Available online at www.sciencedirect.com

Physica E 17 (2003) 232–234

www.elsevier.com/locate/physe

Indium segregation and reevaporation e"ects on thephotoluminescence properties of highly strained

InxGa1−xAs=GaAs quantum wellsB. Ilahi, L. Sfaxi, L. Bouza-.ene, F. Hassen, H. Maaref∗

Laboratoire de Physique des Semiconducteurs et des Composants Electroniques, Facult�e des Sciences de Monastir,Departement de Physique, Avenue de l’Environnement, 5000 Monastir, Tunisia

Abstract

Temperature dependence of the e"ective band gap (BG) energy of strained InxGa1−xAs=GaAs single-quantum well andmulti-quantum well structures grown by solid source MBE at varied substrate temperature is investigated by photoluminescencespectroscopy between 10 K and room temperature. For low-temperature-grown heterostructure, the temperature-induced BGshrinkage exhibits a good correlation with that of unstrained material. However, no consensus is shown to occur for arelatively high-temperature-grown quantum wells (QWs). This discrepancy is interpreted in terms of indium segregation andreevaporation during epitaxy. The low-temperature range, where the well-known Varshni law fails to :t PL peak positions,is found to decrease with increasing QW width and is attributed to the interface-roughness-induced exciton localization.This study was propped by numerical solving of Schr-odinger equation taking into account strain, indium segregation anddesorption e"ects.? 2002 Elsevier Science B.V. All rights reserved.

PACS: 78.60; 78.66; 64.75

Keywords: InxGa1−xAs; QW; Reevaporation; Band gap variation

1. Introduction

InxGa1−xAs-based quantum well (QW) structureshave attracted a great deal of attention owing to theirpotential application in the fabrication of high perfor-mance devices.In this paper, we report PL study, between 10 K

and RT, of InxGa1−xAs=GaAs QW structures grownat di"erent temperatures. The e"ective band gap(BG) variation versus temperature of relatively

∗ Corresponding author. Fax: +216-3-462-873.E-mail address: hassen.maaref@fsm.rnu.tn (H. Maaref).

low-temperature-grown QWs is found to be in goodcorrelation with that of bulk material with the samecomposition. Meanwhile, discrepancy is shown to oc-cur for higher-temperature-grown structures.

2. Experimental procedure

Samples involved in this study are In0:25Ga0:75As=GaAs strained layers grown by SS-MBE on semi-insulating GaAs (0 0 1) substrates and constituted bytwo sets of multi-quantum wells (MQWs) of four dif-ferent nominal well widths (10, 15, 20 and 25 ML)grown at 480◦C (MQW1) and at 520◦C (MQW2)

1386-9477/03/$ - see front matter ? 2002 Elsevier Science B.V. All rights reserved.doi:10.1016/S1386-9477(02)00771-3

B. Ilahi et al. / Physica E 17 (2003) 232–234 233

0 50 100 150 200 250 3001.28

1.30

1.32

1.34

1.36

1.38MQW1

B-E

Varshni

Lw = 25ML

Lw = 15ML

Lw = 20ML

Lw = 10ML

Ene

rgy

(eV

)

Temperature (K)

Fig. 1. Variation of the PL peak positions versus temperature ofMQW1. Arrows indicate the delocalization temperature.

and two sets of single-quantum well (SQWs) with thesame well width (25 ML), grown at 480◦C (SQW1)and at 520◦C (SQW2).

3. Results and discussion

We have noted an appreciable shift of the 10 KPL peak positions between the same structures grownat di"erent temperatures. Numerical solution ofShr-odinger equation has been carried out taking intoaccount strain and indium surface segregation [1] ef-fects. The results show that the PL peak energies ofQWs grown at 480◦C correlate with calculated val-ues. However, for those grown at 520◦C, predictedvalues of fundamental transition energies are lowerthan those provided by PL experiments. Agreementhas been reached after the modi:cation of the segre-gation model as follows:

xn = rx0(1− Rn) (16 n6N ′w; well);

xn = rx0(1− RN ′w)Rn−N

′w (n¿N ′

w; barrier); (1)

where “r” is the sticking coeLcient, recently reportedto be 0.6 at 520◦C [2].Fig. 1 shows PL peak positions versus temper-

ature of MQW1 as well as the least-squares :tof the experimental data to both Varshni [3] and

Bose–Einstein equations [4] (B–E). At low tempera-tures (T ¡ 60 K), the two theoretical models fail to:t the experimental data. At 10 K, the shift betweentheoretical (Varshni) and experimental values liesbetween 5 meV for the narrower QW (10 ML) and2 meV for the largest one (25 ML) for all samples.This may be due to the exciton localization at theinterface roughness. The typical exciton binding en-ergies on these defects lies in the 2–5 meV range [5]which agrees with the observed red shift of the PLpeak position when compared to the expected valuesfor free excitons [6]. The range of temperature whereno consensus is found to spread out is high as theQW is narrow. For high-temperature range, the BGvariation is well :tted by both equations. The aver-aged bulk material BG variation with temperature isgiven by [7]

MEg(x; T ) =−(1− x) �GaAsT 2

T + �GaAs− x �InAST

2

T + �InAs: (2)

In Fig. 2, we compare the averaged BG shrinkage ver-sus temperature predicted by Eq. (2) for unstrainedIn0:25Ga0:75As (with and without taking reevaporationa"ects into account) with that measured in this workfor di"erent Tg, notably at 480◦C (�=5:8×10−4 eV=Kand �= 270 K) and at 520◦C (�= 4:9× 10−4 eV=Kand �=150 K). For Tg=480◦C the QWs e"ective BGvariation with temperature exhibits good correlationwith that of bulk material with a similar composition.But a discrepancy is shown to occur for 520◦C-grownQWs. We expect this discrepancy to be caused by thee"ects of increasing Tg on incorporated In composi-tion. Using the new indium fraction (x=rx0), we havereached correlation (Fig. 2). Therefore, the discrep-ancy between the e"ective BG variations with tem-perature of the same QWs structure grown at di"erentTg is probably due to the simultaneous change in theQWs shape and In composition.

4. Conclusion

Temperature-induced BG variation of relativelylow-temperature-grown QWs exhibits good correla-tion with that of unstrained material with similar nomi-nal indium composition. However, deviation from thatof high-temperature-grown QWs has been attributedto either In segregation and reevaporation. Thelow-temperature range where the Varshni law fails to

234 B. Ilahi et al. / Physica E 17 (2003) 232–234

0 50 100 150 200 250 300

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

Temperature K

∆Eg e

V

strained In0.25

Ga0.75

As (Tg=480°C)

Bulk In0.25

Ga0.75

As (equation 2)

Bulk In0.15

Ga0.85

As (equation 2)

strained In0.25

Ga0.75

As (Tg=520°C)

Fig. 2. BG shrinkage with temperature of: strained In0:25Ga0:75As=GaAs (grown at 480◦C and at 520◦C) and bulk InxGa1−xAs as predictedby Eq. (2) (for x = 0:25 and 0.15).

:t the experimental data has been found to decreasewith increasing QWs width, showing the dominationof the QW PL by the interface-roughness-inducedexciton localization.

References

[1] K. Muraki, S. Fukatsu, Y. Shiraki, R. Ito, Appl. Phys. Lett.61 (5) (1992) 557.

[2] S. Fujimoto, M. Aoki, Y. Horikoshi, Jpn. J. Appl. Phys. 38(1999) 1872.

[3] K.P. Varshni, Physica 34 (1967) 149.[4] P. Lantenshalager, M. Garriga, S. Logothetidis, M. Cardona,

Phys. Rev. B 35 (1987) 9174.[5] H.I. Jeon, M.S. Jeon, H.W. Shim, Y.G. Shin, K.Y. Lim,

E.K. Suh, H.J. Lee, J. Cryst. Growth 171 (1997) 349, andreferences therein.

[6] J.R. Botha, A.W.R. Leitch, J. Electron Mater. 19 (12) (2000)1362, and references therein.

[7] S. Paul, J.B. Roy, P.K. Basu, J. Appl. Phys. 69 (2) (1991) 827.