Optical properties of GaN grown on porous silicon substrate

phys. stat. sol. (a) 201, No. 3, 582–587 (2004) / DOI 10.1002/pssa.200306740

© 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Optical properties of GaN grown on porous silicon substrate

T. Boufaden*, 1, A. Matoussi2, S. Guermazi2, S. Juillaguet3, A. Toureille4, Y. Mlik2, and B. El Jani1 1 Unité de Recherche sur les Hétéro-Epitaxies et Applications, Faculté des Sciences de Monastir, Tunisia 2 Laboratoire LaMaCoP de Sfax, Institut Préparatoire aux Etudes d’Ingénieurs de Sfax, Tunisia 3 Groupe d’Etudes des Semi-conducteurs, USTL, Montpellier, France 4 Laboratoire d’Electrotechnique de Montpellier, USTL, Montpellier, France

Received 16 July 2003, revised 22 October 2003, accepted 10 November 2003 Published online 22 January 2004

PACS 68.37.Hk, 68.55.Jk, 71.35.–y, 78.55.Cr, 81.15.Gh

A photoluminescence (PL) study of GaN grown on Si(100) substrate using porous silicon (PS) as an in-termediate layer is reported. The samples were characterized using PL for the temperature range 5–300 K under various excitation powers from 5 to 50 mW. For growth temperatures below 800 °C, the room tem-perature PL shows a broad peak located around cubic GaN emission. This is in clear contradiction with previous scanning electron microscopy and X-ray measurements. At low PL temperature, the observed lines located at 3.306 and 3.364 eV have a narrow full width at half maximum of about 6 and 10 meV, re-spectively. When the excitation power was varied, no peak shift was observed. These peaks were assigned as deeply localized excitons related to stacking faults near the PS/GaN interface. Quantum confinement (type I or II) due to the presence of nanometric cubic inclusions is another possible explanation for the low-temperature PL.


1 Introduction

The integration of GaN-based devices on the well-developed Si electronic technology is a great chal-lenge. Si substrate is available with large size, of high quality and at low cost. It is also stable under GaN hard growth conditions. To date, GaN properties still remain poor compared to those obtained on sap-phire. The Si presents a large lattice and thermal coefficient mismatch with GaN, which leads to a high density of defects, a residual tensile strain and many cracks in the subsequent GaN layer. Much effort has been made to overcome these problems of heteroepitaxy by the use of compliant layers [1–3]. A review of intermediate layers used between GaN and Si is reported by Krost and Dadgar [3]. Recently research-ers explored GaN growth on Si and SiC using a porous Si (PS) intermediate layer [4–10]. A nano-epitaxial lateral overgrowth is expected to occur on the nano-patterned surface of the PS layer reducing the density of defects and the residual strain without dislocation formation. Experimental observation of GaN on porous SiC revealed that misfit dislocations can terminate at a pore or bend laterally into the GaN grown over a pore [11]. Si, SiC and GaN porous layers are obtained by anodization or by reactive ion etching [4, 12]. Recently Gosh et al. [10] obtained PS GaN by nitridation of a 2 µm thick GaAs layer deposited on Si(111). In our work, for the growth of GaN on PS/Si(100) [4–6], we are actively pursuing several issues to improve growth. We are testing several PS porosities and AlN layer thicknesses. There are two main directions of study. The first deals with the use of PS layers to obtain a bi-dimensional growth process and to reduce the dislocation density to less than 109 cm–2. The second concerns polycrystalline GaN on

* Corresponding author: e-mail: [email protected], Phone: 216 73 500 278, Fax: 216 73 500 276

phys. stat. sol. (a) 201, No. 3 (2004) / www.pss-a.com 583


PS/Si substrate. Polycrystalline GaN, facilely obtained on a wide variety of large-size and low-cost sub-strates [13, 14], is promising for the fabrication of large-area and low-cost photonic devices operating in the blue and ultraviolet regions [13]. Recently Bour et al. [15] succeeded in the fabrication of a polycrys-talline InGaN/GaN light emitting diode (LED). The band edge photoluminescence (PL) efficiency of polycrystalline GaN is higher than that obtained on sapphire. In some cases the PL of polycrystalline GaN is dominated by narrow deep emissions related to structurally perturbed regions. The origin of these emissions is not clear. It seems to depend on the growth process, the growth temperature and the sub-strate used. These deep lines dominate the low-temperature PL of our GaN samples grown on PS. A detailed study of the optical properties of GaN/PS/Si is reported in this article. Sample temperature was varied from 5 to 300 K with different excitation power densities.

2 Results and discussion

The growth was performed by metal organic vapour-phase epitaxy (MOVPE) at atmospheric pressure. In order to work on the same PS layer porosity and thickness, the substrates used were cut from the same PS/Si wafer. We tested the stability of the PS layer under high GaN growth temperature. Scanning elec-tron microscopy (SEM) revealed no change in the pore distributions, forms and dimensions. GaN was grown directly on PS/Si without any NH3 treatment. The growth temperature was varied from 450 to 1050 °C. Details have been presented elsewhere [4–6]. In this report typical sample thick-nesses are in the range 1–2 µm. PL measurements from 5 to 300 K were made using an argon laser with doubled frequency (244 nm) as the excitation source with power density ranging from 5 to 50 mW. A Jobin Yvon triax spectrometer and cooled CCD camera completed the set-up. Figure 1 displays the room temperature PL spectra of the grown samples. The laser power was 10 mW. Two sets of peaks are clearly discernable. The first set is located on the near band edge lumines-cence of hexagonal GaN. The second set is located around 3.24 eV coinciding with cubic GaN emission. For growth temperatures of 480, 600, 800 and 1050 °C, the peaks are located at 3.25, 3.24, 3.424 and 3.396 eV with a full width at half maximum (FWHM) of 98, 102, 160 and 143 meV, respectively. Based on the SEM and X-ray (θ – 2θ) analyses performed on these samples [4, 5], the PL result obtained for the second set of peaks is surprising. In fact X-ray analysis and the particular surface morphology (SEM) indicate a wurtzite structure for samples grown below 800 °C (480, 500 and 600 °C). In all cases this does not exclude possible cubic inclusions as noted in many reports [16]. The same room temperature PL behaviour was reported for GaN grown by MOVPE on different sapphire orientations [17–19]. We performed a study of PL versus temperature from 5 to 300 K. Low-temperature (5 K) emission of the samples is presented in Fig. 2. As for room temperature PL, two regions are also clearly observed. For high growth temperature, hexagonal excitons are present and are denoted as free excitons (FE) at about 3.5 eV and bound excitons (BE) at around 3.48 eV. For low growth temperature, a very narrow line appears at 3.364 eV followed by another line at 3.306 eV. These lines are labelled I3 and I4, respec-

3.0 3.1 3.2 3.3 3.4 3.5 3.6

600˚C (x5)

480˚C (x5)

800˚C (x5)

1050˚C

Inte

nsity

(a.u

)

Energy (eV)

Fig. 1 Room temperature PL spectra of samples grown on PS/Si(100) at various temperatures.

584 T. Boufaden et al.: Optical properties of GaN grown on porous silicon substrate


3,3 3,4 3,5

3.3 3.4 3.5

3.31 eV

3.364 eV

WZ band edge

Inte

nsi

ty(a

.u)

Energy (eV)

480 ˚C

600 ˚C

800 ˚C

Inte

nsity

(a.u

)

Energy (eV)

3.1 3.2 3.3 3.4

150 K (x100)

100 K (x 100)

77 K

50 K

30 K

20 K

15 K

5 K

Inte

nsity

(a.u

)

Energy (eV)

tively [18–22]. Martinez-Criado et al. [23] associated these peaks with the presence of triangular cubic defects. If we believe that room temperature PL is from cubic GaN, we will expect to observe excitonic emis-sions at low temperature of cubic crystals at around 3.261 eV [24, 25]. The I3 and I4 lines are situated far from cubic excitons. Wetzel et al. [21] varied the pressure up to 4 GPa and studied its effect on the low-temperature PL. No significant shift of the I3 and I4 lines is observed in contrast to the shift of cubic exci-tons, which is about 190 meV at 4.4 GPa. This feature rules out the cubic nature of the I3 and I4 lines and indicates that the emission at room temperature is an overlap of these emissions (broadened and shifted). In addition, in our SEM images we did not see any triangular defects [23] or grains exhibiting distinct square facets [26] characteristic of SEM images of cubic GaN. For the samples grown at 480 and 600 °C, we note the absence of any emission in the yellow band region, while samples grown at 800 and 1050 °C reveal a high-intensity yellow emission. The evolution of the I3 and I4 lines as a function of temperature is presented in Fig. 3. At a temperature of 5 K the FWHM of I3 and I4 are about 6 and 10 meV, respectively. As the temperature increases, two other peaks appear. The first is located at 3.23 eV and the second at 3.16 eV. The spacing between the I4 and 3.23 eV lines and between the 3.23 and 3.16 eV lines is about 70 meV. This value coincides with that reported by Chen et al. [17] and associated with TO coupling. The 3.23 and 3.16 eV lines are the first and second TO replicas of the I4 line. The origin of this peak (I4) is a subject of controversy. Hong et al. [27] assigned it to a donor–acceptor pair transition (DAP). When the excitation power was increased, they observed a clear high-energy shift of the I4 line. In the present work, we also varied the excitation power between 5 and 50 mW. No shift was observed for either I3 or I4 lines in accordance with other reports [17, 23]. We think that this range of excitation is sufficient to see at least a small shift towards high or low energy. Figure 4 shows Arrhenius plots of the integrated intensities of the I3 and I4 lines versus 1/T. When the temperature is increased from 5 K a rapid quenching of the two line intensities is observed. We used the well-known thermal relation (I) to fit the obtained data:

0

B

,1 exp ( / )i i

i

II

A E k T=

+ −∑ (1)

where I0 is the intensity at 0 K, Ai are the transition probabilities of non-radiative processes, Ei are the activation energies and T is the temperature of the sample. This fitting gives two activation energies of

Fig. 2 Low-temperature (5 K) PL of samples grown on PS/Si(100) at various temperatures. The excitation power is 10 mW. The inset (PL in another zone) shows the appearance of a weak WZ band edge emission.

Fig. 3 Evolution of PL spectra as a function of tempera-ture for the sample grown at 600 °C. The same behaviour was observed for the sample grown at 480 °C.



0,00 0,05 0,10 0,15 0,200

100

200

300

400

500

Inte

grat

edin

tens

ity(a

.u)

Inverse temperature (K-1)

Exp fit 3.364 lineExp Fit 3.306 line

about 1.3 and 19 meV for the I3 line and an energy of about 26 meV for the I4 line. The first very low energy is in accordance with the capture of a free exciton. In fact Eckey et al. [28] observed in low-temperature PL transient decay a first 20 ps rise time consistent with the capture of a free exciton. The activation energy of the I4 line is close to activation energies found for FB or bound excitons and is different from the activation energy of DAP-related PL lying between 130 and 250 meV [21] and is close to the thermal depth of the acceptor [29]. We must note here that Mah et al. [30] reported activation energies of 25 and 230 meV at low and high temperature, respectively, for the 3.278 eV line (DAP). In our case, the narrowness and no shift versus excitation power of the I4 line clearly deny the DAP nature. There are two possible explanations for the nature of the I3 and I4 peaks. The first explanation was pro-posed by many authors [17–19, 21, 23]. They assigned strongly localized exciton nature to the I3 and I4 lines. When the temperature is increased from 5 to 150 K the I3 line moves towards low energies by about 24 meV while the red shift of the I4 line is only about 7 meV. This small I4 shift (compared to the I3 line) is close to that observed for free to bound features [17]. This would support the attribution of the FB nature to the I4 line rather than the strongly localized exciton. The presence of the strongly localized excitons in the samples grown on PS/Si at low growth temperature is associated with a high density of extended defects essentially in the GaN/PS interface. The rough surface of the PS layer can also enhance the formation of stacking faults generating (111) and (11–20) facets in accordance with PL reported for GaN grown on different sapphire orientations [17, 18]. The density of these defects seems to be ther-mally disfavoured. In fact, the bound excitons appear up to a growth temperature of 800 °C. Above this temperature their intensities quench. This may be related to the change of GaN structural morphology from a columnar-like structure at 600 °C to large crystallites of about 4 µm above 800 °C [5]. Sun et al. [31] reported the 3.3 eV line in a cross-sectional cathodoluminescene imaging experiment performed on a 5 µm thick sample. This emission is located above the substrate surface (1–1.5 µm) indicative that this emission is not limited to defects in the interface and can have its origins due to the existence of cubic-phase GaN domains in the hexagonal phase. Recently the quantum confinement of carriers in cubic inclusions in the hexagonal phase analogous to a type I and II quantum well was pro-posed by Mah et al. [22] for the (I3, I4) lines and by Lu et al. [32] and Albrecht et al. [33] for the 3.17 and 3.41 eV lines, respectively. This model can be a second potential explanation of our low-temperature PL, knowing that at low growth temperature possible nanometric cubic inclusions may exist on the hexago-nal GaN matrix. Further studies are needed to clarify the nature of the I3 and I4 emissions. Transmission electron microscopy of these samples to detect nanometric cubic inclusions may be helpful in this regard.

3 Conclusion

A PS layer was used as an intermediate layer for the growth of GaN on Si(100) substrate. Samples were characterized using SEM, X-ray analysis and PL (5–300 K). For a growth temperature below 800 °C, the room temperature PL spectra are dominated by an emission around 3.24 eV, while for higher growth

Fig. 4 Evolution of the integrated intensity of the I3 and I4

peaks versus inverse temperature.

586 T. Boufaden et al.: Optical properties of GaN grown on porous silicon substrate


temperatures luminescence peaks are located around the band edge of WZ GaN. The presence of the 3.24 eV peak is in clear disagreement with the X-ray analysis, which indicates a WZ crystalline or microcrystalline character. Low-temperature PL shows the presence of intense and sharp lines labelled I3 and I4. No shift is observed towards higher energy versus excitation power. This indicates a signature of deeply bound excitons for the I3 line and a FB signature for the I4 line. Type I and II quantum confine-ment of the carriers is another possible explanation of our low-temperature PL. The intensity of these lines quenches when the growth temperature is increased above 800 °C. The high growth temperature reduces the density of the extended defects responsible for the deeply localized excitons.

Acknowledgements The authors gratefully acknowledge support from the DGRST and the CMCU project under number 01/F1310.

References

[1] M. A. Sanchez-Garcia, J. Ristic, E. Calleja, A. Perez-Rodriguez, C. Serre, A. Romano-Rodriguez, J. Morante, R. Koegler, W. Skorupa, A. Trampert, and K. H. Ploog, phys. stat. sol. (a) 192, 401 (2002).

[2] E. Feltin, B. Beaumont, M. Laügt, P. De Mierry, P. Vennéguès, M. Leroux, and P. Gibart, phys. stat. sol. (a) 188, 531 (2001).

[3] A. Krost and A. Dadgar, phys. stat. sol. (a) 194, 361 (2002). [4] A. Missaoui, M. Saadoun, H. Ezzaouia, B. Bessaïs, T. Boufaden, A. Rebey, and B. El Jani, phys. stat. sol. (a)

182, 189 (2000). [5] A. Matoussi, T. Boufaden, A. Missaoui, S. Guermazi, B. Bessaïs, Y. Mlik, and B. El Jani, Microelectron. J. 32,

995 (2001). [6] T. Boufaden, N. Chaaben, M. Christophersen, and B. El Jani, Microelectron. J. 34, 843 (2003). [7] T. W. Kang and S. H. Park, J. Mater. Res. 15, 2602 (2000). [8] J. Liang, S.-K. Hong, N. Koukin, R. Beresford, and J. M. Xu, Appl. Phys. Lett. 83, 1752 (2003). [9] M. Mynbaeva, A. Tikov, A. Kryzhanovski, I. Koutousova, A. S. Zubrilov, V. V. Ratnikov, V. Yu. Davydov,

N. Kuznetsov, K. Mynbaev, D. V. Tsvetkov, S. Stepanov, A. Cherenkov, and V. A. Dmitriev, MRS Internet J. Nitride Semicond. Res. 4, 14 (1999).

[10] B. K. Ghosh, T. Tanikawa, A. Hashimoto, A. Yamamoto, and I. Ito, J. Cryst. Growth 249, 422 (2003). [11] C. K. Inoki, T. S. Kuan, A. Sagar, C. D. Lee, R. M. Feenstra, D. D. Koleske, D. J. Diaz, P. W. Bohn, and I. Ade-

sida, in: Proc. 5th Int. Conf. Nitride Semiconductors, Nara, Japan, phys. stat. sol. (a) 200, No. 1, 36 (2003). [12] M. G. Mynbaeva and D. Tsvetkov, paper presented at 23rd Int. Symp. Compound Semiconductors, St. Peters-

burg, Russia, 23–27 Sept. 1996, Inst. Phys. Conf. 155, Chap. 4. [13] H. Asahi, H. Tampo, K. Yamada, K. Ohnishi, Y. Imanishi, and K. Asami, phys. stat. sol. (a) 188, 601 (2001). [14] Seong-Eun Park, Dae-Jin Kim, and O. Byungsung, J. Cryst. Growth 252, 87 (2003). [15] D. P. Bour, N. M. Nickel, C. G. Van de Walle, M. S. Kneissl, B. S. Krusor, Ping Mei, and N. M. Johnson,

Appl. Phys. Lett. 76, 2182 (2000). [16] T. Yodo, H. Ando, H. Tsuchiya, D. Nosei, M. Shimeno, and Y. Harada, J. Cryst. Growth 227, 431 (2001). [17] Zhen Chen, Da-Cheng Lu, Xianglin Liu, Xiaohui Wang, Peide Han, Du Wang, Hairong Yuan, ZhanGuo

Wang, GuoHua Li, and Zaili Fang, J. Appl. Phys. 93, 316 (2003). [18] M. Cazzanelli, D. Cole, J. F. Donegan, J. G. Lunney, P. G. Middleton, K. P. O’Donnell, C. Vinegoni, and

L. Pavesi, Appl. Phys. Lett. 73, 3390 (1998). [19] M. Cazzanelli, C. Vinegoni, D. Cole, J. G. Lunney, P. G. Middleton, C. Trager-Cowan, K. P. O’Donnell, and

L. Pavesi, Mater. Sci. Eng. B 59, 137 (1999). [20] R. Dai, S. Fu, J. Xie, G. Hu, and C. Klingshirn, J. Phys. C 15, 393 (1982). [21] C. Wetzel, S. Fisher, J. Krüger, E. E. Haller, R. J. Molnar, T. D. Moustakas, E. N. Mokhov, and P. G. Baranov,

Appl. Phys. Lett. 68, 2556 (1996). [22] K. W. Mah, J.-P. Mosnier, E. McGlynn, M. O. Henry, D. O’Mahony, and J. G. Lunney, Appl. Phys. Lett. 80,

3301 (2002). [23] G. Martinez-Criado, A. Cros, A. Cantarera, R. Dimitrov, O. Ambacher, and M. Stutzman, J. Appl. Phys. 88,

3470 (2000). [24] O. C. Noriega, A. Tabata, J. A. N. T. Soares, S. C. P. Rodrigues, J. R. Leite, E. Ribeiro, J. R. L. Fernandez,

E. A. Meneses, F. Cerdeira, D. J. As, D. Schikora, and K. Lischka, J. Cryst. Growth 252, 208 (2003). [25] H. Yaguchi, J. Wu, B. Zhang, Y. Segawa, H. Nagasawa, K. Onabe, and Y. Shiraki, J. Cryst. Growth 195, 323

(1998).



[26] J. Wu, H. Yaguchi, B. P. Zhang, Y. Segawa, K. Onabe, and Y. Shiraki, phys. stat. sol. (a) 180, 403 (2000). [27] C. H. Hong, D. Pavlidis, S. W. Brown, and S. C. Rand, J. Appl. Phys. 77, 1705 (1995). [28] L. Eckey, J.-Ch. Holst, P. Maxim, R. Heitz, A. Hoffmann, I. Broser, B. K. Meyer, C. Wetzel, E. N. Mokhov,

and P. G. Baranov, Appl. Phys. Lett. 68, 415 (1996). [29] M. Leroux, N. Grandjean, B. Beaumont, G. Nataf, F. Semond, J. Massies, and P. Gibart, J. Appl. Phys. 86,

3721 (1999). [30] K. W. Mah, E. McGlynn, J. Castro, J. G. Lunney, J.-P. Mosnier, D. O’Mahony, and M. O. Henry, J. Cryst.

Growth 222, 497 (2001). [31] X. L. Sun, S. H. Goss, J. L. Brillson, D. C. Look, and R. J. Molnar, phys. stat. sol. (b) 228, 441 (2001). [32] X. H. Lu, P. Y. Yu, L. X. Zheng, S. J. Xu, M. H. Xie, and S. Y. Tong, Appl. Phys. Lett. 82, 1033 (2003). [33] M. Albrecht, S. Christiansen, G. Salvati, C. Zanotti-Fregonara, Y. T. Rebane, Y. G. Shreter, M. Mayer,

A. Pelzmann, M. Kamp, K. J. Ebeling, M. D. Bremser, R. F. Davis, and H. P. Sttrunk, Mater. Res. Soc. Symp. Proc. 468, 293 (1997).

Documents

Optical properties of GaN grown on porous silicon substrate