Chapter 3 InGaN/GaN Single-heterostructures (SH)
3.5 Discussions
We have demonstrated the variations of strain and indium composition of an
InGaN layer in InGaN/GaN single-heterostructures and double-heterostructures based
on the RSM and PL measurements. In the SH samples, InGaN with low (bottom) and
high (top) indium contents can be identified. On the other hand, only a thin layer of
strain relaxed InGaN in the middle in a DH sample can be observed. In the SEM results,
we observe that when strain starts to relax, the surface roughness increases. The quality
of epitaxy decreases. In the DH samples, formation of indium droplets has been
observed and confirmed based on HRTEM study. Indium droplet formation due to
phase separation will lead to poor electric property. In another HRTEM result of DH
samples, we find that as the thickness of the InGaN layer decreases, the density of
indium droplet is also decreased. Therefore, it is necessary to grow thinner InGaN
layers to avoid the formation of indium droplet. However, one layer may not be enough
for sufficient absorption for solar cell applications. In the next chapter, we will
demonstrate our measurements of multi-InGaN-layer structures. Such indium
composition distribution can be used for fabricating multi-thin film solar cells.
Fig. 3.1.1 Structures of SH and DH samples.
Fig. 3.2.1(a) RSM plot of a SH sample with 200-nm InGaN grown at 700
o
C.
Fig. 3.2.1(b) RSM plot of a DH sample with 200-nm InGaN grown at 700
o
C.
Fig. 3.3.1 X-ray diffraction pattern (ω-2θ scan curves) of SH samples
with four different thicknesses.
Fig. 3.3.2(a) Cross-section HRTEM images of SH 25nm.
Fig. 3.3.2(b) Cross-section HRTEM images of SH 50nm.
Fig. 3.3.2(c) Cross-section HRTEM images of SH 100nm.
Fig. 3.3.2(d) Cross-section HRTEM images of SH 200nm.
Fig. 3.3.3(a) Plane–view SEM image of SH 25 nm.
Fig. 3.3.3(b) Plane–view SEM image of SH 50 nm.
Fig. 3.3.3(c) Plane–view SEM image of SH 100 nm.
Fig. 3.3.3(d) Plane–view SEM image of SH 200 nm.
Fig. 3.3.4(a) PL spectra of sample SH 200 nm with excitation from the top.
Fig. 3.3.4(b) PL spectra of sample SH 200 nm with excitation from the
bottom.
Fig. 3.3.5(a) The direction (red arrow) of the eight points that were measured by EDX in SH 200 nm.
Fig. 3.3.5(b) The trend of the relative indium content of SH 200 nm.
Fig. 3.4.1 X-ray diffraction pattern (ω-2θ scan curves) of DH samples
with four different thicknesses.
Fig. 3.4.2 Cross-section HRTEM images of DH 200 nm and indium
droplets.
Fig. 3.4.3 The four points (labeled by 1~4) for EDX measurement in DH 50
nm.
Table 3.4.1 Relative indium contents of the four points in Fig. 3.4.3 from
EDX measurement.
Fig. 3.4.4(a) Low-magnification cross-section HRTEM images of DH 25
nm.
Fig. 3.4.4(b) Low-magnification cross-section HRTEM images of DH 50
nm.
Fig. 3.4.4(c) Low-magnification cross-section HRTEM images of DH 100
nm.
Fig. 3.4.4(d) Low-magnification cross-section HRTEM images of DH 200
nm.
Fig. 3.4.5 Higher-magnification cross-section HRTEM images of DH
25nm.
Fig. 3.4.6(a) Plane–view SEM image of DH 25 nm.
Fig. 3.4.6(b) Plane–view SEM image of DH 50 nm.
Fig. 3.4.6(c) Plane–view SEM image of DH 100 nm.
Fig. 3.4.6(d) Plane–view SEM image of DH 200 nm.
Fig. 3.4.7(a) PL spectra of sample DH 200 nm with excitation from the top.
Fig. 3.4.7(b) PL spectra of sample DH 200 nm with excitation from the
bottom.
Fig. 3.4.8(a) The direction (the red arrow) of the nine points for EDX measurement in DH 200 nm.
Fig. 3.4.8(b) The trend of the relative indium content of DH 200 nm.
Chapter 4
Comparison between InGaN/GaN Multi-layer Samples under Different Growth Conditions
In this chapter, we will compare the nanostructures and optical properties of two
InGaN/GaN multi-layer samples of different growth conditions (named as 10nm-5P and
8nm-6P hereafter). The main differences in structure between the two samples are the
thicknesses of the InGaN layer and GaN barrier and the number of periods.
4.1 Sample Growth and Measurement Conditions
Both samples were grown by metalorganic chemical vapor deposition (MOCVD).
The 10nm-5P (10nm wells, 5 periods) and 8nm-6P (8nm wells, 6 periods) samples were
prepared in the following manner. First, a 2~3 μm undoped-GaN layer was grown on a
(0001) c-plane sapphire substrate at 1080 °C. Then, the multi-layer wells were grown.
The InGaN wells of the 10nm-5P (8nm-6P) samples were grown at 675 °C. The growth
temperature was then ramped to 800 °C for depositing GaN barriers of 15 nm (20 nm)
in thickness. A 15 nm (20 nm) GaN caplayer was deposited at 800 °C after the
multi-layer wells . The detailed structures of these two samples are shown in Fig. 4.1.1.
Temperature-dependent PL measurements for the effective band gaps of these two
samples by using a 35mW He-Cd (325 nm) laser for excitation were performed.
HRTEM measurements under two-beam condition were performed to observe the
indium distribution within the InGaN wells of these two samples. EDX was measured to
obtain the elemental distribution in the horizontal and vertical direction in each InGaN
layer. Strain state analysis (SSA) results provide more local information about a well,
such as average indium content and indium composition fluctuation.
4.2 PL and HRTEM Results
Low-temperature PL spectra of the 10nm-5P and 8nm-6P samples from 13 to 300
K are shown in Figs. 4.2.1(a) and (b), respectively. The PL intensities drop with
increasing temperature due to the increasing capture probability of carriers by defects.
The 10nm-5P sample shows a broad peak centered around 535 nm with no distinct
individual peaks within. One can clearly observe two spectral peaks in the 8nm-6P
sample. The long-wavelength spectral peak location of the 8nm-6P sample was also
centered around 535 nm. The short-wavelength spectral peak red-shifts from 442 nm to
458 nm as the temperature rises from 30 to 300K for the 8nm-6P sample.
HRTEM images are needed for understanding the indium distributions among
the InGaN well layers to see the nano-structure differences leading to the variations in
the PL spectras. Low-magnification cross-section HRTEM images of the 10nm-5P and
8nm-6P samples are shown in Fig. 4.2.2. Here, we can see the general structures of the
two samples, including the InGaN well and GaN barrier. The white arrow indicates the
direction of (0001) from the substrate. The G1 glue can be seen at the very top. In
general, the variations of the contrast in the HRTEM images represent the indium
composition fluctuations. HRTEM images of higher magnification of the first through
fifth wells from the bottom of the 8nm-6P and 10nm-5P samples are shown in Figs.
4.2.3(a) and (b) through 4.2.7(a) and (b), respectively. One can observe nano-scale
clusters in these images. We notice that the bottom well of the 8nm-6P sample have the
best quality. Then, we compare the indium distributions within the five InGaN wells.
We find that the indium distribution is more uniform and the interfaces between the
InGaN wells and GaN barriers are clearer in the 8nm-6P sample, when compared with
the 10nm-5P sample.
4.3 EDX and SSA Results
EDX is used to estimate indium distribution in the horizontal and vertical
directions of each InGaN layer. The indium distribution of the horizontal (red arrow)
and vertical directions (yellow arrow) measured by EDX in 10nm-5P and 8nm-6P
samples are shown in Figs. 4.3.1(a) and (b), respectively. The white arrow indicates the
direction of (0001) from the substrate. The data at these points are used to draw a
diagram for comparing the relative indium contents as functions of distance. The trend
of the relative indium content along the horizontal direction in each InGaN layer of the
10nm-5P and 8nm-6P samples are shown in Figs. 4.3.2(a) and (b), respectively. The
nearest left point along the arrow of the InGaN layer is labeled as 0 nm in distance.
Then, we average the relative indium contents at these points along the horizontal
direction of each layer to obtain the average indium content of each InGaN layer. The
variation of the average indium content along the vertical direction of the 10nm-5P and
8nm-6P samples are shown in Figs. 4.3.3(a) and (b), respectively. We also calculate the
standard deviation at these points in each point array (as shown in Figs. 4.3.2 and 4.3.3).
In the horizontal direction, we find that the indium composition fluctuation is stronger
in the 10nm-5P sample, when compared with the 8nm-6P sample. In the vertical
direction, we find that the 8nm-6P sample has a more uniform indium composition in
each layer, when compared with the 10nm-5P sample.
For more detail investigation, we used a package program called the DALI
(Digital Analysis of Lattice Images) to analyze the two-beam images of the InGaN
layers and obtain the SSA images for the composition mapping of the InGaN layers.
SSA can exclude some uncertain factors, such as thickness, defocus, and strain
relaxation. In such SSA images, the deep blue color represents GaN and the red color
corresponds to a d value larger than 1.1 or indium content higher than 21 %. The five
InGaN layers of the 10nm-5P sample are shown in Figs. 4.3.4(a) through (e). The color
coding for various d-factor values are shown below the image in Fig. 4.3.4(a). In this
image, one can see that the layer shape can still be identified although a cluster exists
and indium atoms out-diffuse into the upper barrier. The horizontal and vertical
line-scan profiles show that the local indium composition inside the cluster can be as
large as 23.1%. The six InGaN layers of the 8nm-6P sample are shown in Figs. 4.3.5(a)
through 4.3.5(f). As indicated with the line-scan results, the relatively constant indium
content and more uniform indium distributions in each InGaN well of the 8nm-6P can
also be seen in these SSA images.
We calculate the crest-to-valley value in each well of samples 10nm-5P and
8nm-6P. From the results, one can see the variation trend of indium composition
fluctuation range (crest-to-valley). The fifth and sixth wells from the bottom in samples
10nm-5P and 8nm-6P have the largest composition fluctuation ranges, respectively.
Other variation trends were shown in Table 4.3.1. We also evaluate the average indium
content of each well in samples 10nm-5P and 8nm-6P by integrating the indium
composition distribution in an SSA image and dividing by the designated InGaN well
area. We assume that all the indium atoms distributed around a well in an SSA image
belong to the InGaN well. The results of the average indium contents of those wells are
listed in Table 4.3.2. These results are consistent with the EDX data.
4.4 Discussions
We have demonstrated the effective band gaps and indium distributions of
InGaN/GaN multi-layers with different structures based on PL and HRTEM
measurements. In HRTEM results, we found that the indium distribution was more
uniform and the interfaces between InGaN wells and GaN barriers were clearer in the
sample 8nm-6P, when compared with sample 10nm-5P. This might be why PL result
did not show clear peaks in the sample 10nm-5P. PL measurement clearly showed two
spectral peaks in the sample 8nm-6P. Those results implied higher optical quality in the
sample 8nm-6P than in the sample 10nm-5P. In the other measurements, EDX and SSA
results both showed that the sample 10nm-5P had a larger indium composition
fluctuation than the sample 8nm-6P.
Based on the data and discussions above, one can conclude that the structure of
the sample 8nm-6P has a better crystal quality when compared with the sample
10nm-5P. The only growth difference was that sample 8nm-6P had a thinner InGaN
layer and thicker GaN barrier.
Fig. 4.1.1 Structures of InGaN/GaN multi-layer samples (10nm-5P and
8nm-6P).
Fig. 4.2.1(a) PL spectra of sample 10nm-5P.
Fig. 4.2.1(b) PL spectra of sample 8nm-6P.
Fig. 4.2.2 Low-magnification cross-section HRTEM images of 10nm-5P
and 8nm-6P.
(a)
(b) Fig. 4.2.3(a) and (b) HRTEM image of sample 8nm-6P and 10nm-5P,
respectively (the first well from the bottom).
a cluster
(a)
(b) Fig. 4.2.4(a) and (b) HRTEM image of sample 8nm-6P and 10nm-5P,
respectively (the second well from the bottom).
a cluster
(a)
(b) Fig. 4.2.5(a) and (b) HRTEM image of sample 8nm-6P and 10nm-5P,
respectively (the third well from the bottom).
(a)
(b) Fig. 4.2.6(a) and (b) HRTEM image of sample 8nm-6P and 10nm-5P,
respectively (the fourth well from the bottom).
(a)
(b) Fig. 4.2.7(a) and (b) HRTEM image of sample 8nm-6P and 10nm-5P,
respectively (the fifth well from the bottom).
Fig. 4.3.1(a) The indium distribution of horizontal (red arrow) and vertical
direction (yellow arrow) as measured by EDX in 10nm-5P.
Fig. 4.3.1(b) The indium distribution of horizontal (red arrow) and vertical
direction (yellow arrow) as measured by EDX in 8nm-6P.
Fig. 4.3.2(a) Trend of the relative indium content along the horizontal
direction in each InGaN layer of 10nm-5P.
Fig. 4.3.2(b) Trend of the relative indium content along the horizontal
direction in each InGaN layer of 8nm-6P.
Fig. 4.3.3(a) Trend of the average indium content along the vertical direction of 10nm-5P.
Fig. 4.3.3(b) Trend of the average indium content along the vertical
direction of 8nm-6P.
Fig. 4.3.4(a) An SSA image of 10nm-5P (the first well from the bottom).
Fig. 4.3.4(b) An SSA image of 10nm-5P (the second well from the
bottom).
Fig. 4.3.4(c) An SSA image of 10nm-5P (the third well from the bottom).
Fig. 4.3.4(d) An SSA image of 10nm-5P (the fourth well from the bottom).
Fig. 4.3.4(e) An SSA image of 10nm-5P (the fifth well from the bottom).
Fig. 4.3.5(a) An SSA image of 8nm-6P (the first well from the bottom).
Fig. 4.3.5(b) An SSA image of 8nm-6P (the second well from the bottom).
Fig. 4.3.5(c) An SSA image of 8nm-6P (the third well from the bottom).
Fig. 4.3.5(d) An SSA image of 8nm-6P (the fourth well from the bottom).
Fig. 4.3.5(e) An SSA image of 8nm-6P (the fifth well from the bottom).
Fig. 4.3.5(f) An SSA image of 8nm-6P (the sixth well from the bottom).
Table 4.3.1 Indium composition fluctuation range (crest-to-valley) of the wells in the 10nm-5P and 8nm-6P samples.
Table 4.3.2 Calibrated average indium contents of the wells in the 10nm-5P
and 8nm-6P samples.
Chapter 5 Conclusions
In the first part of this research, we applied HRTEM and SEM techniques to study
the nanostructures of two series samples including single-heterostructures (SH) and
double-heterostructures (DH). HRTEM results showed formations of indium droplets
depicting a trend where the density of indium droplets decreased as the InGaN thickness
decreased. In addition, SEM results showed that as the strain started to relax the surface
roughness of these samples rose. RSM and PL measurements also demonstrated the
variations of strain and indium composition of an InGaN layer in these two series
samples. In the SH samples, InGaN with low (bottom) and high (top) indium contents
can be identified. On the other hand, only a thin layer of relaxed InGaN in the middle of
the DH samples can be observed. The indium composition trend, as measured by EDX,
is consistent with the PL results. Owing to the above properties and sufficient
absorption for solar cell applications, one can conclude that it is necessary to grow
thinner and multiple InGaN layers.
The second part of this research compared the nanostructures of two InGaN/GaN
multi-layers of different structures. PL measurements showed that the 10nm-5P sample
only displayed a broad spectrum, whereas the 8nm-6P sample clearly showed two
spectral peaks. This result implied a better optical quality in the 8nm-6P sample than in
the 10nm-5P sample. From the cross-section HRTEM images, one can see that the
indium distribution was more uniform and the interfaces between the InGaN wells and
GaN barriers were clearer in the 8nm-6P sample than in the 10nm-5P sample. The SSA
calibrated average indium contents and indium composition fluctuation, which gave us
local information about the two samples. Based on the SSA results, we find that the
10nm-5P sample has a larger indium composition fluctuation than the 8nm-6P sample.
The EDX results are consistent with the SSA results. We concluded that the 8nm-6P
sample (thinner InGaN thickness and thicker GaN barrier) generally had better crystal
quality than the 10nm-5P sample.
Reference
Chapter 1
[1] S. Nakamura and G. Fasol, The Blue Laser Diode (springer, Berlin, 1997).
[2] H. Morkoç, Nitride Semiconductors and Devices (Springer, 1999).
[3] J.H. Edgar, S. Stritite, I. Akasaki, H. Amano and C. Wetzel, Gallium Nitride and Related Semiconductor (1999).
[4] X. Y. Sun, R. Bommena, D. Burckel, A. Frauenglass, M. N. Fairchild, and S. R. J. Brueck, G. A. Garrett and M. Wraback, S. D. Hersee,
“Defect reduction mechanisms in the nanoheteroepitaxy of GaN on SiC,” J. Appl. Phys. 95, 1450 (2004)
[5] Valcheva, T. Paskova, I. G. Ivanov, R. Yakimova, Q. Wahab, S. Savage, N. Nordell, and C. I. Harris, “B implantation in 6H--SiC: Lattice damage recovery and implant activation upon high-temperature annealing,” J. Vac. Sci. Technol. B 17, 1040 (1999)
[6] S. Nakamura, Jpn. J. Appl. Phys. 30, L1705 (1991).
[7] J. H. EDGAR, S. STRITE, I. AKASAKI, H. AMANO and C. WETZEL,
Properties, Processing and Applications of Gallium Nitride and
Related Semiconductor (1999).
[8] D. V. Lang and C. H. Henry, Phys. Rev. Lett. 35, 1525 (1975).
[9] S. Tomiya, E. Morita, M. Ukita, H. Okuyama, S. Itoh, K. Nakano, and A. Ishibashi, Appl. Phys. Lett. 66, 1208 (1995).
[10] T. Hino, S. Tomiya, T. Miyajima, K. Yamashima, S. Hashimoto and M.Ikeda, Appl. Phys. Lett. 76, 3421 (2000).
[11] X. H. Wu, C. R. Elsass, A. Abare, M. Mack, S. Keller, P. M. Petroff, S.
P. DenBaars, J. S. Speck, and S. J. Rosner, Appl. Phys. Lett. 72, 692 (1998).
[12] D. Hull and D.J. Bacon, Introduction to Dislocations (2001).
[13] G. Martin, A. Botchkarev, A. Rockett, and H. Morkoc, Appl. Phys. Lett.
68, 2541 (1996).