**1. Introduction**

188 Atomic Force Microscopy – Imaging, Measuring and Manipulating Surfaces at the Atomic Scale

*Sensitivity* 7.308 0.042 0.225 0.233 *RSS mmm*

The last specification examines the relationship between the level of impurity (%HMX) and sensitivity. The regression analysis demonstrates that there is no statistically significant relationship between these two variables. For further analysis of shock sensitivity and its

Atomic force microscopy can be used to obtain a large number of data observations at the nanometric level. These data can statistically be used to investigate and establish quantitative relationships between various variables. This work demonstrates that surface characteristics data obtained from topographical scans can be used in investigating the relationship between the shock sensitivity of the materials at the macroscale and their surface roughness characteristics at the nanoscale. Statistical analysis can be used not only to show that there is a statistical relationship, but with the help of regression techniques, can precisely estimate any such relationships. As demonstrated in this chapter, the surface roughness variation on the surface of the particle has a substantial negative impact on the shock sensitivity in RDX materials. A one unit increase in the average standard deviation (Sm) reduces the shock sensitivity by 0.225 GPa. The statistical analysis can also be used to demonstrate absence of any meaningful relationship. For instance, the results demonstrate that the level of HMX impurity

This work was supported by the Office of Naval Research (ONR) and the Naval Surface Warfare Center (NSWC) at Indian Head, MD. The authors express their thanks to Mary Sherlock, Robert Raines, Tina Woodland and Philip Thomas for helpful discussions and for

Doherty, R..M. & Watts, D.S. (2008). Relationship between RDX Properties and Sensitivity.

Bellitto, V.J. & Melnik M.I. (2010). Surface defects and their role in the shock sensitivity of cyclotrimethylene-trinitramine. *Applied Surface Science,* Vol.256, pp. 3478-3481 Bellitto, V.J.; Melnik, M.I.; Sorensen, D.N. & Chang, J.C. (2010). Predicting the Shock

Levine, D.M.; Stephan, D. F.; Krehbiel, T.C. & Berenson, M.L. (2011). *Statistics for Managers* 

Long, J.S. & L.H. Ervin (2000). Using Heteroscedasticity Consistent Standard Errors in the Linear Regression Model. *The American Statistician* Vol. 54, pp. 217-224 MacKinnon, J.G. & White, H. (1985). Some heteroskedasticity consistent covariance matrix

Sensitivity of Cyclotrimethylene-Trinitramine. *Journal of Thermal Analysis and* 

estimators with improved finite sample properties. *Journal of Econometrics*, Vol . 29,

determinants in RDX materials see Bellitto and Melnik (2010) and Bellitto et al (2010).

does not impact the shock sensitivity in the studied RDX materials.

*Propellants, Explosives, Pyrotechnics,* Vol.33, pp. 4-13

Greene, W. H. (2003). *Econometric Analysis*, 5th edition, Prentice Hall,

*using Microsoft Excel*, 6th edition, Prentice Hall

*Calorimetery*, Vol. 102, pp. 557-562

**4. Conclusion** 

**5. Acknowledgements** 

providing the samples.

pp. 53-57

**6. References** 

The nitride family is an exciting material system for optoelectronics industry. Indium nitride (InN), gallium nitride (GaN) and aluminium nitride are all direct bandgap materials, and their energy gaps cover a spectral range from infrared (IR) to deep ultraviolet (UV). This means that by using binary and ternary alloys of these compounds, emission at any visible wavelength should be achievable.

Atomic force microscopy (AFM) is a powerful tool to study the III-nitride surface morphology, crystal growth evolution and devices characteristics. In this chapter, we provide an overview of AFM application in AlInGaN based materials and devices.

Typical surface morphologies of GaN materials characterized by AFM are presented in §1. In additional, three types of threading dislocations, including edge, screw and mixed threading dislocation, are studied by AFM in this section. In §2, V-shape defects and other features in InN, InGaN film and InGaN/GaN multiple quantum wells are summarized. It is not easy to grow high quality AlN and AlGaN films, materials for UV light emitting diode and high electron mobility transistor, which usually have high dislocation density and three dimensional growth mode. Growth condition optimization of high quality, crack-free and smooth Al(Ga)N, with the assistance of AFM, is reviewed in §3. Applications of AFM in GaN based devices are discussed in §4, including the patterned sapphire substrates, asgrown LED surface, backside polished surface, ITO surface investigation.

### **2. AFM study of GaN**

In recent years, the III-nitride-based alloy system has attracted special attention since highbrightness blue, green and white light-emitting diodes (LEDs), and blue laser diodes (LDs) became commercially available.

Good topography and crystal quality of GaN films are some of the key factors to improve devices performance. AFM helps researchers to observe the surface morphology of GaN, to study dislocations in GaN film, and to further understand and optimize the growth condition of the GaN.

AFM Application in III-Nitride Materials and Devices 191

(a) (b)

Fig. 2. 5 × 5 μm� AFM scans of the (a) GaN sample with lot of hillocks; (b) GaN sample with

Fig. 3. AFM image of one GaN film showing three types of pits, corresponding to three

dislocations, edge, screw and mixed types are usually observed in wurtzite GaN epitaxial layers, with the corresponding Burgers vectors, confirmed by TEM (Follstaedt et al. 2003 ;

b 1/3 edge 1120

single hillock and several pits.

kinds of dislocations in GaN.

Datta et al. 2004).

#### **2.1 AFM study of GaN surface morphology**

Fig. 1(a) is a 10 × 10 μm� AFM image of typical high-quality, fully coalesced GaN film on sapphire grown by metal organic chemical vapor deposition (MOCVD). Well-defined, uniform and long crystallographic steps could be observed on the surface. The sample topography exhibited atomic terraces with a measured height of 0.3 nm, in close agreement with the 0.26 nm bi-layer spacing of GaN. Fig. 1(b) is 20 × 20 μm� AFM image of poorquality, uncoalesced GaN film for comparison. The biggest void (hole) was formed at the coalescence boundaries of the nanoisland growth fronts. The width of the observed hole is up to 2.5 μm and the depth could be several microns because the holes result from the poor coalescent at the beginning of the growth.

Fig. 1. (a) 10 × 10 μm� AFM scan of high quality, fully coalesced GaN film; (b) 20 × 20 μm� AFM scan of un- coalesced GaN film.

The intersection of a screw-component dislocation with the film surface creates an atomic step termination that may lead to spiral step procession and hillock formation (Burton et al 1951). At the top of these hillocks AFM reveals the presence of either single screw or clustered screw and mixed-type dislocations, as shown in Fig. 2(a). The presence of clustered defects with the screw component of the Burgers vector is a reason of formation of the growth hillocks.

Spiral growth hillocks, with pits located at the top and some pits located away from the hillock peaks, are shown in Fig. 2(b). Pits at the center of a spiral growth hillock are expected to have a screw component, while dislocation pits away from hillock peaks without terminated steps are of pure edge character.

#### **2.2 Study of dislocation in GaN by AFM**

Fig. 3 is a 2 × 2 μm� AFM image of a GaN film, in which three kinds of pits of different sizes correspond to three different dislocations were observed. Three types of threading

Fig. 1(a) is a 10 × 10 μm� AFM image of typical high-quality, fully coalesced GaN film on sapphire grown by metal organic chemical vapor deposition (MOCVD). Well-defined, uniform and long crystallographic steps could be observed on the surface. The sample topography exhibited atomic terraces with a measured height of 0.3 nm, in close agreement with the 0.26 nm bi-layer spacing of GaN. Fig. 1(b) is 20 × 20 μm� AFM image of poorquality, uncoalesced GaN film for comparison. The biggest void (hole) was formed at the coalescence boundaries of the nanoisland growth fronts. The width of the observed hole is up to 2.5 μm and the depth could be several microns because the holes result from the poor

(a) (b)

The intersection of a screw-component dislocation with the film surface creates an atomic step termination that may lead to spiral step procession and hillock formation (Burton et al 1951). At the top of these hillocks AFM reveals the presence of either single screw or clustered screw and mixed-type dislocations, as shown in Fig. 2(a). The presence of clustered defects with the screw component of the Burgers vector is a reason of formation of

Spiral growth hillocks, with pits located at the top and some pits located away from the hillock peaks, are shown in Fig. 2(b). Pits at the center of a spiral growth hillock are expected to have a screw component, while dislocation pits away from hillock peaks without

Fig. 3 is a 2 × 2 μm� AFM image of a GaN film, in which three kinds of pits of different sizes correspond to three different dislocations were observed. Three types of threading

Fig. 1. (a) 10 × 10 μm� AFM scan of high quality, fully coalesced GaN film; (b) 20 ×

**2.1 AFM study of GaN surface morphology** 

coalescent at the beginning of the growth.

20 μm� AFM scan of un- coalesced GaN film.

terminated steps are of pure edge character.

**2.2 Study of dislocation in GaN by AFM** 

the growth hillocks.

Fig. 2. 5 × 5 μm� AFM scans of the (a) GaN sample with lot of hillocks; (b) GaN sample with single hillock and several pits.

Fig. 3. AFM image of one GaN film showing three types of pits, corresponding to three kinds of dislocations in GaN.

dislocations, edge, screw and mixed types are usually observed in wurtzite GaN epitaxial layers, with the corresponding Burgers vectors, confirmed by TEM (Follstaedt et al. 2003 ; Datta et al. 2004).

$$\mathbf{b}\_{\text{edge}} = 1 / \, 3 \left[ \, 11 \overline{2} 0 \right],$$

AFM Application in III-Nitride Materials and Devices 193

Fig. 4. AFM topography of V-shape defect in InGaN film. A representation of a dislocation

It is usually accepted that the high defect density in GaN leads to poor optical property and also affects the structural and optical quality of the active layer composed of the InGaN/GaN MQWs. It has been reported that threading dislocations disrupt the InGaN/GaN MQW and initiate the V defect using transmission electron microscopy (TEM) and atomic force microscopy (AFM) (Sharma et al., 2000 ; Lin et al., 2000). Several research groups have reported that there is always a threading dislocation (TD) connected with the bottom of V defect and the cause of V-defect formation is the increased strain energy and the reduced Ga incorporation on the [101�1] pyramid planes compared with the [0001] plane (Sun et al., 1997). Cho et al. investigate the V defects in InGaN/GaN MQWs by TEM (Fig. 5) and found that the origin of V defects are not only connected to TD, but also generated from the stacking mismatch boundaries (SMBs) induced by stacking faults (SFs) shown in Fig.

Fig. 5. Cross-sectional bright-field TEM images of the In���Ga���N/GaN MQWs.

terminating in a pit at the (0001) surface is shown.

6(a) and 6(b) (Cho et al., 2001).

$$\mathbf{b}\_{\text{mixed}} = 1 / \,\mathrm{3} \left[ \,\mathrm{11}\overline{\mathbf{2}}\,\mathrm{3} \right]$$

$$\mathbf{b}\_{\text{screw}} = \left[ \,\mathrm{0001} \right]$$

According to the thermodynamics of pit formation (Sangwal, 1987), it follows that the potential difference (Δμ of a stable nucleus of a pit depends inversely on the elastic energy (Eel) of the dislocation: 2 2 μ el 2π Ωγ /E where: Eel is elastic energy of dislocation, γ is edge free energy, Ω is molecular volume. The elastic energy value for screw, edge and mixed type of dislocations are:

$$\begin{aligned} \mathbf{E}\_{\text{screw}} &= \mathbf{G} \mathbf{b}^2 \mathbf{a} \\\\ \mathbf{E}\_{\text{edge}} &= \mathbf{G} \mathbf{b}^2 \mathbf{a} (\frac{1}{1-\mathbf{v}}) \\\\ \mathbf{E}\_{\text{mixed}} &= \mathbf{G} \mathbf{b}^2 \mathbf{a} (1 - \mathbf{v} \cos \theta \,/\,(1-\mathbf{v})) \end{aligned}$$

(where: G is shear modulus, b is Burgers vector, α is geometrical factor, ν is Poisson's constant and θ is the angle between screw and edge components of the Burgers vector of mixed dislocations) (Hull et al., 1984).

Large differences in the magnitude of Burgers vectors, especially between edge type and screw/mixed type dislocations, imply that the size of pits should be different depending on the type of dislocation, i.e. the largest pits are formed on screw-, intermediate size pits on mixed- and the smallest ones on edge-type dislocations (Weyher et al., 2004). The densities of the pits in the sample shown in Fig. 3 with median and larger sizes are 7.5 × 10�/cm�, in agreement with the expected density of screw component dislocations. The majority of dislocations are of pure edge character, related to the smallest pits, with a density of 1 × 10�/cm�, in agreement with the expected pure edge dislocation density determined by TEM examination.

## **3. AFM study for In(Ga)N**

#### **3.1 V-shape defect in InGaN**

Bulk InGaN or InGaN/GaN multiple quantum wells (MQWs) have been used as active layers for near UV, blue, green and white LEDs, laser diodes and solar cells due to the tunable band-gap energy of InGaN, from 0.7 to 3.4 eV through indium composition variation. Therefore it is important to understand the role of microstructural and compositional inhomogeneities in the InGaN layers on the optical emission.

InGaN MQWs structures often have a so-called V-shape defect (Wu et al., 1998; Kim et al., 1998; Northrup et al. 1999; Duxbury et al., 2000; Scholz et al., 2000; Kobayashi et al., 1998) that consists of a threading dislocation terminated by a pit in the shape of an inverted hexagonal pyramid with [101�1] sidewalls, as shown in Fig. 4. Pit formation creates 6 equivalent [101�1] facets. The depth of the pit is h=1.63a. The angle of the hexagonal inverted pyramid defects is 61°, which could be measured accurately by AFM shown in Fig. 4.

b 1/3 mixed 1123

b [0001] screw

According to the thermodynamics of pit formation (Sangwal, 1987), it follows that the potential difference (Δμ of a stable nucleus of a pit depends inversely on the elastic energy (Eel) of the dislocation: 2 2 μ el 2π Ωγ /E where: Eel is elastic energy of dislocation, γ is edge free energy, Ω is molecular volume. The elastic energy value for screw, edge and mixed type

<sup>2</sup> E Gb screw α

2

<sup>1</sup> E Gb <sup>α</sup>( ) <sup>1</sup> <sup>ν</sup>

<sup>2</sup> E Gb mixed α(1 νcosθ /(1 ν))

(where: G is shear modulus, b is Burgers vector, α is geometrical factor, ν is Poisson's constant and θ is the angle between screw and edge components of the Burgers vector of

Large differences in the magnitude of Burgers vectors, especially between edge type and screw/mixed type dislocations, imply that the size of pits should be different depending on the type of dislocation, i.e. the largest pits are formed on screw-, intermediate size pits on mixed- and the smallest ones on edge-type dislocations (Weyher et al., 2004). The densities of the pits in the sample shown in Fig. 3 with median and larger sizes are 7.5 × 10�/cm�, in agreement with the expected density of screw component dislocations. The majority of dislocations are of pure edge character, related to the smallest pits, with a density of 1 × 10�/cm�, in agreement with the expected pure edge dislocation density

Bulk InGaN or InGaN/GaN multiple quantum wells (MQWs) have been used as active layers for near UV, blue, green and white LEDs, laser diodes and solar cells due to the tunable band-gap energy of InGaN, from 0.7 to 3.4 eV through indium composition variation. Therefore it is important to understand the role of microstructural and

InGaN MQWs structures often have a so-called V-shape defect (Wu et al., 1998; Kim et al., 1998; Northrup et al. 1999; Duxbury et al., 2000; Scholz et al., 2000; Kobayashi et al., 1998) that consists of a threading dislocation terminated by a pit in the shape of an inverted hexagonal pyramid with [101�1] sidewalls, as shown in Fig. 4. Pit formation creates 6 equivalent [101�1] facets. The depth of the pit is h=1.63a. The angle of the hexagonal inverted

pyramid defects is 61°, which could be measured accurately by AFM shown in Fig. 4.

compositional inhomogeneities in the InGaN layers on the optical emission.

edge

of dislocations are:

mixed dislocations) (Hull et al., 1984).

determined by TEM examination.

**3. AFM study for In(Ga)N 3.1 V-shape defect in InGaN** 

Fig. 4. AFM topography of V-shape defect in InGaN film. A representation of a dislocation terminating in a pit at the (0001) surface is shown.

It is usually accepted that the high defect density in GaN leads to poor optical property and also affects the structural and optical quality of the active layer composed of the InGaN/GaN MQWs. It has been reported that threading dislocations disrupt the InGaN/GaN MQW and initiate the V defect using transmission electron microscopy (TEM) and atomic force microscopy (AFM) (Sharma et al., 2000 ; Lin et al., 2000). Several research groups have reported that there is always a threading dislocation (TD) connected with the bottom of V defect and the cause of V-defect formation is the increased strain energy and the reduced Ga incorporation on the [101�1] pyramid planes compared with the [0001] plane (Sun et al., 1997). Cho et al. investigate the V defects in InGaN/GaN MQWs by TEM (Fig. 5) and found that the origin of V defects are not only connected to TD, but also generated from the stacking mismatch boundaries (SMBs) induced by stacking faults (SFs) shown in Fig. 6(a) and 6(b) (Cho et al., 2001).

Fig. 5. Cross-sectional bright-field TEM images of the In���Ga���N/GaN MQWs.

AFM Application in III-Nitride Materials and Devices 195

Fig. 8. 5 × 5 μm� AFM scans of a InGaN/GaN MQW structure. Two kinds of V-defects with

should depend on the magnitude of the Burgers vector of dislocations. The bigger TD diameter results in a bigger pit, thus, the bigger V-pits on the InGaN/GaN MQWs are rooted to screw or mixed TD, while the smaller V-pits connected to edge TDs. The diameter of the bigger V-pits in Fig. 8 is about 300 nm, with the density around 1.2 × 10�/cm�, corresponding to the density of the screw TD, while diameter and density of the smaller Vpits in Fig. 8 is about 180 nm and 1.0 × 10�/cm� , corresponding to the density of the edge TD. The size of the V-pits is bigger than InGaN/GaN SL structure shown in the previous Fig. 7(a) because V-defects grow bigger and bigger with a thicker InGaN layer and a higher Indium composition. The density of the V-pits usually is lower than the TD density in GaN underlayer because not every TD develops to a V-pit. In additional, only one kind of V-pit with the same size was observed for some InGaN/GaN sample, because the V-pits

InN has a band gap value of 0.67–0.8 eV, which potentially extends the spectral range covered by group-III nitrides to the near-infrared. In addition, InN has a very small electron effective mass and a high electron drift velocity (O'Leary et al., 1998). Recent theoretical calculations predict an ultimate room temperature electron mobility for InN to reach 14,000 cm2/Vs. (Polyakov et al., 2006). However, it is very difficult to grow InN materials due to the thermal instability of InN and the large lattice mismatches between InN and substrates. A high V/III flux ratio and a low growth temperature are usually required to suppress InN decomposition, which often results in unsatisfactory crystal quality and undesired threedimensional (3D) surfaces roughness. Several technologies have been applied to improve InN film quality and surface morphology. However, AFM has been the most common tools

Fig. 9 is an AFM scan showing the surface morphology of a typical InN film grown by MOCVD on GaN. A low density of large islands with widths up to about 1 μm and heights of 300–400 nm can be observed, which might be indium droplets (N¨orenberg et al., 2002).

different sizes were observed.

connected to screw TDs is easier to be opened.

for surface morphology evaluation and improvement.

**3.2 InN growth condition optimization** 

Fig. 6. Schematic models for V-defect formation connected with (a) a threading dislocation and (b) a SMB induced by stacking faults.

The AFM images in Fig. 7 shows the 5 × 5 μm� scans of the InGaN/GaN superlattice (SL) structure, which is used as the strain release layer in LEDs and LDs. Sample A, with In% composition of 8% is pictured in Fig. 7(a) and sample B, with In% composition of 2% is shown in Fig. 7(b). Remarkable morphological differences are noticeable. A slightly lower density of pits with a wider range of pit diameters is noted for sample A (Fig. 7(a)] when compared to sample B (Fig. 7(b)). In addition, the typical inclusions are shown for sample A as bright, irregular white features. The V-pits density in Fig. 7(a) is around 3 × 10�/cm�, with the average diameter of around 120 nm, and inclusion heights up to 120 Å. The defect density in sample B (less Indium) is around 4 × 10�/cm� with diameters in the 50 nm range.

Fig. 7. 5 × 5 μm� AFM scans of the InGaN/GaN superlattice (SL) structure with (a) indium composition of 7% (b) indium composition of 2%.

The AFM images in Fig. 8 show the 5 × 5 μm� scans of a InGaN/GaN MQW structure. As previously described in Fig. 4, a V-pit usually connected to a TD and the diamter of the TD

Fig. 6. Schematic models for V-defect formation connected with (a) a threading dislocation

The AFM images in Fig. 7 shows the 5 × 5 μm� scans of the InGaN/GaN superlattice (SL) structure, which is used as the strain release layer in LEDs and LDs. Sample A, with In% composition of 8% is pictured in Fig. 7(a) and sample B, with In% composition of 2% is shown in Fig. 7(b). Remarkable morphological differences are noticeable. A slightly lower density of pits with a wider range of pit diameters is noted for sample A (Fig. 7(a)] when compared to sample B (Fig. 7(b)). In addition, the typical inclusions are shown for sample A as bright, irregular white features. The V-pits density in Fig. 7(a) is around 3 × 10�/cm�, with the average diameter of around 120 nm, and inclusion heights up to 120 Å. The defect density in sample B (less Indium) is around 4 × 10�/cm� with diameters in the 50 nm range.

(a) (b) Fig. 7. 5 × 5 μm� AFM scans of the InGaN/GaN superlattice (SL) structure with (a) indium

The AFM images in Fig. 8 show the 5 × 5 μm� scans of a InGaN/GaN MQW structure. As previously described in Fig. 4, a V-pit usually connected to a TD and the diamter of the TD

and (b) a SMB induced by stacking faults.

composition of 7% (b) indium composition of 2%.

Fig. 8. 5 × 5 μm� AFM scans of a InGaN/GaN MQW structure. Two kinds of V-defects with different sizes were observed.

should depend on the magnitude of the Burgers vector of dislocations. The bigger TD diameter results in a bigger pit, thus, the bigger V-pits on the InGaN/GaN MQWs are rooted to screw or mixed TD, while the smaller V-pits connected to edge TDs. The diameter of the bigger V-pits in Fig. 8 is about 300 nm, with the density around 1.2 × 10�/cm�, corresponding to the density of the screw TD, while diameter and density of the smaller Vpits in Fig. 8 is about 180 nm and 1.0 × 10�/cm� , corresponding to the density of the edge TD. The size of the V-pits is bigger than InGaN/GaN SL structure shown in the previous Fig. 7(a) because V-defects grow bigger and bigger with a thicker InGaN layer and a higher Indium composition. The density of the V-pits usually is lower than the TD density in GaN underlayer because not every TD develops to a V-pit. In additional, only one kind of V-pit with the same size was observed for some InGaN/GaN sample, because the V-pits connected to screw TDs is easier to be opened.

#### **3.2 InN growth condition optimization**

InN has a band gap value of 0.67–0.8 eV, which potentially extends the spectral range covered by group-III nitrides to the near-infrared. In addition, InN has a very small electron effective mass and a high electron drift velocity (O'Leary et al., 1998). Recent theoretical calculations predict an ultimate room temperature electron mobility for InN to reach 14,000 cm2/Vs. (Polyakov et al., 2006). However, it is very difficult to grow InN materials due to the thermal instability of InN and the large lattice mismatches between InN and substrates. A high V/III flux ratio and a low growth temperature are usually required to suppress InN decomposition, which often results in unsatisfactory crystal quality and undesired threedimensional (3D) surfaces roughness. Several technologies have been applied to improve InN film quality and surface morphology. However, AFM has been the most common tools for surface morphology evaluation and improvement.

Fig. 9 is an AFM scan showing the surface morphology of a typical InN film grown by MOCVD on GaN. A low density of large islands with widths up to about 1 μm and heights of 300–400 nm can be observed, which might be indium droplets (N¨orenberg et al., 2002).

AFM Application in III-Nitride Materials and Devices 197

homojunction LED operating at 210 nm was realized (Taniyasu et al., 2006). The first AlN metal-semiconductor-metal photodetector with a peak responsivity at 200 nm was. obtained (Li et al., 2003). Significant improvement of the AlN crystal quality using Lateral Epitaxial Overgrowth (LEO) of AlN on sapphire substrates was also demonstrated (Chen et al., 2006). This resulted in 214 nm stimulated emission, the shortest wavelength stimulated emission

AlN is widely used as the buffer layer for Ultraviolet (UV) Light Emitting Diodes (LED) and High Electron Mobility Transistors (HEMT) grown on SiC substrates currently due to its small lattice constant, wide bandgap and high thermal conductivity (Chen et al. 2009a; Chen et al. 2009b). It is therefore necessary to develop an uncomplicated method for growing high quality, thick and crack-free AlN on SiC substrates. AFM images of the AlN grown in 3D and 2D modes are shown in Figs. 11(a) and 11(b), respectively. Islands are present in Fig. 11(a), showing that this material has a 3D growth mode. Steps with a height difference of one (0001) AlN monolayer can be seen in Fig. 11(b), showing that the growth mode of this material is 2D. The RMS surface roughness of the 3D AlN over 5 × 5 μm� area is 0.517 nm,

Fig. 11. AFM images of AlN grown at different growth modes: a) three-dimensional, b) two-

High quality AlN films were grown by switching between the established 2D and 3D AlN growth modes, a method we call modulation growth (MG) (Chen et al. 2008;). The structure of a MG AlN sample is shown in Fig.12. First, a 300 nm 3D AlN layer was grown on the SiC substrate. Then a 2D 200 nm AlN layer was grown. Subsequently, this 3D-2D period was repeated twice. The total thickness of the resulting film is 1.5 µm. The surface of the AlN grown by 3D-2D modulation growth is very smooth. Fig. 13 shows a 5 × 5 μm� AFM scan of the surface with an RMS surface roughness of 0.132 nm. Well defined steps and terraces indicate a step-flow growth mode, and can be observed in all parts of the wafer. The height difference between terraces corresponds to one monolayer of (0001) AlN. No step terminations were observed over the scanned area indicating a low density of Threading

reported in semiconductor materials (Shatalov et al., 2006).

while that of the 2D AlN is 0.151 nm.

dimensional.

Fig. 9. AFM micrographs of typical InN layer.

InN is usually grown by MBE due to its low growth temperature. Chad et al. grew InN films in two different growth regimes and characterized the surface morphology by AFM, as shown in Fig. 10(a) (the N-rich regime) and Fig. 10(b) (the In-droplet regime). Clear growth steps were observed for InN grown in N-rich regime and Indium droplets were observed when it was grown in the In-droplet regime (Gallinat et al., 2006).

Fig. 10. AFM micrographs of (a) a 1.5- μm -thick InN layer grown in the In-droplet regime and (b) a 1- μm -thick InN layer grown in the N-rich regime.
