Superlattices and Microstructures 84 (2015) 13–23

Contents lists available at ScienceDirect

Superlattices and Microstructures

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o ca t e / s u p e r l a t t i c es

Photoreflectance study of GaN grown on SiNtreated sapphire substrate by MOVPE

http://dx.doi.org/10.1016/j.spmi.2015.04.0300749-6036/� 2015 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: elbouzidimed16@yahoo.com (M. Bouzidi).

M. Bouzidi ⇑, Z. Benzarti, I. Halidou, Z. Chine, A. Bchetnia, B. El JaniUniversité de Monastir, Faculté des Sciences, Unité de recherche sur les Hétéro-Epitaxies et Applications (URHEA), 5000Monastir, Tunisia

a r t i c l e i n f o

Article history:Received 23 April 2015Accepted 28 April 2015Available online 5 May 2015

Keywords:GaNMOVPESiN treatmentPhotoreflectanceExciton transition

a b s t r a c t

GaN films were grown on silicon nitride (SiN) treated c-planesapphire substrates in a home-made vertical reactor by atmo-spheric pressure metalorganic vapor phase epitaxy (MOVPE). Inorder to obtain different thickness layers, the growth procedurewas interrupted at diverse stages using in-situ laser reflectometry.The structural and optical properties of obtained samples wereinvestigated by high resolution X-ray diffraction (HRXRD) and pho-toreflectance (PR). In the 0.7–2 lm epilayer thickness range, thedislocation density decreases and remains roughly constant abovethis range. For fully coalesced layers, PR measurements at 11 Kreveal the presence of well resolved excitonic transitions relatedto A, B and C excitons. A strong correlation between dislocationdensity and exciton linewidths is observed. Based on theoreticalapproaches and experimental results, the electronic band structuremodification of GaN films due to isotropic biaxial strain wasinvestigated. The valence band deformation potentials D3 and D4,interband hydrostatic deformation potentials a1 and a2, spin–orbitDso and crystal field Dcr parameters were re-examined and foundto be 8.2 eV, �4.1 eV, �3.8 eV, �12 eV, 15.6 meV and 16.5 meV,respectively.

� 2015 Elsevier Ltd. All rights reserved.

14 M. Bouzidi et al. / Superlattices and Microstructures 84 (2015) 13–23

1. Introduction

Because of its large direct band-gap energy and good thermal stability, GaN and related materialshave attracted tremendous interest owing to a large variety of potential applications covering optical,optoelectronic, and electronic devices [1,2]. Extensive research and development efforts have beenexpended in III-nitrides materials over the past decades. These efforts were crowned by the fabricationand commercialization of a variety of devices such as lasers and light-emitting diodes,modulation-doped field-effect transistors and solar-blind ultraviolet photodetectors [3–6]. Despiteall of this, the full potential of GaN-based devices has yet to be achieved due to the GaN material itselfsuffering from a high density of defects inherent to heteroepitaxy on foreign substrates necessitatedby the lack of bulk GaN substrates. Further improvements in high performance devices call for thereduction of dislocation densities which are known to affect optical and electrical properties ofGaN-based devices by trapping point defects or forming complexes with them [6–8]. In this regard,various strategies were employed and many methods have been developed to reduce the dislocationdensity such as epitaxial lateral overgrowth (ELOG) [9–12], pendeo-epitaxy (PENDO) [13,14], facetcontrolled epitaxial lateral overgrowth (FACELO) [15,16], and SiN treatment [17–22]. In contrast tothe overgrowth techniques which have shown a high degree of complication, the SiN treatment pro-cedure is simpler and has been proven to significantly reduce threading dislocation density in GaN.The threading dislocation reduction mechanism is based on the change in growth mode to 3D islandformation on the SiN treated GaN surface and the half-loop formation between the bent-over thread-ing dislocations that occurs during the lateral overgrowth [21]. Different growth parameters wereshown to influence the final quality of the GaN films, such as growth temperature, V/III ratio, carriergas and film thickness [17–27]. In this context, many authors have studied the thickness effect on thestructural, electrical and luminescence properties of GaN layers and they have shown a large influenceon such physical properties [22–25]. However, few reports have been devoted to investigate the evo-lution of free exciton transitions during the smoothing process and their progress with dislocationdensity.

In this paper, the structural and optical properties of GaN layers grown by MOVPE on SiN treated(0001) sapphire substrate at different stages of the growth process are investigated using HRXRD andlow temperature PR spectroscopy. A strong correlation between structural defects and excitonic prop-erties during the coalescence process is observed.

2. Experiment

GaN epilayers were grown on (0001) sapphire substrate in atmospheric MOVPE home-made ver-tical reactor. Trimethylgallium (TMG) and NH3 were used as the precursors of gallium and nitrogen.The carrier gas was a mixture of N2 and H2. After a cleaning procedure of the sapphire substrate,the growth started by a nitridation step under NH3 þN2 þH2 atmosphere for 10 min at 1080 �C.Then, the SiN treatment was carried out by in-situ deposition of a thin SiN mask on the sapphire.SiN coating is obtained by introducing silane (SiH4) in the vapor phase at the end of the nitridationstep of the sapphire substrate. Afterward, the temperature was decreased to 600 �C in order to deposit30 nm GaN buffer layer. GaN epilayer was finally grown after a temperature ramp from 600 to1120 �C. Details of the process and optimum growth conditions can be found in references [18,20].The samples used in the present work were grown under nominally identical conditions except forthe thickness which was chosen in such a way that it covers all the different stages of film coalescence,i.e., samples S1 (0.1 lm), S2 (0.3 lm), S3 (0.7 lm), S4 (0.8 lm), S5 (1.7 lm), S6 (2 lm), S7 (3 lm) andS8 (5 lm). A Brucker X-ray diffractometer with Cu Ka radiation was used to determine the crystalquality of GaN films. PR measurements were carried out by employing a standard setup with the325 nm line of He–Cd laser as the pump light which was mechanically chopped at 280 Hz. The probelight was obtained from a 75 W Xe lamp dispersed with a 275 mm focal length monochromator. Thereflected light was dispersed by a 500 mm focal length grating monochromator and was detectedusing Hamamatsu R-928 photomultiplier.

M. Bouzidi et al. / Superlattices and Microstructures 84 (2015) 13–23 15

3. Results and discussion

Fig. 1 shows a typical real time record of reflectivity signal during the complete SiN treatmentgrowth process. The different steps are described in reference [27], where correlation between thereflectivity signal evolution and atomic force microscopy (AFM) images is observed. In fact, the layermorphology evolves from surface covered with discontinued islands to smooth continued surface, i.e.a transition from 3D to 2D growth mode occurs. As a result, this process decreases the surface rough-ness and then the root mean square (RMS) is reduced from 121 nm to 0.5 nm. At the interruptedgrowth stages labeled by S1 to S8 (Fig. 1), the effect of the transition growth mode from 3D to 2Don the dislocation densities and the electronic band structure modification of GaN films is analyzed.

HRXRD measurements were performed to study the thickness effect on the structural properties ofGaN thin films. Symmetric ð0002Þ and asymmetric ð10 �13Þ rocking curve of the samples S1, S4, S6 andS8 are shown in Fig. 2. It is seen that the full width at half maximum (FWHM) of both ð0002Þ andð10 �13Þ reflections decreases with increasing layer thickness and reaches lower values of 269 arcsecfor ð0002Þ and 369 arcsec for ð10 �13Þ when the 2D growth mode is fully established. The symmetricrocking curves are narrow in comparison to the asymmetric ones, since they are not influenced by thepresence of edge dislocations [26]. In order to estimate the dislocation densities of GaN films, we haveused the following equations [22]:

Fig. 1.

Dedge ¼FWHM2

ð10�13Þ

9b2edge

ð1aÞ

Dscrew ¼FWHM2

ð0002Þ

9b2screw

ð1bÞ

Ddis ¼ Dscrew þ Dedge ð1cÞ

where Dedge and Dscrew represent the edge and screw dislocation densities, respectively. b is the Burgersvector length (bedge ¼ 0:3189 nm and bscrew ¼ 0:5185 nm). The variation of the dislocation density withthe film thickness is shown in Fig. 3. One can see a monotonic decrease of the dislocation density dur-ing the transition from 3D to 2D growth mod which can be correlated to improvement in the crystalquality. For a thickness above 2 lm (2D growth phase), the dislocation density remains almost

0 600 1200 1800 2400 3000 3600 4200 4800

0

50

100

150

200

250

300

350

400

Ref

lect

ivity

(a.u

.)

Time (s)

S1

S2

S4

S5 S6 S7

S8

S3

Reflectivity versus growth time of GaN. The points labeled S1–S8 indicate the stage at which the growth was interrupted.

0-2000 -1000 1000 2000

S1 S4 S6 S8

Nor

mal

ised

Inte

nsity

Δω (arcsec)

(0002)(a)

0-3000 -2000 -1000 1000 2000 3000

S1 S4 S6 S8

Nor

mal

ised

inte

nsity

Δωω (arcsec)

(1013) (b)

Fig. 2. Rocking curve of symmetric (0002) (a) and asymmetric (10 �13) (b) reflections of samples S1, S4, S6 and S8.

16 M. Bouzidi et al. / Superlattices and Microstructures 84 (2015) 13–23

constant at about 5� 108 cm�2. In fact, nucleation of GaN during the initial stage of growth is consid-ered as the primary source of threading dislocation generation and the formation of small angle grainboundaries. The quality of GaN films improves along with thickness evolution as the dislocation den-sity originating from the GaN/sapphire interface is reduced. This improvement can also be attributedto the high interaction between screw and edge dislocations that would lead to their eventual anni-hilation [22]. For comparison purpose, the result obtained from a GaN sample grown in the same con-ditions but without SiN treatment is included in Fig. 3. One can conclude that the dislocation density isremarkably lower in GaN layers grown after SiN treatment compared to the density in layers obtainedusing the standard process. Indeed, the SiN treatment changes the surface morphology of GaN bufferlayer by forming selective GaN islands which promote a kind of ELOG as stated in a previous paper[27]. Engl et al. noted that during coalescence of the GaN islands, the dislocation cores are laterallyovergrown and bent into the (0001) basal plane [28].

Fig. 4 depicts the PR spectra of GaN samples measured at 11 K. Thin GaN films with thickness below0.7 lm show no PR signal with positive and negative parts. This behavior is probably due to the poorsurface quality of these films during the first stages of growth [27]. For thicker layers however, PRspectra show several features that become more resolved with thickness evolution. As reported in lit-erature [29–33], these features are due to the free A, B and C excitons resonance (XA, XB and XC) relatedto the CV

9 ! CC7, CV

7 ! CC7 (upped band) and CV

7 ! CC7 (lower band) interband transitions, respectively.

0 1 2 3 4 5

0

20

40

60

80

100 with SiN treatment without SiN treatment

Dis

loca

tion

dens

ity (1

08 cm

-2)

Thickness (µm)

S1

S2

S4S6 S7 S8

Fig. 3. Dislocation density dependence on epilayer thickness. Dislocation density of reference sample grown without SiNtreatment is included for comparison.

M. Bouzidi et al. / Superlattices and Microstructures 84 (2015) 13–23 17

These transitions only appear in wurtzite GaN of high quality. In order to extract the energy and thebroadening parameter of observed excitonic transitions, experimental data were least-square fitted tothe first derivative Lorentzian line shape functional form [31–34] given by:

DRR¼ Re

Xj

k¼1

½AkeiukðE� Ek þ iCkÞ��m ð2Þ

where, j is the number of the transitions and spectral functions used in the fitting procedure, E is thephoton energy, Ak, uk, Ek and Ck are the amplitude, phase, energy and the broadening parameter of thetransition, respectively. The exponent m is a characteristic parameter, which equals 2 for excitonictransition. A best fit of the experimental data to Eq. (2) was obtained using 4 spectral functions.The obtained resonance energies and broadening parameters are presented in Table 1. For all samples,three transition energies are unambiguously indentified to the ground state of the free excitons XA, XB

and XC. The fourth transition observed between XB and XC at about 17 meV above the XA transition, isassociated to the first excited state of the A-exciton (An=2) [26,27]. Assuming that the hydrogenicmodel based on the effective mass approximation is applicable, such identification permits a directestimation of the XA binding energy from the separation between the n = 1 and n = 2 states. Indeed,according to references [31,32], the XA binding energy (Eb) is given by:

Eb ¼43ðE2 � E1Þ ð3Þ

where E1 and E2 are the ground and the first excited state energies of XA, respectively. From Eq. (3) abinding energy of about 23 meV is found. This value agrees well with the published XA binding ener-gies range [35].

The dependence of exciton linewidths on the epilayer thickness is presented in Fig. 5. We noticethat with increasing layer thickness from 0.7 to 2 lm, the free exciton linewidths decrease from 5to 2.5 meV for XA, from 5.9 to 2.9 meV for XB and from 10 to 4.2 meV for XC. Such result indicates aclear improvement of the optical properties through the smoothing process. For the epilayer thicknessabove 2 lm, the linewidth remains almost constant at about 2.5, 2.9 and 4.2 meV for XA, XB and XC,respectively which is the state of the art for GaN grown on sapphire. Many works have been reportedon the broadening mechanisms of the exciton linewidth. The excitons phonons interaction, potentialfluctuations induced by ionized impurities, surface scattering, native defects and various other imper-fections in the crystal have been suggested as responsible for the broadening of exciton transitions

3.44 3.48 3.52

Exp Fit

BC

A

Energy (eV)

S3

S4 Exp Fit

BC

A

S5 Exp Fit B

C

A

ΔR

/R (a

.u)

S7 Exp Fit

B

C

A

S8

C

B

Exp Fit

A

Fig. 4. Low temperature photoreflectance spectra of samples S3, S4, S5, S7 and S8. Solid lines are the least-square fits of theexperimental data (open circles) using Eq. (2).

18 M. Bouzidi et al. / Superlattices and Microstructures 84 (2015) 13–23

[33,34,36]. In our case and for thin layers (incomplete coalesced layers), the high density of defectsobserved by HRXRD can be considered as the principal source of the exciton linewidth broadening.As the layer thickness increases, the density of defects decreases resulting in a narrowing of the exci-ton linewidth. In Fig. 5, the values of the exciton linewidths RXA ¼ 5:5 meV, RXB ¼ 6:7 meV andRXC ¼ 10:5 meV measured from the reference sample grown without SiN treatment are also inserted.It is clearly shown from Fig. 5 that the SiN treatment reduces the exciton linewidths. Thus, thelinewidth-decrease of about 50% compared to the reference one proves that the SiN treatment signif-icantly improves the optical properties of GaN. It should be noted that the exciton linewidth and thedislocation density are observed to be strongly correlated. Schenk et al. [25] have shown a strong

Table 1Excitonic resonance energies EA , EB and EC (corresponding to the A, B and C excitons respectively) obtained from PR spectra at 11 K.EAðn ¼ 2Þ represents the first excited resonance energy of the A exciton.

Sample EA (eV) EB (eV) EC (eV) EA(n = 2) (eV) CA (meV) CB (meV) CC (meV)

S3 3.474 3.483 3.506 3.493 5 5.9 10S4 3.475 3.484 3.506 3.492 4.7 5.7 9S5 3.482 3.490 3.517 3.498 3.6 4.3 6S6 3.484 3.492 3.521 3.501 2.7 3.1 4.5S7 3.484 3.492 3.521 3.500 2.5 2.9 4.3S8 3.480 3.487 3.514 3.497 2.5 2.9 4.2

0.0 0.8 1.6 2.4 3.2 4.0 4.8 5.62

3

4

5

6

7

8

9

10

11

12 A-exciton B-exciton C-excitonRXA

RXB

RXC

Exci

ton

linew

idth

(meV

)

Thickness (µm)

Fig. 5. Exciton linewidths dependence on epilayers thickness. Closed square, circle and triangle are the A, B and C-excitonlinewidths. RXA, RXB and RXC are the A, B and C-exciton linewidth of the reference sample respectively (open symbols).

-0.002 -0.001 0.000 0.001 0.002 0.0033.40

3.45

3.50

3.55present work

EA

EB

EC

Ener

gy (e

V)

εxx

After Ref.[30]EAEBEC

After Ref.[44]EA

EB

EC

After Ref.[31]EA

EB

EC

Theo.lineEC

EB

EA

Fig. 6. Experimental results (symbols) and theoretical calculation (solid lines) of the strain dependence of the three excitonictransition energies EA , EB and EC in c-plane wurtzite GaN.

M. Bouzidi et al. / Superlattices and Microstructures 84 (2015) 13–23 19

correlation between cathodo-luminescence and photoluminescence intensity, and exciton linewidthfrom one hand, and dislocation density from other hand. They reported that the intensity ofcathodo-luminescence and photoluminescence spectra increased and the exciton lines narrowed withdecreasing dislocation density.

20 M. Bouzidi et al. / Superlattices and Microstructures 84 (2015) 13–23

The energy positions of the A, B and C exciton transitions vary slightly from sample to sample asindicated in Table 1. This behavior is attributed to the variation of residual strain in the different lay-ers. Fig. 6 shows the measured free-exciton transition energies for the GaN samples as a function ofin-plane strain determined by HRXRD measurements. It is shown that the resonance energy values,EA, EB and EC increase along with the strain increasing. Based on theoretical approaches described inreferences [29,30], for wurtzite GaN structure, the evolution of the three excitons transition energyexpressed with the isotropic in-plane strain exx is given as follows:

EA ¼ Eg0 � Eb þ 2 2C13

C33az � at

� �þ C13

C33D1 � D2

� �þ C13

C33D3 � D4

� �� �exx ð4aÞ

EB ¼ Eg0 � Eb þD1 þ 3D2

2þ 2

C13

C33az � at

� �þ 2

C13

C33D1 � D2

� �þ C13

C33D3 � D4

� �� �exx

� 12

D1 � D2 � 2C13

C33D3 � 2D4

� �exx

� �2

þ 8D23

" #12

ð4bÞ

EC ¼ Eg0 � Eb þD1 þ 3D2

2þ 2

C13

C33az � at

� �þ 2

C13

C33D1 � D2

� �þ C13

C33D3 � D4

� �� �exx

� 12

D1 � D2 � 2C13

C33D3 � 2D4

� �exx

� �2

þ 8D23

" #12

ð4cÞ

where Eg0 is the energy gap between the conduction and the A-valence bands in the unstrained crystal.azand at are the conduction band deformation potential constants parallel and perpendicular to the zdirection while D1;D2;D3 and D4 are the valence band deformation potential constants. The quantityD1 ¼ Dcr is called the crystal–field splitting and the quantity 3D2 ¼ Dso is called the spin–orbit splittingand equal to 3D3 under the quasicubic approximation [29,30]. C13 and C33 are the elastic stiffness coef-ficients, they connect the out-of plane strain ezz with the in plane strain exx as follows [29–31]:

ezz ¼ �2C13

C33exx ð4dÞ

The literature provides a great number of papers on the experimental and theoretical data ofD1;D2;D3;D4;Dso and Dcr [29–31,35,37–42]. However, such rich database suffers from scattering.This dispersion can be explained by numerous experimental factors and identification methodologiesused, influencing the assessment accuracy of GaN physical parameters [35]. In order to provideenhanced GaN material parameters, this study proposes a calculation method that takes into accountsolely the measured experimental data in this work. Thus, by avoiding the random use of some scat-tered values published in the literature (elastic stiffness constants, exciton binding energy) [35], forthe determination of remaining parameters, the valence band deformation potential constants, thecrystal field and the spin–orbit splitting parameters are accurately derived using a stepwise method.Such procedure can be reached by the introduction of different fitting equations which can reduce thenumber of fitting parameters at each step. Therefore, we use the value of the exciton binding energyðEb ¼ 23 meVÞ determined in the previous section. After comparing our X-ray diffraction measure-ment results with the published values of C13 and C33, we found that, within experimental uncertainty,our results are consistent with the values C13 ¼ 103 GPa and C33 ¼ 405 GPa given by Wright [43]. Inaddition, it is reasonable to set az ¼ at ¼ a ¼ 0 since we are only concerned with the energy differenceof the conduction band and the valence band rather than with their absolute positions. Such procedureis synonym to fixing the energy position of the CB minimum. In the first step, we extract the deforma-tion potentials constants D1;D2;D3 and D4 by the least square fitting method and, in the second step,D1 and D2 were adjusted in order to fit the evolution of the excitonic energies as a function of exx.

Eqs. (4a), (4b) and (4c) give:

EC þ EB � EA ¼ Eg0 � Eb þ D1 þ 3D2 þ 2C13

C33D1 � D2

� �exx ð5aÞ

Table 2Numeriliteratu

Ref.

PresChuaShikSuzuFu [2Vurg

M. Bouzidi et al. / Superlattices and Microstructures 84 (2015) 13–23 21

EC þ EB � 2EA ¼ D1 þ 3D2 þ 2 D4 �C13

C33D3

� �exx ð5bÞ

Eqs. (5a) and (5b) has been fitted to the experimental data using the least-square method, and weget:

D4 �C13

C33D3 ¼ �6:14 eV ð6aÞ

D2 �C13

C33D1 ¼ 11:05 eV ð6bÞ

D1 þ 3D2 ¼ 0:032 eV ð6cÞ

Eg0 � Eb þ D1 þ 3D2 ¼ 3:505 eV ð6dÞ

The value of unstrained band gap energy Eg0 ¼ 3:496 eV is deduced from Eqs. (6c) and (6d). In thequasicubic approximation, the valence band deformation potential constants are related by [29,39]:

D1 � D2 ¼ �D3 ¼ 2D4 ð7Þ

Then, D3 ¼ 8:2 eV and D4 ¼ �4:1 eV can be immediately determined from Eqs. (6a) and (7). Thesevalues agree with those derived by Fu et al. [29] and Ghosh et al. [39]. Next, D1 and D2 are deducedfrom Eqs. (6b) and (7) to be 3.8 eV and 12 eV, respectively. Then, the resulting interband hydrostaticdeformation potentials are a1 ¼ a� D1 ¼ �3:8 eV and a2 ¼ a� D2 ¼ �12 eV, which are very close tothose reported in references [29,39]. After substituting the deformation potential values obtainedherein into Eqs. (4a), (4b) and (4c), the crystal field and spin–orbit splitting parameters are adjustedin order to fit the evolution of the excitonic energies as a function of biaxial strain. The best fit tothe experimental data is obtained for Dcr ¼ 16:5 meV and Dso ¼ 15:6 meV as shown in Fig. 6. It isnoticed that the crystal field and spin–orbit splitting values are determined within a sensitivity of0.5 meV. Table 2 summarizes the values of Dcr and Dso obtained from the current work and those ref-erenced in the literature for comparison. It is found that the obtained Dcrvalue is consistent with thatderived by Chuang and Chang [37], but higher than the value 9.2 meV reported by Fu et al. [29]. Thecrystal field value in reference [38] seems to be too large. The present value of Dso agrees well with theones reported by Suzuki et al. [38] and Shikanai et al. [30]. Such Dso value for wurtzite GaN is close tothat obtained for zinc-blende GaN [35] as in the case for many other semiconductors [30]. It should benoted that the values of D1 to D4, Eg0, Dso and Dcr obtained in this work accurately reproduce not onlyour experimental results but also the most of data reported by Shan et al. [31] and Chichibu et al. [44]as observed in Fig. 6. The weak discrepancy in the C-exciton energy values, compared to the A andB-excitons, may be attributed to the uncertainty in assignment of the resonance energies caused bythe weak signature and unresolved character of the C-exciton, which make it difficult to preciselyextract its related transition energy as in the case of A and B-excitons.

cal values of the spin– orbit splitting Dso and the crystal–field splitting Dcr obtained from this work and those reported inre for GaN.

Structure Dso Dcr

ent work Wurtzite 15.6 16.5ng [37] Wurtzite 12 16

anai [30] Wurtzite 15 22ki [38] Wurtzite 15.6 72.99] Wurtzite 18.9 9.2aftman [35] Zinc-blende 17 –

22 M. Bouzidi et al. / Superlattices and Microstructures 84 (2015) 13–23

4. Conclusion

We have investigated the thickness effects on structural and optical properties of GaN grown onSiN treated sapphire substrate by MOVPE. The process was interrupted in all stages of the growthin order to analyze the structural and optical properties evolution. The results show that the disloca-tion density and the exciton linewidths have the same behavior toward the variation of epilayer thick-ness. Compared to the standard process, the specific effect of the SiN treatment in improving thestructural and optical properties is confirmed. Based on theoretical approaches and experimentalresults, we have investigated the electronic band structure modification of GaN films due to isotropicin-plane strain. The valence band deformation potentials, interband hydrostatic deformation poten-tials, spin–orbit and crystal field parameters are accurately re-extracted and the obtained values arecompared to the previously reported ones.

Acknowledgement

The authors acknowledge financial support from DGRST.

References

[1] S. Nakamura, Solid State Commun. 102 (1997) 237.[2] S.T. Kim, Y.J. Lee, D.C. Moon, C.H. Hong, T.K. Yoo, J. Cryst. Growth 194 (1998) 37.[3] H. Morkoç, Nitride Semiconductors and Devices, Springer, Berlin, 1999.[4] Ahmad Hadi Ali, Ahmad Shuhaimi Abu Bakar, Takashi Egawa, Zainuriah Hassan, Superlattices Microstruct. 60 (2013) 201.[5] A.Y. Polyakov, N.B. Smirnov, E.B. Yakimov, A.S. Usikov, H. Helava, K.D. Shcherbachev, A.V. Govorkov, Yu.N. Makarov, In-

Hwan Lee, J. Alloys Compd. 617 (2014) 200.[6] S. Zhou, Z. Lin, H. Wang, T. Qiao, L. Zhong, Y. Lin, W. Wang, W. Yang, G. Li, J. Alloys Compd. 610 (2014) 498.[7] M.J. Kappers, M.A. Moram, D.V. Sridhara Rao, C. McAleese, C.J. Humphreys, J. Cryst. Growth 312 (2010) 363.[8] F. Yun, Y.-T. Moon, Y. Fu, K. Zhu, Ü. Ozgür, H. Morkoç, C.K. Inoki, T.S. Kuan, Ashutosh Sagar, R.M. Feenstra, J. Appl. Phys. 98

(2005) 123502.[9] H. Kim, H. Rho, L.W. Jang, In-Hwan Lee, Curr. Appl. Phys. 13 (2013) 267.

[10] K.R. Song, J.H. Lee, S.H. Han, H.R. Yi, S.N. Lee, Mater. Res. Bull. 48 (2013) 5088.[11] W.K. Fong, K.K. Leung, C. Surya, J. Cryst. Growth 318 (2011) 488.[12] S.I. Kim, B. Kim, S. Jang, A.y. Kim, J. Park, D. Byun, J. Cryst. Growth 326 (2011) 200.[13] K. Pakula, R. Bozek, J.M. Baranowski, J. Jasinski, Z. Lilientalweber, J. Cryst. Growth 267 (2004) 1.[14] Z. Liliental-Weber, J. Jasinski, D. Cherns, M. Baines, R. Davis, Mat. Res. Soc. Proc. 693 (2002) 309.[15] K. Hiramatsu, K. Nishiyama, M. Onishi, H. Mizutani, M. Narukawa, A. Motogaito, H. Miyake, Y. Iyechika, T. Maeda, J. Cryst.

Growth 221 (2000) 316.[16] S. Nitta, J. Yamamoto, Y. Koyama, Y. Ban, K. Wakao, K. Takahashi, J. Cryst. Growth 272 (2004) 438.[17] I. Halidou, Z. Benzarti, Z. Bougrioua, T. Boufaden, B. El Jani, Superlattices Microstruct. 40 (2006) 490.[18] Z. Benzarti, I. Halidou, T. Boufaden, B. El Jani, S. Juillaguet, M. Ramonda, Phys. Stat. Sol. (a) 201 (2004) 502.[19] S. Haffouz, P.R. Hageman, V. Kirilyuk, L. Macht, J.L. Weyher, P.K. Larsen, Mater. Sci. Eng., B 97 (2003) 9.[20] I. Halidou, Z. Benzarti, T. Boufaden, B. El Jani, S. Juillaguet, M. Ramonda, Mater. Sci. Eng., B 110 (2004) 251.[21] S. Sakai, T. Wang, Y. Morishima, Y. Naoi, J. Cryst. Growth 221 (2000) 334.[22] A. Bchetnia, A. Toure, T.A. Lafford, Z. Benzarti, I. Halidou, M.M. Habchi, B. El Jani, J. Cryst. Growth 308 (2007) 283.[23] A. Toure, A. Bchetnia, T.A. Lafford, Z. Benzarti, I. Halidou, Z. Bougrioua, B. El Jani, Phys. Stat. Sol. (a) 205 (2008) 2042.[24] I. Halidou, A. Touré, L. Nguyen, A. Bchetnia, B. El Jani, Opt. Mater. 35 (2013) 988.[25] H.P.D. Schenka, P. Vennéguès, O. Tottereau, T. Riemann, J. Christen, J. Cryst. Growth 258 (2003) 232.[26] E. Arslan, Ö. Duygulu, A. Arslan Kaya, A. Teke, S. Özçelik, E. Ozbayf, Superlattices Microstruct. 46 (2009) 846.[27] I. Halidou, Z. Benzarti, H. Fitouri, W. Fathallah, B. El Jani, Phys. Stat. Sol. (c) 4 (2007) 129.[28] K. Engl, M. Beer, N. Gmeinwieser, U.T. Schwarz, J. Zweck, W. Wegscheider, S. Miller, A. Miler, H.-J. Lugauer, G. Brüderl, A.

Lell, V. Harle, J. Cryst. Growth 289 (2006) 6.[29] D. Fu, R. Zhang, B. Wang, Z. Zhang, B. Liu, Z. Xie, X. Xiu, H. Lu, Y. Zheng, G. Edwards, J. Appl. Phys. 106 (2009) 023714.[30] A. Shikanai, T. Azuhata, T. Sota, S. Chichibu, A. Kuramata, K. Horino, S. Nakamura, J. Appl. Phys. 81 (1997) 417.[31] W. Shan, A.J. Fischer, S.J. Hwang, B.D. Little, R.J. Hauenstein, X.C. Xie, J.J. Song, D.S. Kim, B. Goldenberg, R. Horning, S.

Krishnankutty, W.G. Perry, M.D. Bremser, R.F. Davis, J. Appl. Phys. 83 (1998) 455.[32] D.O. Dumcenco, S. Levcenco, Y.S. Huang, C.L. Reynolds Jr., J.G. Reynolds, K.K. Tiong, T. Paskova, K.R. Evans, J. Appl. Phys. 109

(2011) 123508.[33] R. Kudrawiec, M. Rudzinski, J. Serafinczuk, M. Zajac, J. Misiewicz, J. Appl. Phys. 105 (2009) 093541.[34] Y.S. Huang, Fred H. Pollak, S.S. Park, K.Y. Lee, H. Morkoç, J. Appl. Phys. 94 (2003) 899.[35] I. Vurgaftman, J.R. Meyer, J. Appl. Phys. 94 (2003) 3675.[36] D.Y. Song, M. Basavaraj, S.A. Nikishin, M. Holtz, V. Soukhoveev, A. Usikov, V. Dmitriev, J. Appl. Phys. 100 (2006) 113504.[37] S.L. Chuang, C.S. Chang, Phys. Rev. B 54 (1996) 2491.[38] M. Suzuki, T. Uenoyama, A. Yanase, Phys. Rev. B 52 (1995) 8132.[39] S. Ghosh, P. Waltereit, O. Brandt, H.T. Grahn, K.H. Ploog, Phys. Rev. B 65 (2002) 075202.

M. Bouzidi et al. / Superlattices and Microstructures 84 (2015) 13–23 23

[40] B. Gil, O. Briot, R.L. Aulombard, Phys. Rev. B 52 (1995) R1728.[41] B. Gil, P. Lefebvre, H. Morkoç, Strain effects in GaN epilayers, C. R. Acad. Sci. Paris, t. 1, Série IV, 2000, p. 51–60.[42] N.V. Edwards, S.D. Yoo, M.D. Bremser, T.W. Weeks Jr., O.H. Nam, R.F. Davis, H. Liu, R.A. Stall, M.N. Horton, N.R. Perkins, T.F.

Kuech, D.E. Aspnes, Appl. Phys. Lett. 70 (1997) 2001.[43] A.F. Wright, J. Appl. Phys. 82 (1997) 2833.[44] S. Chichibu, T. Azuhata, T. Sota, H. Amano, I. Akasaki, J. Appl. Phys. 70 (1997) 2085.