phys. stat. sol. (c) 2, No. 7, 2145–2148 (2005) / DOI 10.1002/pssc.200461515

© 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Effects of indium incorporation in AlGaN on threading dislocation density

H. Kang1, S. Kandoor2, S. Gupta1, I. Ferguson1∗, S. P. Guo3, and M. Pophristic3 1 School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA 30332,

USA 2 School of Material Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA 3 EMCORE Corporation, 145 Belmont Drive, Somerset, NJ 08873, USA

Received 13 July 2004, revised 2 August 2004, accepted 10 November 2004 Published online 8 February 2005

PACS 61.10.Nz, 68.37.Ps, 68.55.Ln, 81.15.Gh

A comparison of dislocation densities in AlGaN and InAlGaN with approximately similar alloy composi-tions was completed. A systematic series of the AlGaN layers with concentration of 17% Aluminum were grown by metal-organic chemical vapor deposition with trace amounts of indium incorporated into the layers. X-ray diffraction analysis by Williamson Hall plot and reciprocal space mapping was employed to investigate columnar structure in these layers. It was found that lateral coherence length, related to threa-ding dislocation, was systematically varied with Indium content. The lateral coherence length increased with the consequence that the threading dislocation density decreased as Indium content increased, which indicated that even small amounts of indium incorporation could improve crystalline quality. The results are in good agreement with etch pit density study using AFM.

© 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction

AlGaN alloys grown by Metalorganic Chemical Vapor Deposition (MOCVD) with high Aluminum content are used for promising ultraviolet (UV) optoelectronic devices. However, these AlGaN layers are generally grown on sapphire, so that the large lattice mismatch between them increases threading dislo-cation (TD) density that degrades radiative efficiency [1–3]. Therefore, it is crucial to reduce the TD density in these AlGaN epilayers in order to develop UV sources with high efficiency. Recently, InAl-GaN quaternary alloy has attracted much attention for the advantage of separately selecting bandgap energy and smaller lattice mismatch. It has been reported that even a small amount of indium incorpora-tion enhances efficiencies of their applications [4–7]. In this work, we focused on the structural analysis of the AlGaN epilayers with different amounts of indium incorporation for reduction of dislocation den-sity. X-ray diffraction (XRD) measurements were utilized to investigate the densities of screw (Nscrew) and edge (Nedge) dislocations in these layers. Two techniques, namely, the Williamson-Hall (WH) plot and Reciprocal Space Mapping (RSM), were employed to distinguish and determine these dislocations in these layers. In addition, atomic force microscopy (AFM) measurements was performed to study the morphology of the layer surface and enumerate the dislocation density by means of etch pit density.

2 Experimental details

The AlGaN epilayers were grown by MOCVD in an EMCORE D180 rotating disk reactor on basal plane. Trimethyl gallium (TMGa), trimethyl aluminium (TMAl), and NH3 were used as the source pre-

∗ Corresponding author: e-mail: ianf@ece.gatech.edu

2146 H. Kang et al.: Effects of indium incorporation in AlGaN

© 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

cursors for Ga, Al and N, respectively. The epilayers were grown on sapphire (0001) substrate with an intermediate growth of low-temperature (LT) AlN nucleation layer at 600 °C followed by a high-temperature (HT) AlN nucleation layer at 1075 °C. In order to improve the surface morphology and the film quality, variable flow rates of trimethyl indium (TMIn) (0, 50 and 500 ccm) were introduced during the growth for the incorporation of indium. XRD measurements were performed by using the Philips X’pert MRD triple-axis diffractometer. Ω-scans were carried out on the symmetric (000l, l=2, 4, and 6) reflection planes. Φ-scans were also completed on each sample. RSMs were performed on symmetric and asymmetric reflections. In addition, AFM measurements using PSIA XE100 instrument were per-formed after a chemical wet etch to obtain the etch-pit count.

3 Results and discussion

In order to investigate the effects of indium concentration on the structural properties of AlGaN layers, three samples A, B, and C were used with different Indium compositions of 0, 0.04 and 0.15% but with same Al composition (17 %). Coherence length, Lll (average measure of the lateral size of the domain structure), tilt angle, αtilt and twist angle, αtwist (the average mis-orientation between the domains) of these samples can be obtained through the use of W-H plots. Figure 1 (a) shows the W-H plots used to obtain the lateral coherence length and tilt angle that are the dominant parameters in calculating TD density. These parameters are used to determine the physical sizes of the columnar structure and the strain com-ponents as presented in Fig. 1 (b).

0.2 0.3 0.4 0.5 0.6

1.0

1.5

2.0

2.5

3.0

3.5

FW

HM

*si

nθ/λ

(1

0-3

) [A

-1

]

sinθ/λ [A-1

]

A

B

C

(a)0.00 0.05 0.10 0.15

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1100

1150

1200

1250

1300

1350 L

ll

Lat

eral

Coher

ence

Len

gth

(L

ll)

[µm

]

In Compositions [%]

Tilt an

gle (

αtilt ) [arcsec]

αtilt

(b)

Fig. 1 (a) Williamson-Hall plots for sample A, B, and C (Lll = 0.9 / 2Yo [8, 9], where FWHM, θ, λ, and YO are the full width at half maximum of Ω-scan, Bragg reflection angle, X-ray wavelength, and the y-intercept of the fitted line respectively. The slope of the fitted line corresponds to αtilt, and (b) variation of lateral coherence length (Lll) and tilt angle (αtilt) with In composition obtained by these plots

Figure 1 (b) indicates that the lateral coherence length and the tilt angle increase with Indium compositi-on. However, the lateral coherence length more dominantly increases than the tilt angle. This result can be explained by basis of the unit cell parameters of the component GaN, AlN, and InN. InN has larger lattice parameters (a = 3.644, c = 5.718 Å) than GaN and AlN. Therefore, the physical sizes (i.e. lateral coherence length) of the InAlGaN columnar structures with higher indium concentration would be larger than those with lower Indium concentration, while the angular orientation (i.e. tilt angle) of the columnar structures is relatively less affected with respect to the growth direction. Thus, it can be stated that small amount of Indium affects the geometric size of columnar structure with minimal effect on the strain component. However, Indium is incorporated as a surfactant during growth, which may give rise to the nucleation of less rough growth and hence a larger domain structure associated with the columnar growth.

The twist angles to estimate edge dislocation density were obtained directly from Φ-scans on asym-metric reflection planes. Figure 2 shows the variation in dislocation densities with Indium variance.

phys. stat. sol. (c) 2, No. 7 (2005) / www.pss-c.com 2147

© 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0.5 um 0.5 um 0.5 um

0.00 0.05 0.10 0.15

1E9

1E10

Dis

loca

tio

n d

ensi

ty [

cm-2

]

In composition [%]

Nedge

Nscrew

Fig. 2 Variation of edge dislocation (Nedge) and screw dislocation (Nscrew) densities with In composition. Nedge, with the Burgers vector |ba| and Nscrew, with the Burgers vector |bc| are determined by Nedge = αtwist / (2.1 |ba| Lll

) and Nscrew = αtilt

2/ ( 4.35 |bc|2 ) [8, 9,14, 15].

The screw dislocation density does not change much with Indium contents. However, the edge disloca-tion density reduces noticeably with the incorporation of small amounts of Indium in the layer. Thus, there is an effective decrease in the total TD density, which is the combination of these two dislocation densities. The tilt angle and twist angle are the dominant causes of defects, while the lateral coherence length affects the annihilation of dislocation density. Therefore, it was found that small amount of Indi-um incorporation increases the physical size of columnar structure in these epilayers without degrading crystalline quality of these layers. However, excess amount of Indium incorporation might increase the defects by forming relaxed layers.

Fig. 3 AFM images of (a) sample A, (b) sample B and (c) sample C after etching by H3PO4

The surface morphologies of the samples are shown in Fig. 3. In order to verify the effect of small amount of Indium incorporation, AFM measurements on these samples were carried out after wet chemi-cal etching using phosphoric acid (H3PO4). The dislocation density was estimated by counting the etch-pits over a specific area. The results show that the number of etch-pits is decreased, as Indium is incorpo-rated as shown in Table 1. RSM can also be used to investigate the columnar structure. The intensity distribution of the X-ray reflection can be characterized to explain two broadening mechanisms that occur because of the physical size and the angular misalignment of the columnar structure [10–13]. The values of physical parameters and the dislocation density obtained by these two XRD techniques are tabulated in Table 1. Although there is good agreement between the trends observed by both techniques, there is a discrepancy between the absolute numbers. This is due to interpolation of c-spacing parameters of columnar structure by Wil-liamson-Hall plot, where as RSM gives specific values for columnar structure on a specific reflection

2148 H. Kang et al.: Effects of indium incorporation in AlGaN

© 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

plane. In Williamson-Hall plot, the columnar structure parameters are determined by estimating a trend among different reflection planes, but in RSM, a specific value on a specific reflection plane is used to estimate the parameters of columnar structures. Thus, these differences are derived from non uniform columnar structure when measuring on different reflection planes.

Table 1 Summary of columnar structure factors of AlGaN with different indium content.

Method Sample A Sample B Sample C

W-H plot 0.26 0.58 1.25 Lll [µm] RSM 0.08 0.076 0.103

W-H plot 1222 1233 1259 αtilt [aresec]

RSM 1475 1433 1361 Φ scan 3750 3539 5592

αtwist [arcsec] RSM 3413 3438 4680

W-H plot 3.00E+09 3.05E+09 3.19E+09 Nscew [cm–2]

RSM 4.37E+09 4.13E+09 3.72E+09 W-H plot 1.40E+10 4.42E+09 3.24E+09

Nedge [cm–2] RSM 3.10E+10 3.27E+10 3.29E+10

TD [cm–2] Etch Pits 2.20E+09 2.00E+09 1.70E+09

4 Conclusion

In this study, it is found that a small amount of Indium incorporation appears to improve the structural properties of the columnar growth structures in AlGaN. This results in reduction of TD density with trace Indium incorporation. This means that integration of columnar structure with larger size can relax the TD density in epilayers. In other words, the local incorporation of Indium causes the annihilation of the de-fects in the epilayers, thereby improving the growth quality of the films.

Acknowledgements This work was funded by AFOSR (Lt. Col. Todd Steiner) under contract F49620-03-1-0294.

References

[1] W. Perry, M. Bremser, T. Zheleva, K. Linticum, and R. Davis, Thin Solid Films 324, 107 (1998). [2] M. Kneissl, D. Treat, M. Tepe, N. Miyasita, and N. Johnson, Appl. Phys. Lett. 82, 4441 (2003). [3] S. Terao, M. Iwaya, T. Sanoa, T. Nakamura, S. Kamiyama, H. Amano, and I. Akasaki, J. Cryst. Growth 237-

239, 947 (2002). [4] C. Chen, J. Zhang, M. Gaevski, H. Wang, W. Sun, R. Fareed, J. Yang, and M. Khana, Appl. Phys. Lett. 81,

4961 (2002). [5] S. Mochizuki, T. Detchprohm, S. Sano, T. Nakamura, and I. Akasaki, J. Cryst. Growth 237-239, 1065 (2002). [6] H. Hirayama, M. Ainoya, A. Kinoshita, and Y. Aoyagi, Appl. Phys. Lett. 80, 2057 (2002). [7] H. Kang, N. Spencer, D. Nicol, Z.C. Feng, I. Ferguson, S. Guo, M. Pophristic, and B. Peres Mat. Res. Soc.

Symp. Proc. 743, L6.12.1 (2002). [8] T. Metzger, R. Hopler, E. Born, and O. Ambacher, Philos. Mag. A 77, 1013 (1998). [9] H.Wang, J. Zhang, C. Chen, Q. Fareed, and J. Yang, Appl. Phys. Lett. 81, 605 (2002). [10] P. Fewster, X-Ray Scattering from Semiconductor (Imperial Press, London, 2000), p. 345. [11] P. Fewster, X-Ray and Neutron Dynamical Diffraction: Theory and Applications, NATO ASI Series, Ser. B

357, 321 (1996). [12] P. Fewster, N. L. Andrew, and C.T. Foxon, J. Cryst. Growth 230, 404 (2001). [13] D. Bowen, High Resolution X-ray Diffractometry and Topography (Talyor & Francis, 2001), p. 67. [14] G. Williamson and W. Hall, Acta. Metall. 1, 22 (1953). [15] De Keijser et al., J. Appl. Crystallogr. 16, 309 (1983).