot of *ln[−ln(1−x)]* vs *ln(t−τ)* for Sb2Te3 films. Points – experiment, lines – results from fittings using JMAK equation (3). Insert shows the rate constant *K* as function of the reciprocal temperature.

#### **9. Conclusions**

This chapter presents some of the results concerning the crystallization properties of the most commonly employed materials on the GeTe–Sb2Te3 pseudo-binary line for phase

Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 455

temperature. All these factors and the possible existence of a local composition fluctuation can be responsible for the appearance of Ge1Sb4Te7 nuclei in the process of crystallization in

The clarification of crystallization mechanisms in alloys that lie on the GeTe–Sb2Te3 pseudobinary line will allow the production of phase-change materials with better recording

Bahgat, A. A., Mahmoud, E. A., Abd, Rabo, A. S. & Mahdy, I. A. (2006). Study of the

Caravati, S., Bernasconi, M. & Parrinello, M. First principles study of the optical contrast in

Claudio, D., Gonzalez-Hernandez, J., Licea, O., Laine, B., Prokhorov, E. & Trapaga, G. (2006)

Christian, J. W. *The Theory of Transformation in Metals and Alloys*, (1975), Pergamon Press,

Coombs, J. H., Jongenelis, A. P. J. M., Van Es-Spiekman, W. & Jacobs, B. A. (1995). Laser

Fan, Z., Wang, L. & Laughlin, D. E. (2004). Modeling of Crystallization Activation Energy

Fantini, A., Sousa, V. Perniola, L., Gourvest, E., Bastien, J. C., Maitrejean, S., Braga, S.,

Gervacio, Arciniega, J. J., Prokhorov, E., Espinoza, Beltran, F. J. & Gonzalez-Hernandez, J.

Gonzalez-Hernandez, J., Chao, B. S., Strand, D., Ovshinsky, S. R., Pawlik, D. & Gasiorowski,

*Physics Communications*, 11, No. 4, (April 1992) 557-581, ISSN 0277-9374. Henderson, D.W. (1979). Thermal analysis of non-isotermal crystallization kinetics in glass

978-1-4424-7418-5, San Francisco, CA. USA, December 2010.

*Letters*, 97, No. 6, (August 2010) 063504, ISSN 0003-6951.

ferroelectric properties of Ge–Sb–Te alloys. *Physica B: Condensed Matter*, 382 No. 1-2,

phase change materials. *Journal of Physics: Condensed Matter*, 22, No. 31, (July 2010)

An analytical model to represent crystallization kinetics in materials with metastable phase formation. *Journal of Non-Crystalline Solids*, 352, No. 1, (January

induced crystallization phenomena in GeTe based alloys. II. Composition dependence of nucleation and growth. *Journal of Applied Physics*, 78, No. 8, (October

for GeTe-Sb2Te3 Based Phase Change Materials. *Proceedings of SPIE*, ISBN 0-8194-

Pashkov, N., Bastard, A., Hyot, B., Roule, A., Persico, A., Feldis, H., Jahan, C., Nodin, J. F., Blachier, D., Toffoli, A., Reimbold, G., Fillot, F., Pierre, F., Annunziata, R., Benshael, D., Mazoyer, P., Vallee, C., Billon, T., Hazart, J., De Salvo, B. & Boulanger, F. (2010). N-doped GeTe as Performance Booster for Embedded Phase-Change Memories. *Proceedings of International Electron Device Meeting (IEDM)*. ISBN

(2010). Ferroelectric properties of Ge2Sb2Te5 phase-change films. *Applied Physics* 

P. (1992). Crystallization studies of GeSbTe optical memory materials. *Applied* 

forming liquits. *Journal of Non-Crystalline Solids*, 30, No. 3, (July 1979) 301-315, ISSN

Ge2Sb2Te5 and Ge1Sb2Te4 materials.

(April 2006) 271-278, ISSN 0921-4526.

Oxford. ISBN 978-0-08-044019-4, United Kingdom.

315801, ISSN 0953-8984.

2006) 51–55, ISSN 0022-3093.

1995) 4918-4956, ISSN 0021-8979.

0022-3093.

5293-9, Monterey CA. USA, April 2004.

properties.

**10. References** 

change memory application. Despite the fact that these alloys have found application in already commercialized optical disc and workable electrical phase change memory devices, their properties are not well known so far. This leads to many contradictory and controversial results in the literature.

As has been shown in this chapter, the crystallization properties of GeTe, Ge4Sb1Te5, Ge1Sb4Te7, and Sb2Te3 cannot be analyzed using the classical Johnson–Mehl–Avrami-Kolmogorov model for isothermal annealing due to the long incubation times. This limitation can be avoided by using a modification of the JMAK model, which takes into account that after incubation time the rate constant of crystallization can be considered as independent from time. Such model gave reasonable values of crystallization energy, which proved compatible with crystallization energy obtained by other methods. The difference in interpretation of isothermal measurements, in addition to the difference in preparation methods, can be responsible for the dispersion of crystallization parameters reported in the literature. But this modified model can be applied if during incubation time non-neglible amounts of crystalline phase material are observed. In the case of Ge2Sb2Te5 and Ge1Sb2Te4 films during incubation times, the nuclei fraction is of about 10% and has a composition corresponding to Ge1Sb4Te7, which is different from the nominal value for the amorphous matrix. That is why the crystallization in Ge2Sb2Te5 and Ge1Sb2Te4 materials during isothermal annealing can be considered as a process that takes place in two stages: in the first stage, nuclei of a metastable Ge1Sb4Te7 phase appear; in the second stage, the nuclei transforms into the equilibrium NaCl-type stoichiometric structures correspondent to the material composition. Such crystallization process can be described by a model in which the stable crystalline phase is preceded by the formation of a metastable phase. The most questionable issue is to explain these crystallization properties.

It is necessary to note that crystallization with phase separation has been observed in materials close to the GeTe–Sb2Te3 pseudo-binary line with excess of Ge (Ge2+xSb2Te5 x=0.5 (Privitera et al., 2003) and Sb (Ge2Sb2+xTe5 , 0<x<1, (Yamada et al., 2000), and Ge2Sb2.3Te5 (Yao et al., 2003)). In all of these materials, separation occurs during the crystallization phase, with segregation of small amounts of the excess elements, which remain in the amorphous state at the grain boundaries. By increasing the annealing temperature, the residual amorphous material can convert into another polycrystalline NaCl-type structure with a slightly lower lattice parameter. As a result, after the formation of the first phase, the crystallization rate is strongly reduced and a further conversion of the film into another crystalline structure that crystallizes at higher temperatures can occur (Privitera et al., 2003).

Furthermore, Ge2Sb2Te5, Ge1Sb2Te4 and Ge1Sb4Te7 materials crystallize into the same NaCltype (*Fm* 3 *m*) structure (Matsunaga et al., 2004, Matsunaga et al., 2006) with lattice constants of 6.001, 6.044 and 6.0876 Ǻ respectively (Morales-Sanchez et al., 2005). In materials with excess Sb, as the content of Sb increases, the lattice constant also increases (Yamada et al., 2000). Thus, during crystallization, atoms in the amorphous state must travel less distance to take their position in the Ge1Sb4Te7 crystal lattice than in any other ternary alloys. This effect is responsible for the highest crystallization speed of Ge1Sb4Te7 when compared with Ge2Sb2Te5 and Ge1Sb2Te4 materials. Also, Ge1Sb4Te7 has the lowest crystallization temperature. All these factors and the possible existence of a local composition fluctuation can be responsible for the appearance of Ge1Sb4Te7 nuclei in the process of crystallization in Ge2Sb2Te5 and Ge1Sb2Te4 materials.

The clarification of crystallization mechanisms in alloys that lie on the GeTe–Sb2Te3 pseudobinary line will allow the production of phase-change materials with better recording properties.

## **10. References**

454 Crystallization – Science and Technology

change memory application. Despite the fact that these alloys have found application in already commercialized optical disc and workable electrical phase change memory devices, their properties are not well known so far. This leads to many contradictory and

As has been shown in this chapter, the crystallization properties of GeTe, Ge4Sb1Te5, Ge1Sb4Te7, and Sb2Te3 cannot be analyzed using the classical Johnson–Mehl–Avrami-Kolmogorov model for isothermal annealing due to the long incubation times. This limitation can be avoided by using a modification of the JMAK model, which takes into account that after incubation time the rate constant of crystallization can be considered as independent from time. Such model gave reasonable values of crystallization energy, which proved compatible with crystallization energy obtained by other methods. The difference in interpretation of isothermal measurements, in addition to the difference in preparation methods, can be responsible for the dispersion of crystallization parameters reported in the literature. But this modified model can be applied if during incubation time non-neglible amounts of crystalline phase material are observed. In the case of Ge2Sb2Te5 and Ge1Sb2Te4 films during incubation times, the nuclei fraction is of about 10% and has a composition corresponding to Ge1Sb4Te7, which is different from the nominal value for the amorphous matrix. That is why the crystallization in Ge2Sb2Te5 and Ge1Sb2Te4 materials during isothermal annealing can be considered as a process that takes place in two stages: in the first stage, nuclei of a metastable Ge1Sb4Te7 phase appear; in the second stage, the nuclei transforms into the equilibrium NaCl-type stoichiometric structures correspondent to the material composition. Such crystallization process can be described by a model in which the stable crystalline phase is preceded by the formation of a metastable phase. The most

It is necessary to note that crystallization with phase separation has been observed in materials close to the GeTe–Sb2Te3 pseudo-binary line with excess of Ge (Ge2+xSb2Te5 x=0.5 (Privitera et al., 2003) and Sb (Ge2Sb2+xTe5 , 0<x<1, (Yamada et al., 2000), and Ge2Sb2.3Te5 (Yao et al., 2003)). In all of these materials, separation occurs during the crystallization phase, with segregation of small amounts of the excess elements, which remain in the amorphous state at the grain boundaries. By increasing the annealing temperature, the residual amorphous material can convert into another polycrystalline NaCl-type structure with a slightly lower lattice parameter. As a result, after the formation of the first phase, the crystallization rate is strongly reduced and a further conversion of the film into another crystalline structure that crystallizes at higher

Furthermore, Ge2Sb2Te5, Ge1Sb2Te4 and Ge1Sb4Te7 materials crystallize into the same NaCltype (*Fm* 3 *m*) structure (Matsunaga et al., 2004, Matsunaga et al., 2006) with lattice constants of 6.001, 6.044 and 6.0876 Ǻ respectively (Morales-Sanchez et al., 2005). In materials with excess Sb, as the content of Sb increases, the lattice constant also increases (Yamada et al., 2000). Thus, during crystallization, atoms in the amorphous state must travel less distance to take their position in the Ge1Sb4Te7 crystal lattice than in any other ternary alloys. This effect is responsible for the highest crystallization speed of Ge1Sb4Te7 when compared with Ge2Sb2Te5 and Ge1Sb2Te4 materials. Also, Ge1Sb4Te7 has the lowest crystallization

controversial results in the literature.

questionable issue is to explain these crystallization properties.

temperatures can occur (Privitera et al., 2003).


Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 457

Liu, D., Xu, L., Liao, Y. B., Dai, M., Zhao, L., …Xu, J., Wu, L. C., Ma, Z. Y. & Chen, K. J.

Lu Q. M. & Libera, M. (1995). Microstructural measurements of amorphous GeTe

Lv, B., Hu, S., Li, W., Di, X., Feng, L., Zhang, J., Wu, L., Cai, Y., Li, B. & Lei, Z. (2010).

Matsunaga, T. & Yamada, N. (2004). Structural investigation of GeSb2Te4 : A high-speed

Matsushita, T., Nakau, T., Suzuki, A. & Okuda, M. (1989). Measurements of activation

Morales-Sanchez, E., Lain, B., Prokhorov, E., Hernaґndez-Landaverde, M. A., Trapaga, G. &

Morales-Sanchez, E., Prokhorov, E., Gonzalez-Hernandez, J. & Mendoza-Galvan, A. (2005).

Naito, M., Ishimaru, M. & Hirotsu, Y. (2004). Local structure analysis of Ge-Sb-Te phase

Nobukuni N., Takashima, M., Ohno, T. & Horiea, M. (1995). Microstructural changes in

Films. *Vacuum*, 84, No. 7, (March 2010) 877–881, ISSN 0042-207X.

*Films*, 471, No. 1-2, (January 2005) 243-247, ISSN 0040-6090.

*Crystalline Solids*, 112, No. 1-3, (October 1989) 211-214, ISSN 0022-3093. Miao, X. S., Shi, L. P., Lee, H. K., Li, J. M., Zhao, R., Tan, P. K., Lim, K. G., Yang, H. X. &

*Physics*, 48, No. 12, (December 2009) 121104, ISSN 0021-4922.

(January 1995) 517-521, ISSN 0021-8979.

2241, ISSN 0020-1669.

1098-0121.

ISSN 0021-4922.

8979.

4922.

(2006). Comparison of Thermal Stability and Electrical Characterization between Ge2Sb2Te5 and Ge1Sb2Te4 Phase-Change Materials. *Japanese Journal of Applied* 

crystallization by hot-stage optical microscopy. *Journal of Applied Physics*, 77, No. 2,

Preparation and Characterization of Sb2Te3 Thin Films by Coevaporation*. International Journal of Photoenergy*, Article ID 476589 (June 2010), ISSN 1110-662X. Matsunaga, T., Kojima, R., Yamada, N., Kifune, K., Kubota, Y., Tabata, Y. & Takata, M.

(2006). Single Structure Widely Distributed in a GeTe-Sb2Te3 Pseudobinary System: A Rock Salt Structure is Retained by Intrinsically Containing an Enormous Number of Vacancies within Its crystal, *Inorganic Chemistry*, 45, No. 5, (February 2006) 2235-

phase-change material. *Physical Rewiew B*, 69, No. 10, (March 2004) 104111, ISSN

energy in the initialization process of amorphous GeTe films. *Journal of Non-*

Chong, T. C. (2006). Temperature Dependence of Phase-Change Random Access Memory Cell. *Japanese Journal of Applied Physics*, 45, No. 5A, (May 2006), 3955–3958,

Gonzalez-Hernandez, J. (2010). Crystallization process in Ge2Sb2Te5 amorphous

Structural, electric and kinetic parameters of ternary alloys of GeSbTe. *Thin Solid* 

change materials using high-resolution electron microscopy and nanobeam diffraction. *Journal of Applied Physics*, 95, No. 12, (June 2004) 8130-8135, ISSN 0021-

GeSbTe film during repetitious overwriting in phase change optical recording. *Journal of Applied Physics*, 78, No. 12, (December 1995) 6980-6988, ISSN 0021-8979. Ohta, T., Inoue, K., Uchida, M., Yoshida, K., Akiyama, T., Furakawa, S., Nagata, K. &

Nakamura, S. Phase change disk media having rapid cooling structure. *Japanese Journal of Applied Physics*, 28, Suppl. 28, No. 3, (March 1989) 123-128, ISSN 0021-


Huang, S. M., Huang, S. Y., Zhao, Z. J. & Sun, Z. (2006). Investigation of phase changes in

Im, J., Eom, J. H., Park, C., Park, K., Suh, D. S., Kim, K., Kang, Y. S., Kim, C., Lee, T. Y.,

Kalb, J., Spaepen, F. & Wuttig, M. (2004). Atomic force microscopy measurements of crystal

Kato, N., Takeda, Y., Fukano, T., Motohiro, T., Kawai, S. & Kuno, H. (1999). Compositional

Khulbe P. K., Wright, E. M. & Mansuripur, M. (2000). Crystallization behavior of as-

Kim, M.Y. & Oh, T. S. (2010). Crystallization behavior and thermoelectric characteristics of

Kolobov, A. V., Fons, P., Frenkel, A. I., Ankudinov, A. L., Tominaga, J. & Uruga, T. (2004).

Kolobov, A. V., Fons, P., Tominaga, J., Ankudinov, A. L., Yannopoulos, S. N. &

Kolobov, A. V., Fons, P., Tominaga, J. & Uruga, T. (2006). Why DVDs work the way they do:

*Letters*, 84, No. 25, (June 2004), 5240–5242, ISSN 0003-6951.

*International*, 34, No. 4, (May 2008) 1043–1046. ISSN 0272-8842.

*Materials*, 3, (September 2004) 703-708, ISSN 1476-1122.

2006) 529–533, ISSN 0947-8396.

2010) 6550-6553, ISSN 0040-6090.

S5103–S5108, ISSN 0953-8984.

(March 1993) 2272-2282, ISSN 0021-8979.

4922.

(November 2008) 205205, ISSN 1098-0121.

Ge1Sb4Te7 films by single ultra-fast laser pulses. *Applied Physics A*, 82, No. 3, (March

Khang, Y., Yoon, Y. G. & Ihm, J. (2008). Hierarchical structure and phase transition of (GeTe)*n*(Sb2Te3)*<sup>m</sup>* used for phase-change memory. *Physical Rewiew B*, 78, No. 20,

nucleation and growth rates in thin films of amorphous Te alloys. *Applied Physics* 

Dependence of Optical Constants and Microstructures of GeSbTe Thin Films for Compact-Disc-Rewritable (CD-RW) Readable with Conventional CD-ROM Drives *Japanese Journal of Applied Physics*, 38, No. 3B, (March 1999) 1707-1708, ISSN 0021-

deposited, melt quenched, and primed amorphous states of Ge2Sb2.3Te5 films. *Journal of Applied Physics*, 88, No. 7, (October 2000) 3926-3933, ISSN 0021-8979. Kim M. S., Cho, S. H., Hong, S. K., Roh, J. S. & Choi. D. J. (2008). Crystallization

characteristics of nitrogen-doped Sb2Te3 films for PRAM application. *Ceramics* 

the electrodeposited Sb2Te3 thin films. *Thin Solid Films*, 518, No. 22, (September

Understanding the phase-change mechanism of rewritable optical media. *Nature* 

Andrikopoulos, K. S. (2004). Crystallization-induced short-range order changes in amorphous GeTe. *Journal of Physics: Condensed Matter*, 16, No. 44, (October 2004)

The nanometer-scale mechanism of phase change in Ge–Sb–Te alloys. *Journal of Non-Crystalline Solids*, 352, No. 9-20, (March 2006) 1612–1615, ISSN 0022-3093. Kyrsta, S., Cremer, R., Neuschutz, D., Laurenzis, M., Bolivar, P. H. & Kurz, H. (2001).

Deposition and characterization of Ge–Sb–Te layers for applications in optical data storage. *Applied Surface Science*, 179, No.1-4, (July 2001) 55-60, ISSN 0169-4332. Laine, B., Trapaga, G., Prokhorov, E., Morales-Sanchez, E. & Gonzalez-Hernandez, J. (2003).

Model for isothermal crystallization kinetics with metastable phase formation. *Applied Physics Letters*, 83, No. 24, (December 2003) 4969–4971, ISSN 0003-6951. Libera M. & Chen, M. (1993). Time-resolved reflection and transmission studies of

amorphous Ge-Te thin-film crystallization. *Journal of Applied Physics*, 73, No. 5,


Crystallization of Ge:Sb:Te Thin Films for Phase Change Memory Application 459

Senkader S. & Wright, C. D. (2004). Models for phase-change of Ge2Sb2Te5 in optical and

Sian, T. S., Mittal, S., Goel, A., Chakraborty, S., Gautam, S., Budhwar, A., Kapil, R., Achar, P.

Strand, D. (2005). Ovonics: from science to products. *Journal of Optoelectronics and Advanced* 

Tominaga, J., Shima, T., Kuwahara, M., Fukaya, T., Kolobov, A. &Nakano, T. (2004).

Trappe, C., Bechvet, B., Hyot, B., Winkler, O., Facso, S. & Kurz, H. (2000). Recrystallization

Wei J. & F. Gan. (2003). Theoretical explanation of different crystallization processes

Yamada, N. (1996). Erasable phase-change optical material, *MRS Bulletin*, 21, No. 9, (October

Yamada, N. & Matsunaga, T. (2000) Structure of laser-crystallized Ge2Sb2+xTe5 sputtered thin

Yamada, N., Ohno, E., Nishiuchi, K., Akahira, N. & Takao, M. (1991). Rapid-phase transitions

*Journal of Applied Physics*, 69, No. 5, (March 1991) 2849-2856, ISSN 0021-8979. Yao, H. B., Shi, L. P., Chong, T. C., Tan, P. K. & Miao, X. S. (2003). Optical Transition of

Yin, Y., Sone H. & S. Hosaka. (2007). Characterization of nitrogen-doped Sb2Te3 films and

Zhai, F., Wang, Y., Wu, Y. & Gan, F. (2009). Phase Transition Kinetics of Sb2Te3 Phase

Zhang, T., Cheng, Y., Song. Z., Liu, B., Feng, S., Han, X., Zhang, Z. & Chen, B. (2008).

*Solid Films*, 441, No. 1-2, (September 2003) 292-297, ISSN 0040-6090.

*Materials,* 7, No. 4, (August 2005) 1679–1690, ISSN 1454-4164.

*15, No. 5, (January 2004) 411-415, ISSN 0957-4484.* 

511, ISSN 0021-8979.

936–941, ISSN 0021-4922.

3176, ISSN 0021-8979.

1996) 48-50, ISSN 0883-7694.

2000) 7020-7028, ISSN 0021-8979.

2B, (February 2003) 828-831, ISSN 0021-4922.

(September 2007) 064503, ISSN 0021-8979.

November 2008.

electrical memory devices. *Journal of Applied Physics*, 95, No. 2, (January 2004) 504-

C. & Nyati, G. (2008). Correlations between electrical parameters of rewritable digital versatile disc and optical and thermal properties of phase change alloy: effect of sputter power. *Japanese Journal of Applied Physics*, 47, No. 2, (February 2008)

Ferroelectric catastrophe: beyond nanometre-scale optical resolution*. Nanotechnology,* 

Dynamics of Phase Change Optical Disks with a Nitrogen Interface Layer. *Japanese Journal of Applied Physics*, 39, No. 2B, (February 2000) 766-769, ISSN 0021-4922. Wamwangi, D., Njoroge, W. K. & Wuttig, M. (2002). Crystallization kinetics of Ge4Sb1Te5 films. *Thin Solid Films*, 408, No. 1-2, (March 2002) 310–315, ISSN 0040-6090. Weidenhof, V., Friedrich, I., Ziegler, S. & Wuttig, M. (2001). Laser induced crystallization of

amorphous Ge2Sb2Te5 films. *Journal of Applied Physics*, 89, No. 6, (March 2001) 3168–

between as-deposited and melt-quenched amorphous Ge2Sb2Te5 thin films. *Thin* 

films for use in optical memory. *Journal of Applied Physics*, 88, No. 12, (December

of GeTe-Sb, Te, pseudobinary amorphous thin films for an optical disk memory,

Chalcogenide Phase-Change Thin Films. *Japanese Journal of Applied Physics*, 42, No.

their application to phase-change memory. *Journal of Applied Physics*, 102, No. 6,

Change Thin Films. *Proceedings of SPIE*, ISBN 0-8194-5293-9 Wuhan, China,

Comparison of the crystallization of Ge–Sb–Te and Si–Sb–Te in a constant-


Okura, M., Naito, H. & Matsushita, T. (1992). Discussion on the Mechanism of Reversible

Ovshinsky S. R. (1968). Reversible electrical switching phenomena in disordered structures, *Physical Review Letters*, 21, No. 20, (November 1968) 1450–1453, ISSN 0031-9007. Ovshinsky, S. & P. Klose, Reversible High-Speed High-Resolution Imaging in Amorphous

Paesler, M. A., Baker, D. A., Lucovsky, G., Edwards, A. E. & Taylor, P.C. (2007). EXAFS

Park, J., Kim, M. R., Choi, W. S., Seo, H. & Yeon, C. (1999). Characterization of Amorphous

Perniola, L., Sousa, V., Fantini, A., Arbaoui, E., Bastard, A., Armand, M., Fargeix, A., Jahan,

Popescu, M. A. Ovonic Materials. (2005). *Journal of Ovonic Research*, 1, No. 6, (December

Privitera, S., Rimini, E., Bongiorno, C., Zonca, R., Pirovano, A. & Bez, R. (2003).

Rabe, K. & Joannopoulos, J. (1987). Theory of the structural phase transition of GeTe. *Physical Rewiew B*, 36, No. 12, (October 1987) 6631-6639, ISSN 1098-0121. Raoux, S., Shelby, R. M., Jordan-Sweet, J., Munoz, B., Salinga, M., Chen, Y. C., Shih, Y.H.,

Reddy, G. B., Dhar, A., Malhotra L. K. & Sharmila E. K. (1992). Comparative study of

techniques. *Thin Solid Films*, 220, No. 1-2 , (1992) 111-115, ISSN 0040-6090. Ruitenberg, G., Petford-Long, A. K. & Doole, R. C. (2002). Determination of the isothermal

*Physics*, 94, No. 7, (October 2003) 4409-4413, ISSN 0021-8979.

Crystallization and phase separation. in Ge2+xSb2Te5 thin Films. *Journal of Applied* 

Lai, E. K. & Lee, M.H. (2008). Phase change materials and their application to random access memory technology. *Microelectronic Engineering*, 85, No. 12, (August

crystallization processes in Sb,Te, films using laser and thermal annealing

nucleation and growth parameters for the crystallization of thin Ge2Sb2Te5 films. *Journal of Applied Physics*, 92, No. 6, (September 2002) 3166-3123, ISSN 0021-8979. Ruiz, Santos, R., Prokhorov, E., Espinoza, Beltran, F. J., Trapaga, Martnez, L. G. &

Gonzalez-Hernandez, J. (2010). Crystallization and ferroelectric properties of Ge4Sb1Te5 films. *Journal of Non-Crystalline Solids*, 356, No. 52-54, (July 2010) 3026-

(February 1992) 466-470, ISSN 0021-4922.

*Technical Papers*, First Edition, 58-61 (1971).

0022-3697.

ISSN 0741-3106.

4775–4779, ISSN 0021-4922.

2005), 69-76, ISSN 1584–9953.

2008) 2330–2333, ISSN 0167-9317.

3031, ISSN 0022-3093.

Phase Change Optical Recording. *Japanese Journal of Applied Physics*, 31, No. 2B,

Semiconductors, *Society for Information Display International Symposium Digest of* 

study of local order in the amorphous chalcogenide semiconductor Ge2Sb2Te5. *Journal of Physics and Chemistry of Solids*, 68, No. 5-6, (March 2007) 873-877, ISSN

Phases of Ge2Sb2Te5 Phase-Change Optical Recording Material on Their Crystallization Behavior. *Japanese Journal of Applied Physics*, 38, No. 8, (May 1999)

C., Nodin, J.-F., Persico, A., Blachier, D., Toffoli, A., Loubriat, S., Gourvest, E., Betti, Beneventi, G., Feldis, H., Maitrejean, S., Lhostis, S., Roule, A., Cueto, O., Reimbold, G., Poupinet, L., Billon, T., De Salvo, B., Bensahel, D., Mazoyer, P., Annunziata, R., Zuliani, P. & Boulanger, F. (2010). Electrical Behavior of Phase-Change Memory Cells Based on GeTe. *Electron Device Letters IEEE*, 31, No. 5, (April 2010) 488-490,


**17** 

*India* 

**Metal Induced Crystallization** 

*School of Physics, University of Hyderabad, Hyderabad,* 

Ahamad Mohiddon Mahamad and Ghanashyam Krishna Mamidipudi

Polycrystalline silicon thin films have attracted the attention of semiconductor industries in the past few decades due to their wide applications in thin film transistors, solar cells, display units and sensors (Schropp & Zeman, 1998; Choi et al., 2005; Mahan et al., 2008). Polycrystalline Si thin films are generally fabricated by crystallizing amorphous Si (a-Si) thin films, because these can render larger grains compared to the conventional poly Si film deposition. As a consequence, a variety of methods for lowering the crystallization temperature of a-Si have been developed. Excimer laser annealing is one of the promising ways to achieve large grain size poly Si films at lower substrate temperatures. Its high costs and nonuniform grain size, however, are significant obstacles that prevent its wide use (Parr et al., 2002). The other promising technique is the solid phase crystallization method. But this technique is essentially a high-temperature process and many substrates, including most forms of glass, cannot withstand the thermal processing. In order to achieve lower costs and have a wider range of application, inexpensive materials such as glass and special polymers have to substitute quartz or Pyrex™ substrates. In the case of glass substrates, all of the processing steps need to be limited to temperatures below 550 °C. The other known technique is rapid thermal annealing (RTA). In RTA infrared radiation is used as a heating source, and has the advantage of the high heating speed (up to 60 ºC/s) that reduces the crystallization time. In RTA radiation is applied in pulses to heat the sample without heating the glass substrate (which is transparent to the infrared radiation). However, the grain size

obtained in the crystallization of a-Si is also in the range of a few micrometers.

In an effort to reduce the crystallization temperature and crystallization time, and to increase the grain size, metal-induced crystallization (MIC) has been investigated as an alternative crystallization process for thin-film device fabrication. The MIC process involves the deposition of a-Si films on top of which a layer of suitable metal is deposited. This bilayer of metal and Si is then annealed in a furnace at temperatures ranging from 150 to 700 °C for durations between one minute to several hours leading to crystallization of the a-Si. As a result the Si nanocrystals grow in the metal/metal silicide/metal oxide matrix and their growth rate depends on the annealing conditions of the bilayer. Thus, to grow the Si nanocrystals, both thermodynamics and kinetics of the process have to be understood in detail. In the MIC process, metals like aluminum, nickel, gold, silver etc are used to decrease the crystallization temperature below 700 °C. Among the metals employed for the study of the MIC (Jin et al.,1999; Kumar et al.,2007; Kumar et al.,2008; Mohiddon & Krishna, 2011), the preferred metal up to date has been nickel (Ni) due to its low residual metal

**1. Introduction** 

temperature annealing process. *Scripta Materialia*, 58, No. 11, (February 2008) 977– 980, ISSN 1359-6462.

Zhu, M., Wu, L., Rao, F., Song, Z., Peng, C., Li, X., Yao, D., Xi, W. & Feng. S. Phase Change Characteristics of SiO2 Doped Sb2Te3 Materials for Phase Change Memory Application. *Electrochemical and Solid-State Letters*, 14, No. 10, (July 2011) H404- H407, ISSN 1099-0062.

## **Metal Induced Crystallization**

Ahamad Mohiddon Mahamad and Ghanashyam Krishna Mamidipudi *School of Physics, University of Hyderabad, Hyderabad,* 

## *India*

#### **1. Introduction**

460 Crystallization – Science and Technology

Zhu, M., Wu, L., Rao, F., Song, Z., Peng, C., Li, X., Yao, D., Xi, W. & Feng. S. Phase Change

980, ISSN 1359-6462.

H407, ISSN 1099-0062.

temperature annealing process. *Scripta Materialia*, 58, No. 11, (February 2008) 977–

Characteristics of SiO2 Doped Sb2Te3 Materials for Phase Change Memory Application. *Electrochemical and Solid-State Letters*, 14, No. 10, (July 2011) H404-

> Polycrystalline silicon thin films have attracted the attention of semiconductor industries in the past few decades due to their wide applications in thin film transistors, solar cells, display units and sensors (Schropp & Zeman, 1998; Choi et al., 2005; Mahan et al., 2008). Polycrystalline Si thin films are generally fabricated by crystallizing amorphous Si (a-Si) thin films, because these can render larger grains compared to the conventional poly Si film deposition. As a consequence, a variety of methods for lowering the crystallization temperature of a-Si have been developed. Excimer laser annealing is one of the promising ways to achieve large grain size poly Si films at lower substrate temperatures. Its high costs and nonuniform grain size, however, are significant obstacles that prevent its wide use (Parr et al., 2002). The other promising technique is the solid phase crystallization method. But this technique is essentially a high-temperature process and many substrates, including most forms of glass, cannot withstand the thermal processing. In order to achieve lower costs and have a wider range of application, inexpensive materials such as glass and special polymers have to substitute quartz or Pyrex™ substrates. In the case of glass substrates, all of the processing steps need to be limited to temperatures below 550 °C. The other known technique is rapid thermal annealing (RTA). In RTA infrared radiation is used as a heating source, and has the advantage of the high heating speed (up to 60 ºC/s) that reduces the crystallization time. In RTA radiation is applied in pulses to heat the sample without heating the glass substrate (which is transparent to the infrared radiation). However, the grain size obtained in the crystallization of a-Si is also in the range of a few micrometers.

> In an effort to reduce the crystallization temperature and crystallization time, and to increase the grain size, metal-induced crystallization (MIC) has been investigated as an alternative crystallization process for thin-film device fabrication. The MIC process involves the deposition of a-Si films on top of which a layer of suitable metal is deposited. This bilayer of metal and Si is then annealed in a furnace at temperatures ranging from 150 to 700 °C for durations between one minute to several hours leading to crystallization of the a-Si. As a result the Si nanocrystals grow in the metal/metal silicide/metal oxide matrix and their growth rate depends on the annealing conditions of the bilayer. Thus, to grow the Si nanocrystals, both thermodynamics and kinetics of the process have to be understood in detail. In the MIC process, metals like aluminum, nickel, gold, silver etc are used to decrease the crystallization temperature below 700 °C. Among the metals employed for the study of the MIC (Jin et al.,1999; Kumar et al.,2007; Kumar et al.,2008; Mohiddon & Krishna, 2011), the preferred metal up to date has been nickel (Ni) due to its low residual metal

Metal Induced Crystallization 463

oxide interface layer, depending on the layer sequence, to allow the inter diffusion of the materials. The Si oxide layer is transformed by the Al into a mixture of Al oxide and an Al*x*Si phase. This newly formed Al*x*Si phase provides a diffusion channel for the Si and Al atoms (Nast & Wenham, 2000). The evaporated Al films are of polycrystalline nature, there are four different diffusion paths available for the diffusion of the Si atoms. 1) Diffusion inside the Al grains, 2) Diffusion along Al grain boundaries, 3) Diffusion along the Al/*a*-Si

The diffused Si atoms start crystallizing near the Al/*a*-Si interface. The Al grain boundaries are preferential nucleation sites for the dissolved Si atoms, since they are the sites of low critical free energy for nucleus formation. According to the random successive nucleation model, at early stage of crystallization, Si nuclei are formed at preferred sites located at the Al/Si interface. The released heat (crystallization energy and strain energy) leads to rise the local temperature in ambient regions. The temperature field propagates fast, since the thermal diffusion is faster than the atomic diffusion, so that the heat flow can stimulate new nuclei appearing randomly in nearby region. These nuclei of the next generation also produce a local temperature rise. This process repeats many time during annealing,

Silicon atoms, dissolved in the Al layer, have a high mobility and, hence, diffuse quickly within the film and/or along the interfaces towards the growing Si grains (McCaldin & Sankur, 1971). It is this fast growth within the Al layer compared to any other growth normal to the layer structure that leads to the formation of a continuous poly-Si film. At an early stage of the crystallization process the newly formed Si grains within the Al matrix are far apart and do not influence each other. The nucleation rate strongly depends on the defects and grain boundaries in the Al layer as well as the temperature and corresponding silicon concentration. Si solute depletion occurs up to the effective diffusion distance in the vicinity of the growing grain (Zener, 1949). In this region the possibility of new nucleation decreases with increasing depletion towards the advancing Al/Si grain interface. When the effective diffusion distances of adjacent grains begin to overlap, competition for the available Si atoms dissolved in the Al layer occurs. At this point, the possibility of new nuclei formation decreases. This type of crystallization pattern, where isolated grains start to interfere at an early stage of the process, is different to solid-phase crystallization of

The grain-size distribution is dependent on the ratio of the grain growth to the nucleation rate. This ratio increases with decreasing annealing temperature (Nast & Wenham, 2000). The decrease in grain growth due to a lower annealing temperature is less than the decrease in nucleation rate. The effective diffusion distance is longer at lower annealing temperatures leading to larger depletion areas around the growing grains, which prevents further nucleation. Therefore, the grains grow to a larger size before impingement occurs. The nucleations as well as the diffusion-controlled growth are thermally activated processes. Once the size of the Si grains is equivalent to the thickness of the Al layer, they solely advance laterally since they are constrained by the substrate and the *a*-Si/Al interface. Throughout the poly-Si growth process the Al layer is gradually displaced. The continuous supply of Si atoms from a-Si layer and gradual displacement of Al atoms leads to the growth of poly Si film with the large grain size. As an alternative member of this group of

interfaces, and 4) Diffusion along the glass/Al interface.

resulting in a fractal formation.

amorphous silicon (Spinella et al., 1998).

metals, we have investigated the tin induced a-Si crystallization.

contamination in the poly-Si region (Wang & Wong, 2001). The mobility and the transfer characteristics of the poly Si TFT are improved by reducing the leakage current caused by the metal agents. To achieve this, metal-induced lateral crystallization (MILC) has been introduced. Ni induced MILC TFTs' are now widely used in large display applications. Large number of metals have been investigated to crystallize the a-Si at possible low temperature. These metals are classified in to two groups basing on the mechanism they follow in crystallizing the a-Si matrix. One group of metals forms eutectic with a-Si (follows Layer exchange mechanism) and the other reacts with a-Si and forms silicide. The eutectic phase forming metals can produce extended continuous poly-Si films with large grain size of 10 μm (Oliver & Hartmann, 2000). However, these metals contaminate the so produced *c*-Si and degrade its properties. Thus, the search for metals that would lead to crystallization of *a*-Si thin film at low temperatures and also reduce the contamination is a subject of recent research.

The aim of this review is to present an overview of the current state of understanding of the mechanisms involved in MIC and to discuss some of the recent work of the current authors in this area. The motivation is to crystallize a-Si films using metals that have not been investigated in detail so far, in addition to Ni which is a well studied metal used in the MIC process. The two other metals that were selected for this study are Tin (Sn) and Chromium (Cr). The former is known to form a eutectic alloy with the Si whereas the latter is a silicide/oxide forming metal, similar to Nickel. Furthermore, results on two different types of geometries are presented; the first involves the deposition of metal on top of the a-Si film, while in the second geometry the a-Si film is deposited on top of the metal film. This provides very interesting insights in to the mechanisms that cause crystallization of a-Si.

## **2. Mechanisms of metal induced crystallization**

The main mechanisms of metal induced crystallization are discussed in this section.

## **2.1 Layer exchange mechanism**

The proposed formation of poly-Si films on foreign substrates by eutectic forming metals relies on the overall layer exchange of adjacent Si and metal films during the transformation of amorphous to polycrystalline Si. The phenomenon is named as "layer exchange mechanism". The general driving force behind metal-induced crystallization is the reduction of the free energy of the silicon material during the transformation of the amorphous to the crystalline phase. When *a*-Si is in contact with certain metals, electronic screening of the covalent bonding in the Si material occurs, which weakens the Si bonds, and therefore, facilitates the inter diffusion of the metal and silicon atoms (Hiraki, 1980). The representatives of the eutectic forming metals are Au, Ag and Al. Aluminum-induced crystallization [AIC] is extensively studied metal to form extended continuous poly-Si films of large-grained material on glass, which is desirable for thin film solar cells.

The AIC comprises the diffusion of Si atoms into the Al layer, which occurs due to the fact that adjacent Al and Si layers are not in thermal equilibrium at elevated temperatures. According to the Al/Si phase diagram, up to 1.5 at.% of Si can be dissolved in Al at temperatures below the eutectic temperature of 577 °C (Murray & McAlister, 1984). The initial interaction of the Al and *a*-Si layers involves the partial dissolution of the Al or Si

contamination in the poly-Si region (Wang & Wong, 2001). The mobility and the transfer characteristics of the poly Si TFT are improved by reducing the leakage current caused by the metal agents. To achieve this, metal-induced lateral crystallization (MILC) has been introduced. Ni induced MILC TFTs' are now widely used in large display applications. Large number of metals have been investigated to crystallize the a-Si at possible low temperature. These metals are classified in to two groups basing on the mechanism they follow in crystallizing the a-Si matrix. One group of metals forms eutectic with a-Si (follows Layer exchange mechanism) and the other reacts with a-Si and forms silicide. The eutectic phase forming metals can produce extended continuous poly-Si films with large grain size of 10 μm (Oliver & Hartmann, 2000). However, these metals contaminate the so produced *c*-Si and degrade its properties. Thus, the search for metals that would lead to crystallization of *a*-Si thin film at low temperatures and also reduce the contamination is a

The aim of this review is to present an overview of the current state of understanding of the mechanisms involved in MIC and to discuss some of the recent work of the current authors in this area. The motivation is to crystallize a-Si films using metals that have not been investigated in detail so far, in addition to Ni which is a well studied metal used in the MIC process. The two other metals that were selected for this study are Tin (Sn) and Chromium (Cr). The former is known to form a eutectic alloy with the Si whereas the latter is a silicide/oxide forming metal, similar to Nickel. Furthermore, results on two different types of geometries are presented; the first involves the deposition of metal on top of the a-Si film, while in the second geometry the a-Si film is deposited on top of the metal film. This provides very interesting insights in to the mechanisms that cause crystallization of a-Si.

The main mechanisms of metal induced crystallization are discussed in this section.

of large-grained material on glass, which is desirable for thin film solar cells.

The proposed formation of poly-Si films on foreign substrates by eutectic forming metals relies on the overall layer exchange of adjacent Si and metal films during the transformation of amorphous to polycrystalline Si. The phenomenon is named as "layer exchange mechanism". The general driving force behind metal-induced crystallization is the reduction of the free energy of the silicon material during the transformation of the amorphous to the crystalline phase. When *a*-Si is in contact with certain metals, electronic screening of the covalent bonding in the Si material occurs, which weakens the Si bonds, and therefore, facilitates the inter diffusion of the metal and silicon atoms (Hiraki, 1980). The representatives of the eutectic forming metals are Au, Ag and Al. Aluminum-induced crystallization [AIC] is extensively studied metal to form extended continuous poly-Si films

The AIC comprises the diffusion of Si atoms into the Al layer, which occurs due to the fact that adjacent Al and Si layers are not in thermal equilibrium at elevated temperatures. According to the Al/Si phase diagram, up to 1.5 at.% of Si can be dissolved in Al at temperatures below the eutectic temperature of 577 °C (Murray & McAlister, 1984). The initial interaction of the Al and *a*-Si layers involves the partial dissolution of the Al or Si

subject of recent research.

**2. Mechanisms of metal induced crystallization** 

**2.1 Layer exchange mechanism** 

oxide interface layer, depending on the layer sequence, to allow the inter diffusion of the materials. The Si oxide layer is transformed by the Al into a mixture of Al oxide and an Al*x*Si phase. This newly formed Al*x*Si phase provides a diffusion channel for the Si and Al atoms (Nast & Wenham, 2000). The evaporated Al films are of polycrystalline nature, there are four different diffusion paths available for the diffusion of the Si atoms. 1) Diffusion inside the Al grains, 2) Diffusion along Al grain boundaries, 3) Diffusion along the Al/*a*-Si interfaces, and 4) Diffusion along the glass/Al interface.

The diffused Si atoms start crystallizing near the Al/*a*-Si interface. The Al grain boundaries are preferential nucleation sites for the dissolved Si atoms, since they are the sites of low critical free energy for nucleus formation. According to the random successive nucleation model, at early stage of crystallization, Si nuclei are formed at preferred sites located at the Al/Si interface. The released heat (crystallization energy and strain energy) leads to rise the local temperature in ambient regions. The temperature field propagates fast, since the thermal diffusion is faster than the atomic diffusion, so that the heat flow can stimulate new nuclei appearing randomly in nearby region. These nuclei of the next generation also produce a local temperature rise. This process repeats many time during annealing, resulting in a fractal formation.

Silicon atoms, dissolved in the Al layer, have a high mobility and, hence, diffuse quickly within the film and/or along the interfaces towards the growing Si grains (McCaldin & Sankur, 1971). It is this fast growth within the Al layer compared to any other growth normal to the layer structure that leads to the formation of a continuous poly-Si film. At an early stage of the crystallization process the newly formed Si grains within the Al matrix are far apart and do not influence each other. The nucleation rate strongly depends on the defects and grain boundaries in the Al layer as well as the temperature and corresponding silicon concentration. Si solute depletion occurs up to the effective diffusion distance in the vicinity of the growing grain (Zener, 1949). In this region the possibility of new nucleation decreases with increasing depletion towards the advancing Al/Si grain interface. When the effective diffusion distances of adjacent grains begin to overlap, competition for the available Si atoms dissolved in the Al layer occurs. At this point, the possibility of new nuclei formation decreases. This type of crystallization pattern, where isolated grains start to interfere at an early stage of the process, is different to solid-phase crystallization of amorphous silicon (Spinella et al., 1998).

The grain-size distribution is dependent on the ratio of the grain growth to the nucleation rate. This ratio increases with decreasing annealing temperature (Nast & Wenham, 2000). The decrease in grain growth due to a lower annealing temperature is less than the decrease in nucleation rate. The effective diffusion distance is longer at lower annealing temperatures leading to larger depletion areas around the growing grains, which prevents further nucleation. Therefore, the grains grow to a larger size before impingement occurs. The nucleations as well as the diffusion-controlled growth are thermally activated processes. Once the size of the Si grains is equivalent to the thickness of the Al layer, they solely advance laterally since they are constrained by the substrate and the *a*-Si/Al interface. Throughout the poly-Si growth process the Al layer is gradually displaced. The continuous supply of Si atoms from a-Si layer and gradual displacement of Al atoms leads to the growth of poly Si film with the large grain size. As an alternative member of this group of metals, we have investigated the tin induced a-Si crystallization.

Metal Induced Crystallization 465

(Yoon et al., 2001). The other mechanism of FAMIC considers the bombardment of Ni atoms by electrons traveling in an *a*-Si layer. Nickel atoms are pushed toward the positive electrode under a large number of low-momentum collisions (electron wind). Also, energetic electrons are capable of breaking bonds by exciting the bonding electrons through the collision. In fact both field-enhanced diffusion under electric force and electron bombardment effects are responsible for the enhanced crystallization in FAMIC. In both mechanisms, the diffusion of metal is a common argument. Grisenti et al. (Grisenti et al., 2008) reported the Extended x-ray absorption fine structure (EXAFS) analysis on Ni induced *a*-Si produced by co-sputtering. They came to conclusion that, Ni segregates as NiSi2 even at temperatures as low as 200 °C and this NiSi2 enhances the MIC of *a*-Si. The transformation from one kind of silicide to other is well explained by Ni diffusion through Ni/Si interface

In case of MILC first metal silicides are formed at the interface between metal and Si layers at the lower annealing temperatures. On further annealing, nucleation of c-Si on metal silicide precipitate starts, which is then followed by migration of the metal silicide precipitates throughout the a-Si thin film, resulted in crystallization of the entire a-Si film. This mechanism is supported by the following experimental observations. Jin et al. (Jin et al, 1998) studied the XPS depth profile of Ni (10 nm) covered a-Si films (70 nm) which was annealed at 500 C for 1 hr. Ar ions with energy of 4 keV were used for sputtering during depth profiling. They reported the concentration variation of selected elements (Ni, O and Si) across the 100 nm stack. From their results it is found that the Ni concentration first decreased with Ar sputtering then stabilized to about 4% in the bulk of the film and finally increased to about 10% near the interface of the film. No Si was detected on the surface of the stack before the Ar sputtering, indicating the diffusion of Si in Ni and NiO*<sup>x</sup>* was slow at the heat treatment temperature of 500 °C. The presence of excess metallic Ni at the bottom of the stack was attributed to the diffusion of the NiSi2 nodules. Hayzelden et al. (Hayzelden & Batstone, 1993) reported an in-situ transmission electron microscopy study on Ni induced *a*-Si crystallization and suggested a similar mechanism for the diffusion of NiSi2 in *a*-Si matrix. According to the mechanism proposed by these authors, Si first crystallizes on one of the eight faces of NiSi2 crystalline seeds and then it dissociates as Ni and Si atoms at the *c*-Si/ NiSi2 interface. The Nickel atoms then diffuse and react again with *a*-Si to form NiSi2 at *a*-Si/ NiSi2 interface. The cause of such dissociation, diffusion and reformation of NiSi2 was explained by chemical potential free energy of the reactions at different interfaces. The chemical potential of Ni is lower at NiSi2/a-Si interface, while the chemical potential of Si is lower at the NiSi2/c-Si interface. Thus Ni moves to the NiSi2/a-Si interface and Si atoms are froced to diffuse to the NiSi2/c-Si interface. The consumption of a-Si at the NiSi2/a-Si interface and diffusion of the Ni atoms result in the growth of needle like Si crystallites. The diffusion of Ni in crystalline Si is much faster than that in a-Si. This supports the fast diffusion of Ni from the c-Si interface to the a-Si interface through a NiSi2 crystallite. Hwang et al. (Hwang & Li, 2005) studied the Auger depth profile on Ni induced MIC and came to a similar conclusion. Our recent work on the transmission electron microscopy study of Ni induced MIC of a-Si films also supports the NiSi2 diffusion mechanism. The details of the

and this mechanism is widely accepted in the semiconductor technology.

experiment and result are presented in the section 4.2 of the chapter.

**2.3 Metal Induced Lateral Crystallization (MILC)** 

## **2.2 Mechanism of diffusion assisted crystallization**

Two main mechanisms of diffusion assisted crystallization are proposed in literature. One is based on the diffusion and movement of Ni in the *a*-Si matrix causing crystallization of Si around dispersed nanostructured NiSi2 seeds (Ni diffusion assisted MIC); the other is based on the formation of the silicides at the interface between Ni and Si layers, which then moves into the *a*-Si matrix, leading to crystallization of Si (MILC)

According to the first mechanism, the metal (Ni) at first diffuses through *a*- Si matrix and then forms NiSi2 at a sufficient high temperature. This silicide will then act as seed for the crystallization of *a*-Si. The concept of metal diffusion is proposed on the basis of the experimental observations reported in the literature. Park et al. [Park et al., 2001] reported secondary ion mass spectroscopy (SIMS) and transmission electron microscope studies on excimer laser annealed Ni induced crystallization of poly Si. They measured the melting temperatures as a function of Ni content and reported that *a*-Si, just below the Ni layer, melted at lower temperature and the melting point increases with movement away from the Ni rich layer. The change in melting point was attributed to change in Ni concentration. Their explanation for the observed behaviour was that, the Ni diffuses through *a*-Si matrix only when a sufficiently high temperature is reached. Ferri et al. [Ferri et al., 2001] studied diluted metal contaminated *a*-Si system by detailed Raman spectroscopy measurement and came to similar conclusions.

The knowledge of formation, structure and electrical behaviour of a metal-semiconductor interfaces plays an important role in semiconductor technology. The nickel monosilicide (NiSi) is a low resistance silicide in Ni-Si binary system, and is a key material to reduce the contact resistance of gate and source-drain regions. I. H. Hong et al. (Hong et al., 2006) carried out scanning photoelectron spectromicroscopy on the a-Si/Ni films, which were deposited by chemical vapor deposition at ultra high vacuum pressure. They state that when the Si/Ni interface is heated, Ni2Si, which is an unstable phase with high resistivity, is first formed at 200-300 °C. At around 300 °C or above, Ni2Si starts transformation into a low resistivity phase NiSi. The Ni2Si phase disappears at above 400°C. Another high resistivity phase, NiSi2, nucleates at above 750 °C. Another method of enhancing metal diffusion is electric field assisted MIC (FAMIC). In FAMIC, the crystallization of a-Si is enhanced by applying an electric field to the metal/Si binary system.

The two major driving forces for the MIC process are (1) the free-energy gradient between silicon and metal silicides, (2) the concentration gradient caused by atomic diffusion in metal/silicon stacked films. In addition to these driving forces, the applied electrical field introduces an additional driving force to lower the activation energy for crystallization. Lee (Lee et al., 2000) suggested in the study of a Cu/Si FAMIC system that the electrons can charge up at the a-Si surface of the Cu silicide/a-Si interface under an applied voltage. The Cu ions with positive charges could then migrate at a faster speed, resulting in a directional and rapid lateral movement of Cu silicide/a-Si interface and so the crystallized poly-Si can be formed at the enhanced speed. The similar enhanced growth rate is reported in Ni/Si system. It has been postulated that the enhanced growth rate is a result of field-enhanced diffusion of nickel atoms (Park et al., 1999; Yoon et al., 2001). Having an effective charge of - 0.3299e in the Si lattice, Ni atoms move toward the positive electrode under an electric field

Two main mechanisms of diffusion assisted crystallization are proposed in literature. One is based on the diffusion and movement of Ni in the *a*-Si matrix causing crystallization of Si around dispersed nanostructured NiSi2 seeds (Ni diffusion assisted MIC); the other is based on the formation of the silicides at the interface between Ni and Si layers, which then moves

According to the first mechanism, the metal (Ni) at first diffuses through *a*- Si matrix and then forms NiSi2 at a sufficient high temperature. This silicide will then act as seed for the crystallization of *a*-Si. The concept of metal diffusion is proposed on the basis of the experimental observations reported in the literature. Park et al. [Park et al., 2001] reported secondary ion mass spectroscopy (SIMS) and transmission electron microscope studies on excimer laser annealed Ni induced crystallization of poly Si. They measured the melting temperatures as a function of Ni content and reported that *a*-Si, just below the Ni layer, melted at lower temperature and the melting point increases with movement away from the Ni rich layer. The change in melting point was attributed to change in Ni concentration. Their explanation for the observed behaviour was that, the Ni diffuses through *a*-Si matrix only when a sufficiently high temperature is reached. Ferri et al. [Ferri et al., 2001] studied diluted metal contaminated *a*-Si system by detailed Raman spectroscopy measurement and

The knowledge of formation, structure and electrical behaviour of a metal-semiconductor interfaces plays an important role in semiconductor technology. The nickel monosilicide (NiSi) is a low resistance silicide in Ni-Si binary system, and is a key material to reduce the contact resistance of gate and source-drain regions. I. H. Hong et al. (Hong et al., 2006) carried out scanning photoelectron spectromicroscopy on the a-Si/Ni films, which were deposited by chemical vapor deposition at ultra high vacuum pressure. They state that when the Si/Ni interface is heated, Ni2Si, which is an unstable phase with high resistivity, is first formed at 200-300 °C. At around 300 °C or above, Ni2Si starts transformation into a low resistivity phase NiSi. The Ni2Si phase disappears at above 400°C. Another high resistivity phase, NiSi2, nucleates at above 750 °C. Another method of enhancing metal diffusion is electric field assisted MIC (FAMIC). In FAMIC, the crystallization of a-Si is enhanced by

The two major driving forces for the MIC process are (1) the free-energy gradient between silicon and metal silicides, (2) the concentration gradient caused by atomic diffusion in metal/silicon stacked films. In addition to these driving forces, the applied electrical field introduces an additional driving force to lower the activation energy for crystallization. Lee (Lee et al., 2000) suggested in the study of a Cu/Si FAMIC system that the electrons can charge up at the a-Si surface of the Cu silicide/a-Si interface under an applied voltage. The Cu ions with positive charges could then migrate at a faster speed, resulting in a directional and rapid lateral movement of Cu silicide/a-Si interface and so the crystallized poly-Si can be formed at the enhanced speed. The similar enhanced growth rate is reported in Ni/Si system. It has been postulated that the enhanced growth rate is a result of field-enhanced diffusion of nickel atoms (Park et al., 1999; Yoon et al., 2001). Having an effective charge of - 0.3299e in the Si lattice, Ni atoms move toward the positive electrode under an electric field

**2.2 Mechanism of diffusion assisted crystallization** 

into the *a*-Si matrix, leading to crystallization of Si (MILC)

applying an electric field to the metal/Si binary system.

came to similar conclusions.

(Yoon et al., 2001). The other mechanism of FAMIC considers the bombardment of Ni atoms by electrons traveling in an *a*-Si layer. Nickel atoms are pushed toward the positive electrode under a large number of low-momentum collisions (electron wind). Also, energetic electrons are capable of breaking bonds by exciting the bonding electrons through the collision. In fact both field-enhanced diffusion under electric force and electron bombardment effects are responsible for the enhanced crystallization in FAMIC. In both mechanisms, the diffusion of metal is a common argument. Grisenti et al. (Grisenti et al., 2008) reported the Extended x-ray absorption fine structure (EXAFS) analysis on Ni induced *a*-Si produced by co-sputtering. They came to conclusion that, Ni segregates as NiSi2 even at temperatures as low as 200 °C and this NiSi2 enhances the MIC of *a*-Si. The transformation from one kind of silicide to other is well explained by Ni diffusion through Ni/Si interface and this mechanism is widely accepted in the semiconductor technology.

#### **2.3 Metal Induced Lateral Crystallization (MILC)**

In case of MILC first metal silicides are formed at the interface between metal and Si layers at the lower annealing temperatures. On further annealing, nucleation of c-Si on metal silicide precipitate starts, which is then followed by migration of the metal silicide precipitates throughout the a-Si thin film, resulted in crystallization of the entire a-Si film. This mechanism is supported by the following experimental observations. Jin et al. (Jin et al, 1998) studied the XPS depth profile of Ni (10 nm) covered a-Si films (70 nm) which was annealed at 500 C for 1 hr. Ar ions with energy of 4 keV were used for sputtering during depth profiling. They reported the concentration variation of selected elements (Ni, O and Si) across the 100 nm stack. From their results it is found that the Ni concentration first decreased with Ar sputtering then stabilized to about 4% in the bulk of the film and finally increased to about 10% near the interface of the film. No Si was detected on the surface of the stack before the Ar sputtering, indicating the diffusion of Si in Ni and NiO*<sup>x</sup>* was slow at the heat treatment temperature of 500 °C. The presence of excess metallic Ni at the bottom of the stack was attributed to the diffusion of the NiSi2 nodules. Hayzelden et al. (Hayzelden & Batstone, 1993) reported an in-situ transmission electron microscopy study on Ni induced *a*-Si crystallization and suggested a similar mechanism for the diffusion of NiSi2 in *a*-Si matrix. According to the mechanism proposed by these authors, Si first crystallizes on one of the eight faces of NiSi2 crystalline seeds and then it dissociates as Ni and Si atoms at the *c*-Si/ NiSi2 interface. The Nickel atoms then diffuse and react again with *a*-Si to form NiSi2 at *a*-Si/ NiSi2 interface. The cause of such dissociation, diffusion and reformation of NiSi2 was explained by chemical potential free energy of the reactions at different interfaces. The chemical potential of Ni is lower at NiSi2/a-Si interface, while the chemical potential of Si is lower at the NiSi2/c-Si interface. Thus Ni moves to the NiSi2/a-Si interface and Si atoms are froced to diffuse to the NiSi2/c-Si interface. The consumption of a-Si at the NiSi2/a-Si interface and diffusion of the Ni atoms result in the growth of needle like Si crystallites. The diffusion of Ni in crystalline Si is much faster than that in a-Si. This supports the fast diffusion of Ni from the c-Si interface to the a-Si interface through a NiSi2 crystallite. Hwang et al. (Hwang & Li, 2005) studied the Auger depth profile on Ni induced MIC and came to a similar conclusion. Our recent work on the transmission electron microscopy study of Ni induced MIC of a-Si films also supports the NiSi2 diffusion mechanism. The details of the experiment and result are presented in the section 4.2 of the chapter.

Metal Induced Crystallization 467

deposited BSG/Sn/*a*-Si stack, shows sharp diffraction peaks along with a broad diffuse hump around 2θ = 27o, due to the amorphous substrate and *a*-Si film. The peaks can be assigned to the (200), (101), (220), (211), (112) and (312) planes of the tetragonal phase of metallic Sn (the peaks are indexed according to the PCPDF file no- 23456). There is no

On annealing this stack at 300 °C for 1 hr, there is a very interesting transformation in the crystalline behavior of the films. Significantly, all the peaks due to metallic Sn are completely suppressed and the x-ray diffraction pattern resembles that of an amorphous film except for a very diffuse peak at 2θ = 33.58o assigned to the (111) plane of the diamond cubic form of Si. The intensity of this peak increases with increasing annealing temperature up to 500 °C as shown in Fig. 2 indicating that the extent of crystallization of Si is improved by increasing the annealing temperature. However, there is no evidence for crystalline Sn in the films. The reason for this is, most probably, partial oxidation of Sn. The peak at 2θ = 33.58o assigned to the (111) plane of diamond cubic silicon (PCPDF file no- 23345) is shifted by 0.45o which is attributable to strain in the film. Evidently, the onset of crystallization of *a*-Si occurs at 300 °C by contacting with the Sn metal. M Joen et al. have also reported Sn induced crystallization of *a*-Si at 300 °C (Jeon et al., 2010). Significantly, there is no evidence for the formation of silicides in the Sn-Si system even after annealing at 500 °C. Deconvolution of the peak at 2θ = 33.58o provides more insight into the mechanism of crystallization of *a-*Si. The peak was deconvoluted by smoothening and best fitting with a Gaussian function. The data is shown in the inset of Fig.2 for the samples annealed at 300, 400 and 500 °C. Closer observation of the peak corresponding to the sample annealed at 500 °C shows that the peak can be resolved into two peaks centered at 33.7 and 34o. The deconvoluted peaks fitted using Gaussian function is shown in the inset of the Fig 2. The peak centered at 2θ=34° is assigned to the (113) plane of orthorhombic SnO2 phase (PCPDF file no- 781063). By eliminating the effect of SnO2 phase in 500 °C annealed samples, the crystallite size is estimated as 18 nm. The average nanocrystal size calculated by Scherrer's equation (Mohiddon & Yadav, 2008) is found to increase from 5 nm to 14 nm with increasing the annealing temperature from 300 to 400 °C. The solid solubility of Sn in Si is known to be nil, *i.e.* they are immiscible in the solid state (Leonard & Koch, 1992). Hence it is not expected to form a silicide. X-ray diffraction data presented earlier indicates that this is the case even in the present samples. The crystallographic evolution of the c-Si phase in the current case is traced by studying the dependence of the crystallite size of Sn and Si on the annealing temperature. This is plotted in Fig. 3, from which it is evident that the crystallite size of Si is zero in the as-deposited film while that of the Sn phase is high. Interestingly, there is a cross-over region below 300 °C when the crystallite sizes of both phases are low. This indicates that at this temperature, the film is essentially amorphous. On increasing the annealing temperature to 400 and 500 °C, there is further increase in the crystallite size of Si while crystallite size of the Sn phase is invariant at zero. This behavior is very similar to that observed in the case of Al induced crystallization (AIC) of a-Si (McCaldin & Sanku, 1971; Zener,1949;Spinella et al.,1998) with one very crucial difference. No peaks corresponding to crystalline Sn are observed on annealing the films unlike the AIC case. This is evidently due to oxidation of Sn during the annealing process as revealed by the deconvolution of the (111) peak of Si in the inset of Fig.2. Thus we conclude from this work that *a*-Si crystallizes when it is brought in contact with Sn metal layer without involving the formation of a

evidence for the presence of crystalline Si.

silicide, exactly as followed by eutectic forming metals at 300 °C.

### **3. Tin induced a-Si crystallization**

Tin is a representative of the eutectic forming group along with Al and Au. The Sn–Si alloy has a relatively low eutectic temperature of 232 °C (Jeon et al., 2010) compared other members of the group. Our recent work on Tin induced a-Si revealed that it follows the eutectic forming metal mechanism well below the Al-Si system (Mohiddon & Krishna, 2011). Jeon et al. (Jeon et al., 2010) studied the Sn-Si system for growing Si nanowires and quoted that Sn is a favorable catalyst for low temperature synthesis of Si nanowires.

The experimental details are as follows. Sn films of 500 nm thickness were deposited onto Borosilicate Glass (BSG) substrates by resistive thermal evaporation. Si films of 500 nm thickness were deposited by electron beam evaporation over the Sn layers. Fig. 1 shows the block diagram of experimental deposition setup of electron beam evaporator and block diagram of film annealing setup. The starting materials were granular pure silicon powder (99.999%) and Sn (99.99% pure). A pressure 5 × 10−6 Torr was maintained throughout the depositions. The depositions were carried out at ambient temperature and in all cases the substrate to source distance was kept constant at 10 cm. The thickness of the films was measured after deposition using a surface profilometer (model XP-1 of Ambios Technology, USA). The films were annealed in a furnace atmosphere at different temperatures for 1 hr. X-ray diffraction patterns were recorded on a powder x-ray diffractometer (CPS120 of Inel, France) machine equipped with a Co K*α* x-ray source (wavelength = 0.178896 nm) and gas phase position sensitive detector.

Fig. 1. Schematic diagram of (a) film deposition by Electron beam evaporator (b) annealing of the film in muffle furnace

X-ray diffraction patterns of as deposited BSG/Sn/*a*-Si stacks and stacks post deposition annealed at different temperatures from 300 °C to 500 °C each for 1 hour are presented in Fig. 2. XRD pattern of Sn film deposited on BSG and annealed at 500 °C for 15 hrs is also presented in the same figure. It is evident from the XRD pattern of the annealed Sn film that, it is well crystallized and the pattern can be identified as belonging to the orthorhombic phase of SnO2 (the peaks were indexed according to the PCPDF file no-781063). The as

Tin is a representative of the eutectic forming group along with Al and Au. The Sn–Si alloy has a relatively low eutectic temperature of 232 °C (Jeon et al., 2010) compared other members of the group. Our recent work on Tin induced a-Si revealed that it follows the eutectic forming metal mechanism well below the Al-Si system (Mohiddon & Krishna, 2011). Jeon et al. (Jeon et al., 2010) studied the Sn-Si system for growing Si nanowires and

The experimental details are as follows. Sn films of 500 nm thickness were deposited onto Borosilicate Glass (BSG) substrates by resistive thermal evaporation. Si films of 500 nm thickness were deposited by electron beam evaporation over the Sn layers. Fig. 1 shows the block diagram of experimental deposition setup of electron beam evaporator and block diagram of film annealing setup. The starting materials were granular pure silicon powder (99.999%) and Sn (99.99% pure). A pressure 5 × 10−6 Torr was maintained throughout the depositions. The depositions were carried out at ambient temperature and in all cases the substrate to source distance was kept constant at 10 cm. The thickness of the films was measured after deposition using a surface profilometer (model XP-1 of Ambios Technology, USA). The films were annealed in a furnace atmosphere at different temperatures for 1 hr. X-ray diffraction patterns were recorded on a powder x-ray diffractometer (CPS120 of Inel, France) machine equipped with a Co K*α* x-ray source (wavelength = 0.178896 nm) and gas

Fig. 1. Schematic diagram of (a) film deposition by Electron beam evaporator (b) annealing

X-ray diffraction patterns of as deposited BSG/Sn/*a*-Si stacks and stacks post deposition annealed at different temperatures from 300 °C to 500 °C each for 1 hour are presented in Fig. 2. XRD pattern of Sn film deposited on BSG and annealed at 500 °C for 15 hrs is also presented in the same figure. It is evident from the XRD pattern of the annealed Sn film that, it is well crystallized and the pattern can be identified as belonging to the orthorhombic phase of SnO2 (the peaks were indexed according to the PCPDF file no-781063). The as

quoted that Sn is a favorable catalyst for low temperature synthesis of Si nanowires.

**3. Tin induced a-Si crystallization** 

phase position sensitive detector.

of the film in muffle furnace

deposited BSG/Sn/*a*-Si stack, shows sharp diffraction peaks along with a broad diffuse hump around 2θ = 27o, due to the amorphous substrate and *a*-Si film. The peaks can be assigned to the (200), (101), (220), (211), (112) and (312) planes of the tetragonal phase of metallic Sn (the peaks are indexed according to the PCPDF file no- 23456). There is no evidence for the presence of crystalline Si.

On annealing this stack at 300 °C for 1 hr, there is a very interesting transformation in the crystalline behavior of the films. Significantly, all the peaks due to metallic Sn are completely suppressed and the x-ray diffraction pattern resembles that of an amorphous film except for a very diffuse peak at 2θ = 33.58o assigned to the (111) plane of the diamond cubic form of Si. The intensity of this peak increases with increasing annealing temperature up to 500 °C as shown in Fig. 2 indicating that the extent of crystallization of Si is improved by increasing the annealing temperature. However, there is no evidence for crystalline Sn in the films. The reason for this is, most probably, partial oxidation of Sn. The peak at 2θ = 33.58o assigned to the (111) plane of diamond cubic silicon (PCPDF file no- 23345) is shifted by 0.45o which is attributable to strain in the film. Evidently, the onset of crystallization of *a*-Si occurs at 300 °C by contacting with the Sn metal. M Joen et al. have also reported Sn induced crystallization of *a*-Si at 300 °C (Jeon et al., 2010). Significantly, there is no evidence for the formation of silicides in the Sn-Si system even after annealing at 500 °C. Deconvolution of the peak at 2θ = 33.58o provides more insight into the mechanism of crystallization of *a-*Si. The peak was deconvoluted by smoothening and best fitting with a Gaussian function. The data is shown in the inset of Fig.2 for the samples annealed at 300, 400 and 500 °C. Closer observation of the peak corresponding to the sample annealed at 500 °C shows that the peak can be resolved into two peaks centered at 33.7 and 34o. The deconvoluted peaks fitted using Gaussian function is shown in the inset of the Fig 2. The peak centered at 2θ=34° is assigned to the (113) plane of orthorhombic SnO2 phase (PCPDF file no- 781063). By eliminating the effect of SnO2 phase in 500 °C annealed samples, the crystallite size is estimated as 18 nm. The average nanocrystal size calculated by Scherrer's equation (Mohiddon & Yadav, 2008) is found to increase from 5 nm to 14 nm with increasing the annealing temperature from 300 to 400 °C. The solid solubility of Sn in Si is known to be nil, *i.e.* they are immiscible in the solid state (Leonard & Koch, 1992). Hence it is not expected to form a silicide. X-ray diffraction data presented earlier indicates that this is the case even in the present samples. The crystallographic evolution of the c-Si phase in the current case is traced by studying the dependence of the crystallite size of Sn and Si on the annealing temperature. This is plotted in Fig. 3, from which it is evident that the crystallite size of Si is zero in the as-deposited film while that of the Sn phase is high. Interestingly, there is a cross-over region below 300 °C when the crystallite sizes of both phases are low. This indicates that at this temperature, the film is essentially amorphous. On increasing the annealing temperature to 400 and 500 °C, there is further increase in the crystallite size of Si while crystallite size of the Sn phase is invariant at zero. This behavior is very similar to that observed in the case of Al induced crystallization (AIC) of a-Si (McCaldin & Sanku, 1971; Zener,1949;Spinella et al.,1998) with one very crucial difference. No peaks corresponding to crystalline Sn are observed on annealing the films unlike the AIC case. This is evidently due to oxidation of Sn during the annealing process as revealed by the deconvolution of the (111) peak of Si in the inset of Fig.2. Thus we conclude from this work that *a*-Si crystallizes when it is brought in contact with Sn metal layer without involving the formation of a silicide, exactly as followed by eutectic forming metals at 300 °C.

Metal Induced Crystallization 469

The second group of metals used for MIC, reacts with silicon and forms silicides. e.g.: Ni, Cr, Pd, Mo etc. A very small concentration of metals contamination crystallizes large part of *a*-Si and hence they are highly useful in electronic applications. In this section we review our

Nickel induced MIC is one of best representatives of this group. To investigate Nickel based MIC, two types of experiments were carried out. In the first case, a 400 nm *a*-Si film is deposited on a fused silica (FS) substrate followed by 200 nm Ni film forming a FS/*a*-Si(400nm)/Ni(200nm) (FSN) stack. The depositions were carried out at different substrate temperatures from 200 to 400 °C. After the deposition is completed, the films were annealed in the vacuum, in the same chamber without removing the vacuum for 30 min at the deposition temperature. The second growth process involves the deposition of 200 nm Ni films on fused silica (FS) substrates followed by deposition of 400 nm amorphous Si films at

All films were deposited by the electron beam evaporation technique. The starting materials were granular pure silicon powder (99.999%) and nickel powder (99.99% pure). The vacuum chamber was evacuated using a diffusion–rotary pump combination equipped with a liquid nitrogen trap. A pressure 5 × 10−6 Torr was maintained throughout the depositions. In all cases the substrate to source distance was kept constant at 10 cm. The thickness of the film is measured after deposition using a surface profilometer (model XP-1 Ambios Tech., USA). Xray diffraction patterns were recorded on a powder x-ray diffractometer (CPS120 of Inel, France) machine equipped with a Co K*α* = 0.178896nm and gas phase position sensitive detector. XAFS measurements on the FNS stacks were performed at the Ni K-edge at 8333 eV, in fluorescence mode at two different incidence geometries. The two incident angles are 2° and total reflection (TR) geometries. In total reflection geometry only few nanometers of top layer are monitored (10 to 20 nm). At 2 degree incident angle, the entire thickness of the film is under focus. The radiation source was the European Synchrotron Radiation Facility (ESRF, Grenoble, France) and XAS measurements were performed at the BM08 (GILDA) beamline, with an average storage ring current of 180 mA. Data are collected in uorescence

Figure 4 shows the X-ray diffraction pattern (XRD) of FS/ *a*-Si / Ni (FSN) stack deposited at different substrate temperatures. The FSN stack deposited at 200 and 300 °C, shows similar XRD pattern. These XRD patterns, indexed according to the PCPDF file no –701849 and refined using PowderX software, are identified as belonging to face centred cubic (FCC) structure of metallic Nickel. The refined FCC unit cell parameter is *a=*3.53Å. No change in the unit cell parameter is observed for the samples deposited at 200 and 300 °C. In the FSN stack deposited at 400 °C, it is observed that, a new set of XRD peaks are grown along with the metallic Ni XRD peaks. These new set of peaks belong to the face centre cubic structure of NiSi2 indexed according to the PCPDF file no-652974. The FCC unit cell parameter of metallic Ni calculated using (111) peak is found to be 3.52 Å and that of NiSi2 using (111) peak is 5.36 Å. There is no change in the unit cell parameter of metallic Ni for the samples

work on Nickel and Chromium induced crystallization of amorphous Si films.

200 °C substrate temperature, forming a FS/Ni(200nm)/a-Si(400nm) (FNS) stack.

**4. Nickel and Chromium induced crystallization** 

mode using a 13- element hyper pure Ge detector.

**4.1 Nickel induced crystallization** 

Fig. 2. X-ray diffraction pattern of (*i*) BSG/Sn film annealed at 500 °C for 15 hours. (*ii*) asdeposited BSG/Sn/a-Si stack and BSG/Sn/a-Si stack annealed at (*iii*) 300, (*iv*) 400 and (*v*) 500 °C along with the expanded and smoothened (111) peak of *c*-Si for BSG/Sn/a-Si stack, deconvoluted to show the presence of SnO2 (Mohiddon et al., 2012).

Fig. 3. Variation in crystallite size of Si and Sn phases as a function of annealing temperature (Mohiddon et al., 2012).

Fig. 2. X-ray diffraction pattern of (*i*) BSG/Sn film annealed at 500 °C for 15 hours. (*ii*) asdeposited BSG/Sn/a-Si stack and BSG/Sn/a-Si stack annealed at (*iii*) 300, (*iv*) 400 and (*v*) 500 °C along with the expanded and smoothened (111) peak of *c*-Si for BSG/Sn/a-Si stack,

Fig. 3. Variation in crystallite size of Si and Sn phases as a function of annealing temperature

deconvoluted to show the presence of SnO2 (Mohiddon et al., 2012).

(Mohiddon et al., 2012).

## **4. Nickel and Chromium induced crystallization**

The second group of metals used for MIC, reacts with silicon and forms silicides. e.g.: Ni, Cr, Pd, Mo etc. A very small concentration of metals contamination crystallizes large part of *a*-Si and hence they are highly useful in electronic applications. In this section we review our work on Nickel and Chromium induced crystallization of amorphous Si films.

## **4.1 Nickel induced crystallization**

Nickel induced MIC is one of best representatives of this group. To investigate Nickel based MIC, two types of experiments were carried out. In the first case, a 400 nm *a*-Si film is deposited on a fused silica (FS) substrate followed by 200 nm Ni film forming a FS/*a*-Si(400nm)/Ni(200nm) (FSN) stack. The depositions were carried out at different substrate temperatures from 200 to 400 °C. After the deposition is completed, the films were annealed in the vacuum, in the same chamber without removing the vacuum for 30 min at the deposition temperature. The second growth process involves the deposition of 200 nm Ni films on fused silica (FS) substrates followed by deposition of 400 nm amorphous Si films at 200 °C substrate temperature, forming a FS/Ni(200nm)/a-Si(400nm) (FNS) stack.

All films were deposited by the electron beam evaporation technique. The starting materials were granular pure silicon powder (99.999%) and nickel powder (99.99% pure). The vacuum chamber was evacuated using a diffusion–rotary pump combination equipped with a liquid nitrogen trap. A pressure 5 × 10−6 Torr was maintained throughout the depositions. In all cases the substrate to source distance was kept constant at 10 cm. The thickness of the film is measured after deposition using a surface profilometer (model XP-1 Ambios Tech., USA). Xray diffraction patterns were recorded on a powder x-ray diffractometer (CPS120 of Inel, France) machine equipped with a Co K*α* = 0.178896nm and gas phase position sensitive detector. XAFS measurements on the FNS stacks were performed at the Ni K-edge at 8333 eV, in fluorescence mode at two different incidence geometries. The two incident angles are 2° and total reflection (TR) geometries. In total reflection geometry only few nanometers of top layer are monitored (10 to 20 nm). At 2 degree incident angle, the entire thickness of the film is under focus. The radiation source was the European Synchrotron Radiation Facility (ESRF, Grenoble, France) and XAS measurements were performed at the BM08 (GILDA) beamline, with an average storage ring current of 180 mA. Data are collected in uorescence mode using a 13- element hyper pure Ge detector.

Figure 4 shows the X-ray diffraction pattern (XRD) of FS/ *a*-Si / Ni (FSN) stack deposited at different substrate temperatures. The FSN stack deposited at 200 and 300 °C, shows similar XRD pattern. These XRD patterns, indexed according to the PCPDF file no –701849 and refined using PowderX software, are identified as belonging to face centred cubic (FCC) structure of metallic Nickel. The refined FCC unit cell parameter is *a=*3.53Å. No change in the unit cell parameter is observed for the samples deposited at 200 and 300 °C. In the FSN stack deposited at 400 °C, it is observed that, a new set of XRD peaks are grown along with the metallic Ni XRD peaks. These new set of peaks belong to the face centre cubic structure of NiSi2 indexed according to the PCPDF file no-652974. The FCC unit cell parameter of metallic Ni calculated using (111) peak is found to be 3.52 Å and that of NiSi2 using (111) peak is 5.36 Å. There is no change in the unit cell parameter of metallic Ni for the samples

Metal Induced Crystallization 471

presented above. Thus the EXAFS work carried on the FNS spectra support the Ni diffusion

Fig. 5. (a) X-ray absorption spectra along with its derivative near XANES in the inset (b) F.T of EXAFS part of FS/Ni/a-Si stack deposited at 200 °C measured in total reflectance and 2

As discussed in Sec 2.3, in case of MILC first metal silicides are formed at the interface between metal and Si layers at the lower annealing temperatures. On further annealing, nucleation of c-Si on metal silicide precipitate starts, which is then followed by migration of the metal silicide precipitates throughout the a-Si thin film, resulted in crystallization of the entire a-Si film. We have carried a detailed transmission electron microscope study and found similar observation. The details of the experiment are as follows. The Ni and Si films are deposited using electron beam evaporation at 4x10-6 Torr and ambient temperature onto Si (311) substrates. The starting materials were granular pure silicon powder (99.999%) and nickel powder (99.99% pure). The (311) oriented Si wafer was well cleaned and subjected to heat treatment of 1000 °C for 2 hours in normal furnace atmosphere to form a thick SiO2 layer, before the deposition is started. The SiO2 layer is expected to act as a barrier to stop the diffusion of Ni atom into the Si substrate. X-ray diffraction patterns were recorded in grazing incidence of 0.5 degree on a powder x-ray diffractometer (CPS120 of Inel, France) machine. A 50 nm film of Ni was deposited on the *c*-Si/SiO2 substrate, followed by the deposition of the 400 nm Si film without breaking vacuum in the deposition chamber. The thickness of the films was measured after deposition using a surface profilometer (model XP-1 Ambios Tech., USA). The films were subsequently annealed in a furnace (in air) at 600 °C for 1 hr. Transmission electron micrographs were obtained by a Tecnai 20 G2 STwin, FEI electron microscope, operated at 200 kV. Electron diffraction patterns (EDPs) were recorded with a Gatan CCD camera. A 10 nm gold film deposited on the grid was used for purpose of camera length calibration. The samples for TEM measurement were prepared by scratching

degree along with metallic Ni reference compound.

**4.2 Ni induced Lateral Crystallization (NILC)** 

the film and transferring it on to the grid.

assisted MIC mechanism.

deposited at 300 and 400 °C. The small change of 0.01Å is considered within the range of instrumental measurement error. The disagreement of our result with that of I. H. Hong et al. (Hong et al., 2006) work is expected to be due to the ultra high vacuum pressure deposition conditions, which may facilitate the reaction of the diffused metallic Ni atoms at lower temperature to form different silicide phases.

Fig. 4. XRD pattern of FS/Si/Ni stack deposited at different substrate temperature

Figure 5(a) compares the experimental XAFS spectrum of Ni K edge in the FNS stack, which were measured in two different incident angles (TR and 2 degree), along with the reference metal Ni foil spectra. The first derivative near the XANES region of the spectra is presented in the inset of the Fig. 5(a). The observation of the inset figure shows that the XANES spectra of FNS stack deposited at 200 °C and measured in TR and 2 degree has a resemblance with that of the metallic Ni reference spectra. The qualitative analysis of the XAFS spectra shows that the nature of Ni impurity measured in TR and 2 degree are similar and has close resemblance with that of the metallic Ni. Thus at 200 °C annealing temperature the metallic Ni, which is deposited at the bottom, is diffused through 400 nm a-Si layer. The similar observation is found in the EXAFS part of the spectra. Figure 5 (b) compares the Fourier Transform (FT) of the EXAFS signal at the K edge of Ni presented for the FNS stack deposited at 200 °C and measured in two different incident angles, along with the metallic Ni. It shows that the FNS stack deposited at 200 °C and measured in two different incident angles has the features of metallic Ni. The detailed quantitative analyses were carried by FEFF8 and FEFFIT theoretical code and are presented elsewhere (Mohiddon et al., 2011). The conclusions of the work are in good agreement with that of the qualitative analysis

deposited at 300 and 400 °C. The small change of 0.01Å is considered within the range of instrumental measurement error. The disagreement of our result with that of I. H. Hong et al. (Hong et al., 2006) work is expected to be due to the ultra high vacuum pressure deposition conditions, which may facilitate the reaction of the diffused metallic Ni atoms at

> **(111)**

**^**

**\***

**(220)**

**^**

**(111)**

**(200)**

**\***

**\***

**(311)**

**NiSi2 ^ Ni metallic**

**400 C**

**300 C**

**200 C**

**^**

**^**

**(200)**

**30 40 50 60 70**

Figure 5(a) compares the experimental XAFS spectrum of Ni K edge in the FNS stack, which were measured in two different incident angles (TR and 2 degree), along with the reference metal Ni foil spectra. The first derivative near the XANES region of the spectra is presented in the inset of the Fig. 5(a). The observation of the inset figure shows that the XANES spectra of FNS stack deposited at 200 °C and measured in TR and 2 degree has a resemblance with that of the metallic Ni reference spectra. The qualitative analysis of the XAFS spectra shows that the nature of Ni impurity measured in TR and 2 degree are similar and has close resemblance with that of the metallic Ni. Thus at 200 °C annealing temperature the metallic Ni, which is deposited at the bottom, is diffused through 400 nm a-Si layer. The similar observation is found in the EXAFS part of the spectra. Figure 5 (b) compares the Fourier Transform (FT) of the EXAFS signal at the K edge of Ni presented for the FNS stack deposited at 200 °C and measured in two different incident angles, along with the metallic Ni. It shows that the FNS stack deposited at 200 °C and measured in two different incident angles has the features of metallic Ni. The detailed quantitative analyses were carried by FEFF8 and FEFFIT theoretical code and are presented elsewhere (Mohiddon et al., 2011). The conclusions of the work are in good agreement with that of the qualitative analysis

Fig. 4. XRD pattern of FS/Si/Ni stack deposited at different substrate temperature

**2 (degrees)**

lower temperature to form different silicide phases.

**(111)**

**\***

**Intensity (arb. units)**

presented above. Thus the EXAFS work carried on the FNS spectra support the Ni diffusion assisted MIC mechanism.

Fig. 5. (a) X-ray absorption spectra along with its derivative near XANES in the inset (b) F.T of EXAFS part of FS/Ni/a-Si stack deposited at 200 °C measured in total reflectance and 2 degree along with metallic Ni reference compound.

### **4.2 Ni induced Lateral Crystallization (NILC)**

As discussed in Sec 2.3, in case of MILC first metal silicides are formed at the interface between metal and Si layers at the lower annealing temperatures. On further annealing, nucleation of c-Si on metal silicide precipitate starts, which is then followed by migration of the metal silicide precipitates throughout the a-Si thin film, resulted in crystallization of the entire a-Si film. We have carried a detailed transmission electron microscope study and found similar observation. The details of the experiment are as follows. The Ni and Si films are deposited using electron beam evaporation at 4x10-6 Torr and ambient temperature onto Si (311) substrates. The starting materials were granular pure silicon powder (99.999%) and nickel powder (99.99% pure). The (311) oriented Si wafer was well cleaned and subjected to heat treatment of 1000 °C for 2 hours in normal furnace atmosphere to form a thick SiO2 layer, before the deposition is started. The SiO2 layer is expected to act as a barrier to stop the diffusion of Ni atom into the Si substrate. X-ray diffraction patterns were recorded in grazing incidence of 0.5 degree on a powder x-ray diffractometer (CPS120 of Inel, France) machine. A 50 nm film of Ni was deposited on the *c*-Si/SiO2 substrate, followed by the deposition of the 400 nm Si film without breaking vacuum in the deposition chamber. The thickness of the films was measured after deposition using a surface profilometer (model XP-1 Ambios Tech., USA). The films were subsequently annealed in a furnace (in air) at 600 °C for 1 hr. Transmission electron micrographs were obtained by a Tecnai 20 G2 STwin, FEI electron microscope, operated at 200 kV. Electron diffraction patterns (EDPs) were recorded with a Gatan CCD camera. A 10 nm gold film deposited on the grid was used for purpose of camera length calibration. The samples for TEM measurement were prepared by scratching the film and transferring it on to the grid.

Metal Induced Crystallization 473

Fig. 6. Selected Bright field image of Ni/Si film annealed at 600 °C for 1 hr along with the

Fig. 7. STEM image of Ni/Si film annealed at 600 °C for 1 hr, selected the same region as

that selected for bright field image in Fig. 6 (Mohiddon et al., 2012).

diffraction pattern in the dark and bright regions (Mohiddon et al., 2012).

Figure 6 is a bright-field transmission electron micrograph obtained from a region of the Ni/Si thin film crystallized by the MIC process. The typical microstructure shown in Fig. 6 consists of dark region with an irregular shape. The microstructure consists of dark dendrite like nanowire structures spread over the bright looking matrix. To find the nature of nanostructures at two different regions of the sample, selected area electron diffraction (SAED) patterns were recorded at both the dark and bright regions of the sample. A typical diffraction pattern from the dark part of the image is shown in the inset of the Fig. 6. The diffraction pattern consists mainly of bright diffraction spots radially positioned on rings. The diffraction spots arise from the crystalline part of the sample, whereas the ring originates from the polycrystalline parts of the sample. The inset of the Fig 6 shows that, the diffraction spots A, B and C lie on the polycrystalline ring of radius 0.314 nm, which is indexed to the (111) plane of the Silicon diamond cubic phase. The spot D has a *d*=0.185 nm which can be indexed as belonging to the (2 2 0) plane of the diamond cubic system [JCPDS-892955]. However, an alternative assignment is possible for the same diffraction spots (and the higher order diffraction spots), because they can be equally assigned to the face centred cubic *c*-NiSi2, with *d* values of 0.312 nm, 0.191 nm for (111) and (2 2 0) planes, respectively [JCPDS-652974]. Hence, the diffraction taken in dark region of the sample shown in Fig 6 may belong either to diamond cubic Si or to face centre cubic NiSi2.

The diffraction pattern from the bright area region is shown in other inset of Fig.6. Even at the higher magnifications, the microstructure of this bright region is seen to contain dark dendritic lines. It was thus inpossible to eliminate the presence of the darker regions. Hence, the diffraction pattern shown in second inset of Fig 6 has a combined effect, with a small contribution from dark nanostructures and large contribution from the bright area. The diffraction pattern is slightly different from that of dark region diffraction pattern. It consists of one diffuse diffraction rings with d=0.314 nm and a bright diffraction spots with d = 0.192 nm (indicated by E). There are, however, no diffraction spots disposed on the ring, thus suggesting that the dark region is better crystallized compared to bright region although both the regions show similar diffraction patterns.

To exactly identify the material(s) that are contributing to the diffraction from the bright and dark regions of the samples, Z (atomic number)-contrast scanning transmission electron microscopy (STEM) study was carried out. Fig. 7 is a Z-contrast STEM image of the same selected area that is shown in Fig. 6. The change in contrast in the image is due to change in Z value of Ni (Z= 28) and Si (Z= 14). This observation suggests that the central high contrast region is due to the contribution from Ni component, i.e. NiSi2 and the low contrast region is from Si. Hence, it can be inferred that the NiSi2, after its formation, starts spreading through the matrix of *a*-Si like a dendrite and leading to its crystallization. The results presented in this part of the study supports the mechanism of movement of NiSi2 dendrites in the *a*-Si matrix as the cause for crystallization. Similar movement of Cu3Si dendrites with time of annealing was reported using in-situ bright field transmission electron microscopy, but no diffraction study was reported (Russell et al.,1991). It is evident from the study of SAED and STEM, that when Ni makes contact with the Si layer at high temperature it forms NiSi2. This silicide starts diffusing as dendrites into the *a*-Si matrix and crystallizes the silicon at the nanoscale. This strongly supports the mechanism of silicide movement as the basis for crystallization of *a*-Si by the MILC process using silicide forming metals.

Figure 6 is a bright-field transmission electron micrograph obtained from a region of the Ni/Si thin film crystallized by the MIC process. The typical microstructure shown in Fig. 6 consists of dark region with an irregular shape. The microstructure consists of dark dendrite like nanowire structures spread over the bright looking matrix. To find the nature of nanostructures at two different regions of the sample, selected area electron diffraction (SAED) patterns were recorded at both the dark and bright regions of the sample. A typical diffraction pattern from the dark part of the image is shown in the inset of the Fig. 6. The diffraction pattern consists mainly of bright diffraction spots radially positioned on rings. The diffraction spots arise from the crystalline part of the sample, whereas the ring originates from the polycrystalline parts of the sample. The inset of the Fig 6 shows that, the diffraction spots A, B and C lie on the polycrystalline ring of radius 0.314 nm, which is indexed to the (111) plane of the Silicon diamond cubic phase. The spot D has a *d*=0.185 nm which can be indexed as belonging to the (2 2 0) plane of the diamond cubic system [JCPDS-892955]. However, an alternative assignment is possible for the same diffraction spots (and the higher order diffraction spots), because they can be equally assigned to the face centred cubic *c*-NiSi2, with *d* values of 0.312 nm, 0.191 nm for (111) and (2 2 0) planes, respectively [JCPDS-652974]. Hence, the diffraction taken in dark region of the sample shown in Fig 6

The diffraction pattern from the bright area region is shown in other inset of Fig.6. Even at the higher magnifications, the microstructure of this bright region is seen to contain dark dendritic lines. It was thus inpossible to eliminate the presence of the darker regions. Hence, the diffraction pattern shown in second inset of Fig 6 has a combined effect, with a small contribution from dark nanostructures and large contribution from the bright area. The diffraction pattern is slightly different from that of dark region diffraction pattern. It consists of one diffuse diffraction rings with d=0.314 nm and a bright diffraction spots with d = 0.192 nm (indicated by E). There are, however, no diffraction spots disposed on the ring, thus suggesting that the dark region is better crystallized compared to bright region although

To exactly identify the material(s) that are contributing to the diffraction from the bright and dark regions of the samples, Z (atomic number)-contrast scanning transmission electron microscopy (STEM) study was carried out. Fig. 7 is a Z-contrast STEM image of the same selected area that is shown in Fig. 6. The change in contrast in the image is due to change in Z value of Ni (Z= 28) and Si (Z= 14). This observation suggests that the central high contrast region is due to the contribution from Ni component, i.e. NiSi2 and the low contrast region is from Si. Hence, it can be inferred that the NiSi2, after its formation, starts spreading through the matrix of *a*-Si like a dendrite and leading to its crystallization. The results presented in this part of the study supports the mechanism of movement of NiSi2 dendrites in the *a*-Si matrix as the cause for crystallization. Similar movement of Cu3Si dendrites with time of annealing was reported using in-situ bright field transmission electron microscopy, but no diffraction study was reported (Russell et al.,1991). It is evident from the study of SAED and STEM, that when Ni makes contact with the Si layer at high temperature it forms NiSi2. This silicide starts diffusing as dendrites into the *a*-Si matrix and crystallizes the silicon at the nanoscale. This strongly supports the mechanism of silicide movement as the basis for

may belong either to diamond cubic Si or to face centre cubic NiSi2.

crystallization of *a*-Si by the MILC process using silicide forming metals.

both the regions show similar diffraction patterns.

Fig. 6. Selected Bright field image of Ni/Si film annealed at 600 °C for 1 hr along with the diffraction pattern in the dark and bright regions (Mohiddon et al., 2012).

Fig. 7. STEM image of Ni/Si film annealed at 600 °C for 1 hr, selected the same region as that selected for bright field image in Fig. 6 (Mohiddon et al., 2012).

Metal Induced Crystallization 475

the [1 2 13] zone axis. These diffraction spots were diffracted from a grain which has a hexagonal structure and oriented in the [1 2 13] direction and the spots A, B and C were diffracted from the (1 0 0), (l 1 1), and (O 1 1) planes, respectively. The corresponding lattice parameters were calculated as *a*=0.296nm, *c*=0.629nm with *c*/*a* ratio of 2.13. The angle calculated between these planes using *c*/*a* value is found to be exactly matching with the measured values suggesting that Si crystallizes in the hexagonal structure. For diamond cubic Si crystals the interplanar distances of the lattice planes should be 0.33 1 nm, 0.33 1 nm and 0.331 nm, respectively and the angles between transmitted beam and the spots will be 60º, 60º and 60º, with the diffraction pattern showing the expected six fold symmetry. But in our case the angles between the dots are AOB=65º, BOC=55º and AOC=120º respectively, clearly not cubic Si. Thus, the TEM analysis shows that the BSG/Ni/Si stack after heat treatment consists of 40 nm particles of hexagonal Silicon with *c*/*a* ratio of 2.11.

Fig. 8. Electron diffraction pattern along with indexing of selected w-Si part of BSG/Ni/Si stack, inset shows the bright field image of the above diffraction area (Mohiddon et al., 2011).

Additional evidence for the formation of *w*-Si is provided in the form of Raman spectra of the films. It has earlier been used to characterize porous Si (Sui et al., 1992) and Si nanostructures (Kozlowski et al., 1991). The Raman shift and the shape of the Raman peak yield information on the degree of crystallinity achieved in Si by MIC process. Fig. 9 shows Raman spectra from the Silicon film annealed at 550 ºC for 1 hrs. A sharp peak at 504 cm-1 and weak broad peaks around 300 and 950 cm-1, are associated with optical phonons, two transverse acoustic (2TA) phonons, and two transverse optical (2TO) phonons of *w-*Si (Mohiddon & Kirshna, 2011) respectively. The asymmetric nature of the peak at 504 cm-1 can

#### **4.3 Stabilization of wurtzite Si**

Silicon (Si) usually crystallizes in the cubic diamond structure with fourfold coordinated symmetry. Numerous high-pressure experiments have been performed, revealing no less than 12 different polymorphs of silicon between the well characterized diamond cubic phase and theoretically intractable amorphous phase. With the release of pressure several metastable phases are observed (Wu et al., 2000). For example, a nonmetal to metal transition of cubic-diamond phase, which occur by changing pressure from ambient atmosphere pressure to 10–13 GPa to form a β-tin structure, which further transforms to rhombohedral phase on slow pressure release. The latter transforms reversibly at a pressure of 2 GPa to a body-centered-cubic phase (bc8). Hexagonal-wurtzite silicon (*w-*Si) can be formed by heating the bc8 phase to above 200 °C, or directly from the cubic-diamond phase in the presence of shear stresses at twin intersections, or a nonhydrostatic stress of 8 GPa in indentation experiments (Kailer et al., 1997). *w-*Si material is rarely studied by spectroscopic measurements, because it cannot be obtained in a stable phase. Zhang et al. (Zhang et al., 1999) produced stable *w-*Si phase by laser ablation. Its identification by electron diffraction has been confirmed by micro Raman spectroscopy. Bandet et al. (Bandet et al., 2002) deposited *w-*Si during elaboration of SiO2 thin films and reported that oxygen seems to play a crucial role in the stabilization of the uncommon metastable structure. Kim et al. (Kim & Lee, 1996) reported that the micrometer-sized diamond cubic silicon (*c-*Si) crystals contain minority part of hexagonal silicon, when *a*-Si films crystallized by a pulsed laser. In the present work we have crystallized the silicon film with a metal contact and found that silicon nanocrystals are in wurtzite phase. The experimental details are as follows.

Nickel and Silicon films were deposited by electron beam evaporation on to BoroSilicate Glass (BSG) substrates in high vacuum of the order of 10-6 Torr. First a 50 nm thin nickel blanket bottom layer was deposited on BSG substrates maintained at ambient temperature. This is followed by the deposition of a 700 nm thick Si film, without breaking vacuum, to form a BSG/Ni/Si stack. The thickness of the films is measured after deposition using a surface profilometer (model XP-1 Ambios Tech., USA). The films were annealed in a furnace atmosphere (in air) at 550 °C for 1 hr. Transmission electron micrographs were obtained by a Tecnai 20 G2 STwin, FEI electron microscope, operated at 200 kV. Electron diffraction patterns (EDPs) were recorded with a Gatan CCD camera. A 10 nm gold film deposited on the grid was used for camera length calibration purposes. The Raman spectra were recorded in air using an Nd-YAG 532nm laser in the back scattering geometry in a CRM spectrometer equipped with a confocal microscope and 100× objective (1 *μ*m diameter focal spot size) with a CCD detector (model alpha 300 of WiTec Germany). The phase content with in the samples was investigated in a spectral region 200–1500 cm−1.

Figure 8 shows the electron diffraction pattern along with a bright-field transmission electron micrograph obtained from a region in the silicon thin film crystallized by a MIC process in the inset. The typical microstructure shown in inset of Fig. 8 consists of spherical particles with approximate diameter of 40 nm. The diffraction spots can be identified as belonging to well defined orientations. The calculated interplanar distances of the lattice planes at different spots of the diffraction pattern assigned by symbols A, B and C in Fig. 8 are A=0.256nm, B=0.237nm and C=0.237nm respectively. From the calculated interplanar distances the diffraction spots were indexed as (1 0 0), (l 1 1), and (O 1 1), which belong to

Silicon (Si) usually crystallizes in the cubic diamond structure with fourfold coordinated symmetry. Numerous high-pressure experiments have been performed, revealing no less than 12 different polymorphs of silicon between the well characterized diamond cubic phase and theoretically intractable amorphous phase. With the release of pressure several metastable phases are observed (Wu et al., 2000). For example, a nonmetal to metal transition of cubic-diamond phase, which occur by changing pressure from ambient atmosphere pressure to 10–13 GPa to form a β-tin structure, which further transforms to rhombohedral phase on slow pressure release. The latter transforms reversibly at a pressure of 2 GPa to a body-centered-cubic phase (bc8). Hexagonal-wurtzite silicon (*w-*Si) can be formed by heating the bc8 phase to above 200 °C, or directly from the cubic-diamond phase in the presence of shear stresses at twin intersections, or a nonhydrostatic stress of 8 GPa in indentation experiments (Kailer et al., 1997). *w-*Si material is rarely studied by spectroscopic measurements, because it cannot be obtained in a stable phase. Zhang et al. (Zhang et al., 1999) produced stable *w-*Si phase by laser ablation. Its identification by electron diffraction has been confirmed by micro Raman spectroscopy. Bandet et al. (Bandet et al., 2002) deposited *w-*Si during elaboration of SiO2 thin films and reported that oxygen seems to play a crucial role in the stabilization of the uncommon metastable structure. Kim et al. (Kim & Lee, 1996) reported that the micrometer-sized diamond cubic silicon (*c-*Si) crystals contain minority part of hexagonal silicon, when *a*-Si films crystallized by a pulsed laser. In the present work we have crystallized the silicon film with a metal contact and found that

silicon nanocrystals are in wurtzite phase. The experimental details are as follows.

samples was investigated in a spectral region 200–1500 cm−1.

Nickel and Silicon films were deposited by electron beam evaporation on to BoroSilicate Glass (BSG) substrates in high vacuum of the order of 10-6 Torr. First a 50 nm thin nickel blanket bottom layer was deposited on BSG substrates maintained at ambient temperature. This is followed by the deposition of a 700 nm thick Si film, without breaking vacuum, to form a BSG/Ni/Si stack. The thickness of the films is measured after deposition using a surface profilometer (model XP-1 Ambios Tech., USA). The films were annealed in a furnace atmosphere (in air) at 550 °C for 1 hr. Transmission electron micrographs were obtained by a Tecnai 20 G2 STwin, FEI electron microscope, operated at 200 kV. Electron diffraction patterns (EDPs) were recorded with a Gatan CCD camera. A 10 nm gold film deposited on the grid was used for camera length calibration purposes. The Raman spectra were recorded in air using an Nd-YAG 532nm laser in the back scattering geometry in a CRM spectrometer equipped with a confocal microscope and 100× objective (1 *μ*m diameter focal spot size) with a CCD detector (model alpha 300 of WiTec Germany). The phase content with in the

Figure 8 shows the electron diffraction pattern along with a bright-field transmission electron micrograph obtained from a region in the silicon thin film crystallized by a MIC process in the inset. The typical microstructure shown in inset of Fig. 8 consists of spherical particles with approximate diameter of 40 nm. The diffraction spots can be identified as belonging to well defined orientations. The calculated interplanar distances of the lattice planes at different spots of the diffraction pattern assigned by symbols A, B and C in Fig. 8 are A=0.256nm, B=0.237nm and C=0.237nm respectively. From the calculated interplanar distances the diffraction spots were indexed as (1 0 0), (l 1 1), and (O 1 1), which belong to

**4.3 Stabilization of wurtzite Si** 

the [1 2 13] zone axis. These diffraction spots were diffracted from a grain which has a hexagonal structure and oriented in the [1 2 13] direction and the spots A, B and C were diffracted from the (1 0 0), (l 1 1), and (O 1 1) planes, respectively. The corresponding lattice parameters were calculated as *a*=0.296nm, *c*=0.629nm with *c*/*a* ratio of 2.13. The angle calculated between these planes using *c*/*a* value is found to be exactly matching with the measured values suggesting that Si crystallizes in the hexagonal structure. For diamond cubic Si crystals the interplanar distances of the lattice planes should be 0.33 1 nm, 0.33 1 nm and 0.331 nm, respectively and the angles between transmitted beam and the spots will be 60º, 60º and 60º, with the diffraction pattern showing the expected six fold symmetry. But in our case the angles between the dots are AOB=65º, BOC=55º and AOC=120º respectively, clearly not cubic Si. Thus, the TEM analysis shows that the BSG/Ni/Si stack after heat treatment consists of 40 nm particles of hexagonal Silicon with *c*/*a* ratio of 2.11.

Fig. 8. Electron diffraction pattern along with indexing of selected w-Si part of BSG/Ni/Si stack, inset shows the bright field image of the above diffraction area (Mohiddon et al., 2011).

Additional evidence for the formation of *w*-Si is provided in the form of Raman spectra of the films. It has earlier been used to characterize porous Si (Sui et al., 1992) and Si nanostructures (Kozlowski et al., 1991). The Raman shift and the shape of the Raman peak yield information on the degree of crystallinity achieved in Si by MIC process. Fig. 9 shows Raman spectra from the Silicon film annealed at 550 ºC for 1 hrs. A sharp peak at 504 cm-1 and weak broad peaks around 300 and 950 cm-1, are associated with optical phonons, two transverse acoustic (2TA) phonons, and two transverse optical (2TO) phonons of *w-*Si (Mohiddon & Kirshna, 2011) respectively. The asymmetric nature of the peak at 504 cm-1 can

Metal Induced Crystallization 477

ambient substrate temperature, remains in metallic form up to an annealing temperature of 500 °C (Mohiddon et al.,2012). As an extension of this work we deposited Cr/Si stacks at higher substrate temperature. The details of the experiment are as follows. A Cr(200nm)/Si(400nm) stack was deposited over a fused silica substrate at 400 °C substrate temperature and a similar stack of Cr(50nm)/Si(400nm) was deposited at ambient substrate temperature by electron beam evaporation. We have carried out a detailed EXAFS analysis on these set of samples (Mohiddon et al., 2012) and concluded that, in case of film deposited at ambient substrate temperature, major part of Cr is in metallic form apart from being oxidized. In the case of the stack deposited at 400 °C, the entire Cr is turned into either Cr2O3 or CrSi2. After the EXAFS measurements were completed, these samples were heat treated at different temperatures. Samples were annealed at 600, 650 and 700 °C for 1 hr. in excess Nitrogen flux. At each heat treatment, the samples were cooled to room temperature and then the Raman spectra were collected at three to four different position of the each sample. The details of the Raman measurements are discussed in the section 4.3 of the chapter.

Figure 10(a) and 10(b) compares the Raman spectra of the FS/Cr(200nm)/Si(400nm) stack deposited at 400 °C substrate temperature and FS/Cr(50nm)/Si(400nm) stack deposited ambient substrate temperature then annealed at different temperature respectively. Fig. 10(a) shows that the as deposited stack contains a broad diffuse peak around 480 cm-1 that can be assigned to the a-Si phase. After the heat treatment at 600 °C, a sharp peak, around 520 cm-1, appears that can be assigned to c-Si. The intensity of the c-Si peak increases with annealing temperature up to 650 °C. No such change in the Raman spectra is observed in the Fig. 10 (b). By combining the EXAFS and Raman results we conclude that the FCS stack deposited at 400°C possesses CrSi2 in the as deposited condition. When this stack is further heat treated the CrSi2 acts as a seed for the Si crystallization. In the FCS stack deposited at minimum substrate temperature, no CrSi2 is observed from the EXAFS analysis. Thus even

after heat treatment up to 700 C, no Si crystallization is observed.

Fig. 10. Raman spectra of (a) FS/Cr(200nm)/Si(400nm) stack deposited at 400 C; (b) FS/Cr(50nm)/Si(400nm) stack deposited at minimum substrate temperature

be attributed to the presence of small amounts of *a*-Si, which has its broad peak at 480 cm-1. In general, when Si crystallizes in cubic diamond structure the Raman peak occurs at 520 cm-1. The broad peaks around 283 cm-1 in our observation is may belong to NiSi, NiSi2 or to wurtzite TA mode. However, this observation implies that, at 550 ºC nickel silicides are present in the samples. Tan et al. (Tan et al., 1981) reported a stress-induced metastable form of hexagonal silicon, which has the wurtzite structure with the c/a ratio close to 1.63, the interplanar distances and angular relationships obtained from our experiment were compared with those of silicon which has the wurtzite structure. However, for silicon with the wurtzite structure, no exact coincidence was obtained. Kim et al. reported a similar observation and suggested a hexagonal phase of Si with *a*=0.382nm, *c*=1.024nm and *c*/*a* ratio of 2.68 (Kim & Lee, 1996). Parson and Hoelke suggested similar hexagonal structure for Ge with *c*/*a* ratio of 2.17 (Parsons & Hoelke, 1983). Zhang *et al*. (Zhang et al., 1994) have reported wurzite silicon by laser ablation, and they observed Raman peaks at 516 and 518 cm*−*1 due to the hexagonal silicon. Kumar et al. observed *w*-Si in chromium induced Si crystallization evidenced by a peak at 495 cm-1 (Kumar & Krishna, 2008). Kailer et al. (Kailer et al., 1997) performed indentation Raman investigations and reported that, shear deformation leads to *w*-Si phase formation. They also stated that high pressure phases leads to the formation of *w*-Si at moderate temperatures and to the reversal to the original diamond structure at higher temperatures.

Fig. 9. Raman spectra of BSG/Ni/Si stack annealed at 550 ºC for 1 hr (Mohiddon et al., 2011).

#### **4.4 Chromium induced crystallization**

The search for new metal in the MIC process to reach the best device needs is still continuing. Chromium is another member of the silicide forming group. Chromium induced MIC has attracted our attention due to its ability to crystallize the *a*-Si into wurtzite phase (Kumar et al., 2008). Our recent work on Cr induced MIC has revealed that Cr deposited at

be attributed to the presence of small amounts of *a*-Si, which has its broad peak at 480 cm-1. In general, when Si crystallizes in cubic diamond structure the Raman peak occurs at 520 cm-1. The broad peaks around 283 cm-1 in our observation is may belong to NiSi, NiSi2 or to wurtzite TA mode. However, this observation implies that, at 550 ºC nickel silicides are present in the samples. Tan et al. (Tan et al., 1981) reported a stress-induced metastable form of hexagonal silicon, which has the wurtzite structure with the c/a ratio close to 1.63, the interplanar distances and angular relationships obtained from our experiment were compared with those of silicon which has the wurtzite structure. However, for silicon with the wurtzite structure, no exact coincidence was obtained. Kim et al. reported a similar observation and suggested a hexagonal phase of Si with *a*=0.382nm, *c*=1.024nm and *c*/*a* ratio of 2.68 (Kim & Lee, 1996). Parson and Hoelke suggested similar hexagonal structure for Ge with *c*/*a* ratio of 2.17 (Parsons & Hoelke, 1983). Zhang *et al*. (Zhang et al., 1994) have reported wurzite silicon by laser ablation, and they observed Raman peaks at 516 and 518 cm*−*1 due to the hexagonal silicon. Kumar et al. observed *w*-Si in chromium induced Si crystallization evidenced by a peak at 495 cm-1 (Kumar & Krishna, 2008). Kailer et al. (Kailer et al., 1997) performed indentation Raman investigations and reported that, shear deformation leads to *w*-Si phase formation. They also stated that high pressure phases leads to the formation of *w*-Si at moderate temperatures and to the reversal to the original

Fig. 9. Raman spectra of BSG/Ni/Si stack annealed at 550 ºC for 1 hr (Mohiddon et al., 2011).

The search for new metal in the MIC process to reach the best device needs is still continuing. Chromium is another member of the silicide forming group. Chromium induced MIC has attracted our attention due to its ability to crystallize the *a*-Si into wurtzite phase (Kumar et al., 2008). Our recent work on Cr induced MIC has revealed that Cr deposited at

diamond structure at higher temperatures.

**4.4 Chromium induced crystallization** 

ambient substrate temperature, remains in metallic form up to an annealing temperature of 500 °C (Mohiddon et al.,2012). As an extension of this work we deposited Cr/Si stacks at higher substrate temperature. The details of the experiment are as follows. A Cr(200nm)/Si(400nm) stack was deposited over a fused silica substrate at 400 °C substrate temperature and a similar stack of Cr(50nm)/Si(400nm) was deposited at ambient substrate temperature by electron beam evaporation. We have carried out a detailed EXAFS analysis on these set of samples (Mohiddon et al., 2012) and concluded that, in case of film deposited at ambient substrate temperature, major part of Cr is in metallic form apart from being oxidized. In the case of the stack deposited at 400 °C, the entire Cr is turned into either Cr2O3 or CrSi2. After the EXAFS measurements were completed, these samples were heat treated at different temperatures. Samples were annealed at 600, 650 and 700 °C for 1 hr. in excess Nitrogen flux. At each heat treatment, the samples were cooled to room temperature and then the Raman spectra were collected at three to four different position of the each sample. The details of the Raman measurements are discussed in the section 4.3 of the chapter.

Figure 10(a) and 10(b) compares the Raman spectra of the FS/Cr(200nm)/Si(400nm) stack deposited at 400 °C substrate temperature and FS/Cr(50nm)/Si(400nm) stack deposited ambient substrate temperature then annealed at different temperature respectively. Fig. 10(a) shows that the as deposited stack contains a broad diffuse peak around 480 cm-1 that can be assigned to the a-Si phase. After the heat treatment at 600 °C, a sharp peak, around 520 cm-1, appears that can be assigned to c-Si. The intensity of the c-Si peak increases with annealing temperature up to 650 °C. No such change in the Raman spectra is observed in the Fig. 10 (b). By combining the EXAFS and Raman results we conclude that the FCS stack deposited at 400°C possesses CrSi2 in the as deposited condition. When this stack is further heat treated the CrSi2 acts as a seed for the Si crystallization. In the FCS stack deposited at minimum substrate temperature, no CrSi2 is observed from the EXAFS analysis. Thus even after heat treatment up to 700 C, no Si crystallization is observed.

Fig. 10. Raman spectra of (a) FS/Cr(200nm)/Si(400nm) stack deposited at 400 C; (b) FS/Cr(50nm)/Si(400nm) stack deposited at minimum substrate temperature

Metal Induced Crystallization 479

Jin, Z., Bhat, G.A., Yeung, M., Kwok, H.S., & Wong, M. (1998). Nickel induced crystallization

Kailer, A., Gogotsi, Y.G., & Nickel, K.G. (1997). Phase transformation of silicon caused by

Kim, J.H., & Lee, J.Y. (1996). Hexagonal silicon formation by pusled laser beam annealing.

Kozlowski, F., Petrova-Koch, V., Kux, A., Stadler, W., Fleischmann, A., & Sigmund, H.

Kumar, K.U.M., Brahma, R., Krishna, M.G., Bhatnagar, A.K., & Dalba, G. (2007). An optical

Kumar, K.U.M., & Krishna, M.G., (2008). Chromium-induced nanocrystallization of a-Si thin

Lee, C.J., Lee, J.B., Chung, Y.C., & Choi D.K. (2000). Lateral crystallization of amorphous silicon films. *Jpn. J. Appl. Phys*. 39, (2000), pp. (6191-6195), ISSN 1347-4065 Leonard, R.T., & Koch, C.C. (1992). Nanoscale composites of Si/Sn and Ge/Sn synthesized

Mahan, A.H., Ahrenkiel, S.P., Schropp, R.E.I., Li, H., & Ginley, D.S. (2008). A comparison of

Mohiddon, M.A., & Krishna, M.G. (2012) Growth and optical properties of Sn-Si nanocomposite thin films *J. Material Science*, DOI: 10.1007/s10853-012-6647-0 Mohiddon, M.A., & Yadav, K. L. (2008). Effect of 90° domain on ferroelectric properties of

Mohiddon, M.A., Naidu, K.L., Dalba, G., Rocca, F., & Krishna, M.G. (2012) unpublished Mohiddon, M.A., Naidu, K.L., Dalba, G., Rocca, F., & Krishna, M.G. (2012) Cr induced

crystallization process, *J Nanopart Res* 13 (June 2011), pp. (5999–6004). Mohiddon, M.A., Krishna, M.G., Dalba, G. & Rocca, F. (2012) Transmission electron

Mohiddon, M.A., Naidu, K.L., Dalba, G., Rocca, F., & Krishna, M.G. (2011) Growth, optical,

Murray, J. L., & McAlister, A.J. (1984). Al-Si. *Bull. Alloy Phase Diagram*, 5, 1 (January 1984),

*Materials Letters* 27, 6, (August 1996), pp. (275-279), ISSN 0167-577X

138, 1, (January 1991), pp. (91-94), ISSN 0022-3093

(December 2007), pp. (496208-1-11), ISSN 1361-648X

(April 2011), pp. (2672-2677), ISSN 1573-4803

ISSN 0021-8979.

478), ISSN: 0965-9773

ISSN 1361-6463

2012), pp. (1493–1495 )1493–1495

10.1016/j.mseb.2012.05.018

pp. (74-84) ISSN 0197-0216.

of amorphous silicon thin film. *J. Appl. Phys.* 84, 1, (April 1998), pp. (194-200)

contact loading. *J. Appl. Phys.*, 81, 7, (January 1997), pp. (3057-3063), ISSN 0021-8979

(1991). Structural instability of Si-nanocrystals in a (SiH2)n. *J. Non-Cryst. Solids*, 137-

study of Ni induced crystallization of a-Si thin films. *J. Phys. Condens. Matter* 19, 49,

films into the wurtzite structure. *J. Nanomater.* 2008, (February 2008), pp. (736534-1-6)

by mechanical attrition. *Nanostructured Materials,* 1, 6, (December 1992), pp. (471-

grain nucleation and grain growth during crystallization of HWCVD and PECVD a-Si:H films. *Thin Solid Films,* 516, 5, (January 2008), pp. (529-532), ISSN 0040-6090 McCaldin, J. O., & Sankur, H. (1971). Diffusivity and solubility of Si in the Al metallization of integrated circuits. *Appl. Phys. Lett*., 19, 12, (July 1971), pp. (524-527), ISSN 0003-6951 Mohiddon, M.A., & Krishna, M.G. (2011*).* Nanocrystalline wurtzite Si-nickel silicide

composite thin films with large band gap and high resistivity. *J Mater. Sci*., 46, 8,

alkali modified SBN. *J. Phys. D: Appl. Phys.,* 41, 22, (October 2008), pp. (225406-1-7),

nanocrystallization of *a*-Si thin films: its mechanism *Physica Status Solidi C,* 9 (April

and electrical properties of silicon films produced by the metal-induced

microscopy study of Ni-Si nanocomposite films *Mate. Sci. Eng. B*, DOI:

## **5. Conclusions**

An overview of the metal induced crystallization of amorphous Silicon films has been presented. The possible mechanisms of crystallization have been discussed. Two metals Sn and Cr, which have not been used frequently for MIC, have been shown to be as attractive as Ni for metal induced crystallization of a-Si. Our results clearly show that these are very useful materials for the microelectronics industry. The crystallization occurs at temperatures ranging from 200 to 600 °C, which is much lower than the normal crystallization temperature of Si (1100 °C). The stabilization of the wurtzite form of Si has also been demonstrated.

## **6. Acknowledgements**

The authors acknowledge discussions with Prof. F. Rocca and Prof. G. Dalba of University of Trento and funding for this work through the DST-ITPAR program. Facilities provided by the CAS programme of the School of Physics and DST funded Centre for Nanotechnology at the University of Hyderabad are also acknowledged.

## **7. References**


An overview of the metal induced crystallization of amorphous Silicon films has been presented. The possible mechanisms of crystallization have been discussed. Two metals Sn and Cr, which have not been used frequently for MIC, have been shown to be as attractive as Ni for metal induced crystallization of a-Si. Our results clearly show that these are very useful materials for the microelectronics industry. The crystallization occurs at temperatures ranging from 200 to 600 °C, which is much lower than the normal crystallization temperature of Si

The authors acknowledge discussions with Prof. F. Rocca and Prof. G. Dalba of University of Trento and funding for this work through the DST-ITPAR program. Facilities provided by the CAS programme of the School of Physics and DST funded Centre for

Choi, J.H., Cheon, J.H., Kim, S.K., & Jang, J. (2005). Giant-grain silicon (GGS) and its

Bandet, J., Despax, B., & Caumont, M. (2002). Vibrational and electronic properties of

Ferri, F.A., Zanatta, A.R., & Chambouleyron, I. (2006). Metal-induced nanocrystalline

Grisenti, R., Dalba, G., Fornasini, P., Rocca, F., Kumar, K.U.M., & Krishna, M.G. (2008).

Hayzelden, C., & Batstone, J.L. (1993). Silicide formation and silicide-mediated

Hiraki, A. (1980). A model on the mechanism of room temperature interfacial intermixing

*Electrochem. Soc.* 127, 12, (December 1980), pp. (2662-2665), ISSN 1945-7111 Hong, I., Hsu, T.C., Yen, S.C., Lin, F.S., Huang, M.L., & Chen, C.H. (2006). Proc. 8th Int.

Hwang, J.D., & Lin, J.W. (2005). Nanostructural silicon films prepared by metal induced

Jeon, M., Jeong, C., & Kamisako, K. (2010). Tin induced crystallization of hydrogenated

application to stable thin film transistor. *Displays* 26, 3, (June 2005), pp. (137-142),

stabilized wurtzite like silicon. *Journal of Physics D: Applied Physics*, 35, 3, (January

structures in Ni-containing amorphous silicon thin films. *J. Appl. Phys.,* 100, 9,

XAFS study on Ni surroundings in metal induced crystallization of thin film amorphous silicon. Solid State Comm., 147, 9-10, (September 2008), pp. (401-404)

crystallization of nickel-implanted amorphous silicon thin films. J. Appl. Phys., 73,

reaction in various metal-semiconductor couples: what triggers the reaction?. *J.* 

growth using RTCVD system. *J. Electrochem. Soc.*, 152, 1, (Novermber 2005), pp.

amoprphous silico, K., Kowk, H.S., & Wang, M. (1999), The effects of extended heat treatment on Ni induced lateral crystallization of amorphous silicon thin films. *IEEE Trans. Electron Devices,* 46, 1, (January 1999), pp. (78-82), ISSN 0018-9383

(1100 °C). The stabilization of the wurtzite form of Si has also been demonstrated.

Nanotechnology at the University of Hyderabad are also acknowledged.

(November 2006), pp. (094311-094317), ISSN 0021-8979

12, (February 1993), pp. (8279-8289), ISSN 0021-8979

Conf. X-ray Microscopy, IPAP Conf. Series 7 (2006) 270.

**5. Conclusions** 

**6. Acknowledgements** 

ISSN 0141-9382

ISSN 0038-1098

(G35-G39), ISSN 1945-7111

2002), pp. (234-239), ISSN 1361-6463

**7. References** 


**18** 

*Malaysia* 

**ArF Excimer Laser Annealing of** 

**Polycrystalline Silicon Thin Film** 

*Laser Research Group, Advanced Photonic Science Institute, Faculty of Science,* 

Crystallization of amorphous silicon (a-Si) using excimer laser annealing (ELA) has been reported since 1994 by Watanabe group. It is known as the best method to fabricate a good poly-silicon because it can heat the film up to the melting point and, at the same time no thermal damage occur into the glass substrate (Carluccio et al., 1997; Matsumura and Oh, 1999). ELA technique is widely used to increase the grain size and changes the microstructure of polysilicon thin film which is the most important characteristics of excellent built-in polysilicon devices (Palani et al., (2008). The principle advantage of excimer lasers is the strong absorption of ultraviolet (UV) light in silicon beside having larger beam size and high energy density than other laser light sources (Watanabe, et al. 1994). Another major advantage of the excimer lasers is it low-temperature polysilicon annealing. Excimer laser with 308 nm wavelength for example has been reported can transform 50-nm-thin layers of amorphous silicon into high-quality polycrystalline silicon with greatly enhanced electron mobility, for use in flat-panel displays for mobile phones and flat-screen televisions (Delmdahl, 2010). In the low-temperature annealing of polysilicon, excimer lasers with UV output energies of over 1 J per pulse and output powers of 600 W are used to manufacture liquid-crystal and organic LED backplanes at a rate of 100 cm2 s–1. A new VYPER/LB750 line beam annealing system enables volume production of low-temperature polysilicon (LTPS) on large generation 6 glass panels. LTPS is the key material for high-resolution liquid crystal displays (LCDs) and organic light-

Today, the most prominent applications of excimer lasers are in semiconductor chip manufacturing. This relies on the deep-UV emission wavelength of 193 nm, with semiconductor chip manufacturing capitalizing on the optical resolution of excimer lasers. For example, high repetition rate line-narrowed excimer laser models with 10 mJ per pulse, pulse frequencies of up to 6 kHz and narrowed bandwidths of 0.35 pm are used in advanced

Ongoing miniaturization in microelectronics and the trend towards thin-film technologies demands increasing lateral resolution and selective machining. Functional structures and active layers are often only tens of nanometres thick and has to be annealed, patterned and removed in a selective manner without damaging underlying layers or substrates. Because excimer lasers provide the shortest wavelength of all laser technologies, they will continue

emitting diode (OLED) displays for smartphones, tablet PCs and TVs.

photolithography to produce computer chips with feature sizes of 45 nm.

**1. Introduction** 

Noriah Bidin and Siti Noraiza Ab Razak

*Universiti Teknologi Malaysia Johor Malaysia,* 


## **ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film**

Noriah Bidin and Siti Noraiza Ab Razak

*Laser Research Group, Advanced Photonic Science Institute, Faculty of Science, Universiti Teknologi Malaysia Johor Malaysia, Malaysia* 

## **1. Introduction**

480 Crystallization – Science and Technology

Nast, O., & Wenham, S.R. (2000). Elucidation of the layer exchange mechanism in the

Oliver, N., & Hartmann, A. J. (2000). Influence of interface and Al structure on layer

Park, S.H., Jun, S.I., Song, K.S., Kim, C.K., & Choi, D.K. (1999). Field aided lateral

Park, K.C., Song, I.H., Jeon, J.H., & Han, M.K. (2001). Excimer laser annealing effect on

Parr, A.A., Gill, K., Gardiner, D., Holyland, J., Sands, D., Brunson, K., & Carline, R. (2002). A

Parsons, J.R., & Hoelke, C.W. (1983). Crystallography Of hexagonal germanium. *Nature*, 30,

Russell, S. W., Jian, Li., & Mayer, J. W. (1991). In situ observation of fractal growth during a-

Schropp, R.E.I., & Zeman, M. (1998). *Amorphous and Microcrystalline Silicon Solar* 

Spinella, C., Lombardo, S., & Priolo, F. (1998). Crystal grain nucleation in amorphous silicon.

Sui, Z., Leong, P.P., Herman, I.P., Higashi, G.S., & Temkin, H. (1992). Raman analysis of

Tan, T.Y., Foll, H., Hu, S.M. (1981). On the diamond-cubic to hexagonal phase transformation in silicon. *Philos. Mag. A,* 44, 1, (July 1981), pp. (127-140), ISSN 1478-6443 Wang, M., & Wong, M. (2001). Characterization of an individual grain boundary in metal-

*Electron Devices,* 48, 8, (August 2001), pp. (1655-1660), ISSN 0018-9383 Wu, B.R. (2000). First- principles study on the high-pressure behavior of the zone-center modes of lonsdaleite silicon. *Phys. Rev. B*, 61, 1, (January 2000), pp. (5-8), ISSN 1550-235X Yoon, S. Y., Park, S. J., Kim, K. H., & Jang, J. (2001). Metal induced crystallization of

*J. Appl. Phys.,* 84, 10, (July 1998), pp. (5383-5415), ISSN 0021-8979

*Physics,* 88, (January 2000), pp. (124-131), ISSN 0021-8979

*Phys.* 88 (April 2000), pp. (716-724) ISSN 0021-8979

(August 2001), pp. (G563-G565), ISSN 1945-7111

5901, (1983), pp. (591-592), ISSN 1476-4687

5155), ISSN 0021-8979

2088), ISSN 0003-6951

0040-6090

ISSN 0003-6951

(3016-3019), ISSN 0021-8979

978-0-7923-8317-8, Boston.

*Technol.* 17, 1, (January 2002), pp. (47-54), ISSN 1361-6641

1999), pp. (L108-L109), ISSN 1347-4065

formation of polycrystalline silicon by aluminum induced crystallization. *J. Appl.* 

exchange during aluminum induced crystallization of amorphous silicon. *J. Appl.* 

crystallization of amorphous silicon thin film. *Jpn. J. Appl. Phys.*, 38, 2, (January

Nickel-induced crystallized polycrystalline silicon film. *J. Electrochem. Soc.,* 148, 10,

comparison of laser- and furnace- annealed poly-silicon structure. *Semicond. Sci.* 

Si crystallization in a Cu3Si matrix. *J. Appl. Phys.,* 70, 9, (August 1991), pp. (5153-

*Cells:Modeling, Materials and Device Technology,* Kluwer Academic Publishers, ISBN

light emitting porous silicon. *Appl. Phys. Lett.,* 60, 17, (February 1992), pp. (2086-

induced laterally crystallized polycrystalline silicon thin-film device. *IEEE Trans.* 

amorphous silicon. *Thin Solid Films,* 383, 1-2, (February 2001), pp. (34-38), ISSN

phase by laser ablation. *Appl. Phys. Lett.,* 75, 18, (September 1999), pp. (2758-2760),

frequency region of p- type porous silicon. *J. Appl. Phys.* 76, 5, (May 1994), pp.

Zener, C. (1949). Theory of growth of spherical precipitates from solid solution. *J. Appl.* 

Zhang, Y., Iqbal, Z., Vijayalakshmi, S., & Grebel, H. (1999). Stable hexagonal wurtzite silicon

Zhang, S.L., Wang, X., Ho, K., Li, J.J., Diao, P., & Cai, S. (1994). Raman spectra in a broad

*Phys.,* 20, 10, (March 1949), pp. (950-953), ISSN 0021-8979

Crystallization of amorphous silicon (a-Si) using excimer laser annealing (ELA) has been reported since 1994 by Watanabe group. It is known as the best method to fabricate a good poly-silicon because it can heat the film up to the melting point and, at the same time no thermal damage occur into the glass substrate (Carluccio et al., 1997; Matsumura and Oh, 1999). ELA technique is widely used to increase the grain size and changes the microstructure of polysilicon thin film which is the most important characteristics of excellent built-in polysilicon devices (Palani et al., (2008). The principle advantage of excimer lasers is the strong absorption of ultraviolet (UV) light in silicon beside having larger beam size and high energy density than other laser light sources (Watanabe, et al. 1994). Another major advantage of the excimer lasers is it low-temperature polysilicon annealing. Excimer laser with 308 nm wavelength for example has been reported can transform 50-nm-thin layers of amorphous silicon into high-quality polycrystalline silicon with greatly enhanced electron mobility, for use in flat-panel displays for mobile phones and flat-screen televisions (Delmdahl, 2010). In the low-temperature annealing of polysilicon, excimer lasers with UV output energies of over 1 J per pulse and output powers of 600 W are used to manufacture liquid-crystal and organic LED backplanes at a rate of 100 cm2 s–1. A new VYPER/LB750 line beam annealing system enables volume production of low-temperature polysilicon (LTPS) on large generation 6 glass panels. LTPS is the key material for high-resolution liquid crystal displays (LCDs) and organic lightemitting diode (OLED) displays for smartphones, tablet PCs and TVs.

Today, the most prominent applications of excimer lasers are in semiconductor chip manufacturing. This relies on the deep-UV emission wavelength of 193 nm, with semiconductor chip manufacturing capitalizing on the optical resolution of excimer lasers. For example, high repetition rate line-narrowed excimer laser models with 10 mJ per pulse, pulse frequencies of up to 6 kHz and narrowed bandwidths of 0.35 pm are used in advanced photolithography to produce computer chips with feature sizes of 45 nm.

Ongoing miniaturization in microelectronics and the trend towards thin-film technologies demands increasing lateral resolution and selective machining. Functional structures and active layers are often only tens of nanometres thick and has to be annealed, patterned and removed in a selective manner without damaging underlying layers or substrates. Because excimer lasers provide the shortest wavelength of all laser technologies, they will continue

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 483

Excimer lasers are widely used in high-resolution photolithography machines, one of the critical technologies required for microelectronic chip manufacturing. Current state-of-theart lithography tools use deep ultraviolet (DUV) light from the KrF and ArF excimer lasers with wavelengths of 248 and 193 nanometers (the dominant lithography technology today is thus also called "excimer laser lithography), which has enabled transistor feature sizes to shrink below 45 nanometers. Excimer laser lithography has thus played a critical role in the

The most widespread industrial application of excimer lasers has been in deep-ultraviolet photolithography, a critical technology used in the manufacturing of microelectronic devices (i.e., semiconductor integrated circuits or "chips"). Historically, from the early 1960s through the mid-1980s, mercury-xenon lamps had been used in lithography for their spectral lines at 436, 405 and 365 nm wavelengths. However, with the semiconductor industry's need for both higher resolution (to produce denser and faster chips) and higher throughput (for lower costs), the lamp-based lithography tools were no longer able to meet the industry's requirements. This challenge was overcome when in a pioneering development in 1982, deep-UV excimer laser lithography was proposed and demonstrated at I.B.M. by Kanti Jain. With phenomenal advances made in equipment technology in the last two decades, and today microelectronic devices fabricated using excimer laser lithography totaling \$400 billion in annual production, it is the semiconductor industry view that excimer laser lithography has been a crucial factor in the continued advance of Moore's law, enabling minimum features sizes in chip manufacturing to shrink from 0.5 micrometer in 1990 to 32 nanometers in 2010. This trend is expected to continue into this decade for even denser chips, with minimum features approaching 10 nanometers. From an even broader scientific and technological perspective, since the invention of the laser in 1960, the development of excimer laser lithography has been highlighted as one of the major

For micro- and nano-technologies, laser micromachining is currently used in a large number of R&D and industrial applications. The range of applications to which laser methods are applied is continuously expanding, supported also by the development of novel processing techniques. Over the last decades the excimer laser has obtained the key position among lasers in various sectors of micromachining. Excimer lasers have developed into powerful manufacturing tools mainly because of two reasons: (i) The short wavelengths of the excimer laser offers excellent quality of machining and a great versatility in features which can be produced. (ii) The progress in basic excimer laser technology has made the excimer

There are certainly not many types of lasers which have found such broad markets as the excimer laser. Over the last years the main growth results from increasing industrial use followed by medical applications while new sales into R&D applications stay nearly constant. Today the largest known industrial applications of excimer lasers are (i) based on micromaching of different materials as polymers, ceramics and glasses, applied for example in the production of ink jet cartridges by drilling the nozzles, (ii) excimer laser radiation is being used for changing the structure and properties of materials as oxides, silicon or glass in bulk or thin films, as applied for the production of active matrix LCD monitors, fiber Bragg gratings in telecommunication, and high temperature superconducting films, (iii) employing the excimer laser as "short wavelength light bulb" in optical microlithography

continued advance of the so-called Moore's law for the last 20 years.

milestones in the 50-year history of the laser.

lasers to reliable machines suitable for the industrial environment.

to have a crucial role in many industries over the next decade. In years to come we will see high-power applications of excimer lasers, with output energies reaching 2 J and output power levels of 1,000 W and above (Delmdahl, 2010).

The aim of this chapter is to demonstrate the crystallization of polysilicon thin film by excimer laser annealing. Various aspects of excimer laser annealing process will be covered in this chapter including principle and experimental results. The definition and the history of excimer laser annealing are described in introduction section. Since excimer laser is the key component in this annealing process, the detail regarding excimer laser system is discussed. Mechanism of laser annealing is explained under third section of laser ablation which includes the ablation phenomena and photochemical mechanism. The detail sample preparation using solid phase crystallization technique is described under section 4. Section 5 will give information regarding polycrystalline silicon. The basic annealing process relies on transformation temperature in which three important parameters, comprising critical point, recalescence effect and super lateral growth energy are desired to be identified. The crystallization of silicon thin film is characterized via atomic force microscope analysis. Finally the excimer laser annealing on silicon thin film is summarized in section 8.

## **2. Excimer laser**

## **2.1 History and application**

An excimer laser is a form of ultraviolet laser which is commonly used in the production of microelectronic devices (semiconductor integrated circuits or "chips"), eye surgery, and micromachining. The excimer laser was invented in 1970 by Nikolai Basov, V. A. Danilychev and Yu. M. Popov, at the Lebedev Physical Institute in Moscow, using a xenon dimer (Xe2) excited by an electron beam to give stimulated emission at 172 nm wavelength. A later improvement, developed by many groups in 1975 was the use of noble gas halides (originally XeBr).

The excimer laser typically uses a combination of a noble gas (argon, krypton, or xenon) and a reactive gas (fluorine or chlorine). Under the appropriate conditions of electrical stimulation and high pressure, a pseudo-molecule called an excimer (or in the case of noble gas halides, exciplex) is created, which can only exist in an energized state and can give rise to laser light in the ultraviolet range.

Laser action in an excimer molecule occurs because it has a bound (associative) excited state, but a repulsive (dissociative) ground state. This is because noble gases such as xenon and krypton are highly inert and do not usually form chemical compounds. However, when in an excited state (induced by an electrical discharge or high-energy electron beams, which produce high energy pulses), they can form temporarily-bound molecules with themselves (dimers) or with halogens (complexes) such as fluorine and chlorine. The excited compound can give up its excess energy by undergoing spontaneous or stimulated emission, resulting in a strongly repulsive ground state molecule which very quickly (on the order of a picosecond) dissociates back into two unbound atoms. This forms a population inversion. Excimer lasers are usually operated with a pulse repetition rate of around 100 Hz and a pulse duration of ~10 ns, although some operate at pulse repetition rates as high as 8 kHz and some have pulsewidth as large as 30 ns.

to have a crucial role in many industries over the next decade. In years to come we will see high-power applications of excimer lasers, with output energies reaching 2 J and output

The aim of this chapter is to demonstrate the crystallization of polysilicon thin film by excimer laser annealing. Various aspects of excimer laser annealing process will be covered in this chapter including principle and experimental results. The definition and the history of excimer laser annealing are described in introduction section. Since excimer laser is the key component in this annealing process, the detail regarding excimer laser system is discussed. Mechanism of laser annealing is explained under third section of laser ablation which includes the ablation phenomena and photochemical mechanism. The detail sample preparation using solid phase crystallization technique is described under section 4. Section 5 will give information regarding polycrystalline silicon. The basic annealing process relies on transformation temperature in which three important parameters, comprising critical point, recalescence effect and super lateral growth energy are desired to be identified. The crystallization of silicon thin film is characterized via atomic force microscope analysis.

Finally the excimer laser annealing on silicon thin film is summarized in section 8.

An excimer laser is a form of ultraviolet laser which is commonly used in the production of microelectronic devices (semiconductor integrated circuits or "chips"), eye surgery, and micromachining. The excimer laser was invented in 1970 by Nikolai Basov, V. A. Danilychev and Yu. M. Popov, at the Lebedev Physical Institute in Moscow, using a xenon dimer (Xe2) excited by an electron beam to give stimulated emission at 172 nm wavelength. A later improvement, developed by many groups in 1975 was the use of noble gas halides

The excimer laser typically uses a combination of a noble gas (argon, krypton, or xenon) and a reactive gas (fluorine or chlorine). Under the appropriate conditions of electrical stimulation and high pressure, a pseudo-molecule called an excimer (or in the case of noble gas halides, exciplex) is created, which can only exist in an energized state and can give rise

Laser action in an excimer molecule occurs because it has a bound (associative) excited state, but a repulsive (dissociative) ground state. This is because noble gases such as xenon and krypton are highly inert and do not usually form chemical compounds. However, when in an excited state (induced by an electrical discharge or high-energy electron beams, which produce high energy pulses), they can form temporarily-bound molecules with themselves (dimers) or with halogens (complexes) such as fluorine and chlorine. The excited compound can give up its excess energy by undergoing spontaneous or stimulated emission, resulting in a strongly repulsive ground state molecule which very quickly (on the order of a picosecond) dissociates back into two unbound atoms. This forms a population inversion. Excimer lasers are usually operated with a pulse repetition rate of around 100 Hz and a pulse duration of ~10 ns, although some operate at pulse repetition rates as high as 8 kHz

power levels of 1,000 W and above (Delmdahl, 2010).

**2. Excimer laser** 

(originally XeBr).

**2.1 History and application** 

to laser light in the ultraviolet range.

and some have pulsewidth as large as 30 ns.

Excimer lasers are widely used in high-resolution photolithography machines, one of the critical technologies required for microelectronic chip manufacturing. Current state-of-theart lithography tools use deep ultraviolet (DUV) light from the KrF and ArF excimer lasers with wavelengths of 248 and 193 nanometers (the dominant lithography technology today is thus also called "excimer laser lithography), which has enabled transistor feature sizes to shrink below 45 nanometers. Excimer laser lithography has thus played a critical role in the continued advance of the so-called Moore's law for the last 20 years.

The most widespread industrial application of excimer lasers has been in deep-ultraviolet photolithography, a critical technology used in the manufacturing of microelectronic devices (i.e., semiconductor integrated circuits or "chips"). Historically, from the early 1960s through the mid-1980s, mercury-xenon lamps had been used in lithography for their spectral lines at 436, 405 and 365 nm wavelengths. However, with the semiconductor industry's need for both higher resolution (to produce denser and faster chips) and higher throughput (for lower costs), the lamp-based lithography tools were no longer able to meet the industry's requirements. This challenge was overcome when in a pioneering development in 1982, deep-UV excimer laser lithography was proposed and demonstrated at I.B.M. by Kanti Jain. With phenomenal advances made in equipment technology in the last two decades, and today microelectronic devices fabricated using excimer laser lithography totaling \$400 billion in annual production, it is the semiconductor industry view that excimer laser lithography has been a crucial factor in the continued advance of Moore's law, enabling minimum features sizes in chip manufacturing to shrink from 0.5 micrometer in 1990 to 32 nanometers in 2010. This trend is expected to continue into this decade for even denser chips, with minimum features approaching 10 nanometers. From an even broader scientific and technological perspective, since the invention of the laser in 1960, the development of excimer laser lithography has been highlighted as one of the major milestones in the 50-year history of the laser.

For micro- and nano-technologies, laser micromachining is currently used in a large number of R&D and industrial applications. The range of applications to which laser methods are applied is continuously expanding, supported also by the development of novel processing techniques. Over the last decades the excimer laser has obtained the key position among lasers in various sectors of micromachining. Excimer lasers have developed into powerful manufacturing tools mainly because of two reasons: (i) The short wavelengths of the excimer laser offers excellent quality of machining and a great versatility in features which can be produced. (ii) The progress in basic excimer laser technology has made the excimer lasers to reliable machines suitable for the industrial environment.

There are certainly not many types of lasers which have found such broad markets as the excimer laser. Over the last years the main growth results from increasing industrial use followed by medical applications while new sales into R&D applications stay nearly constant. Today the largest known industrial applications of excimer lasers are (i) based on micromaching of different materials as polymers, ceramics and glasses, applied for example in the production of ink jet cartridges by drilling the nozzles, (ii) excimer laser radiation is being used for changing the structure and properties of materials as oxides, silicon or glass in bulk or thin films, as applied for the production of active matrix LCD monitors, fiber Bragg gratings in telecommunication, and high temperature superconducting films, (iii) employing the excimer laser as "short wavelength light bulb" in optical microlithography

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 485

Excimer laser is excited by passing a short, intense electrical pulse through a mixture of gases containing the desired rare gas and halogen. However, the molecule were found more stable when it is on excited state and become less stable on ground state and this properties can be found on argon fluoride (ArF), krypton fluoride (KrF2) and xenon chloride (XeCl2).

Figure 2 shows a graph of potential energy curve as a function of atomic distance between the molecules. Generally, the potential energy is minimized at equilibrium molecule distance. However, for ArF, KrF2 and XeCl2 material, these properties did not occur. They are found not stable even when they are in their ground state. Thus, excitation process cannot directly be done on ground state. Hence, indirect excitation in discharge system is employed. For example, in ArF excitation process, electron reaction process is as followed,

The negative ion produced will combine with positive ion to produce excited molecule

 Ar+ + F─ → (ArF)\* (2) This reaction can produce an efficient excited dimer molecule. The unstable molecules at ground state leads to a small number of population. Hence, it causes the population inversion to easily happen and laser transition will trigger. The nature of dimer material that is not stable on their ground state causes the atom to easily breaks-up (Shah, 2009). Because of the ground state essentially does not exist, there is a population inversion as long as there are molecules in the excited state. This process is performed again and again; this is how the

An argon fluoride ArF excimer laser, manufactured by GAMLaser model EX5/200-110, is employed as a source of ablation. The wavelength of the laser is 193 nm with pulse duration

e + F2 → F- + F (1)

This phenomenon could be explained in detail by referring on Figure 2.

Fig. 2. Excimer laser potential energy curve.

pulse excimer laser is trigger out.

**2.4 External triggering** 

for the production of computer chips with critical dimensions below 0.25 *μ*m (the largest homogeneous market for excimer lasers).

All the widespread applications of excimer lasers in micromachining and medicine are based on the early use of excimer lasers leading to the discovery of the ablation of materials under intense illumination with ultraviolet laser pulses by R. Srinivasan.

#### **2.2 Crystallization source**

The source of energy used for crystallization in this work is an argon fluoride excimer laser. The excimer laser generates deep ultraviolet light measuring 193 nm when high voltage energy is discharged into a cavity containing a mixture of a rare gas argon and a halogen (i.e., fluoride) Almost, 90% or more of the mixture contain buffer rare gas (typically helium) that does not take part in the reaction. This gas mixture is pre-ionized by a set of electrodes before a high-voltage current (about 30,000 eV is applied) resulting in formation of highly unstable rare-gas halide molecules, which rapidly dissociate, emitting UV light whose wavelength is determined by the particular gas mixture chosen. Excimer lasers use in this particular project is comprised of a mixture of argon and fluoride gas. The term excimer is derived from the two words "excited" and "dimer" which are used to describe the reaction in which the laser transfer energy through an ultraviolet beam of light (Schneider, 1998). All emitted powerful pulses lasting in 10 nanoseconds at wavelength of ultraviolet region of 193 nm.

#### **2.3 Excimer laser structure**

The gain of excimer lasers is extremely high, so the output is superradiant. A single rear mirror is employed and an output coupler of 4 to 8% which transmit in the region of interest (which is in UV region) is used (Figure 1). Divergence of the beam is reduced when a full optical cavity is used, and alignment is easy since the laser operates even when cavity mirrors are completely misaligned. The beam is rectangular profile. Quartz cannot be used as an output coupler for argon fluoride (ArF) Excimer laser since fluorine attacks the material, hence magnesium fluoride are used to replace quartz as it absorbs less UV at most wavelength compare to quartz (Csele, 2004). In discharge excitation, electric current flows through the laser medium, typically ranging from a kilovolt (kV) to over tens of kilovolts deliver energy to the laser gas (Hecht, 1992).

Fig. 1. An electrical discharge exciting a gas laser (Hecht, 1992)

for the production of computer chips with critical dimensions below 0.25 *μ*m (the largest

All the widespread applications of excimer lasers in micromachining and medicine are based on the early use of excimer lasers leading to the discovery of the ablation of materials

The source of energy used for crystallization in this work is an argon fluoride excimer laser. The excimer laser generates deep ultraviolet light measuring 193 nm when high voltage energy is discharged into a cavity containing a mixture of a rare gas argon and a halogen (i.e., fluoride) Almost, 90% or more of the mixture contain buffer rare gas (typically helium) that does not take part in the reaction. This gas mixture is pre-ionized by a set of electrodes before a high-voltage current (about 30,000 eV is applied) resulting in formation of highly unstable rare-gas halide molecules, which rapidly dissociate, emitting UV light whose wavelength is determined by the particular gas mixture chosen. Excimer lasers use in this particular project is comprised of a mixture of argon and fluoride gas. The term excimer is derived from the two words "excited" and "dimer" which are used to describe the reaction in which the laser transfer energy through an ultraviolet beam of light (Schneider, 1998). All emitted powerful

The gain of excimer lasers is extremely high, so the output is superradiant. A single rear mirror is employed and an output coupler of 4 to 8% which transmit in the region of interest (which is in UV region) is used (Figure 1). Divergence of the beam is reduced when a full optical cavity is used, and alignment is easy since the laser operates even when cavity mirrors are completely misaligned. The beam is rectangular profile. Quartz cannot be used as an output coupler for argon fluoride (ArF) Excimer laser since fluorine attacks the material, hence magnesium fluoride are used to replace quartz as it absorbs less UV at most wavelength compare to quartz (Csele, 2004). In discharge excitation, electric current flows through the laser medium, typically ranging from a kilovolt (kV) to over tens of kilovolts

under intense illumination with ultraviolet laser pulses by R. Srinivasan.

pulses lasting in 10 nanoseconds at wavelength of ultraviolet region of 193 nm.

homogeneous market for excimer lasers).

**2.2 Crystallization source** 

**2.3 Excimer laser structure** 

deliver energy to the laser gas (Hecht, 1992).

Fig. 1. An electrical discharge exciting a gas laser (Hecht, 1992)

Excimer laser is excited by passing a short, intense electrical pulse through a mixture of gases containing the desired rare gas and halogen. However, the molecule were found more stable when it is on excited state and become less stable on ground state and this properties can be found on argon fluoride (ArF), krypton fluoride (KrF2) and xenon chloride (XeCl2). This phenomenon could be explained in detail by referring on Figure 2.

Fig. 2. Excimer laser potential energy curve.

Figure 2 shows a graph of potential energy curve as a function of atomic distance between the molecules. Generally, the potential energy is minimized at equilibrium molecule distance. However, for ArF, KrF2 and XeCl2 material, these properties did not occur. They are found not stable even when they are in their ground state. Thus, excitation process cannot directly be done on ground state. Hence, indirect excitation in discharge system is employed. For example, in ArF excitation process, electron reaction process is as followed,

$$\text{Fe} + \text{F}^2 \rightarrow \text{F} + \text{F} \tag{1}$$

The negative ion produced will combine with positive ion to produce excited molecule

$$\rm{Ar^{\*}} + \rm{F^{-}} \rightarrow \rm{(ArF)^{\*}} \tag{2}$$

This reaction can produce an efficient excited dimer molecule. The unstable molecules at ground state leads to a small number of population. Hence, it causes the population inversion to easily happen and laser transition will trigger. The nature of dimer material that is not stable on their ground state causes the atom to easily breaks-up (Shah, 2009). Because of the ground state essentially does not exist, there is a population inversion as long as there are molecules in the excited state. This process is performed again and again; this is how the pulse excimer laser is trigger out.

#### **2.4 External triggering**

An argon fluoride ArF excimer laser, manufactured by GAMLaser model EX5/200-110, is employed as a source of ablation. The wavelength of the laser is 193 nm with pulse duration

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 487

The laser is controlled externally using a AFG310 function generator for obtaining laser pulse below than 100 pulses. In this present works, the number of pulse is varied from one to ten pulses. The operational voltage of the function generator is set at 5 Volts 'pulse' output function, and with 'burst' mode. The laser pulse is generated after 2.5 μs upon application of the trigger pulse rising edge with DC charging is operating in external trigger mode. The frequency of the function generator and the excimer laser system is fixed at 20.0 Hz. The silicon copper thin film placed on *x* and *y* translational stage in and vertical position with respect to the excimer laser beam axis. The number of pulses of excimer laser beam

Maximum laser performance can be achieved by aligning the laser mirrors. The mirrors are alligned via the guidance of a visible of helium-neon He-Ne laser. The He-Ne laser is aligned to coaxis with excimer laser beam. The excimer laser is optimized by adjusting the output coupler via the aids of power/energymeter. A bright and rectangular dimension of beam spot is recorded permanently using beam profiler. The images are arranged in order of 3D, 2D and the bright beam spot in the real field such as shown in Figure 5 a, b and c respectively. The dimension of excimer laser beam spot is 4 × 6 mm2. The a-Si films is recrystalized at different energy densities. Finally, nanostructure of the crystallized silicon

Beam splitter

Powermeter

Silicon copper

thin film Atomic force

A

microscope system AFM

exposed on the thin film target is, controlled by the external trigger unit.

Fig. 4. The schematic diagram of excimer laser annealing process.

Fig. 5. Beam profile and beam spot of excimer laser. a. 3D beam profile, b. 2D beam profile,

is examined via atomic force microscope AFM.

External trigger unit

ArF2 excimer laser

He-Ne

and c. The beam spot in the real field.

of 10 ns and repetition rate which can be varied from 20 Hz to 200 Hz. The beam size of the laser output when it hits the sample is 4 mm x 6 mm. The size of the beam also depends on the output energy of the laser beam (GAMLaser Inc., 2003).

The excimer laser is internally operated via the aids of computer using 32 bit Windows software (GAMLaser Inc., 2003). The laser parameters including the frequency, high voltage and number of pulse can be commendable according to the requirement of the experiment. The output energy of the laser can be varied by manipulating the capacitor voltage from 10 kV to 15 kV. The dose number of laser exposure is not limited; however, the minimum number of pulses can be set at 100 pulses. However, to trigger the number of pulse as minimum as 1 pulse, a Sony Tektronix arbitrary function generator model AFG 310 is employed. Figure 3 shows photograph of the excimer laser, which coupled to an arbitrary function generator in the real field. In this particular experiment, only a few pulses of laser exposure are required to obtain the desired optical surface treatment.

Fig. 3. An excimer laser connected to a function generator for external triggering.

The function generator can be used to control the number of pulses lower than 100 pulses. As an external trigger unit, the function generator is set up at '5 Volts', operated at 'pulse' output function, and with 'burst' mode (Sony Tektronix Inc., 2003). The frequency of the function generator is synchronized with the excimer laser system.

### **2.5 Excimer laser annealing**

A GAM LASER Argon Fluoride ArF excimer laser is used to anneal the hydrogenerated silicon thin film. The ArF excimer laser generates ultraviolets laser light at 193 nm with 10 ns pulse duration. The lasers parameters such as number of pulses, repetition rates and pumping voltages can be controlled via a computer program. In this experiment, the laser capacitor voltage is kept constant at 12 kV with repeatition rate of 20 Hz, operating at room temperature. The gas pressure is set at 3200 Torr.

The excimer laser annealing experimental setup is shown in Figure 4. The laser energy (in mJ) for each laser pulses is measured using 13PEM001 Broadband power and energy Meter.

of 10 ns and repetition rate which can be varied from 20 Hz to 200 Hz. The beam size of the laser output when it hits the sample is 4 mm x 6 mm. The size of the beam also depends on

The excimer laser is internally operated via the aids of computer using 32 bit Windows software (GAMLaser Inc., 2003). The laser parameters including the frequency, high voltage and number of pulse can be commendable according to the requirement of the experiment. The output energy of the laser can be varied by manipulating the capacitor voltage from 10 kV to 15 kV. The dose number of laser exposure is not limited; however, the minimum number of pulses can be set at 100 pulses. However, to trigger the number of pulse as minimum as 1 pulse, a Sony Tektronix arbitrary function generator model AFG 310 is employed. Figure 3 shows photograph of the excimer laser, which coupled to an arbitrary function generator in the real field. In this particular experiment, only a few pulses of laser

the output energy of the laser beam (GAMLaser Inc., 2003).

exposure are required to obtain the desired optical surface treatment.

Fig. 3. An excimer laser connected to a function generator for external triggering.

function generator is synchronized with the excimer laser system.

**2.5 Excimer laser annealing** 

temperature. The gas pressure is set at 3200 Torr.

The function generator can be used to control the number of pulses lower than 100 pulses. As an external trigger unit, the function generator is set up at '5 Volts', operated at 'pulse' output function, and with 'burst' mode (Sony Tektronix Inc., 2003). The frequency of the

A GAM LASER Argon Fluoride ArF excimer laser is used to anneal the hydrogenerated silicon thin film. The ArF excimer laser generates ultraviolets laser light at 193 nm with 10 ns pulse duration. The lasers parameters such as number of pulses, repetition rates and pumping voltages can be controlled via a computer program. In this experiment, the laser capacitor voltage is kept constant at 12 kV with repeatition rate of 20 Hz, operating at room

The excimer laser annealing experimental setup is shown in Figure 4. The laser energy (in mJ) for each laser pulses is measured using 13PEM001 Broadband power and energy Meter. The laser is controlled externally using a AFG310 function generator for obtaining laser pulse below than 100 pulses. In this present works, the number of pulse is varied from one to ten pulses. The operational voltage of the function generator is set at 5 Volts 'pulse' output function, and with 'burst' mode. The laser pulse is generated after 2.5 μs upon application of the trigger pulse rising edge with DC charging is operating in external trigger mode. The frequency of the function generator and the excimer laser system is fixed at 20.0 Hz. The silicon copper thin film placed on *x* and *y* translational stage in and vertical position with respect to the excimer laser beam axis. The number of pulses of excimer laser beam exposed on the thin film target is, controlled by the external trigger unit.

Maximum laser performance can be achieved by aligning the laser mirrors. The mirrors are alligned via the guidance of a visible of helium-neon He-Ne laser. The He-Ne laser is aligned to coaxis with excimer laser beam. The excimer laser is optimized by adjusting the output coupler via the aids of power/energymeter. A bright and rectangular dimension of beam spot is recorded permanently using beam profiler. The images are arranged in order of 3D, 2D and the bright beam spot in the real field such as shown in Figure 5 a, b and c respectively. The dimension of excimer laser beam spot is 4 × 6 mm2. The a-Si films is recrystalized at different energy densities. Finally, nanostructure of the crystallized silicon is examined via atomic force microscope AFM.

Fig. 4. The schematic diagram of excimer laser annealing process.

Fig. 5. Beam profile and beam spot of excimer laser. a. 3D beam profile, b. 2D beam profile, and c. The beam spot in the real field.

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 489

explain the ablation mechanisms. In photochemical mechanisms, the photon break the chemical bonds of the material directly where as in photothermal mechanism the material is ablated by heating, melting and vaporizing the material (Chen et al., 1995). It is found that the material removal by laser ablation approximately obeys the Beer lamberts Law at lower laser fluences (Srinivasan, 1982). For photochemical ablation to occur, energy of the photons at that wavelength should overcome the intermolecular bond energies of the target material. The relation between the photon energy of light and laser wavelength is given by

Where is wavelength of light (193 nm), E is energy of photon (eV). The photon energy of the lights depends on the wavelength of the light and as the wavelength increases, the energy decreases. An UV laser with a wavelength of 193 nm has photon energy of 6.45 eV. It can be seen that for photochemical ablation to occur in thin film the photon energy of the light should be greater than the bond energy of the material. While excimer laser does not produce as high average powers as longer wavelength infrared lasers such as Nd:YAG and CO2, their shorter wavelength carried high photon energy. As for example, argon fluoride (ArF) excimer laser with wavelength of 193 nm have photon energy of 6.45 eV, while Nd:YAG with wavelength of 1064 nm have photon energy of 0.18 eV. Hence, only laser with short wavelength can initiate photochemical reaction while the laser with longer wavelength

Apparently, the advantage of photochemical reaction, is that the mechanism can change the molecule structure as well as the refractive index of the material, but not involve in external damage unless the energy deliver is too high. As a result the photochemical mechanism is important effect in the nucleation and the growing of grain size or in the crystallization

Target used to be annealed is an amorphous silicon. In this works silicon is chosen to be the main material and deposited on the glass substrate as the first layer of thin film. Silicon is widely used in semiconductors industry. It can stand at higher temperatures compared to germanium. Furthermore its native oxide is easily grown in a furnace. Plus it can form a

An ordinary glass substrate is conducted to make it possible for low annealing temperature. The glass substrate used in this research has maximum operation temperature in the range of 350°C to 400°C. The thickness of the glass substrate is in between 1.0 mm to 1.2 mm. The glass substrates are cut into small pieces using diamond cutter. Each of them having dimension of 10 10 mm2. All glass substrates are cleaned using Ultrasonic Branson 3210 cleaning machine to ensure the glass surface is free from

Solid phase crystalization (SPC) is prepared by using a vacuum chamber technology. An Edwards 360 thermal evaporation is a vacuum chamber used to fabricate the silicon thin

E = 1.245 / (3)

(Sauerbrey, 1989),

process.

**4. Sample preparation** 

any dust, oil, or contaminations.

**4.1 Thin film fabrication** 

may only lead to heating the material only.

better dielectric interface rather than any other materials.

## **3. Laser ablation**

In laser annealing process the main key for its successful work is the ablation process between optical material and UV light from argon fluoride (ArF) excimer laser. In order to understand the laser-matter interaction, it is better to understand the fundamental mechanism of laser ablation and optical material behavior with respect to the ultraviolet (UV) light illumination. In this section, the photochemical process occurred during ablation process will be discussed.

Ablation is usually described in terms of physical mechanism such as vaporization and shock effects. It is often performed in vacuum or air. Examples of ablation application include polymer ablation, removal of thin metal films from an insulating substrate, and deposition of high-temperature superconducting thin films by ablation of bulk targets (Brannon, 1997).

The laser ablation process requires an intense UV light (λ = 193 nm- 351 nm) produced by an excimer laser. Because of the high absorption of UV light and relatively poor conductivity of many materials, particularly silicon, the energy is deposited in a very thin layer. Whenever the energy density or fluence exceeds the ablation threshold value for the material, chemical bonds are broken, fracturing the material into energetic fragments. The fragments are atoms, groups of atoms, ions and electrons. Because the fragments leave the reaction zone as an energetic gas and solid debris, the ablation process resembles explosive evaporation of the material (Speidell, 1997).

Generally, there are two classes of laser ablation mechanisms: thermal and electronic (nonthermal). Thermal processes rely on an intense laser pulse to very rapidly heat a surface at rates of the order of 1010 K/sec. These processes can produce expansion and vaporization from solid and melted regions.

Laser ablation of given material can involve both thermal and non-thermal mechanisms. For example, during the interaction of the leading edge of the laser pulse with the solid, nonthermal processes may predominate because significant heating has not yet occurred. As the rest of the pulse is absorbed, the temperature rise rapidly and thermally activated mechanism may commence. It should be noted that laser excitation of a strongly absorbing solid will always result in substantial heating regardless of the ablation mechanism. Although non-thermal process may occur, there is no way to "turnoff "the thermal processes. The dominant process is ultimately determined by the detailed chemistry and physics of the solid.

## **3.1 Photochemical**

In the photochemical processes the photon absorption event break a chemical bonding within a molecule forming molecules with smaller number of atoms. The resulting photoproducts occupy a large volume and create pressure inside the irradiated volume that can then convert to the translational energy of ablation (Sato et al., 2001). During this process, thermal and mechanical damage to the surrounding material is minimized, therefore achieving more precise control over the ablated region.

There has been much un-certainty and debate over the fundamental ablation mechanisms in material. Several photochemicals and photothermals models have been developed to

In laser annealing process the main key for its successful work is the ablation process between optical material and UV light from argon fluoride (ArF) excimer laser. In order to understand the laser-matter interaction, it is better to understand the fundamental mechanism of laser ablation and optical material behavior with respect to the ultraviolet (UV) light illumination. In this section, the photochemical process occurred during ablation

Ablation is usually described in terms of physical mechanism such as vaporization and shock effects. It is often performed in vacuum or air. Examples of ablation application include polymer ablation, removal of thin metal films from an insulating substrate, and deposition of high-temperature superconducting thin films by ablation of bulk targets

The laser ablation process requires an intense UV light (λ = 193 nm- 351 nm) produced by an excimer laser. Because of the high absorption of UV light and relatively poor conductivity of many materials, particularly silicon, the energy is deposited in a very thin layer. Whenever the energy density or fluence exceeds the ablation threshold value for the material, chemical bonds are broken, fracturing the material into energetic fragments. The fragments are atoms, groups of atoms, ions and electrons. Because the fragments leave the reaction zone as an energetic gas and solid debris, the ablation process resembles explosive evaporation of the

Generally, there are two classes of laser ablation mechanisms: thermal and electronic (nonthermal). Thermal processes rely on an intense laser pulse to very rapidly heat a surface at rates of the order of 1010 K/sec. These processes can produce expansion and vaporization

Laser ablation of given material can involve both thermal and non-thermal mechanisms. For example, during the interaction of the leading edge of the laser pulse with the solid, nonthermal processes may predominate because significant heating has not yet occurred. As the rest of the pulse is absorbed, the temperature rise rapidly and thermally activated mechanism may commence. It should be noted that laser excitation of a strongly absorbing solid will always result in substantial heating regardless of the ablation mechanism. Although non-thermal process may occur, there is no way to "turnoff "the thermal processes. The dominant process is ultimately determined by the detailed chemistry and

In the photochemical processes the photon absorption event break a chemical bonding within a molecule forming molecules with smaller number of atoms. The resulting photoproducts occupy a large volume and create pressure inside the irradiated volume that can then convert to the translational energy of ablation (Sato et al., 2001). During this process, thermal and mechanical damage to the surrounding material is minimized,

There has been much un-certainty and debate over the fundamental ablation mechanisms in material. Several photochemicals and photothermals models have been developed to

therefore achieving more precise control over the ablated region.

**3. Laser ablation** 

process will be discussed.

material (Speidell, 1997).

physics of the solid.

**3.1 Photochemical** 

from solid and melted regions.

(Brannon, 1997).

explain the ablation mechanisms. In photochemical mechanisms, the photon break the chemical bonds of the material directly where as in photothermal mechanism the material is ablated by heating, melting and vaporizing the material (Chen et al., 1995). It is found that the material removal by laser ablation approximately obeys the Beer lamberts Law at lower laser fluences (Srinivasan, 1982). For photochemical ablation to occur, energy of the photons at that wavelength should overcome the intermolecular bond energies of the target material. The relation between the photon energy of light and laser wavelength is given by (Sauerbrey, 1989),

$$\mathbf{E} = \mathbf{1.245 / \lambda} \tag{3}$$

Where is wavelength of light (193 nm), E is energy of photon (eV). The photon energy of the lights depends on the wavelength of the light and as the wavelength increases, the energy decreases. An UV laser with a wavelength of 193 nm has photon energy of 6.45 eV. It can be seen that for photochemical ablation to occur in thin film the photon energy of the light should be greater than the bond energy of the material. While excimer laser does not produce as high average powers as longer wavelength infrared lasers such as Nd:YAG and CO2, their shorter wavelength carried high photon energy. As for example, argon fluoride (ArF) excimer laser with wavelength of 193 nm have photon energy of 6.45 eV, while Nd:YAG with wavelength of 1064 nm have photon energy of 0.18 eV. Hence, only laser with short wavelength can initiate photochemical reaction while the laser with longer wavelength may only lead to heating the material only.

Apparently, the advantage of photochemical reaction, is that the mechanism can change the molecule structure as well as the refractive index of the material, but not involve in external damage unless the energy deliver is too high. As a result the photochemical mechanism is important effect in the nucleation and the growing of grain size or in the crystallization process.

## **4. Sample preparation**

Target used to be annealed is an amorphous silicon. In this works silicon is chosen to be the main material and deposited on the glass substrate as the first layer of thin film. Silicon is widely used in semiconductors industry. It can stand at higher temperatures compared to germanium. Furthermore its native oxide is easily grown in a furnace. Plus it can form a better dielectric interface rather than any other materials.

An ordinary glass substrate is conducted to make it possible for low annealing temperature. The glass substrate used in this research has maximum operation temperature in the range of 350°C to 400°C. The thickness of the glass substrate is in between 1.0 mm to 1.2 mm. The glass substrates are cut into small pieces using diamond cutter. Each of them having dimension of 10 10 mm2. All glass substrates are cleaned using Ultrasonic Branson 3210 cleaning machine to ensure the glass surface is free from any dust, oil, or contaminations.

#### **4.1 Thin film fabrication**

Solid phase crystalization (SPC) is prepared by using a vacuum chamber technology. An Edwards 360 thermal evaporation is a vacuum chamber used to fabricate the silicon thin

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 491

The laser is absorbed in amorphous silicon thin film surface without heating the substrate. A homogenized excimer laser beam spot is exposed on the thin film surface. Within the laser pulse duration the silicon layer is rapidly heated and melted as shown in Figure 7(a, b). As it cools down the crystallization into poly-silicon occurs (Figure 7c, d). In the process of excimer laser annealing, the amorphous silicon thin film is exposed by multiple pulses of laser. The exposed area is partially melted. The period for heating as well as for freezing is the same as the pulse duration of the excimer laser. The heat release in high speed freezing is more compared to the slow rate once. As a result during cooling the heat released by convection force-liquid flow which break the planar symmetry so that the crystal grow develops along the columns or finger such as shown in Figure 7c and d. In Figure 7c, there is partially melted and partially crystallized growth. The totally crystal growth is illustrated in

Fig. 7. Schematic diagram of excimer laser annealing on amorphous silicon copper thin film.

Silicon films are fabricated with a microstructure tailored to the application depending on economic and performance requirements. Silicon film is typically divided into three

Monocrystalline silicon is a form in which the crystal structure is homogenous throughout the material; the orientation, lattice parameter, and electronic properties are constant throughout the material (Green, 2004). Dopant atoms such as phosphorus and boron are often incorporated into the film to make the silicon n-type or p-type respectively. Monocrystalline silicon is fabricated in the form of silicon wafers, usually by the Czochralski Growth method, and can be quite expensive depending on the radial size of the desired single crystal wafer (Green, 2004). This monocrystalline material is one of the chief expenses where approximately 40% of the final price of the product is attributable to the cost of the

Amorphous silicon has no long-range periodic order. The application of amorphous silicon as a standalone material is somewhat limited by its inferior electronic properties (Streetman and Banerjee, 2000). When paired with microcrystalline silicon in tandem and triple-junction solar cells, however, higher efficiency can be attained than with single-junction solar cells (Shah et al*.,* 2003). This tandem assembly of solar cells allows one to obtain a thin-film

categories: monocrystalline, amorphous, and polycrystalline.

starting silicon wafer used in cell fabrication (Campbell, 2001).

Figure 7d. The crystal grows in the finger formation or poly-crystallization.

**5. Polycrystalline silicon** 

film. Silicon is deposited on a glass substrate. Initially, the source material to be deposited or evaporant is loaded into a molybdenum boat. The glass substrate is placed 10 cm above the evaporant. The chamber is set at a vacuum condition until the pressure achieved down to 10-6 mbar. The current is increased slowly until the evaporant on the molybdenum boat begin to melt. The thickness of film is detected by quartz crystal thickness detector inside the vacuum chamber. The thickness is measured via an Edwards FTM 7 which connected to the vacuum system. To deposit the silicon for about 100 nm thick, 0.3 g of silicon powder is needed. A schematic diagram of the thin film is shown in Figure 6.


Fig. 6. Silicon thin film on glass substrate

## **4.2 Annealing process**

The annealing of silicon thin film is a combination from heat treatment and excimer exposure. Immediately after fabricated the thin film, it is heated via conventional Tube Furnace type F21100. This is important procedure in order to dry out the water inside the thin film so that the hydrogen percentage can be lowered. The hydrogenated thin film is then annealed by using an argon fluoride (ArF) excimer laser.

## **4.2.1 Heat treatment**

The prepared solid phase crystallization of silicon thin film is initially under-went heat treatment. As mention earlier the aim is to extract the water contained in the film during deposition. An encapsulated annealing technique is utilized for this heat treatment in order to avoid vaporization of thin film material. The silicon thin film is placed in a carbon block (both the container and its cover are made from the same element). The capsulated thin film is then placed in the tube furnace. The annealing process is carried out under atmospheric pressure for four hours in air. The annealing is set at a fixed temperature of 350 °C. This furnace has a single set point control which enables the end user to control the furnace up to a preset temperature and remained constant at the set temperature.

#### **4.2.2 Laser annealing**

Nearly all optical materials show moderate to intense absorption in the ultraviolet region. This absorption is usually ascribed to electronic transition from a ground singlet to the first excited state. The unique feature in the UV laser ablation of optical material is encountered only in those wavelength region in which such electronic absorption exist (Srinivasan et al., 1988). Material ablation depends on two key conditions, as shown in Figure 7. First, the material must absorb light strongly at the laser's wavelength. Clean, precise ablation usually requires linear absorption coefficients of at least 104 cm-1. Secondly, ablation occurs only after the material has absorbed a minimum energy per unit volume, i.e., the laser intensity must exceed a threshold value (Brannon et al., 1985).

film. Silicon is deposited on a glass substrate. Initially, the source material to be deposited or evaporant is loaded into a molybdenum boat. The glass substrate is placed 10 cm above the evaporant. The chamber is set at a vacuum condition until the pressure achieved down to 10-6 mbar. The current is increased slowly until the evaporant on the molybdenum boat begin to melt. The thickness of film is detected by quartz crystal thickness detector inside the vacuum chamber. The thickness is measured via an Edwards FTM 7 which connected to the vacuum system. To deposit the silicon for about 100 nm thick, 0.3 g of silicon powder is

The annealing of silicon thin film is a combination from heat treatment and excimer exposure. Immediately after fabricated the thin film, it is heated via conventional Tube Furnace type F21100. This is important procedure in order to dry out the water inside the thin film so that the hydrogen percentage can be lowered. The hydrogenated thin film is

**Glass** 

**Silicon** 

The prepared solid phase crystallization of silicon thin film is initially under-went heat treatment. As mention earlier the aim is to extract the water contained in the film during deposition. An encapsulated annealing technique is utilized for this heat treatment in order to avoid vaporization of thin film material. The silicon thin film is placed in a carbon block (both the container and its cover are made from the same element). The capsulated thin film is then placed in the tube furnace. The annealing process is carried out under atmospheric pressure for four hours in air. The annealing is set at a fixed temperature of 350 °C. This furnace has a single set point control which enables the end user to control the furnace up to

Nearly all optical materials show moderate to intense absorption in the ultraviolet region. This absorption is usually ascribed to electronic transition from a ground singlet to the first excited state. The unique feature in the UV laser ablation of optical material is encountered only in those wavelength region in which such electronic absorption exist (Srinivasan et al., 1988). Material ablation depends on two key conditions, as shown in Figure 7. First, the material must absorb light strongly at the laser's wavelength. Clean, precise ablation usually requires linear absorption coefficients of at least 104 cm-1. Secondly, ablation occurs only after the material has absorbed a minimum energy per unit volume, i.e., the laser intensity

needed. A schematic diagram of the thin film is shown in Figure 6.

then annealed by using an argon fluoride (ArF) excimer laser.

a preset temperature and remained constant at the set temperature.

must exceed a threshold value (Brannon et al., 1985).

Fig. 6. Silicon thin film on glass substrate

**4.2 Annealing process** 

**4.2.1 Heat treatment** 

**4.2.2 Laser annealing** 

The laser is absorbed in amorphous silicon thin film surface without heating the substrate. A homogenized excimer laser beam spot is exposed on the thin film surface. Within the laser pulse duration the silicon layer is rapidly heated and melted as shown in Figure 7(a, b). As it cools down the crystallization into poly-silicon occurs (Figure 7c, d). In the process of excimer laser annealing, the amorphous silicon thin film is exposed by multiple pulses of laser. The exposed area is partially melted. The period for heating as well as for freezing is the same as the pulse duration of the excimer laser. The heat release in high speed freezing is more compared to the slow rate once. As a result during cooling the heat released by convection force-liquid flow which break the planar symmetry so that the crystal grow develops along the columns or finger such as shown in Figure 7c and d. In Figure 7c, there is partially melted and partially crystallized growth. The totally crystal growth is illustrated in Figure 7d. The crystal grows in the finger formation or poly-crystallization.

Fig. 7. Schematic diagram of excimer laser annealing on amorphous silicon copper thin film.

## **5. Polycrystalline silicon**

Silicon films are fabricated with a microstructure tailored to the application depending on economic and performance requirements. Silicon film is typically divided into three categories: monocrystalline, amorphous, and polycrystalline.

Monocrystalline silicon is a form in which the crystal structure is homogenous throughout the material; the orientation, lattice parameter, and electronic properties are constant throughout the material (Green, 2004). Dopant atoms such as phosphorus and boron are often incorporated into the film to make the silicon n-type or p-type respectively. Monocrystalline silicon is fabricated in the form of silicon wafers, usually by the Czochralski Growth method, and can be quite expensive depending on the radial size of the desired single crystal wafer (Green, 2004). This monocrystalline material is one of the chief expenses where approximately 40% of the final price of the product is attributable to the cost of the starting silicon wafer used in cell fabrication (Campbell, 2001).

Amorphous silicon has no long-range periodic order. The application of amorphous silicon as a standalone material is somewhat limited by its inferior electronic properties (Streetman and Banerjee, 2000). When paired with microcrystalline silicon in tandem and triple-junction solar cells, however, higher efficiency can be attained than with single-junction solar cells (Shah et al*.,* 2003). This tandem assembly of solar cells allows one to obtain a thin-film

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 493

that the material scientist can manipulate (Ghosh et al., 1980). Through the methods of crystallization to form polycrystalline silicon, a scientist can control the size of the

The use of polycrystalline silicon in the production of solar cells requires less material and therefore provides for higher profits and increased manufacturing throughput. Polycrystalline silicon does not need to be deposited on a silicon wafer to form a solar cell, rather it can be deposited on other-cheaper materials, thus reducing the cost. Not requiring a silicon wafer alleviates the silicon shortages occasionally faced by the microelectronics industry. An example of not using a silicon wafer is crystalline silicon on glass (CSG)

A primary concern in the photovoltaics industry is cell efficiency. However, sufficient cost savings from cell manufacturing can be suitable to offset reduced efficiency in the field, such as the use of larger solar cell arrays compared with more compact or higher efficiency designs. Designs such as CSG are attractive because of a low cost of production even with reduced efficiency (Basore, 2006). Higher efficiency devices yield modules that occupy less space and are more compacts however the 5-10 % efficiency of typical CSG devices still makes them attractive for installation in large central-service stations, such as a power station (Basore, 2006). The issue of efficiency versus cost is a value decision of whether one requires an "energy dense" solar cell or sufficient area is available for the installation of less expensive alternatives. For instance, a solar cell used for power generation in a remote location might require a more highly efficient solar cell than one used for low-power applications, such as solar accent

Thin film silicon photovoltaics are typically produced by chemical vapor deposition processes yielding an amorphous, polycrystalline, or nanocrystalline film. Conventionally, amorphous silicon thin films are most common. Silicon is usually deposited on glass, plastic, or metallic substrates coated with a transparent conducting oxide material. While chalcogenide-based Cadmium-Tellurium (CdTe) and Copper-Indium-Selenium (CIS) polycrystalline thin films cells have been developed in the lab with great success, there is still industry interest in silicon-based thin film cells. Silicon-based devices exhibit fewer problems than their CdTe and CIS counterparts such as toxicity and humidity issues with CdTe cells and low manufacturing yields of CIS due to material complexity. Additionally, due to political resistance to the use non-"green" materials in solar energy production, there is no stigma in the use of standard silicon. Three major silicon-based module designs dominate: amorphous silicon cells, amorphous or microcrystalline tandem cells, and thin-

Amorphous or microcrystalline silicon consists of a mixed phase of small crystalline regions surrounded by amorphous material. This material typically behaves more like crystalline silicon than the amorphous variety. A 3-month field study has shown that hybrid amorphous or microcrystalline cells degrade roughly to the same degree as triple-junction amorphous cells while maintaining higher conversion efficiencies (7.0% versus 5.0% as measured at the conclusion of the study). This result suggests hybrid designs of this type

polycrystalline grains which will vary the physical properties of the material.

**5.1 Novel ideas for polycrystalline silicon** 

lighting or pocket calculators, or near established power grids.

film polycrystalline silicon on glass (Green, 2003).

may supplant traditional amorphous-based modules (Green, 2003).

materials (Basore, 2006).

material with a bandgap of around 1.12 eV (the same as single-crystal silicon) compared to the bandgap of amorphous silicon of 1.7-1.8 eV bandgap (Shah et al*.,* 2003). Tandem solar cells are then attractive since they can be fabricated with a bandgap similar to single-crystal silicon but with the ease of amorphous silicon.

Polycrystalline silicon is composed of many smaller silicon grains of varied crystallographic orientation (Figure 8). This material can be synthesized easily by allowing liquid silicon to cool using a seed crystal of the desired crystal structure. Additionally, other methods for crystallizing amorphous silicon to form polycrystalline exist such as high temperature chemical vapor deposition (CVD). Each grain is crystalline over the width of the grain. The grain boundary separates the grains where the adjoining grain is at a different orientation than its neighbor. The grain boundary separates regions of different crystal structure thus serving as a center for recombination. '*d*' here is a characteristic grain size, which should be maximized for maximum film efficiency. Typical values of *d* are about 1 micrometer.

Presently, polysilicon is commonly used for the conducting gate materials in semiconductor devices such as MOSFETs; however, it has potential for large-scale photovoltaic devices (Streetman and Banerjee, 2000; Ghosh et al., 1980). The abundance, stability, and low toxicity of silicon, combined with the low cost of polysilicon relative to single crystals makes this variety of material attractive for photovoltaic production (Ghosh et al. 1980). Grain size has been shown to have an effect on the efficiency of polycrystalline solar cells. Solar cell efficiency increases with grain size. This effect is due to reduce recombination in the solar cell. Recombination, which is a limiting factor for current in a solar cell, occurs more prevalently at grain boundaries as shown in Figure 8 (Ghosh et al., 1980).

Fig. 8. Grain boundaries of polysilicon

The resistivity, mobility, and free-carrier concentration in monocrystalline silicon vary with doping concentration of the single crystal silicon. Whereas the doping of polycrystalline silicon does have an effect on the resistivity, mobility, and free-carrier concentration, these properties strongly depend on the polycrystalline grain size, which is a physical parameter that the material scientist can manipulate (Ghosh et al., 1980). Through the methods of crystallization to form polycrystalline silicon, a scientist can control the size of the polycrystalline grains which will vary the physical properties of the material.

## **5.1 Novel ideas for polycrystalline silicon**

492 Crystallization – Science and Technology

material with a bandgap of around 1.12 eV (the same as single-crystal silicon) compared to the bandgap of amorphous silicon of 1.7-1.8 eV bandgap (Shah et al*.,* 2003). Tandem solar cells are then attractive since they can be fabricated with a bandgap similar to single-crystal

Polycrystalline silicon is composed of many smaller silicon grains of varied crystallographic orientation (Figure 8). This material can be synthesized easily by allowing liquid silicon to cool using a seed crystal of the desired crystal structure. Additionally, other methods for crystallizing amorphous silicon to form polycrystalline exist such as high temperature chemical vapor deposition (CVD). Each grain is crystalline over the width of the grain. The grain boundary separates the grains where the adjoining grain is at a different orientation than its neighbor. The grain boundary separates regions of different crystal structure thus serving as a center for recombination. '*d*' here is a characteristic grain size, which should be

maximized for maximum film efficiency. Typical values of *d* are about 1 micrometer.

prevalently at grain boundaries as shown in Figure 8 (Ghosh et al., 1980).

Presently, polysilicon is commonly used for the conducting gate materials in semiconductor devices such as MOSFETs; however, it has potential for large-scale photovoltaic devices (Streetman and Banerjee, 2000; Ghosh et al., 1980). The abundance, stability, and low toxicity of silicon, combined with the low cost of polysilicon relative to single crystals makes this variety of material attractive for photovoltaic production (Ghosh et al. 1980). Grain size has been shown to have an effect on the efficiency of polycrystalline solar cells. Solar cell efficiency increases with grain size. This effect is due to reduce recombination in the solar cell. Recombination, which is a limiting factor for current in a solar cell, occurs more

The resistivity, mobility, and free-carrier concentration in monocrystalline silicon vary with doping concentration of the single crystal silicon. Whereas the doping of polycrystalline silicon does have an effect on the resistivity, mobility, and free-carrier concentration, these properties strongly depend on the polycrystalline grain size, which is a physical parameter

silicon but with the ease of amorphous silicon.

Fig. 8. Grain boundaries of polysilicon

The use of polycrystalline silicon in the production of solar cells requires less material and therefore provides for higher profits and increased manufacturing throughput. Polycrystalline silicon does not need to be deposited on a silicon wafer to form a solar cell, rather it can be deposited on other-cheaper materials, thus reducing the cost. Not requiring a silicon wafer alleviates the silicon shortages occasionally faced by the microelectronics industry. An example of not using a silicon wafer is crystalline silicon on glass (CSG) materials (Basore, 2006).

A primary concern in the photovoltaics industry is cell efficiency. However, sufficient cost savings from cell manufacturing can be suitable to offset reduced efficiency in the field, such as the use of larger solar cell arrays compared with more compact or higher efficiency designs. Designs such as CSG are attractive because of a low cost of production even with reduced efficiency (Basore, 2006). Higher efficiency devices yield modules that occupy less space and are more compacts however the 5-10 % efficiency of typical CSG devices still makes them attractive for installation in large central-service stations, such as a power station (Basore, 2006). The issue of efficiency versus cost is a value decision of whether one requires an "energy dense" solar cell or sufficient area is available for the installation of less expensive alternatives. For instance, a solar cell used for power generation in a remote location might require a more highly efficient solar cell than one used for low-power applications, such as solar accent lighting or pocket calculators, or near established power grids.

Thin film silicon photovoltaics are typically produced by chemical vapor deposition processes yielding an amorphous, polycrystalline, or nanocrystalline film. Conventionally, amorphous silicon thin films are most common. Silicon is usually deposited on glass, plastic, or metallic substrates coated with a transparent conducting oxide material. While chalcogenide-based Cadmium-Tellurium (CdTe) and Copper-Indium-Selenium (CIS) polycrystalline thin films cells have been developed in the lab with great success, there is still industry interest in silicon-based thin film cells. Silicon-based devices exhibit fewer problems than their CdTe and CIS counterparts such as toxicity and humidity issues with CdTe cells and low manufacturing yields of CIS due to material complexity. Additionally, due to political resistance to the use non-"green" materials in solar energy production, there is no stigma in the use of standard silicon. Three major silicon-based module designs dominate: amorphous silicon cells, amorphous or microcrystalline tandem cells, and thinfilm polycrystalline silicon on glass (Green, 2003).

Amorphous or microcrystalline silicon consists of a mixed phase of small crystalline regions surrounded by amorphous material. This material typically behaves more like crystalline silicon than the amorphous variety. A 3-month field study has shown that hybrid amorphous or microcrystalline cells degrade roughly to the same degree as triple-junction amorphous cells while maintaining higher conversion efficiencies (7.0% versus 5.0% as measured at the conclusion of the study). This result suggests hybrid designs of this type may supplant traditional amorphous-based modules (Green, 2003).

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 495

substrates that cannot be exposed to the high temperatures of standard annealing, polymers for instance. Polymer-backed solar cells are of interest for seamlessly integrated power

A third method for crystallizing amorphous silicon is the use of thermal plasma jet. This strategy is an attempt to alleviate some of the problems associated with laser processing – namely the small region of crystallization and the high cost of the process on a production scale. The plasma torch is a simple piece of equipment that is used to thermally anneal the amorphous silicon. Compared to the laser method, this technique is simpler and more cost

Plasma torch annealing is attractive because the process parameters and equipment dimension can be changed easily to yield varying levels of performance. A high level of crystallization (~90%) can be obtained with this method. Disadvantages include difficulty achieving uniformity in the crystallization of the film. While this method is applied frequently to silicon on a glass substrate, processing temperatures may be too high for

Laser annealing on the thin film is a high-speed process. It depends on the pulse duration of the excimer laser normally in nanoseconds region and the threshold energy to nucleate and optimum energy to stop crystallization. Such short duration of annealing process is referred as transient crystallization of amorphous silicon film (Viatella and Singh, 1997). Within transient duration, the annealing process still needs to pass through several critical points.

In order to figure-out the physical change involve in laser annealing it is better to understand the basic concept of phase transformation during heating and cooling of hardening process. There are several important stages in heating and cooling of material. The crucial parameter is temperature at several critical points (Figure 9). The "critical points" are the temperatures at which certain changes in the chemical composition of the material take place, during both heating and cooling. Steel for example at normal temperatures has its carbon (which is its chief hardening element) in a certain form called *pearlite* carbon, and if the steel is heated to a certain temperature, a change occurs and the pearlite becomes *martensite* or hardening carbon. If the steel is allowed to cool slowly, the hardening carbon changes back to pearlite. The points at which these changes occur are the decalescence and recalescence or critical points, and the effect of these molecular changes is as follows: When a piece of steel is heated to a certain point, it continues to absorb heat without appreciably rising in temperature, although its immediate surroundings may be hotter than the steel. This is the *decalescence* point. Similarly, steel cooling slowly from a high heat will, at a certain temperature, actually increase in temperature, although its surroundings may be colder. This takes place at the *recalescence* point. The recalescence point is lower than the decalescence point, and the lower of these points does not manifest itself unless the higher one has first been fully passed. These critical points have a direct relation to the hardening of steel. Unless a temperature sufficient to reach the decalescence point is obtained, so that

In the next section, such phase transformation point will be described in detail.

production schemes that involve placing photovoltaics on everyday surfaces.

effective (Lee et al., 2009).

**6. Transformation temperature** 

polymers.

**6.1 Critical points** 

A new attempt to fuse the advantages of bulk silicon with those of thin-film devices is thin film polycrystalline silicon on glass. These modules are produced by depositing an antireflection coating and doped silicon onto textured glass substrates using plasmaenhanced chemical vapor deposition (PECVD). The texture in the glass enhances the efficiency of the cell by approximately 3% by reducing the amount of incident light reflecting from the solar cell and trapping light inside the solar cell. The silicon film is crystallized by an annealing step, temperatures of 400 – 600 C, resulting in polycrystalline silicon.

These new devices show energy conversion efficiencies of 8% and high manufacturing yields of >90%. Crystalline silicon on glass (CSG), where the polycrystalline silicon is 1-2 micrometers, is noted for its stability and durability; the use of thin film techniques also contributes to a cost savings over bulk photovoltaics. These modules do not require the presence of a transparent conducting oxide layer. This simplifies the production process twofold; not only can this step be skipped, but the absence of this layer makes the process of constructing a contact scheme much simpler. Both of these simplifications further reduce the cost of production. Despite the numerous advantages over alternative design, production cost estimations on a per unit area basis show that these devices are comparable in cost to single-junction amorphous thin film cells (Green, 2003).

#### **5.2 Low temperature induced crystallization of amorphous silicon**

Amorphous silicon can be transformed to crystalline silicon using well-understood and widely implemented high-temperature annealing processes. This typical method is the typical method used in industry but requires high-temperature compatible materials, such as special high temperature glass that is expensive to produce. However, there are many applications for which this is an inherently unattractive production method. Flexible solar cells have been a topic of interest for less conspicuous-integrated power generation than solar power farms. These modules may be placed in areas where traditional cells would not be feasible, such as wrapped around a telephone pole or cell phone tower. In this application a photovoltaic material may be applied to a flexible substrate, often a polymer. Such substrates cannot survive the high temperatures experienced during traditional annealing. Instead, novel methods of crystallizing the silicon without disturbing the underlying substrate have been studied extensively. Aluminum-induced crystallization (AIC) and local laser crystallization are common in the literature, however not extensively used in industry. In both of these methods, amorphous silicon (a-Si or a-Si:H) is grown using traditional techniques such as plasma-enhanced chemical vapor deposition (PECVD). The crystallization methods diverge during post-deposition processing.

Another method of achieving the same result is the use of a laser to heat the silicon locally without heating the underlying substrate beyond some upper temperature limit. An excimer laser or, alternatively, green lasers such as a frequency-doubled Nd:YAG laser is used to heat the amorphous silicon, supplying energy necessary to nucleate grain growth. The laser fluence must be carefully controlled in order to induce crystallization without causing widespread melting. Crystallization of the film occurs as a very small portion of the silicon film is melted and allowed to cool. Ideally, the laser should melt the silicon film through its entire thickness, but not damage the substrate. Toward this end, a layer of silicon dioxide is sometimes added to act as a thermal barrier (Yuan et al., 2009). This allows the use of substrates that cannot be exposed to the high temperatures of standard annealing, polymers for instance. Polymer-backed solar cells are of interest for seamlessly integrated power production schemes that involve placing photovoltaics on everyday surfaces.

A third method for crystallizing amorphous silicon is the use of thermal plasma jet. This strategy is an attempt to alleviate some of the problems associated with laser processing – namely the small region of crystallization and the high cost of the process on a production scale. The plasma torch is a simple piece of equipment that is used to thermally anneal the amorphous silicon. Compared to the laser method, this technique is simpler and more cost effective (Lee et al., 2009).

Plasma torch annealing is attractive because the process parameters and equipment dimension can be changed easily to yield varying levels of performance. A high level of crystallization (~90%) can be obtained with this method. Disadvantages include difficulty achieving uniformity in the crystallization of the film. While this method is applied frequently to silicon on a glass substrate, processing temperatures may be too high for polymers.

## **6. Transformation temperature**

Laser annealing on the thin film is a high-speed process. It depends on the pulse duration of the excimer laser normally in nanoseconds region and the threshold energy to nucleate and optimum energy to stop crystallization. Such short duration of annealing process is referred as transient crystallization of amorphous silicon film (Viatella and Singh, 1997). Within transient duration, the annealing process still needs to pass through several critical points. In the next section, such phase transformation point will be described in detail.

## **6.1 Critical points**

494 Crystallization – Science and Technology

A new attempt to fuse the advantages of bulk silicon with those of thin-film devices is thin film polycrystalline silicon on glass. These modules are produced by depositing an antireflection coating and doped silicon onto textured glass substrates using plasmaenhanced chemical vapor deposition (PECVD). The texture in the glass enhances the efficiency of the cell by approximately 3% by reducing the amount of incident light reflecting from the solar cell and trapping light inside the solar cell. The silicon film is crystallized by an annealing step, temperatures of 400 – 600 C, resulting in polycrystalline

These new devices show energy conversion efficiencies of 8% and high manufacturing yields of >90%. Crystalline silicon on glass (CSG), where the polycrystalline silicon is 1-2 micrometers, is noted for its stability and durability; the use of thin film techniques also contributes to a cost savings over bulk photovoltaics. These modules do not require the presence of a transparent conducting oxide layer. This simplifies the production process twofold; not only can this step be skipped, but the absence of this layer makes the process of constructing a contact scheme much simpler. Both of these simplifications further reduce the cost of production. Despite the numerous advantages over alternative design, production cost estimations on a per unit area basis show that these devices are comparable in cost to

Amorphous silicon can be transformed to crystalline silicon using well-understood and widely implemented high-temperature annealing processes. This typical method is the typical method used in industry but requires high-temperature compatible materials, such as special high temperature glass that is expensive to produce. However, there are many applications for which this is an inherently unattractive production method. Flexible solar cells have been a topic of interest for less conspicuous-integrated power generation than solar power farms. These modules may be placed in areas where traditional cells would not be feasible, such as wrapped around a telephone pole or cell phone tower. In this application a photovoltaic material may be applied to a flexible substrate, often a polymer. Such substrates cannot survive the high temperatures experienced during traditional annealing. Instead, novel methods of crystallizing the silicon without disturbing the underlying substrate have been studied extensively. Aluminum-induced crystallization (AIC) and local laser crystallization are common in the literature, however not extensively used in industry. In both of these methods, amorphous silicon (a-Si or a-Si:H) is grown using traditional techniques such as plasma-enhanced chemical vapor deposition (PECVD).

Another method of achieving the same result is the use of a laser to heat the silicon locally without heating the underlying substrate beyond some upper temperature limit. An excimer laser or, alternatively, green lasers such as a frequency-doubled Nd:YAG laser is used to heat the amorphous silicon, supplying energy necessary to nucleate grain growth. The laser fluence must be carefully controlled in order to induce crystallization without causing widespread melting. Crystallization of the film occurs as a very small portion of the silicon film is melted and allowed to cool. Ideally, the laser should melt the silicon film through its entire thickness, but not damage the substrate. Toward this end, a layer of silicon dioxide is sometimes added to act as a thermal barrier (Yuan et al., 2009). This allows the use of

single-junction amorphous thin film cells (Green, 2003).

**5.2 Low temperature induced crystallization of amorphous silicon** 

The crystallization methods diverge during post-deposition processing.

silicon.

In order to figure-out the physical change involve in laser annealing it is better to understand the basic concept of phase transformation during heating and cooling of hardening process. There are several important stages in heating and cooling of material. The crucial parameter is temperature at several critical points (Figure 9). The "critical points" are the temperatures at which certain changes in the chemical composition of the material take place, during both heating and cooling. Steel for example at normal temperatures has its carbon (which is its chief hardening element) in a certain form called *pearlite* carbon, and if the steel is heated to a certain temperature, a change occurs and the pearlite becomes *martensite* or hardening carbon. If the steel is allowed to cool slowly, the hardening carbon changes back to pearlite. The points at which these changes occur are the decalescence and recalescence or critical points, and the effect of these molecular changes is as follows: When a piece of steel is heated to a certain point, it continues to absorb heat without appreciably rising in temperature, although its immediate surroundings may be hotter than the steel. This is the *decalescence* point. Similarly, steel cooling slowly from a high heat will, at a certain temperature, actually increase in temperature, although its surroundings may be colder. This takes place at the *recalescence* point. The recalescence point is lower than the decalescence point, and the lower of these points does not manifest itself unless the higher one has first been fully passed. These critical points have a direct relation to the hardening of steel. Unless a temperature sufficient to reach the decalescence point is obtained, so that

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 497

There are several researches on the recalescence phenomenon have been reported. Among them are Zhang and Atren, (1992), Armengol et al., (2003) and Mariucci et al., (2003). Zhang and Atren, (1992) have studied the recalescence characteristics in rapid solidification of copper by using a thermokinetic model. They have investigated the effects of the heat transfer coefficient, the melt thickness and the nucleation temperature. Their results had shown that the lower nucleation temperature and thinner melt lead to a longer recalescence effect while larger heat transfer coefficient results in a weaker recalescence effect. A

Mariuci group (2003) had claimed that crystallization less than complete melting point will experience the recalescence effect. Structural properties of thin polycrystalline silicon films, crystallized by single shot excimer laser annealing at different laser energy densities, have been investigated by various researchers (Armengol, et al., 2003,; Mariucci et al., 2003; Razak and Bidin, 2010; Bidin and Razak, 2011a, b). Formation of disk structures has been observed in a wide range of energy densities, from complete melting down to 180 mJ/cm2. These structures have been correlated to the lateral growth of grains starting from the small grains present in the central regions of the disks. They had proposed a new crystallization scenario for energy densities below the complete melting. In this framework, the recalescence effect plays an important role while the super lateral growth-regime is no longer a particular crystallization condition but simply represents the upper energy density limit of partial melt

Similar recalescence effect after bulk solidification in germanium films melted by nanosecond laser pulses had also reported by Armengol, et al., (2003). Rapid solidification

dimensionless number was derived to measure the extent of recalescence.

Fig. 9. Transformation temperature

**6.2 Recalescence effect** 

crystallization regime.

the pearlite carbon is changed into a hardening carbon, no hardening action can take place; and unless the steel is cooled suddenly before it reaches the recalescence point, thus preventing the changing back again from hardening to pearlite carbon, no hardening can take place. The critical points vary for different kinds of material, and must be determined by tests in each case.

Similar phase transformation also experience by other materials during heating and cooling process. The different may be arisen in the aspect of temperature and the time taken to complete the same process. If the steel is replaced with other material like any amorphous silicon thin film and take into account the duration involves in nanosecond time domain then the whole event is considered as a high-speed phenomenon. This includes the rate of absorption and release of heat which occurs in transient time and enough energy to cause the phase change. In case excimer laser ablation which posses in nanoseconds pulse duration and fluence more than 500 J/cm2, having high potential to cause the target thin film experience the same process as hardening steel. This implies excimer laser annealing on thin film should passing through the decalescence and recalescence points to ensure the formation of crystallization.

In general when amorphous silicon thin film is heated and cooling, normal phase change is occurred including the melting and re-solidification transformation. The phenomenon of this phase transformation is similar as illustrated in Figure 9. The red line is indicated the heating process, while the blue line represents the cooling process. In heating process, there are two important points that are decalescence and critical points. Decalescence is a phenomenon that occurs when a material is being heated and there is a sudden slowing in the rate of temperature increase. The slowing rate is due to the change in the internal crystal structure of the material. Meanwhile during cooling down or re-solidification, the phase change is passing through a recalescence point. A sudden spontaneous increase in the temperature of cooling resulting from an exothermic change in crystal structure occurs. This is the critical state in cooling process whereby the formation of crystallization is taken place.

Furthermore, a single pulse of the excimer laser, carried out temperature of 6.42 eV which is higher than threshold to break the bonding of silicon molecules. Once the excimer laser pulse is exposed on the film, it must passing through the decalescence and recalescence points in order to form crystallization. If a lower energy than threshold material or too high energy achieve up to the peak temperature in the transformation process, no crystallization will be formed on silicon film.

In both temperatures, there is latent heat of fusion is liberated from the solid during heating and from liquid during cooling. The latent heat is liberated during both processes to delay either the process of heating or cooling. The coexistent both solid and liquid associated in heating and cooling, resistant the rate of transformation. The advantage of this drawback is given an opportunity to nucleate and grow the crystallization in the silicon.

If the heating is over than optimum energy by exposing with higher number of pulses, the thin film is totally melted. No solid form is left to be as a seed for crystallization. In this case the treatment is achieved up to the peak temperature, passing over the decalescence point. Subsequently re-solidification occurs faster without passing through the recalescence point. Hence no formation of crystallization and end up with homogenous or no change in crystal structure in the excimer laser annealing process.

Fig. 9. Transformation temperature

## **6.2 Recalescence effect**

496 Crystallization – Science and Technology

the pearlite carbon is changed into a hardening carbon, no hardening action can take place; and unless the steel is cooled suddenly before it reaches the recalescence point, thus preventing the changing back again from hardening to pearlite carbon, no hardening can take place. The critical points vary for different kinds of material, and must be determined

Similar phase transformation also experience by other materials during heating and cooling process. The different may be arisen in the aspect of temperature and the time taken to complete the same process. If the steel is replaced with other material like any amorphous silicon thin film and take into account the duration involves in nanosecond time domain then the whole event is considered as a high-speed phenomenon. This includes the rate of absorption and release of heat which occurs in transient time and enough energy to cause the phase change. In case excimer laser ablation which posses in nanoseconds pulse duration and fluence more than 500 J/cm2, having high potential to cause the target thin film experience the same process as hardening steel. This implies excimer laser annealing on thin film should passing through the decalescence and recalescence points to ensure the

In general when amorphous silicon thin film is heated and cooling, normal phase change is occurred including the melting and re-solidification transformation. The phenomenon of this phase transformation is similar as illustrated in Figure 9. The red line is indicated the heating process, while the blue line represents the cooling process. In heating process, there are two important points that are decalescence and critical points. Decalescence is a phenomenon that occurs when a material is being heated and there is a sudden slowing in the rate of temperature increase. The slowing rate is due to the change in the internal crystal structure of the material. Meanwhile during cooling down or re-solidification, the phase change is passing through a recalescence point. A sudden spontaneous increase in the temperature of cooling resulting from an exothermic change in crystal structure occurs. This is the critical state in cooling process whereby the formation of crystallization is taken place. Furthermore, a single pulse of the excimer laser, carried out temperature of 6.42 eV which is higher than threshold to break the bonding of silicon molecules. Once the excimer laser pulse is exposed on the film, it must passing through the decalescence and recalescence points in order to form crystallization. If a lower energy than threshold material or too high energy achieve up to the peak temperature in the transformation process, no crystallization

In both temperatures, there is latent heat of fusion is liberated from the solid during heating and from liquid during cooling. The latent heat is liberated during both processes to delay either the process of heating or cooling. The coexistent both solid and liquid associated in heating and cooling, resistant the rate of transformation. The advantage of this drawback is

If the heating is over than optimum energy by exposing with higher number of pulses, the thin film is totally melted. No solid form is left to be as a seed for crystallization. In this case the treatment is achieved up to the peak temperature, passing over the decalescence point. Subsequently re-solidification occurs faster without passing through the recalescence point. Hence no formation of crystallization and end up with homogenous or no change in crystal

given an opportunity to nucleate and grow the crystallization in the silicon.

by tests in each case.

formation of crystallization.

will be formed on silicon film.

structure in the excimer laser annealing process.

There are several researches on the recalescence phenomenon have been reported. Among them are Zhang and Atren, (1992), Armengol et al., (2003) and Mariucci et al., (2003). Zhang and Atren, (1992) have studied the recalescence characteristics in rapid solidification of copper by using a thermokinetic model. They have investigated the effects of the heat transfer coefficient, the melt thickness and the nucleation temperature. Their results had shown that the lower nucleation temperature and thinner melt lead to a longer recalescence effect while larger heat transfer coefficient results in a weaker recalescence effect. A dimensionless number was derived to measure the extent of recalescence.

Mariuci group (2003) had claimed that crystallization less than complete melting point will experience the recalescence effect. Structural properties of thin polycrystalline silicon films, crystallized by single shot excimer laser annealing at different laser energy densities, have been investigated by various researchers (Armengol, et al., 2003,; Mariucci et al., 2003; Razak and Bidin, 2010; Bidin and Razak, 2011a, b). Formation of disk structures has been observed in a wide range of energy densities, from complete melting down to 180 mJ/cm2. These structures have been correlated to the lateral growth of grains starting from the small grains present in the central regions of the disks. They had proposed a new crystallization scenario for energy densities below the complete melting. In this framework, the recalescence effect plays an important role while the super lateral growth-regime is no longer a particular crystallization condition but simply represents the upper energy density limit of partial melt crystallization regime.

Similar recalescence effect after bulk solidification in germanium films melted by nanosecond laser pulses had also reported by Armengol, et al., (2003). Rapid solidification

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 499

conventional annealing method to anneal the silicon thin film. As reported by previous researchers (Carluccio et. al., 1997, Han et. al., 2003) combination annealing techniques between conventional and laser annealing would produce a better poly-si thin film. Figure 10 shows a different in terms of grain size structure between polysilicon thin film PSTF annealed by (a) conventional annealing and (b) combination of conventional and excimer laser annealing. Polycrystalline silicon thin film (PSTF) has been an interest currently due to

high potential as electrical interconnections in microelectronics and microsensors.

Fig. 10. FESEM images of poly-Si (a) annealed at 600 °C for 24 h (*a*) and the combination

In order to optimize the annealing process, a super lateral growth SLG energy density needs to be identified. The SLG point is known as the critical energy density or laser fluence desires to generate the maximum crystallization. Greater than this energy normally the crystallization will be reduced and soon eliminated. In other word the SLG energy density is equivalent with the critical point in the temperature transformation curve (Figure 9) during heating process. Beyond this energy, the silicon film will totally melt and cause elimination of spontaneous crystallized formation. This implies that calibration needs to be done in

There are two parameters commonly used to quantify the degree of crystallization that are, grain size and surface roughness. The later parameter is due to the melting effect. Both parameters are utilized for measurement of crystallization. If the boundaries of the grain size is expanding or the surface roughness is increasing after being exposed by a series of excimer laser pulses that means the annealing process is within the decalescence and recalescence region. Once the optimum energy density that is super lateral growth energy is achieved that is at the critical point in the temperature transformation curve, no more number of pulses can be added or no more laser energy density can be increased.

annealing with ELA at 470 mJ cm – 2 (Han et. al., 2003)

**6.3 Super lateral growth energy** 

determining this crucial energy.

dynamics in amorphous germanium films melted by nanosecond laser pulses has been analyzed by means of single-shot subnanosecond time resolved reflectivity measurements using a streak camera based setup. The results showed that once a minimum melt depth is induced, a bulk solidification process followed by the release of the solidification enthalpy dominates the solidification process. Moreover, the laser-melted material solidifies completely before being remelted as a consequence of the solidification enthalpy release, something only observed, up to date, upon irradiation with picosecond laser pulses (Armengol et al., 2003).

Bossuyt and Greer, (2004) have been investigated the effects of positive feedback on crystallization kinetics and recalescence. They described that in a series of alloys in the Cu-Ni-Ti-Zr system, with compositions close to the bulk glass forming alloy Cu47Ni8Ti34Zr11 (Vit101), the nucleation density has been shown to be spatially inhomogeneous; in an amorphous matrix there are spherical clusters with a high density of nanocrystals. The implied positive feedback in the nucleation rate was analyzed in terms of recalescence instability, where the latent heat released upon crystallization causes the nucleation rate to increase locally, if the nucleation rate is an increasing function of temperature and thermal diffusivity is low enough to avoid distributing the heat evenly over the sample. In deeply under cooled liquids, the first of these requirements is satisfied, but not for the second requirement.

Bossuyt and Greer have used numerical technique to trace the cooling rate with and without the feedback heat in crystallization formation. They identified a critical cooling rate for recalescence to occur. At slightly faster cooling rates, the heat released by crystallization causes the actual cooling rate to drop to nearly zero, but the heat output is never sufficient to actually raise the temperature. When that does occur, there is an abrupt transition from approximately 40% crystallization at cooling rates just above the critical cooling rate, to full crystallization below the critical cooling rate. Furthermore they claimed that disabling the heat of crystallization eliminates the recalescence effect from the temperature traces, and the transition to full crystallization is now much more gradual, without a clearly defined critical cooling rate.

Other feedback mechanisms have effects similar to those of recalescence: there is always a threshold condition where the acceleration due to feedback offsets the quenching effect of decreasing temperature. At cooling rates above this threshold, the crystallization kinetics did not deviate much from those in the absence of feedback. The effect of feedback becomes noticeable over a relatively narrow range of cooling rates just above the threshold, and then dominant below the threshold, abruptly increasing the degree of crystallization (of at least one of the phases). The decalescence and recalescence effect are shown in Figure 9. Further explanation about the critical points will be discussed on section 6.5.

However not many researchers have been realized about the decalescence effect in the recrystallization process. Without determination the decalescence effect during heating process the recalescence point cannot be achieved. Therefore in order to crystallization of polycrystalline silicon film the emphasized is focusing on these both phenomena (decalescence and recalescence effects). In this project, silicon thin film is crystallization by controlling the number of laser pulses.

In most cases, silicon is deposited onto a corning glass substrate and annealed with high energy excimer laser which requires high fabrication cost for industrial purposes. In this attempt, low energy ArF (wavelength of 193 nm) excimer laser was employed together with

dynamics in amorphous germanium films melted by nanosecond laser pulses has been analyzed by means of single-shot subnanosecond time resolved reflectivity measurements using a streak camera based setup. The results showed that once a minimum melt depth is induced, a bulk solidification process followed by the release of the solidification enthalpy dominates the solidification process. Moreover, the laser-melted material solidifies completely before being remelted as a consequence of the solidification enthalpy release, something only observed, up to date, upon irradiation with picosecond laser pulses

Bossuyt and Greer, (2004) have been investigated the effects of positive feedback on crystallization kinetics and recalescence. They described that in a series of alloys in the Cu-Ni-Ti-Zr system, with compositions close to the bulk glass forming alloy Cu47Ni8Ti34Zr11 (Vit101), the nucleation density has been shown to be spatially inhomogeneous; in an amorphous matrix there are spherical clusters with a high density of nanocrystals. The implied positive feedback in the nucleation rate was analyzed in terms of recalescence instability, where the latent heat released upon crystallization causes the nucleation rate to increase locally, if the nucleation rate is an increasing function of temperature and thermal diffusivity is low enough to avoid distributing the heat evenly over the sample. In deeply under cooled

liquids, the first of these requirements is satisfied, but not for the second requirement.

explanation about the critical points will be discussed on section 6.5.

controlling the number of laser pulses.

Bossuyt and Greer have used numerical technique to trace the cooling rate with and without the feedback heat in crystallization formation. They identified a critical cooling rate for recalescence to occur. At slightly faster cooling rates, the heat released by crystallization causes the actual cooling rate to drop to nearly zero, but the heat output is never sufficient to actually raise the temperature. When that does occur, there is an abrupt transition from approximately 40% crystallization at cooling rates just above the critical cooling rate, to full crystallization below the critical cooling rate. Furthermore they claimed that disabling the heat of crystallization eliminates the recalescence effect from the temperature traces, and the transition to full crystallization is now much more gradual, without a clearly defined critical cooling rate. Other feedback mechanisms have effects similar to those of recalescence: there is always a threshold condition where the acceleration due to feedback offsets the quenching effect of decreasing temperature. At cooling rates above this threshold, the crystallization kinetics did not deviate much from those in the absence of feedback. The effect of feedback becomes noticeable over a relatively narrow range of cooling rates just above the threshold, and then dominant below the threshold, abruptly increasing the degree of crystallization (of at least one of the phases). The decalescence and recalescence effect are shown in Figure 9. Further

However not many researchers have been realized about the decalescence effect in the recrystallization process. Without determination the decalescence effect during heating process the recalescence point cannot be achieved. Therefore in order to crystallization of polycrystalline silicon film the emphasized is focusing on these both phenomena (decalescence and recalescence effects). In this project, silicon thin film is crystallization by

In most cases, silicon is deposited onto a corning glass substrate and annealed with high energy excimer laser which requires high fabrication cost for industrial purposes. In this attempt, low energy ArF (wavelength of 193 nm) excimer laser was employed together with

(Armengol et al., 2003).

conventional annealing method to anneal the silicon thin film. As reported by previous researchers (Carluccio et. al., 1997, Han et. al., 2003) combination annealing techniques between conventional and laser annealing would produce a better poly-si thin film. Figure 10 shows a different in terms of grain size structure between polysilicon thin film PSTF annealed by (a) conventional annealing and (b) combination of conventional and excimer laser annealing. Polycrystalline silicon thin film (PSTF) has been an interest currently due to high potential as electrical interconnections in microelectronics and microsensors.

Fig. 10. FESEM images of poly-Si (a) annealed at 600 °C for 24 h (*a*) and the combination annealing with ELA at 470 mJ cm – 2 (Han et. al., 2003)

## **6.3 Super lateral growth energy**

In order to optimize the annealing process, a super lateral growth SLG energy density needs to be identified. The SLG point is known as the critical energy density or laser fluence desires to generate the maximum crystallization. Greater than this energy normally the crystallization will be reduced and soon eliminated. In other word the SLG energy density is equivalent with the critical point in the temperature transformation curve (Figure 9) during heating process. Beyond this energy, the silicon film will totally melt and cause elimination of spontaneous crystallized formation. This implies that calibration needs to be done in determining this crucial energy.

There are two parameters commonly used to quantify the degree of crystallization that are, grain size and surface roughness. The later parameter is due to the melting effect. Both parameters are utilized for measurement of crystallization. If the boundaries of the grain size is expanding or the surface roughness is increasing after being exposed by a series of excimer laser pulses that means the annealing process is within the decalescence and recalescence region. Once the optimum energy density that is super lateral growth energy is achieved that is at the critical point in the temperature transformation curve, no more number of pulses can be added or no more laser energy density can be increased.

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 501

film is shown in Figure 12. The frames of crystallization are arranged in the increasing order of the energy density. Prior annealing, the grain size of the amorphous silicon thin film is measured. The average nanocrystal size, Gav of the amorphous silicon film before doping and heating as shown in Figure 12(a) is approximately 17 nm. After experienced four hours heat treament via conventional annealing, a new crystal structure is realized. The crystallization pattern almost uniform with an increment up to 56 nm as depicted in Figure 12(b). Immediately after crystalization with a single shot of excimer laser corresponding to energy density of 65.50 mJcm-2, another new pattern nanostructure is realized. The crystallization of the silicon film is accelerated further to a grain size of 75 nm as illustrated in Figure 12(c). The crystallization of thin film is continuously increasing after received the sequential number of pulses from argon fluoride ArF excimer laser. The maximum crystallization of silicon film is notified after received five number of pulses as depicted in Figure 12(g). The enlargement achieved approximately to 143 nm corresponding to the super lateral growth energy of 345 mJcm-2. Beyond the critical energy as shown in Figure

Fig. 12. AFM images of silicon thin film with magnification area of 1000 nm x 1000 nm at

12(h) the crystallization reduced drastically.

different energy density.

Otherwise, the excess energy exposed that would cause the film totally melted and no solid form leaves at the interface between substrate and film to seed the crystallization.

Hence in excimer laser annealing the most crucial process is to identify the SLG energy. Under this particular SLG energy, the heating and the cooling are associated with the decalescence and recalescence effects. This implies that, the phase change is accompanied with the coexistence of liquid and solid forms. This is illustrated in Figure 11. At the interface between film and substrate, there are some particles remain in solid form whereas the rest in liquid phase. The left over the particles are used to seeding the crystallization. This region is similar as the condition achieved at recalescence point in the transformation temperature curve during quenching or cooling period.

Fig. 11. Nearly complete melt of amorphous silicon layer. The liquid/solid interface seeds the crystallization process

It is well known that laser ablation involves photothermal and photochemical process, depending on the nature of the material used and the experimental conditions, for example laser fluence, wavelength, and pulse duration (Sato et al., 2001; Serafetinides et al., 2001; Hahn et al., 1999). A 193 nm wavelength radiation can cause a neat and clean etch region because at this wavelength the molar extinction coefficient of the material is high (~ 104 L mol-1 cm-1) so that 95% of the light is absorbed in the first 300 nm of the substrate. On the other hand, a laser with longer wavelength (~ 500 nm) light melts and damage nearby regions of the sample as a longer wavelength light vibrationally excites the irradiation sites. This process can eventually cause material to desorb from the heated region (Barbara et al.*,* 1984).

## **7. Atomic force microscope analysis**

The crystallization result is quantified based on metallurgical technique. The annealed silicon thin film is examined under the atomic force microscope. The scan area for each sample is fixed to 5000 nm X 5000 nm and zoomed until 1000 nm X 1000 nm to get clearer image of the thin film surface. The typical of the nucleation and the growth of silicon thin

Otherwise, the excess energy exposed that would cause the film totally melted and no solid

Hence in excimer laser annealing the most crucial process is to identify the SLG energy. Under this particular SLG energy, the heating and the cooling are associated with the decalescence and recalescence effects. This implies that, the phase change is accompanied with the coexistence of liquid and solid forms. This is illustrated in Figure 11. At the interface between film and substrate, there are some particles remain in solid form whereas the rest in liquid phase. The left over the particles are used to seeding the crystallization. This region is similar as the condition achieved at recalescence point in the transformation

Fig. 11. Nearly complete melt of amorphous silicon layer. The liquid/solid interface seeds

It is well known that laser ablation involves photothermal and photochemical process, depending on the nature of the material used and the experimental conditions, for example laser fluence, wavelength, and pulse duration (Sato et al., 2001; Serafetinides et al., 2001; Hahn et al., 1999). A 193 nm wavelength radiation can cause a neat and clean etch region because at this wavelength the molar extinction coefficient of the material is high (~ 104 L mol-1 cm-1) so that 95% of the light is absorbed in the first 300 nm of the substrate. On the other hand, a laser with longer wavelength (~ 500 nm) light melts and damage nearby regions of the sample as a longer wavelength light vibrationally excites the irradiation sites. This process can eventually

The crystallization result is quantified based on metallurgical technique. The annealed silicon thin film is examined under the atomic force microscope. The scan area for each sample is fixed to 5000 nm X 5000 nm and zoomed until 1000 nm X 1000 nm to get clearer image of the thin film surface. The typical of the nucleation and the growth of silicon thin

cause material to desorb from the heated region (Barbara et al.*,* 1984).

**7. Atomic force microscope analysis** 

form leaves at the interface between substrate and film to seed the crystallization.

temperature curve during quenching or cooling period.

the crystallization process

film is shown in Figure 12. The frames of crystallization are arranged in the increasing order of the energy density. Prior annealing, the grain size of the amorphous silicon thin film is measured. The average nanocrystal size, Gav of the amorphous silicon film before doping and heating as shown in Figure 12(a) is approximately 17 nm. After experienced four hours heat treament via conventional annealing, a new crystal structure is realized. The crystallization pattern almost uniform with an increment up to 56 nm as depicted in Figure 12(b). Immediately after crystalization with a single shot of excimer laser corresponding to energy density of 65.50 mJcm-2, another new pattern nanostructure is realized. The crystallization of the silicon film is accelerated further to a grain size of 75 nm as illustrated in Figure 12(c). The crystallization of thin film is continuously increasing after received the sequential number of pulses from argon fluoride ArF excimer laser. The maximum crystallization of silicon film is notified after received five number of pulses as depicted in Figure 12(g). The enlargement achieved approximately to 143 nm corresponding to the super lateral growth energy of 345 mJcm-2. Beyond the critical energy as shown in Figure 12(h) the crystallization reduced drastically.

Fig. 12. AFM images of silicon thin film with magnification area of 1000 nm x 1000 nm at different energy density.

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 503

Fig. 14. Grain size as function of energy density

**surface roughness, nm**

**8. Conclusion** 

Fig. 15. Surface roughness of silicon thin film with respect to number of pulses

In general, the excimer laser annealing is capable to crystallize the silicon thin film based on the increment of the grain size as well as the surface roughness. In the process of annealing, the film experience transient heating and freezing. The UV light of excimer laser carried out

The stress developed during solidification due to the melting effect of silicon thin film is quantified based on surface roughness measurement. A typical example of the surface roughness of the silicon copper film is shown in Figure 13. Figure 13(a) indicates the roughness before annealing process. The measurement of surface roughness for the amorphous silicon thin film is 0.3 nm. After annealing the intergration of crystalization formation comprised of the agglomeration and coalesence of grain. Simialr results have also reported by He et al., (2007). The surface roughness of the annealed silicon thin film then increased up to 3.0 nm. The profile of the grain sized and surface roughness measurement are presented in Figure 14 and 15 respectively. However the maximum roughness is not overlapping with maximum grain size. The possible reason for such different may be due to the measured surface roughness is not matched with measured area of the grain size.

Fig. 13. Surface roughness of silicon at different number of pulses: a. o pulses, b. 3 pulses.

The stress developed during solidification due to the melting effect of silicon thin film is quantified based on surface roughness measurement. A typical example of the surface roughness of the silicon copper film is shown in Figure 13. Figure 13(a) indicates the roughness before annealing process. The measurement of surface roughness for the amorphous silicon thin film is 0.3 nm. After annealing the intergration of crystalization formation comprised of the agglomeration and coalesence of grain. Simialr results have also reported by He et al., (2007). The surface roughness of the annealed silicon thin film then increased up to 3.0 nm. The profile of the grain sized and surface roughness measurement are presented in Figure 14 and 15 respectively. However the maximum roughness is not overlapping with maximum grain size. The possible reason for such different may be due to the measured surface roughness is not matched with measured

Fig. 13. Surface roughness of silicon at different number of pulses: a. o pulses, b. 3 pulses.

area of the grain size.

Fig. 14. Grain size as function of energy density

Fig. 15. Surface roughness of silicon thin film with respect to number of pulses

#### **8. Conclusion**

In general, the excimer laser annealing is capable to crystallize the silicon thin film based on the increment of the grain size as well as the surface roughness. In the process of annealing, the film experience transient heating and freezing. The UV light of excimer laser carried out

ArF Excimer Laser Annealing of Polycrystalline Silicon Thin Film 505

Bidin N. and Razak S.N. (2011b). Nucleation and Growth of Cu Thin Films on Silicon Base

Brannon, J.; Lankard, J.R.; Baise, A.; Burns I. and Kaufman F. (1985). Excimer Laser Etching

Brannon, J. (1997). Excimer Laser Ablation and Etching: Making a Mark in Many

Campbell, S. A. (2001).The Science and Engineering of Microelectronic Fabrication (2nd ed.),

Chen,Y.H.; Zheng, H.Y.; Wong, K.S. (1995). Excimer Laser Drilling of Polymers, Proc. SPIE ,

Ghosh, A. K.; Fishman, C. and Feng, T. (1980). Theory of the electrical and photovoltaic

Green, M. A. (2003). Crystalline and thin-film silicon solar cells: state of the art and future potential. Solar Energy 74 (3): 181. doi: 10.1016/S0038-092X (03)00187-7 Green, M. A. (2004). Recent Developments in Photovoltaics. Solar Energy 76 (1–3): 3.

Hahn, C. H.; Lippert, T. and Wokaun, A. (1999). Comparison of the Ablation Behaviour of

Han, W. H.;Kang, C.J. and Yong, S.K (2003). A Novel Structured Polycrystalline Silicon Thin

He M.; Ishihara R.; Metselaar W. and Beenakker K. (2007). Agglomeration of amorphous

Lee, H. S.; Choi, S.; Kim, S. W.; Hong, S. H. (2009). Crystallization of Amorphous Silicon

Mariucci L.; Pecora A.; Fortunato G.; Spinella C. and Bongiorno C. (2003). Crystallization

properties of polycrystalline Silicon. Journal of Applied Physics. 51 (1): 446. doi:

Polymer Films in the IR and UV with Nanosecond and Picosecond pulse. J. Phys.

Film Transistors tha increases the On/Off Current ratio. Semicond. Sci. Technol. 18

silicon film with high energy density excimer laser irradiation. Thin Solid Films. 515(5): 2872 Hecht, J. (1992). Understanding Laser: An Entry –Level Guide. New York: IEEE

Thin Film by Using a Thermal Plasma Jet. Thin Solid Films 517 (14): 4070. doi:

mechanisms in laser irradiated thin amorphous silicon films. Thin Solid Films.

Csele, M. (2004). Fundamental of Light Sources and Lasers. New Jersey: Wiley-Intersciece Carluccio R.; Cina S.; Fortunato G.; Friligkos S.; and Stoemenos J. (1997). Structure of Poly-Si

Films Obtained By Excimer Annealing. Thin Solid Films. 296: 57 Delmdahl, R. (2010). The excimer laser; precision engineering. Nature photonic. 4: 286

symposium on Functional materials, Bali Indonesia, April 27-28, 2011 Bossuyt, A. S. and Greer L. (2004). Effects of Positive Feedback on Crystallization Kinetics and Recalescence. Mat. Res.Soc. Symp. Proc. Vol. 806, Materials Research Society Brendel K.; Nickel N. H.; Lengsfeld P.; Schöpke A.; Sieber I.; Nerding M.; Strunk H. P. and

glass substrates. Thin Solid Films. 427(1-2): 86

New York: Oxford University Press, ISBN 0195136055

GAMLaser Inc. (2003). EX5 Excimer Laser Manual. Florida: User Manual

of Polymide*.* J.Appl. Phys*.* 58:2036

3184:202

10.1063/1.327342

Chem. B. 103: 1287

10.1016/j.tsf.2009.01.138

(845-849).

Press

427(1-2): 91

Applications. *Circuit & Devices*, 97: 11

doi:10.1016/S0038-092X(03)00065-3

Induced By Excimer Laser Annealing (Invited paper). Proc. of International

Fuhs W. (2003). Excimer laser crystallization of amorphous silicon on metal coated

energy of 6.42 eV which is equivalent with 7.5 104 K that is more than enough to melt the thin film. Considering the temporal length of pulse duration each of laser pulse is only 10 ns, this means the silicon copper thin film is experience an ultra-speed heating and freezing. The energetic excimer is enough to control the phase transformation with the decalesence point in which the melted is associated with some solid form to nucleate the crystal growth. The decalescence effect is controlled via the number of pulses. In contrast during rapid resolidification the liquid form is associated with recalescence effect which allows the formation of crystallization. The latent heat is released to delay the transformation from liquid to solid. Inherently the recalescence effect responsible to spontaneously crystallize the silicon film. When the energy density greater than the end limit of decalescence point which equivalent with the super lateral growth energy, the heating achieved the peak temperature where the film totally melted. Consequently no recalescence effect occur which result homogenous solidification without crystallization formation during cooling. Thus, in annealing it is crucial to determine the super lateral energy in order to ensure the process is within the decalescence and recalescence points to allow the formation of crystallization.

In addition, laser annealing process of silicon thin film promising well-arranged and large poly-Si grains. As far as we can see, laser annealing has become the most technique used to fabricate a good quality of poly-Si film for ultra-large scale integrated circuit devices, thin film transistors, active matrix liquid crystal displays, solar cells and other optoelectronic devices for industrial purposes. The employment of excimer laser for low temperature poly-Si (LTPS) thin film fabrication makes it possible to replace glass substrate with plastic substrate. This would certainly reduce the cost for large scale productions. Furthermore the high speed crystallization process will shorten the time to anneal as compare to the conventional method. The combination of time for annealing and the low energy consumption will give advantage for the production of high quality silicon thin film.

## **9. Acknowledgement**

This work has been supported from the government of Malaysia via FRGS grant vote 4F001 and GUP vote 00H10. We would like to acknowledge the support given by all technicians during experimental performances.

## **10. References**


energy of 6.42 eV which is equivalent with 7.5 104 K that is more than enough to melt the thin film. Considering the temporal length of pulse duration each of laser pulse is only 10 ns, this means the silicon copper thin film is experience an ultra-speed heating and freezing. The energetic excimer is enough to control the phase transformation with the decalesence point in which the melted is associated with some solid form to nucleate the crystal growth. The decalescence effect is controlled via the number of pulses. In contrast during rapid resolidification the liquid form is associated with recalescence effect which allows the formation of crystallization. The latent heat is released to delay the transformation from liquid to solid. Inherently the recalescence effect responsible to spontaneously crystallize the silicon film. When the energy density greater than the end limit of decalescence point which equivalent with the super lateral growth energy, the heating achieved the peak temperature where the film totally melted. Consequently no recalescence effect occur which result homogenous solidification without crystallization formation during cooling. Thus, in annealing it is crucial to determine the super lateral energy in order to ensure the process is within the decalescence and recalescence points to allow the formation of crystallization.

In addition, laser annealing process of silicon thin film promising well-arranged and large poly-Si grains. As far as we can see, laser annealing has become the most technique used to fabricate a good quality of poly-Si film for ultra-large scale integrated circuit devices, thin film transistors, active matrix liquid crystal displays, solar cells and other optoelectronic devices for industrial purposes. The employment of excimer laser for low temperature poly-Si (LTPS) thin film fabrication makes it possible to replace glass substrate with plastic substrate. This would certainly reduce the cost for large scale productions. Furthermore the high speed crystallization process will shorten the time to anneal as compare to the conventional method. The combination of time for annealing and the low energy

consumption will give advantage for the production of high quality silicon thin film.

This work has been supported from the government of Malaysia via FRGS grant vote 4F001 and GUP vote 00H10. We would like to acknowledge the support given by all technicians

Armengol, J.; Vega, F.; Chaoui, N.; Solis, J.; Afonso, C. N. (2003). Recalescence after bulk

Barbara, J. and Srinivasan, R., (1984). Microscopic Model for the Ablative

Basore, P. A. (2006). "CSG-2: Expanding the production of a new polycrystalline silicon PV

Bidin N. and Razak S.N. (2011a). Crystallization of Poly-Silicon Film by Different Annealing

http://www.csgsolar.com/downloads/CSG-2%20Basore.pdf.

SIECPC-2011 Riyadh Saudi Arabia, 24-26 April, 2011

solidification in germanium film melt by ns laser pulses. J. of Appl. Phys. 93(3):1505

Photodecomposition of Polymers by Far-Ultraviolet Light Radiation, Appl. Phys.

technology", Proc. of the 21st European Photovoltaic Solar Energy Conference,

Techniques. Proc. of Saudi Int. Electronic, communication and Photonic conference

**9. Acknowledgement** 

**10. References** 

during experimental performances.

Lett. 44(9):849


**19** 

*Japan* 

**Oriented Lateral Growth and** 

Kuninori Kitahara1 and Akito Hara2

*1Shimane University 2Tohoku Gakuin University* 

**Defects in Polycrystalline-Silicon** 

Silicon (Si) thin films on glass substrates have been extensively developed as a semiconductor material for electronic devices. This material is especially useful for largearea panel devices such as thin-film transistors (TFTs) on active-matrix flat panel displays. The most widely used Si films are hydrogenated amorphous Si (a-Si:H), which can be deposited at temperatures lower than the strain point of the substrate. However, improved electronic properties are required to achieve higher device performance. Using polycrystalline Si (poly-Si) films instead of a-Si:H films enhances carrier mobility by two or three orders of magnitude; thus, driver circuits can be incorporated into display panels, as shown in Fig. 1. The application of poly-Si will be extended to mobile displays with large

pixel density, microprocessor–display combined panels, and thin-film solar cells.

Fig. 1. Low-temperature poly-Si liquid crystal display (2 in. diagonal). Driver circuits are

The poly-Si used for TFT must be ≤ 50 nm thick to ensure the desired device performance. Furthermore, the crystalline fraction should be almost 100%. Such thin films cannot be deposited directly on glass; they must be formed by recrystallization of a-Si precursor films. For this purpose, manufactures have employed solid-phase crystallization (SPC) and

integrated at the periphery of the panel.

**1. Introduction** 

**Thin Films on Glass Substrates** 

