**3.3.3 Silver grain microstructure**

72 Sintering of Ceramics – New Emerging Techniques

and growth of metastable crystalline phases and approach to a stable crystalline structure (Pinckney and Beall, 2008). Amorphous phase separation is generally the first stage in glassceramic formation. This phase is highly unstable as a glass and will precipitate primary crystalline nuclei on heating at temperature near the annealing point of the host glass. The next nucleation stage usually involves the heterogeneous nucleation and growth of metastable crystalline phases on the primary crystalline nuclei, resulting in a fine-grained metastable solid solution assemblage. Finally with increasing temperature, this metastable phase assemblage breaks down into stable crystalline phases by means of crystal phase transformation, reaction between metastable phases or a combination of several of these

The effects of sintering temperature on microstructure of glass-ceramic tape are shown in Fig. 6. It is clearly seen that there is inhomogeneous microstructure for the entire sample with some sample showing a big size pore meaning that probably only a few crystallite sizes present in the bulk samples. This feature of the microstructure could be due to the agglomeration within the sample. In the LTCC process, some particles are densely packed and some particles are loosely packed due to the agglomeration of particle at some places. The presence of agglomeration is a common problem in ceramics processing and influenced the microstructure behavior of the whole substrate. As mentioned by Lange (1984) and Hirata et al., (2009), when a laminated substrate or powder compact is heated, the inhomegeneity of the packing provides the different densification rate and the produce a microstructure which is usually not uniform when agglomeration is severe. An inhomogeneous distribution of particles leads to an inhomogeneous liquid distribution such that there is no driving force for redistribution of the liquid, so the densification rate is not

homogeneous and the microstructure development also becomes inhomogeneous.

Based on their research, Deng et al., (2007) studied the microstructure of porous ZrO2 ceramics and found that the agglomeration resulted in localized non-uniform shrinkage and the large pores are believed to originate from the large interagglomerated particles and greater microstructure non-uniformity. The agglomerated tape however leads to lower densities with large shrinkage deviations in particular direction giving a poor quality (Raj and Cannon, 1999). Agglomeration promotes uneven sintering which sometimes results in a mechanically weak and porous product. Thus, to achieve a high density material and good microstructure development, the agglomeration needs to be controlled (Forrester et al.,

The densification process of the glass-ceramic composite can be described by the conventional three-stage liquid phase sintering; particle rearrangement, dissolution and precipitation and solid state sintering. The presence of glass in this composite system, acts as a sintering aid which produces liquid phase formation at a temperature lower than the sintering temperature and may considerably increase the rate of sintering. The viscous liquid which occurs during the sintering process may promote an additional diffusion mechanism of dissolution/precipitation, particle rearrangement and capillary forces and finally affect the densification at high temperature. Compared to the solid state sintering, liquid phase sintering enhances densification by two mechanisms; the densification is caused by mass transport via lattice diffusion from grain boundaries and grain boundary diffusion that should be encouraged to obtain high density when the sintering temperature

mechanisms.

2008).

is increased.

In the LTCC technology, the microstructure changes involves the combination of a substrate and conductor material where LTCC tape materials contain a glass binder and organic solvent for tape casting purpose, thick-film conductor also contain glass frit as adhesion element with the substrate via formation of a vitreous bond (Kuromitsu et al., 1994). However, knowledge about the interaction of ceramic filler and amorphous matrix is very limited. Moreover the understandings of how an amorphous matrix influence the crystal grain growth in the metal film is not yet clear (Liu and Shen, 2004). According to Yajima and Yamaguchi (1984) the densification process of the substrate and printed film are depends upon sintering where the conductor and the substrate consisting of CaO-B2O3-SiO2 sinter simultaneously. So the liquid phase formed is believed to help densification of metal powder to a denser packing of grains (see Fig. 7).

The Effects of Sintering Temperature Variations on Microstructure Changes of LTCC Substrate 75

The tape system as a support substrate for printed metal contain SiO2 and Ca or a combination of CaSiO3 as observed in XRD peak in the samples would cause some defects in microstructure due to the formation of liquid phase. As reported by Louet et al. (2005), the SiO2 content can increase the sintering rate due to the formation of an intragranular vitreous phase while the combination of CaO and SiO2 is responsible for abnormal grain growth. As the contents of SiO2 and Ca or a combination of CaSiO3 increase at some places, they should increase the grain growth kinetics due to the formation of liquid phase which preferentially fill the spaces between the solid particles. It is indicated that the liquid phase has excellent wettability with CaSiO3. Therefore the silver powder was sintered with liquid glass phase. When the CaSiO3 becomes liquid, chemical reactions such as the migration of silver into the liquid frit would occur on the surface of silver. It also believes that the silver conductive thick-film contains about 90 wt% silver and 10wt % glass frit. So, the presence of the liquid phase gives rise to changes of its atomic mobility and is known to be the cause of abnormal grain growth. During sintering process, the glass phase may decompose into liquid phase providing the capillary force for neighboring particles to come closer and react. If the glass

The average grain size of sintered body was measured over 200 grains by the linear intercept method. Fig. 8 was plotted with varied sintering temperature for the average grain size and the density of the substrate. As we can see with increase the firing temperature, the silver grains grow in much good shape up to 900 °C and the average grain size increased from ~4.39µm to ~5.60 µm. The work is consistent with the findings by Huang and Tsai (2001), which noted that the grain size increases with increasing sintering temperature.

> 780 800 820 840 860 880 900 920 **Sintering Temperature**

Fig. 8. Average grain size of printed conductor fired at various sintering temperature.

Grain Size (um) Density after Fired

2.75 2.80 2.85 2.90 2.95 3.00 3.05

**Density (g/cm3)**

phase is not uniformly distributed, the grain growth is not homogeneous.

3.20 3.70 4.20 4.70 5.20 5.70 6.20

**Grain Size (**μ**m)**

The SEM micrograph observation of microstructural changes of silver with sintering temperature is shown in Fig. 7. As observed here, the distribution of silver grain size is not homogenous; some grains have grown pretty well and some grains have not. There are relatively small grains about 2-4 μm in size coexisting with many large grains about 8-12 μm. It was also noted that the grain morphology are non-uniform/irregular shape. There are many curved boundaries and there is little obvious faceting or shape anisotropy in certain areas. Verhoeven (1975), has stated that generally two factors influence the shape of grains; the requirement to fulfill space and to be in a state of metastable equilibrium particularly with regard to the total grain boundary surface area. So the net result is that the three dimensional grains are irregular polyhedron with curved faces. Kang et al., (1985) has revealed that the grain shape changed mainly due to the anisotropic grain growth of grains rather than by contact flattening as suggested by Kingery's model. The distribution of the glass-frit diffusion is an inhomogeneous manner; the distribution of the metal grain will be varied. The quality of silver thick-film is dependent on the properties of the bottom substrate. Rane et al., (2003) reported that the microstructure changes in thick film that occurs during firing depends primarily on the sintering of silver powders as well as on the glass bonded interaction presuming that the substrate material has a slight influence on the sintering and microstructure. In LTCC materials, the LTCC substrate also plays an important role in the development of the microstructure during sintering since it is a cofiring process where the conductor and the substrate were fired together. Moreover the microstructural changes in the conductor film during the co-firing process also depend strongly on the sintering of metal powders, binder material interaction and the decomposition of the binder present in both paste and LTCC tape materials which lead the formation of cavities /voids on the surface of the fired film (Bangali et al., 2008).

Fig. 7. SEM micrograph of silver surface sintered at a) 800 °C, b) 825 °C, c) 850 °C, d) 875 °C and e) 900 °C.

The SEM micrograph observation of microstructural changes of silver with sintering temperature is shown in Fig. 7. As observed here, the distribution of silver grain size is not homogenous; some grains have grown pretty well and some grains have not. There are relatively small grains about 2-4 μm in size coexisting with many large grains about 8-12 μm. It was also noted that the grain morphology are non-uniform/irregular shape. There are many curved boundaries and there is little obvious faceting or shape anisotropy in certain areas. Verhoeven (1975), has stated that generally two factors influence the shape of grains; the requirement to fulfill space and to be in a state of metastable equilibrium particularly with regard to the total grain boundary surface area. So the net result is that the three dimensional grains are irregular polyhedron with curved faces. Kang et al., (1985) has revealed that the grain shape changed mainly due to the anisotropic grain growth of grains rather than by contact flattening as suggested by Kingery's model. The distribution of the glass-frit diffusion is an inhomogeneous manner; the distribution of the metal grain will be varied. The quality of silver thick-film is dependent on the properties of the bottom substrate. Rane et al., (2003) reported that the microstructure changes in thick film that occurs during firing depends primarily on the sintering of silver powders as well as on the glass bonded interaction presuming that the substrate material has a slight influence on the sintering and microstructure. In LTCC materials, the LTCC substrate also plays an important role in the development of the microstructure during sintering since it is a cofiring process where the conductor and the substrate were fired together. Moreover the microstructural changes in the conductor film during the co-firing process also depend strongly on the sintering of metal powders, binder material interaction and the decomposition of the binder present in both paste and LTCC tape materials which lead the

formation of cavities /voids on the surface of the fired film (Bangali et al., 2008).

d

**10** μ**m**

a b c

**10** μ**m 10** μ**m 10** μ**m** 

e

**10** μ**m** 

Fig. 7. SEM micrograph of silver surface sintered at a) 800 °C, b) 825 °C, c) 850 °C, d) 875 °C

and e) 900 °C.

The tape system as a support substrate for printed metal contain SiO2 and Ca or a combination of CaSiO3 as observed in XRD peak in the samples would cause some defects in microstructure due to the formation of liquid phase. As reported by Louet et al. (2005), the SiO2 content can increase the sintering rate due to the formation of an intragranular vitreous phase while the combination of CaO and SiO2 is responsible for abnormal grain growth. As the contents of SiO2 and Ca or a combination of CaSiO3 increase at some places, they should increase the grain growth kinetics due to the formation of liquid phase which preferentially fill the spaces between the solid particles. It is indicated that the liquid phase has excellent wettability with CaSiO3. Therefore the silver powder was sintered with liquid glass phase. When the CaSiO3 becomes liquid, chemical reactions such as the migration of silver into the liquid frit would occur on the surface of silver. It also believes that the silver conductive thick-film contains about 90 wt% silver and 10wt % glass frit. So, the presence of the liquid phase gives rise to changes of its atomic mobility and is known to be the cause of abnormal grain growth. During sintering process, the glass phase may decompose into liquid phase providing the capillary force for neighboring particles to come closer and react. If the glass phase is not uniformly distributed, the grain growth is not homogeneous.

The average grain size of sintered body was measured over 200 grains by the linear intercept method. Fig. 8 was plotted with varied sintering temperature for the average grain size and the density of the substrate. As we can see with increase the firing temperature, the silver grains grow in much good shape up to 900 °C and the average grain size increased from ~4.39µm to ~5.60 µm. The work is consistent with the findings by Huang and Tsai (2001), which noted that the grain size increases with increasing sintering temperature.

Fig. 8. Average grain size of printed conductor fired at various sintering temperature.

The Effects of Sintering Temperature Variations on Microstructure Changes of LTCC Substrate 77

y = -735.32x + 1.5582

*LogD Q R TA* = −( ) / 2.303 1 + (9)

0.0011 0.00112 0.00114 0.00116 0.00118 0.0012 0.00122 0.00124 0.00126 **1/T**

where *k* is the specific reaction rate constant, *Q* is the activation energy, *T* is the absolute temperature, and *R* is the ideal gas constant. The value of *k* however can directly be related

Where *T* is the absolute temperature, *A* is the intercept and *D* is the grain size. By using equation 9, one can obtain a best fitted straight-line plot of grain size where a plot of log D versus the 1/*T* as shown in Fig. 9. From the equation (9), we obtained the slope of the line which is –*Q*/2.303*R* and the value of the activation energy of metal grain growth (Q) can be calculated from the Arrhenius plot which is 14.079kJ/mol which might be ascribed to grain boundary grooving, precipitation, impurity drag or interfacial reaction between the films and the substrate. The obtained values is considered too low compared to the activation energy calculated by Wu et al. (2011) from their studies on the behavior of ZnO-doped silver

A closer look for sample sintered at 800 °C and 825 °C as shown in Fig. 10, we clearly see pronounced layered-structure grain growth/faceted grain microstructure. The large relatively faceted grains have differed orientations between each other. For the oriented grain growth the preferred direction is the same for all directions. So in this case, the anisotropic structure occurs for a restricted number of grains and can be due to the crystallographic effects. For all cases of the direction growth of anisotropic grain is random. Growth anisotropy is produced naturally during the process of grain growth (Uematsu et

to grain size according to Jarcho et al. (1976), which resulted in the equation below:

Fig. 9. Plots of log D versus the reciprocal of absolute temperature (1/T).

0.62 0.64 0.66 0.68 0.7 0.72 0.74 0.76

thick-film and the silver grain growth mechanism.

**3.3.4 Anisotropic grain growth** 

al., 1997).

**Log (D)**

However, the trend of the average grain size (metal surface) and the density (whole substrate) are somewhat opposing trend with increasing sintering temperature. The relation of microstructural changes with decreasing density is not significant in this work because the density is highly extremely extrinsic and macroscopic property while the average grain size is less extrinsic and microscopic average of crystallite size in polycrystalline material.

The average grain size can be increased when increasing the sintering temperature however the density does not always show a similar trend of the average grain size with increase the temperature due to the presence of pores and other defects or stresses within the sample (Alias et al., 2009). The average grain size of a sintered product is influenced by the total shrinkage characteristic during sintering. The glass contained in metal thick-film formed the liquid phase during sintering resulting in faster grain growth speed of the conductor because of its excellent low temperature shrinkage characteristics. The increase of average grain size shows the microstructural evolution of the samples. These can be either be intrinsic effects related to the crystal structure, defect, surface or energy anisotropy of the material or be extrinsic such as pore size, density heterogeneity, partial wetting by the liquid phase and preferential impurity (Feteira and Sinclair, 2008) . The microstructure of printed silver metal can be designed to attain the features such as fine grain size, low porosity and negligible second phase that can improve operation at the particular frequency. The control of microstructure of metal films is very important for a good device performance.

The microstructural evolution of the sintered samples can be described by adapting the sintering mechanism explained by Sameshima et al. (2000). There are 5 stages involved in the process:


Microstructure evolution has a large influence on the material properties generally and can be characterized by abnormal grain growth behavior where a few selected grains grow by replacing the other grains (Ubach et al., 2004). The main cause of the abnormal growth is due to the presence of anisotropy of either the surface energy or the strain density (Paik et al., 2003). Anisotropy of the grain boundary energy has been considered to be origin of the abnormal grain growth observed in bulk materials. It is possible that this may apply to the silver film. The silver grain growth mechanism could be evaluated according to its activation energy (Q). Coble's theory in Coble (1961) mentioned that from the behavior of particle growth, the activation energy of metal grain growth can be predicted using Arrhenius equation as below:

$$d\ln k/dT = \mathbb{Q}\int RT^2\tag{8}$$

Fig. 9. Plots of log D versus the reciprocal of absolute temperature (1/T).

where *k* is the specific reaction rate constant, *Q* is the activation energy, *T* is the absolute temperature, and *R* is the ideal gas constant. The value of *k* however can directly be related to grain size according to Jarcho et al. (1976), which resulted in the equation below:

*LogD Q R TA* = −( ) / 2.303 1 + (9)

Where *T* is the absolute temperature, *A* is the intercept and *D* is the grain size. By using equation 9, one can obtain a best fitted straight-line plot of grain size where a plot of log D versus the 1/*T* as shown in Fig. 9. From the equation (9), we obtained the slope of the line which is –*Q*/2.303*R* and the value of the activation energy of metal grain growth (Q) can be calculated from the Arrhenius plot which is 14.079kJ/mol which might be ascribed to grain boundary grooving, precipitation, impurity drag or interfacial reaction between the films and the substrate. The obtained values is considered too low compared to the activation energy calculated by Wu et al. (2011) from their studies on the behavior of ZnO-doped silver thick-film and the silver grain growth mechanism.

#### **3.3.4 Anisotropic grain growth**

76 Sintering of Ceramics – New Emerging Techniques

However, the trend of the average grain size (metal surface) and the density (whole substrate) are somewhat opposing trend with increasing sintering temperature. The relation of microstructural changes with decreasing density is not significant in this work because the density is highly extremely extrinsic and macroscopic property while the average grain size is less extrinsic and microscopic average of crystallite size in polycrystalline material.

The average grain size can be increased when increasing the sintering temperature however the density does not always show a similar trend of the average grain size with increase the temperature due to the presence of pores and other defects or stresses within the sample (Alias et al., 2009). The average grain size of a sintered product is influenced by the total shrinkage characteristic during sintering. The glass contained in metal thick-film formed the liquid phase during sintering resulting in faster grain growth speed of the conductor because of its excellent low temperature shrinkage characteristics. The increase of average grain size shows the microstructural evolution of the samples. These can be either be intrinsic effects related to the crystal structure, defect, surface or energy anisotropy of the material or be extrinsic such as pore size, density heterogeneity, partial wetting by the liquid phase and preferential impurity (Feteira and Sinclair, 2008) . The microstructure of printed silver metal can be designed to attain the features such as fine grain size, low porosity and negligible second phase that can improve operation at the particular frequency. The control

of microstructure of metal films is very important for a good device performance.

the process:

occur by grain boundary diffusion.

diffusion still playing its role.

Arrhenius equation as below:

transport of sintered body is by lattice diffusion.

without any grain growth of the samples.

The microstructural evolution of the sintered samples can be described by adapting the sintering mechanism explained by Sameshima et al. (2000). There are 5 stages involved in

1. First initial stage where no grain growth and mass transport of the sintered samples

2. Second initial stage occurs afterwards where no grain growth occurs and the mass

3. First intermediate and final stage later takes place where lattice diffusion occurs

4. Second intermediate and final stage where there is grain growth happening and lattice

<sup>2</sup> *d k dT Q RT* ln <sup>=</sup> (8)

5. Final intermediate stage where grain boundary diffusion occur with no grain growth. Microstructure evolution has a large influence on the material properties generally and can be characterized by abnormal grain growth behavior where a few selected grains grow by replacing the other grains (Ubach et al., 2004). The main cause of the abnormal growth is due to the presence of anisotropy of either the surface energy or the strain density (Paik et al., 2003). Anisotropy of the grain boundary energy has been considered to be origin of the abnormal grain growth observed in bulk materials. It is possible that this may apply to the silver film. The silver grain growth mechanism could be evaluated according to its activation energy (Q). Coble's theory in Coble (1961) mentioned that from the behavior of particle growth, the activation energy of metal grain growth can be predicted using

A closer look for sample sintered at 800 °C and 825 °C as shown in Fig. 10, we clearly see pronounced layered-structure grain growth/faceted grain microstructure. The large relatively faceted grains have differed orientations between each other. For the oriented grain growth the preferred direction is the same for all directions. So in this case, the anisotropic structure occurs for a restricted number of grains and can be due to the crystallographic effects. For all cases of the direction growth of anisotropic grain is random. Growth anisotropy is produced naturally during the process of grain growth (Uematsu et al., 1997).

The Effects of Sintering Temperature Variations on Microstructure Changes of LTCC Substrate 79

occurs at low annealing temperatures. The edges of faceted grains become rounded with an increase in annealing temperature and normal grain growth occurs because of an increased contribution of entropy (a reduction of self free energy). Critical amounts of CaO and SiO2 to induce and to suppress abnormal grain growth were also measured. According to the results, when the amounts of CaO and SiO2 are within their solubility limits, normal grain growth occurs. On the other hand, abnormal grain growth occurs if

With increasing temperature, the faceted grain microstructure becomes less prominent. It might be due to the presence of boundary defect. As studied by Maclaren et al. (2003), all faceted microstructure being limited by a paucity of steps, ledges or boundary dislocation to serve as site for atom attachment or detachment. As the sintering temperature increases the faceted or anisotropic microstructure are limited. Although the faceted grain microstructure observed is not large, the effects on surface morphology are significant. Furthermore it should be pointed out that there is geometry limits to observable anisotropies for

As a conclusion, the performance of the laminated sample strongly depends on the starting raw materials and processing method; good technical skills and handling procedures are required to minimize the defect/issue occurring in the LTCC multilayer process. As with other advanced electronic ceramics, the quality of raw materials will have a marked effect on the final properties of a ceramic product. The raw material selection includes characterization of purity, particle size distribution, surface area, consistency and cost. The processing of a ceramic material must be optimized with respect to microstructural characteristics, because only by this way we can produce ceramics with properties that approach the level of intrinsic dielectric properties. The chosen raw material is the first key

The present study is able to explain the effect of the different sintering temperatures on the whole substrate properties and printed thick-film. The microstructural evolution at various sintering temperatures was carefully tracked. The sintering temperature therefore plays a vital role in forming microstructures, porosity and shrinkages of the as-prepared materials. This is more complex than the simple sintering of metal grains. In order to understanding better the microstructure evolution, the study of samples fired should be carried out at the different stages of the sintering process. This work will be very helpful to predict how the microstructural factors and material properties of the LTCC system would affect the microwave-frequency performance of relevant devices. It should be mentioned that the results obtained from this work is not always be consistent with observations from other

The authors wish to thank Telekom Malaysia for their funding support under project IMPACT (RDTC/100745) and Assoc. Prof. Dr. Mansor Hashim, Mr. Ismayadi Ismail and

Mrs. Sabrina Mohd. Shapee for the guidance and support for this research work.

the amount is above the solubility limit.

**4. Conclusion** 

works.

**5. Acknowledgement** 

perpendicular surface in any equilibrium experiment.

to controlling the microstructure and making a consistent product.

Fig. 10. Anisotropic grain microstructure for sample sintered at a) 800 °C and b) 825 °C.

There are some investigations on the anisotropic grain growth as well as the presence of faceting grain of alumina interface in the presence of glass (Simpson and Carter, 1990). The faceting of alumina interface in the presence of a glass is a factor affecting the grain boundary movement during sintering. The geometry and movements of facet during liquid phase sintering process are expected to influence directly the final microstructure of the grain boundaries that ultimately control the properties of the alumina compact.

For a faceted grain in a liquid matrix, the dihedral angle is not uniquely defined because of the torque on a facet, contrary to the case of a rounded grain. When two faceted grains are in contact with a certain angle of crystallographic orientation in a liquid matrix, the shape of the contact boundary in equilibrium must be that of the minimum interfacial energy. These abnormal grains originate from heterogeneities in the chemical composition, particle size and packing density. In the LTCC process, the substrate also plays an important role in the development of microstructure during sintering since it is a co-firing process where the conductor and the substrate are sintered simultaneously. So, the nuclei of abnormal grains may be formed during different stages of sintering. From XRD peak observation for samples sintered at 800 °C and 825 °C, a few phases including CaSiO3 and Ca3(SiO4)O were present suggesting the nucleation rate is not zero. The results found was consistent with the work of Rohrer (2001) which noted that at lower temperature, the anisotropy grain of the surface energy is expected to be higher. The wollastonite phase (CaSiO3) becomes a liquid phase before it decomposes. The liquid phase exists preferentially, filling spaces in small pores and particle contact. The capillary force that is generated by the liquid phase can drag the nearby particle close together. The distribution of the liquid phase may not be uniform because the amount of liquid is small. According to Kang et al, (1991), liquid phase, even in a small amount forms due to the presence of impurities, grains are faceted and abnormal grain growth occurs. Such heterogeneities may trigger the formation of nuclei of abnormal grain. Because the nuclei are small, they have a greater chance to grow at their own pace until they encounter other abnormal grains (Chen and Tuan, 2000). However, when a liquid phase forms due to a high concentration of impurities, the grain shape is faceted and abnormal grain growth occurs at low annealing temperatures. The edges of faceted grains become rounded with an increase in annealing temperature and normal grain growth occurs because of an increased contribution of entropy (a reduction of self free energy). Critical amounts of CaO and SiO2 to induce and to suppress abnormal grain growth were also measured. According to the results, when the amounts of CaO and SiO2 are within their solubility limits, normal grain growth occurs. On the other hand, abnormal grain growth occurs if the amount is above the solubility limit.

With increasing temperature, the faceted grain microstructure becomes less prominent. It might be due to the presence of boundary defect. As studied by Maclaren et al. (2003), all faceted microstructure being limited by a paucity of steps, ledges or boundary dislocation to serve as site for atom attachment or detachment. As the sintering temperature increases the faceted or anisotropic microstructure are limited. Although the faceted grain microstructure observed is not large, the effects on surface morphology are significant. Furthermore it should be pointed out that there is geometry limits to observable anisotropies for perpendicular surface in any equilibrium experiment.
