Journal of Luminescence 131 (2011) 614–619

Contents lists available at ScienceDirect

Journal of Luminescence

0022-23

doi:10.1

n Corr

E-m

journal homepage: www.elsevier.com/locate/jlumin

Temperature-dependent photoluminescence of GaN grownon b-Si3N4/Si (1 1 1) by plasma-assisted MBE

Mahesh Kumar a,b, Mohana K. Rajpalke a, Basanta Roul a,b, Thirumaleshwara N. Bhat a, P. Misra c,L.M. Kukreja c, Neeraj Sinha d, A.T. Kalghatgi b, S.B. Krupanidhi a,n

a Materials Research Centre, Indian Institute of Science, Bangalore 560012, Indiab Central Research Laboratory, Bharat Electronics, Bangalore 560013, Indiac Laser Materials Processing Division, Raja Ramanna Centre for Advanced Technology, Indore 452013, Indiad Office of Principal Scientific Advisor, Government of India, New Delhi 110011, India

a r t i c l e i n f o

Article history:

Received 5 July 2010

Received in revised form

1 November 2010

Accepted 8 November 2010Available online 26 November 2010

Keywords:

Gallium nitride

MBE

Photoluminescence

Stress

13/$ - see front matter & 2010 Elsevier B.V. A

016/j.jlumin.2010.11.001

esponding author. Tel.:+91 80 22932943; fax

ail address: sbk@mrc.iisc.ernet.in (S.B. Krupan

a b s t r a c t

Photoluminescence (PL) of high quality GaN epitaxial layer grown on b-Si3N4/Si (1 1 1) substrate using

nitridation–annealing–nitridation method by plasma-assisted molecular beam epitaxy (PA-MBE) was

investigated in the range of 5–300 K. Crystallinity of GaN epilayers was evaluated by high resolution X-ray

diffraction (HRXRD) and surface morphology by Atomic Force Microscopy (AFM) and high resolution

scanning electron microscopy (HRSEM). The temperature-dependent photoluminescence spectra

showed an anomalous behaviour with an ‘S-like’ shape of free exciton (FX) emission peaks. Distant

shallow donor–acceptor pair (DAP) line peak at approximately 3.285 eV was also observed at 5 K,

followed by LO replica sidebands separated by 91 meV. The activation energy of the free exciton for GaN

epilayers was also evaluated to be �27.870.7 meV from the temperature-dependent PL studies. Low

carrier concentrations were observed �4.572�1017 cm�3 by measurements and it indicates the silicon

nitride layer, which not only acts as a growth buffer layer, but also effectively prevents Si diffusion from

the substrate to GaN epilayers. The absence of yellow band emission at around 2.2 eV signifies the high

quality of film. The tensile stress in GaN film calculated by the thermal stress model agrees very well with

that derived from Raman spectroscopy.

& 2010 Elsevier B.V. All rights reserved.

1. Introduction

Gallium nitride (GaN) and its related materials have been widelystudied for their unique applications in optoelectronic and hightemperature/ high power electronic devices with relatively low powerconsumption [1]. A number of studies are being performed success-fully to grow GaN on different substrates such as Al2O3, SiC, GaAs andSi [2,3]. Silicon is considered as one of the most promising substratesfor the GaN epitaxy because of its many advantages such as highquality, large size, low cost and a well-known existing devicetechnology [4]. However, there exist several hindrances that preventformation of the high quality layers. The difference in the latticeconstant and the thermal-expansion coefficients of GaN and Si, thedeformation fields and corresponding generation of high-densitydislocations at the interface of GaN layers grown on Si (1 1 1) is muchhigher than that of the homoepitaxially grown films. To overcomethese difficulties, the fast works are attempted to grow GaN films byintroducing several intermediate layers on Si substrates such as AlN

ll rights reserved.

: +91 80 23607316.

idhi).

[5,6], GaN [7] and SiC [8], and a defect-induced broad emission, yellowluminescence (YL), was commonly present [9]. By introducing siliconnitride buffer layer, the YL can be suppressed [10] and also it is a veryeffective material for diffusion barrier; so it could reduce the Sidiffusion in the GaN epilayers from substrate [11]. According to Chenget al. [12] report, the dislocation density in heteroepitaxial GaN layerscan also be reduced by using silicon nitride buffer layer. PL is a widelyused technique for a qualitative study of nitride samples since it is asimple, non-destructive and an effective technique [13]. A fairlycomplete investigation on the optical properties of wurtzite GaN onsapphire is accomplished [14], such as accurately measuring the GaNband-gap energy, determining the position of impurities and thephonon modes of GaN, etc. However, more thorough investigations ofGaN on Si substrate, specially using silicon nitride buffer layer, areneeded for a better understanding.

In this article, we have grown high quality GaN epilayers onb-Si3N4 layer formed by nitridation of Si substrate. Also, we havestudied the temperature (5–300 K) dependence of the PL inten-sities of different transitions typically recorded in GaN epilayersand behaviour of FX and DAP peak positions with temperature. Thestress is calculated by Hooke’s law and derived from Ramanspectroscopy.

M. Kumar et al. / Journal of Luminescence 131 (2011) 614–619 615

2. Experiment details

The growth system used in this study was a plasma-assisted MBEsystem (OMICRON) equipped with a radio frequency (RF) plasmasource. The base pressure in the system was below 1�10�10 mbar.The undoped 2-in. Si (1 1 1) substrates were ultrasonically degreasedand boiled in trichloroethylene, acetone and methanol at 70 1C for5 min, respectively, followed by dipping in 5% HF to remove thesurface oxide. The substrates were thermally cleaned at 900 1C for1 h in ultra-high vacuum. The samples were grown by nitridation–annealing–nitridation process, in which first the nitridation of thesubstrate was carried out at 530 1C for 30 min, followed by annealingat 900 1C for 30 min and again nitridation at 700 1C for 30 min. Afternitridation, a low-temperature GaN buffer layer of 20 nm was grownat 500 1C, where the Ga effusion cell temperature was kept at 950 1Cand corresponding beam equivalent pressure (BEP) was maintainedat 5.6�10�7 mbar. Afterwards, a GaN epilayer of thickness 225 nmwas grown on the buffer layer at 700 1C. Nitrogen flow rate andplasma power were kept to 0.5 sccm and 350 W, respectively, for thenitridation, buffer layer and subsequent GaN growth. For reproduci-bility we have grown five numbers of samples using the same growthparameters.

Crystallinity of GaN epilayers was evaluated by HRXRD andsurface morphology by AFM and HRSEM. The PL spectra wererecorded in the temperature range of 5–300 K using a closed cycleoptical cryostat and He–Cd laser of 325 nm excitation wavelengthwith a maximum input power of 30 mW. The Hall effect measure-ments were conducted at room temperature at 0.5 T of magneticfield. Samples of 5�5 mm2 size were cut from the wafers, andaluminum metal dots were vacuum evaporated in the four cornersto obtain electrical contacts in the Van der Pauw geometry. Thesamples were also characterized by micro-Raman spectroscopyusing 532 nm line of the Nd:YAG laser.

3. Results and discussion

Fig. 1 shows the HRXRD 2y-o scans of the GaN films grown on Si(1 1 1) substrate. Besides the Si (1 1 1) and (2 2 2) diffracted peaks

Fig. 1. HRXRD 2y-o scans of GaN on Si (1 1 1) substrate.

at 2y¼28.451 and 58.871, respectively, only a strong GaN (0 0 0 2)peak at 2y¼34.591 and a weak GaN (0 0 0 4) peak at 2y¼731 can beidentified, indicating the epitaxial GaN film to be highly orientedalong the [0 0 0 1] direction of the wurtzite GaN. An in-plane phiscan was also taken by rotating the sample around its surface-normal direction to investigate the in-plane alignment of the GaNfilm. As shown in Fig. 2 the diffraction peaks from the (1 0–1 1)plane of GaN were observed at 601 intervals, clearly confirming thehexagonal structure of the GaN epilayer. In addition, the (1–1 1)plane of Si substrate showed a threefold symmetry as expected.Fig. 3 shows the cross-section and surface SEM images. Thickness ofGaN in all samples measured by cross-section SEM is 24875 nmand the surface of the film is smooth and crack-free. A lower pitdensity (� 4.171�107 cm�2) and small size (range from 10 to35 nm) indicate the good quality of film. Fig. 4 shows 5�5 mm2

AFM images of the GaN layers and the root mean square (RMS)roughness is 3.60570.1 nm. The transport measurements on theGaN films were done using a mobility setup. The Hall effectmeasurements revealed n-type conductivity, with donor concen-trations �4.572�1017 cm�3 and the mobility of 37077 cm2/V sat room temperature.

The core-levels photoelectron spectroscopy was carried out todetermine the chemical bonding states of silicon nitride surface usingAl Ka radiation (hn¼1486.6 eV). Fig. 5 shows Si 2p core-level spectra(CLS), which have been numerically fitted using the Lorentzianconvoluted with Gaussian functions. The background has been takeninto account using a linear profile. In stoichiometric silicon nitride a Siatom is bonded to four N atoms (Si4+) and an N atom is bonded tothree Si atoms [15]; thus only the component with Si4+ is expected. Atthe interface two more coordinations are required (Si3+ and Si1+) foran ideal matching between the silicon nitride and Si (1 1 1) lattices[16]. Fig. 5 also shows the compositions: one bulk and three othercomponents assigned to silicon in Si1+, Si3+ and Si4+ coordination areneeded. The corresponding binding energies are at +0.66, +2.20,+3.05 eV (with respect to the bulk position). For the Si1+, Si3+ and Si4+

components, the measured ratio is 0.133:0.084:0.783 and it indicatesthat silicon nitride is in the form of crystalline b-Si3N4.

At room temperature, the GaN epilayers show a strong band-edge emission peak at 3.428 eV, as shown in Fig. 6. The band-edgeemission peak is slightly red-shifted compared to unstrained bulkGaN (�3.44 eV) due to tensile stress in the film [17]. Roomtemperature PL measurements also illustrate the absence of YL,which is due to the presence of gallium vacancies and deep levelimpurities. Generally, the absence of YL may also attribute to higher

Fig. 2. In-plane phi scan measured from GaN epilayers and Si (1 1 1) substrate

showing a sixfold and a threefold symmetry, respectively.

Fig. 3. SEM images (a) cross-section and (b) surface of GaN on Si (1 1 1) substrate.

Fig. 4. AFM image of GaN surface.

Fig. 5. Si 2p core-level spectra of silicon nitride film.

Fig. 6. Room temperature PL spectra of GaN films.

M. Kumar et al. / Journal of Luminescence 131 (2011) 614–619616

doping, causing carrier contribution in the range of 1019 cm�3.However, the measured values in the present case were almost twoorders of magnitude less. Such situation ruled out the contributionsfor the absence of YL from heavy unintentional doping.

A low-temperature (5 K) PL spectrum of GaN layers is shown inFig. 7(a). From Fig. 7(b) it can be seen that the line fitting for PLspectra contains the following main features: (1) free exciton (FE),neutral shallow donor bound exciton (DBE) and basal planestacking faults (BSFs) [18] peak near 3.501, 3.468 and 3.410 eV,respectively and (2) donor–acceptor pair (DAP) band with the mainpeak at 3.285 eV and a few longitudinal optical (LO) phononreplicas spaced by 91 meV. The PL linewidth of samples is around�50 meV and this is higher than uncompensated Si doped GaN, thePL linewidth at this doping level is about 2 meV in unstrained GaN,and around �10 meV in strained GaN (GaN on sapphire). So it isclear that there must be compensation present in the samples toincrease the PL linewidth, unless there are many other unknowndefects. The peak DBE is the Si-related DBE, with a strain free energyof 3.468 eV and is acceptable in present case because we havegrown on Si substrate at 700 1C. From the substrate, a little amountof Si has diffused in GaN, which was confirmed by X-ray photo-electron spectroscopy (XPS) and the diffusion of Si has decreasedwith increase in thickness of GaN. At higher energies the well-known intrinsic free exciton lines are clearly observed. The DAPpeak has been studied previously in both doped and undoped GaNlayers and time-resolved spectra strongly suggest that these

Fig. 7. (a) Low-temperature (5 K) PL spectrum of GaN layers on Si and (b) line fitting

showing FX, DBE and BSF peaks positions.

Fig. 8. Temperature dependence of peak position for the free excitons of GaN on Si

and the solid curves are fitted Varshni function.

M. Kumar et al. / Journal of Luminescence 131 (2011) 614–619 617

features are due to recombination at donor–acceptor pairs [19]. Themain residual stable shallow donors introduced by Si or oxygen (O)impurities and unstable shallow donors believed to be related to Hor N vacancies [20–21]. The presence of Si cannot be ruled out as wediscussed above. There is also a chance of O impurities as the filmswere grown at high pressures (2.8�10�5 mbar) and a little tracesof O might be present with nitrogen gas. However, we measuredthe O/N ratio by residual gas analyzer (RGA) and was found to bevery less (�1�10�4). We also noticed the absence of yellow band

emission at around 2.2 eV, which established the good opticalquality of film.

The intensity of DBE peak decreased abruptly with increase intemperature, due to delocalization of the bound exciton, while thatof the FX peak decreased gradually with temperature. This is thetypical behaviour of free and bound exciton emissions commonlyobserved in epitaxial GaN [22]. The temperature-dependent PLshows an ‘S-shape’ behaviour and the PL peaks sequential redshift,blueshift or a constant value, and then a redshift with increase intemperature. In order to investigate this effect thoroughly, weextracted more detailed information from temperature-dependentPL such as peak positions and intensities. The peak position for theFX emission lines is summarized in Fig. 8. The behaviour of the DBEwas not considered because some of these lines were not seenclearly at low temperatures and almost disappeared above 100 K. Itis well known that the temperature-dependent energy gap followsVarshni’s equation [23]:

EgðTÞ ¼ Egð0Þ�aT2=ðbþTÞ ð1Þ

where Eg(T) is the transition energy at temperature T , Eg(0) is thecorresponding energy at 0 K and a and b are known as Varshni’sthermal coefficients and the Debye temperature, respectively . Thesolid lines are obtained by least-squares fitting. The Varshniequation does not follow the typical temperature dependence ofthe FX emission and we fitted the measured data with the bestfitting parameters of a¼8.3�10�4 eV K�1 and b¼731.9 K. Theseparameters are in close agreement with those obtained by Shanet al. [24] for GaN layers grown on sapphire, except for a shift inEg(0) due to strain. An ‘S-shape’ behaviour was observed and themeasured data deviated further from the fitted line at a tempera-ture of 20 K. With increase in temperature up to 60 K, an initialdramatic decrease in PL peak energy was observed. Then the PLpeak was almost constant in the temperature range of 60–100 K.Finally, the PL peak energy decreased again with increase intemperature above 140 K and the PL emission peak followed thetemperature dependence described by Varshni’s equation onceagain. At low temperatures, carriers can recombine at energy statesof local potential minima. As the temperature slightly increases,weakly localized carriers are thermally excited and would either

M. Kumar et al. / Journal of Luminescence 131 (2011) 614–619618

recombine nonradiatively or be redistributed to other stronglylocalized states. Thus, the PL peak energy decreases with increase intemperature. After the effect of redistribution is saturated, thethermal energy can excite carriers to higher localized states, andhence the PL peak energy again increases. As the temperatureincreased further, a redshift occurred due to ordinary band-gapshrinkage. A similar behaviour has been reported previously for thetemperature-dependent PL emission energy shift, in GaN nanorodson Si (1 1 1) [25], in III-nitride ternary or quantum well systemssuch as AlxGa1�xN alloys [26] and InGaN/GaN multi-quantum wells[17]. In the present work the fitted lines agree well with themeasured data points except in the range 30–100 K.

The temperature dependence of 3.285 eV peak position isshown in Fig. 9 and it can be observed that a shift in DAP peakto higher energies by a few meV is observed when the temperatureis increased from 5 to 50 K. This type of behaviour has also beenobserved by Lagerstedt and Monemar [27] and is believed to becaused by the enhancement of more closely spaced pairs by ahigher thermal ionization rate for the donors when the tempera-ture is raised. The distance between these two peaks appears to be�1173 meV. If the recombination of free electrons with holes atthe acceptor (free-to-bound emission) is connected with the sameacceptor as the 3.285 eV emissions, the distance between the peaksat, say, 50 K should be

DE¼ ED�EeþEel ð2Þ

where ED is the donor binding energy, Ee is the DAP Coulomb energyat the peak position, which is estimated to be about 13 meV [28]and Eel is a correction for the mean free electron kinetic energy atthe peak, about kT, i.e., 4 meV at 50 K [27]. From the observeddistanceDE¼1173 meV at 50 K, we thus estimate a donor bindingenergy ED¼2074 meV and it has been speculated that this donormight be related to a metastable neutral state of interstitialhydrogen [19,29]. The peak position does not follow the band-gap temperature variation and it is temperature independent up to180 K (Fig. 9). This might partly be explained by the electron kineticenergy correction, which accounts for about half of the shiftexpected. Low-energy phonon interaction could probably alsoinfluence the peak position. The DAP is a DAP only at low

Fig. 9. Shift of transition energy as a function of temperature for PL peaks of

GaN films.

temperatures, above about �50 K it is a conduction band-acceptortransition.

The decrease in PL intensity with increase in temperature can bedemonstrated by the thermal ionization of electron–hole boundstates or localized states and the intensity of the FX transition peaksat different temperatures have been fitted with the generalisedexpression [30]:

IðTÞ ¼ Io=½1þCo expð�Ea=kTÞþC1 expð�Eloc=kTÞ� ð3Þ

where I(T) is the PL intensity at temperature T, Co, C1 and Io are theconstants, k is Boltzmann’s constant and Ea and Eloc are the activationenergy and localization energy in the high- and low-temperatureregime, respectively.

Fig. 10 shows the Arrhenius plots of the peak PL intensities for theFX transition related PL emission as a function of temperature. Weexpect the PL intensity to be dominated by localized excitons at lowtemperatures (To30 K). At higher temperatures (T430 K), the PLintensity is controlled by thermally activated centers. The fit gives alocalization energy (Eloc) of 4.9 meV and a thermal activation energy(Ea) of 27.870.7 meV. These values are in good agreement withthose reported by Monemar (28 meV) [17] and Chichibu et al.(27 meV) [31].

A distinctive feature appeared in the PL signal in terms of aredshift of the free excitons by �12 meV from the value of bulkGaN (3.440 eV) at room temperature [17], which was known to beattributed to the residual stress. It is worth noting that a tensile in-plane thermal strain perpendicular to the c-axis is inevitablygenerated during the cool-down procedure due to large differencein the thermal-expansion coefficients of GaN and Si. Taking intoaccount the GaN films under strain, the in-plane stress can beroughly estimated by Hooke’s law [32]:

s¼ ðaSi�aGaNÞBDT=u ð4Þ

wheres is the in-plane stress, B is the bulk modulus, which is 200 GPafor GaN [33], u is 0.38 for the Poisson ratio of the GaN [34], aSi is thethermal-expansion coefficient of Si, which is 3.59�10�6 K�1 [35],and aGaN is 5.59�10�6 K�1 [36] for the thermal-expansion

Fig. 10. Peak PL intensities of FX transition as a function of temperature of GaN on Si

(1 1 1) substrate.

Fig. 11. Room temperature micro-Raman spectra of GaN on Si (1 1 1) substrate.

M. Kumar et al. / Journal of Luminescence 131 (2011) 614–619 619

coefficient of GaN. DT is the temperature difference between growthtemperature (700 1C) and measured temperature (room temperature�300 1C). By using Eq. (4) the calculated value of the residual stress is�0.420 GPa. We also measured the stress by micro-Raman scatteringat room temperature and Fig. 11 shows a typical Raman spectrum ofGaN on Si (1 1 1) substrate. The E2 (high) mode in the Raman spectracan be used to estimate the stress because it has been proved to beparticularly sensitive to biaxial stress in samples. The redshift of the E2

(high) phonon band �565.77 cm�1 with respect to the standardvalue of �567.7 cm�1 [37] implies a biaxial tensile stress in the GaNlayer. The lattice constant of GaN is smaller than that of Si (the atomicspacing on the (1 1 1) plane of Si is 3.84 A) and its thermal-expansioncoefficient is larger than that of Si. Both lattice and thermalmismatches lead to the tensile stress in the GaN epilayer. Using thesevalues, and recently proposed relationDo ¼4.3sxx cm�1 GPa�1 [38],tensile stress is derived to be �0.448 GPa, which is very close to�0.420 GPa, which is calculated by Hooke’s law.

4. Conclusions

In conclusion, we have investigated the optical properties interms of PL spectroscopy of GaN epilayers grown on Si (1 1 1)substrates by nitridation–annealing–nitridation method usingplasma-assisted molecular-beam epitaxy. In order to confirm thecrystalline quality of the GaN epilayers, we employed HRXRD andsurface morphology by AFM and SEM. Temperature-dependent

photoluminescence spectra showed an abnormal ‘S- shape’ of freeexcitons (FX) and its activation energy was found to be �27.870.7meV. The absence of yellow band emission and low unintentionallydoping indicates the excellent optical properties of GaN films. Thetensile stress in GaN film calculated by the thermal stress modelagrees very well with the one derived by Raman spectroscopy.

References

[1] S. Nakamura, Science 281 (1998) 956.[2] S. Tripathy, S.J. Chua, P. Chen, Z.L. Miao, J. Appl. Phys. 92 (2002) 3503.[3] W.Y. Uen, Z.Y. Li, S.M. Lan, S.M. Li, J. Cryst. Growth 280 (2005) 335.[4] A. Dadgar, F. Schulze, M. Wienecke, A. Gadanecz, J. Blasing, P. Veit, T. Hempel,

A. Diez, J. Christen, A. Krost, New J. Phys. 9 (2007) 389.[5] J.W. Yang, A. Lunev, G. Simin, A. Chitnis, M. Shatalov, M. Asif Khan, J.E. Van

Nostrand, R. Gaska, Appl. Phys. Lett. 76 (2000) 273.[6] A. Dadgar, M. Poschenrieder, J. Blasing, K. Fehse, A. Diez, A. Krost, Appl. Phys.

Lett. 80 (2002) 3670.[7] H. Zhang, Z. Ye, B. Zhao, J. Appl. Phys. 87 (2000) 2830.[8] J.H. Boo, S.A. Ustin, W. Ho, J. Cryst. Growth 189 (1998) 183.[9] W.Y. Uen, Z.Y. Li, S.M. Lan, S.M. Liao, J. Cryst. Growth 280 (2005) 335.

[10] Y. Nakada, I. Aksenov, H. Okumura, Appl. Phys. Lett. 73 (1998) 827.[11] C.L. Wu, J.C. Wang, M.H. Chan, T.T. Chen, S. Gwo, Appl. Phys. Lett. 83 (2003)

4530.[12] K. Cheng, M. Leys, S. Degroote, M. Germain, G. Borghs, Appl. Phys. Lett 92 (2008)

192111.[13] M.A. Reshchikov, R.Y. Korotkov, Phys. Rev. B 64 (2001) 115205.[14] Z. Chen, D.C. Lu, X. Liu, X. Wang, P. Han, D. Wang, H. Yuan, Z. Wang, G. Li, Z. Fang,

Appl. Phys. Lett. 93 (2003) 316.[15] G.L. Zhao, M. Bachlechner, Phys. Rev. B 58 (1998) 1887.[16] J. Kim, H. Yeom, Phys. Rev. B 67 (2003) 035304.[17] B. Monemar, Phys. Rev. B 10 (1974) 676.[18] R. Ravash, J. Blasing, T. Hempel, M. Noltemeyer, A. Dadgar, J. Christen, A. Krost,

Appl. Phys. Lett. 95 (2009) 242101.[19] B. Monemar, P.P. Paskov, J.P. Bergman, M. Iwaya, S. Kamiyama, H. Amano,

I. Akasaki, Physica B 376/377 (2006) 460.[20] B. Monemar, P.P. Paskov, G. Pozina, C. Hemmingsson, J.P. Bergman,

T. Kawashima, H. Amano, I. Akasaki, T. Paskova, S. Figge, D. Hommel,A. Usui, Phys. Rev. Lett. 102 (2009) 235501.

[21] O. Gelhausen, M.R. Phillips, E.M. Goldys, T. Paskova, B. Monemar, M. Strassburg,A. Hoffmann, Phys. Rev. B 69 (2004) 125210.

[22] H.Y. Chen, H.W. Lin, C.H. Shen, S. Gwo, Appl. Phys. Lett. 89 (2006) 243105.[23] Y.P. Varshni, Physica (Amsterdam) 34 (1967) 149.[24] W. Shan, T.J. Schmidt, X.H. Yang, S.J. Hwang, J.J. Song, B. Goldenberg, Appl. Phys.

Lett. 66 (1995) 985.[25] Y.S. Park, T.W. Kang, R.A. Taylor, Nanotechnology 19 (2008) 475402.[26] Y.H. Cho, G.H. Gainer, A.J. Fischer, J.J. Song, S. Keller, U.K. Mishra, S.P. Denbaars,

Appl. Phys. Lett. 73 (1998) 1370.[27] O. Lagerstedt, B. Monemar, J. Appl. Phys. 45 (1974) 2266.[28] R. Dingle, M. Ilegems, Solid State Commun. 9 (1971) 175.[29] B. Monemar, P.P. Paskov, J.P. Bergman, A.A. Toropov, T.V. Shubina,

T. Malinauskas, A. Usui, Phys. Status Solidi (b) 245 (2008) 1723.[30] G. Davies, Phys. Rep. 176 (1989) 83.[31] S. Chichibu, T. Azuhata, T. Sota, S. Nakamura, J. Appl. Phys. 79 (1996) 2784.[32] T. Rieger, H. Metzger, R. Angerer, O. Dimitrov, Ambacher, M. Stutzmann, Appl.

Phys. Lett. 68 (1996) 970.[33] H. Amano, K. Hiramatsu, I. Akasaki, Jpn. J. Appl. Phys. 27 (Part 2) (1988) L1384.[34] T. Detchprohm, K. Hiramatsu, K. Itoh, I. Akasaki, Jpn. J. Appl. Phys. 31 (Part 2)

(1992) L1454.[35] B.H. Bairamov, O. Gurdal, A. Botchkarev, H. Morkoc, G. Irmer, J. Monecke, Phys.

Rev. B 60 (1999) 16741.[36] T. Wang,a, T. Shirahama, H.B. Sun, H.X. Wang, J. Bai, S. Sakai, H. Misawa, J. Appl.

Phys. 76 (2000) 2220.[37] L.A. Falkovsky, W. Knap, J.C. Chervin, P. Wisniewski, Phys. Rev. B 57 (1998)

11349.[38] M. Benyoucef, M. Kuball, B. Beaumont, P. Gibart, Appl. Phys. Lett. 80 (2002)

2275.