Phase separation suppression in InGaN epitaxial layers due to biaxial strain A. Tabata, L. K. Teles, L. M. R. Scolfaro, J. R. Leite, A. Kharchenko, T. Frey, D. J. As, D. Schikora, K. Lischka, J. Furthmüller, and F. Bechstedt Citation: Applied Physics Letters 80, 769 (2002); doi: 10.1063/1.1436270 View online: http://dx.doi.org/10.1063/1.1436270 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/80/5?ver=pdfcov Published by the AIP Publishing This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 186.217.234.225 On: Tue, 14 Jan 2014 12:32:40 http://scitation.aip.org/content/aip/journal/apl?ver=pdfcov http://oasc12039.247realmedia.com/RealMedia/ads/click_lx.ads/www.aip.org/pt/adcenter/pdfcover_test/L-37/2079205716/x01/AIP-PT/APL_ArticleDL_1213/aipToCAlerts_Large.png/5532386d4f314a53757a6b4144615953?x http://scitation.aip.org/search?value1=A.+Tabata&option1=author http://scitation.aip.org/search?value1=L.+K.+Teles&option1=author http://scitation.aip.org/search?value1=L.+M.+R.+Scolfaro&option1=author http://scitation.aip.org/search?value1=J.+R.+Leite&option1=author http://scitation.aip.org/search?value1=A.+Kharchenko&option1=author http://scitation.aip.org/search?value1=T.+Frey&option1=author http://scitation.aip.org/search?value1=D.+J.+As&option1=author http://scitation.aip.org/search?value1=D.+Schikora&option1=author http://scitation.aip.org/search?value1=K.+Lischka&option1=author http://scitation.aip.org/search?value1=J.+Furthm�ller&option1=author http://scitation.aip.org/search?value1=F.+Bechstedt&option1=author http://scitation.aip.org/content/aip/journal/apl?ver=pdfcov http://dx.doi.org/10.1063/1.1436270 http://scitation.aip.org/content/aip/journal/apl/80/5?ver=pdfcov http://scitation.aip.org/content/aip?ver=pdfcov APPLIED PHYSICS LETTERS VOLUME 80, NUMBER 5 4 FEBRUARY 2002 This a Phase separation suppression in InGaN epitaxial layers due to biaxial strain A. Tabata,a) L. K. Teles, L. M. R. Scolfaro, and J. R. Leiteb) Universidade de Sa˜o Paulo, Instituto de Fı´sica, Caixa Postal 66318, 05315-970 Sa˜o Paulo, SP, Brazil A. Kharchenko, T. Frey, D. J. As, D. Schikora, and K. Lischka Universität Paderborn, FB-6 Physik, D-33095 Paderborn, Germany J. Furthmüller and F. Bechstedt Institut für Festkörpertheorie und Theoretische Optik, Friedrich-Schiller-Universita¨t, D-07743, Jena, Germany ~Received 16 July 2001; accepted for publication 28 November 2001! Phase separation suppression due to external biaxial strain is observed in InxGa12xN alloy layers by Raman scattering spectroscopy. The effect is taking place in thin epitaxial layers pseudomorphically grown by molecular-beam epitaxy on unstrained GaN~001! buffers.Ab initio calculations carried out for the alloy free energy predict and Raman measurements confirm that biaxial strain suppress the formation of phase-separated In-rich quantum dots in the InxGa12xN layers. Since quantum dots are effective radiative recombination centers in InGaN, we conclude that strain quenches an important channel of light emission in optoelectronic devices based on pseudobinary group-III nitride semiconductors. ©2002 American Institute of Physics.@DOI: 10.1063/1.1436270# e o - ro e e- oy m im x er aN n e or he p la a uc fo ia y- isc the n ers. rate s. d in the ells. s p- fer t h wth ick, to nt ec- ic di- ther are ned or- aN ation s a ion. on 4 ma The light emission process in optoelectronic devic based on group-III nitride semiconductors is still a matter controversy in the literature.1 However, there is strong evi dence that an important emission mechanism originates f phase-separated quantum dots~QDs! formed by spinodal de- composition taking place in the InGaN alloys, the active m dia in these devices.2–4 Spinodal decomposition occurs b low a critical temperature and for a range of the all composition which defines a miscibility gap at a given te perature. It has been recognized from theory for a long t that the critical temperature lowers significantly due to bia ial strain in coherently grown semiconductor epitaxial lay and the miscibility gap may even be suppressed.5,6 Evidence that strain associated with thin InGaN layers in InGaN/G double heterostructures could suppress phase separatio been recently reported.7 A deep understanding of the rol played by strain on phase separation in InGaN layers highly desirable. The control of strain parameters is imp tant to monitor the QDs formation in the active region of t devices. In this letter, we show that external biaxial strain su press spinodal phase separation in thin InGaN epitaxial ers pseudomorphically grown on thick unstrained cubic~c! GaN~001! buffer layers. The InGaN films are terminated by top GaN layer forming GaN/InGaN/GaN double heterostr tures. By monitoring the alloy composition and thickness a fixed growth temperature, we control the presence of b ial strain induced by the rigid GaN buffer in the InGaN la ers. We start by first showing fromab initio calculations of the alloy free energytaking strain into accountthat the bi- axial strain is expected to induce a suppression of the m a!Permanent address: Universidade Estadual Paulista, Caixa Postal 17033-360 Bauva, S.P., Brazil. b!Author to whom correspondence should be addressed; electronic jrleite@macbeth.if.usp.br 7690003-6951/2002/80(5)/769/3/$19.00 rticle is copyrighted as indicated in the article. Reuse of AIP content is sub 186.217.234.225 On: Tue s f m - - e - s has is - - y- - r x- i- bility gap leading to a single homogeneous phase for InGaN alloys. We use high resolution x-ray diffractio ~HRXRD! reciprocal space maps to select the strained lay We have shown recently that micro-Raman is an accu tool to observe separate phases in InGaN epitaxial layer4,8 Micro-Raman spectroscopy measurements are also use this work to demonstrate conclusively the suppression of spinodal phase separation process in strained quantum w The c-GaN/InxGa12xN/GaN double heterostructure were grown on GaAs~001! substrates by molecular-beam e itaxy using a rf plasma nitrogen source. The GaN buf layers were grown atT5720 °C with thicknesses of abou 400 nm. Thec-InGaN films were deposited at lower growt temperatures of 600 °C. The films were deposited at gro rates of 40 nm/h. The GaN cap layers, of about 30 nm th were grown at low temperatures of about 600 °C in order reduce In desorption and interdiffusion. The growth fro was continuously monitored by reflection high-energy el tron diffraction and the diffraction patterns exhibited a cub symmetry along all major azimuths. We selected two sets of samples~436, 437 and 438, 439! where the InGaN layers were grown under identical con tions but two different thicknesses, one above and the o below the critical one. The characteristics of these layers shown in Table I. Two double heterostructures were plan to contain relaxed thicker alloy layers~samples 436 and 438!. The other two were tailored to comprise pseudom phic strained thinner InGaN layers~samples 437 and 439!. According to our theoretical predictions, the relaxed InG layers in samples 436 and 438 undergo phase separ while in samples 437 and 439 the biaxial strain stabilize single homogeneous phase against spinodal decomposit In order to investigate the effects of a biaxial strain the miscibility of the InxGa12xN alloy in a microscopic scale 73, il: © 2002 American Institute of Physics ject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: , 14 Jan 2014 12:32:40 In-rich 770 Appl. Phys. Lett., Vol. 80, No. 5, 4 February 2002 Tabata et al. This a TABLE I. InxGa12xN alloy compositionx as obtained from micro-Raman and XRD measurements.LW is the layer thickness, LO and S are the longitudinal optical phonon mode frequencies of the layers, and of the separated phases, respectively. Sample No. LW ~nm! LO ~cm21! x ~Raman! x ~XRD! S ~cm21! x ~Raman! x ~XRD! 436 30 685 0.36 0.33 630 0.61 0.54 438 30 701 0.26 0.27 621 0.65 0.55 437 3 672 0.45 0.45 ¯ ¯ ¯ 439 3 689 0.35 0.37 ¯ ¯ ¯ e o m ng ed th io i- or t cu the iax- loy tic un- the on s in red s of n iven aN ted ion si-ub to IP: e we calculated the Helmholtz free energyF(x,T) which al- lowed us to access theT2x phase diagram and obtain th critical temperature for miscibility. We express the therm dynamic potentialF of the alloy as F(x,T)5F0(x,T) 1DF(x,T), where F0(x,T)5(12x)FGaN(T)1xFInN(T), andDF(x,T)5DU(x,T)2TDS(x,T). F0 describes the free energy of a macroscopic mixture of the GaN and InN co ponents whose free energies areFGaN andF InN , respectively. DF gives the mixing free energy as a sum of the mixi enthalpy of the alloy (DU) and the mixing free entropy (DS). The calculation of the mixing free energy was carri out by combining the cluster expansion method within framework of the generalized quasichemical approximat andab initio density functional theory-local density approx mation. Details of the calculations are given in Ref. 9. Figure 1 shows the mixing free energyDF(x,T) for the InxGa12xN alloy as a function of composition calculated f the temperature range of interest. Results are shown for alloys in two extreme strain conditions:@Fig. 1~a!# fully re- laxed ~unstrained! and @Fig. 1~b!# pseudomorphically grown on a rigid GaN~001! buffer layer. Figure 1~a! shows that for the alloy growth temperature,'870 K, the composition of our relaxed layersx'0.4 lies inside a wide miscibility gap region. The critical temperature extracted from our cal FIG. 1. Mixing free energy (DF) of inhomogeneously strained c-InxGa12xN alloys as a function of compositionx for different tempera- tures.~a! unstrained and~b! pseudomorphically grown on a rigid GaN~001! buffer. rticle is copyrighted as indicated in the article. Reuse of AIP content is s 186.217.234.225 On: Tu - - e n he - lated T2x phase diagram is about 1295 K, thus above alloy growth temperature. On the other hand, when the b ial strain effects are introduced,DF in Fig. 1~b! exhibits a pronounced single minimum in the entire range of the al composition. TheT2x phase diagram undergoes a drama change. Spinodal decomposition taking place in our strained samples is predicted to be fully suppressed in samples under biaxial strain. HRXRD measurements with a Philips High Resoluti Diffractometer are used to investigate the strain condition our samples. Figure 2 depicts the distribution of the scatte x-ray intensity in reciprocal space@reciprocal space map ~RSM!# of the asymmetric (1̄1̄3) Bragg reflexes of sample 436 and 437~RSM’s of 438 and 439 are similar!. Also in- cluded are the positions of the Bragg reflexes of GaN and fully strained~pseudomorphic! and fully relaxed InN as well as the relaxation lines~dashed!, which indicate the position of Bragg reflexes of partially relaxed InGaN with a give FIG. 2. Distribution of the scattered x-ray intensity in reciprocal space~re- ciprocal space maps! of the asymmetric (1̄1̄3) Bragg reflexes of c-GaN/InxGa12xN/GaN double heterostructures~a! sample 436 and~b! sample 437. The width of the alloy layers in these heterostructures is g in Table I. The position of the maximum intensity of the GaN and InG Bragg reflexes are indicated by dots. The full lines show the calcula positions of the Bragg reflexes of strained InxGa12xN and relaxed InxGa12xN layers of varying In content. The dashed lines show the posit of the Bragg reflex of a partially relaxed InGaN layer of a given compo tion. ject to the terms at: http://scitation.aip.org/termsconditions. Downloaded , 14 Jan 2014 12:32:40 xe e he a p ng be i e - th al if p t y o s ng of th , at d o t g ve tra at 5 LO n D o e fi rly e a ou th to 9. er hi ce ose re not tra- osi- , as al- s- nd ro, pl. s, - D. s - 771Appl. Phys. Lett., Vol. 80, No. 5, 4 February 2002 Tabata et al. This a ub to IP: composition. The RSM of sample 436 shows Bragg refle of InGaN with two different compositions~x50.33 and 0.54!. The position of thex50.33 reflex reveals that th degree of relaxation in this layer is about 50% while t in-plane lattice parameter of the strained In-rich phase is most equal to that of GaN. Since the InGaN layer of sam 437 is only 3 nm thick the intensity of the correspondi Bragg reflex is low. However, a careful analysis of a num of line scans reveals only one intensity maximum which indicated in Fig. 2~b!. From its position, the In content of th layer is found to bex50.4560.03. The in-plane lattice pa rameter is equal to the GaN lattice spacing indicating that InGaN layer of this heterostructure is fully strained. The loy composition in our samples as obtained from x-ray d fraction ~XRD! measurements is shown in Table I. In order to clearly demonstrate that biaxial strain su press spinodal decomposition in our samples, we use micro-Raman spectroscopy technique recently adopted b to investigate the structural and optical properties c-InxGa12xN epitaxial layers.4,8 We showed that the alloy compositionx in the layer and in the In-rich separated pha can be obtained by measuring the frequencies of the lo tudinal optical~LO! phonon propagating in these regions the sample. Particularly, the LO phonon propagating in In-rich separated phase~QDs!, labeled by us as S phonon allows the identification of this phase and an approxim determination of its compositionx. Our Raman scattering measurements were performe room temperature with the Jobin–Yvon T64000 micr Raman system. Figure 3 shows the Raman spectra for c-GaN/InGaN/GaN samples, recorded in backscattering ometry. Assuming Lorentzian line shapes, the obser peaks in each spectrum were fitted after background sub tion and their frequencies were determined and are indic by arrows in Fig. 3. Thec-GaN phonon frequencies, 55 cm21 transverse optical~TO! and 741 cm21 ~LO! are clearly identified in the spectrum of each sample.10 The other peaks are originated from thec-InxGa12xN alloy and correspond to the TO and LO phonon modes of the layer and to the phonon propagating in the In-rich separated phases~S!. The fact that the LO phonon frequency ofc-InxGa12xN varies linearly with x allows us to determine the alloy compositio in our layers and in the In-rich phases.8 The results are shown in Table I. They are in good agreement with the XR data. The remarkable difference between the spectra samples 438, 436, and 437, 439 in Fig. 3 is the absenc the S peak for the latter. For the 438, 436 samples, the gerprint ~S! of the phase-separated In-rich QDs is clea observed as pronounced peaks between the TO and LO p of the InGaN layers. On the other hand, the S peak dis pears from the spectra of the samples 437 and 439. It c be argued that phase separation is correlated to the leng time required to grow the InGaN films. Since the time grow the thicker layers is about ten times~50 min! than that to grow the thinner ones~5 min!, this would explain the absence of separated phases in samples 437 and 43 check this point, we performed a set of annealing exp ments on samples 437 and 439, in steps of 1 h up to 10 h at 600 °C, each step followed by Raman experiments. Witrticle is copyrighted as indicated in the article. Reuse of AIP content is s 186.217.234.225 On: Tue s l- le r s e - - - he us f e i- e e at - he e- d c- ed f of n- aks p- ld of To i- n the detection limit of our equipment, there was no eviden for the presence of the S peak in the spectra of th samples. In conclusion, we show that in the two samples whe the InGaN layers are strained, the phase separation did take place. We consider this fact, the undoubted demons tion that phase separation induced by spinodal decomp tion in InGaN layers can be suppressed by biaxial strain expected from theory. The authors acknowledge Dr. E. Silveira for the anne ing experiments and Dr. A. Zunger for stimulating discu sions. This work was supported by FAPESP, CNPq, a DFG. 1T. Wang, J. Bai, S. Sakai, and J. K. Ho, Appl. Phys. Lett.78, 2617~2001!. 2S. Chichibu, T. Azuhata, T. Sota, and S. Nakamura, Appl. Phys. Lett.69, 4188 ~1996!; ibid. 70, 2822~1997!. 3K. P. O’Donnell, R. W. Martin, and P. G. Middleton, Phys. Rev. Lett.82, 237 ~1999!. 4V. Lemos, E. Silveira, J. R. Leite, A. Tabata, R. Trentin, L. M. R. Scolfa T. Frey, D. J. As, D. Schikora, and K. Lischka, Phys. Rev. Lett.84, 3666 ~2000!. 5A. Zunger, in Handbook of Crystal Growth, edited by D. T. J. Hurle ~Elsevier, New York, 1994!, Vol. 3, p. 998. 6S. Y. Karpov, MRS Internet J. Nitride Semicond. Res.3, 16 ~1998!. 7R. Singh, D. Doppalapudi, T. D. Moustakas, and L. T. Romano, Ap Phys. Lett.70, 1089~1997!. 8E. Silveira, A. Tabata, J. R. Leite, R. Trentin, V. Lemos, T. Frey, D. J. A D. Schikora, and K. Lischka, Appl. Phys. Lett.75, 3602~1999!. 9L. K. Teles, J. Furthmu¨ller, L. M. R. Scolfaro, J. R. Leite, and F. Bech stedt, Phys. Rev. B62, 2475~2000!. 10A. Tabata, R. Enderlein, J. R. Leite, S. W. da Silva, J. C. Galzerani, Schikora, M. Kloidt, and K. Lischka, J. Appl. Phys.79, 4137~1996!. FIG. 3. Raman spectra forc-GaN/InxGa12xN/GaN double heterostructure recorded for the excitation energyEL52.4 eV. The arrows indicate the pho non frequencies of the TO and LO modes ofc-GaN andc-InGaN layers and of the LO mode ascribed to the In-rich separated phase~S!. ject to the terms at: http://scitation.aip.org/termsconditions. Downloaded , 14 Jan 2014 12:32:40