Optical investigation of GaAs1−xNx/GaAs heterostructure properties

phys. stat. sol. (b) 241, No. 2, 428–433 (2004) / DOI 10.1002/pssb.200301915

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

Optical investigation of GaAs1–xNx/GaAs heterostructure properties

F. Saidi1, F. Hassen1, H. Maaref*,1 L. Auvray2, H. Dumont2, and Y. Monteil2 1 Laboratoire de Physique des Semiconducteurs et des Composants Electroniques Faculté des Sciences,

5000 Monastir, Tunisie 2 Laboratoire Multimatériaux et Interfaces, Université Claude Bernard Lyon I, 43 Bd 11 Novembre 1918,

69622, France

Received 27 February 2003, revised 2 September 2003, accepted 21 September 2003 Published online 27 January 2004

PACS 78.55.Cr, 78.67.De

We have investigated, by photoluminescence (PL), the optical properties of GaAs1–xNx/GaAs epilayers and GaAs1–xNx/GaAs quantum well (QW) structures grown by metal organic vapor phase epitaxy (MOVPE) on (001)-oriented GaAs substrates. Different behaviors have been observed for the bulk epilayer and for the QW structures, respectively: (i) a blue shift of the PL bands in both kinds of structures when increas-ing the excitation density, (ii) an S-shaped PL peak energy versus temperature dependence has been ob-served for the GaAsN epilayer but a usual behavior is obtained for the QWs. Based in these experimental results, we have suggested that the carrier recombination mechanisms in the epilayer and in the quantum well structure are different. An enhanced exciton-localization-like mode for the epilayer is observed.


1 Introduction

Recently, dilute III–V–N alloys and superlattices have attracted great attention in the optoelectronic technology. The optical fiber communication requires an emission wavelength between 1.3 and 1.55 µm. The ternary GaAs1–xNx should offer the possibility of covering the whole spectral wavelength range from infrared to ultraviolet by one material system [1, 2]. Experimental and theoretical studies have shown that dilute GaAs1–xNx (low nitrogen concentration) exhibits many unusual properties when compared to normal semiconductor alloys [3–6]. For instance, this material exhibits a large bandgap reduction with increasing nitrogen content [1–6]. Nitrogen-induced localized states above the GaAs conduction band are involved for carrier recombination pathways [3]. An unusual dependence of the conduction band effective mass on the nitrogen concentration was observed as a result of the formation of a nitrogen-related impurity band [9, 15]. However, both experimental and theoretical investigations of the GaAs1–xNx alloy seem to indicate that the nitrogen-induced perturbation to the valence band is suffi-ciently small, in the dilute nitrogen region up to 4% [9]. In this composition region, a bowing parameter of the order of 10–20 eV has to be introduced to describe the bandgap reduction [1–6]. A recent study by Tisch et al. [7] reproduces more accurately the bandgap evolution with composition by fitting the bowing parameter with a double exponential. A few studies have been focused on the study of the photo-luminescence origin in GaAsN/GaAs epilayers and QWs [7–11]. In this paper, we have investigated the PL emission of GaAsN epilayers on GaAs substrate and of GaAsN/GaAs single quantum well structures with different well widths and for two nitrogen composi-

* Corresponding author: e-mail: [email protected], Tel: +216-3-500-274, Fax: +216-3-500-278

phys. stat. sol. (b) 241, No. 2 (2004) / www.pss-b.com 429


Table 1 Sample characteristics used in this work.

samples structures nitrogen concentration

GaAsN/GaAs width (nm)

Tg (°C)

annealing conditions

R3-443 epilayer 2% 140 550 750 °C/30 min R3-436 SQW 13.5% 10 550 – R3-437 SQW 2% 25 550 750 °C/30 min

tions. We have exploited the optical properties versus temperature and excitation density. The experi-mental results indicate that the carrier recombination mechanisms in the epilayer and in the quantum well structure are different. An enhanced exciton-localization-like mode for the epilayer is observed.

2 Experimental details

The GaAsN/GaAs heterostructures were grown by metal organic chemical vapor phase epitaxy (MOVPE) on vicinal GaAs (001) substrates 2° off (±0.05°) towards [100]. Prior to growth, a 0.1-µm thick GaAs buffer layer was grown at 660 °C. The growth of 0.1–0.3-µm thick GaAs1–xNx epilayers and of GaAsN/GaAs single quantum well (SQWs) structures with 25- and 10-nm well width, was carried out at 550 °C, with a mixture of N2 and H2 (ratio 4:1) as a carrier gas. In order to improve the luminescence intensity of the two kinds of structures (epilayer sample and SQWs) a GaAs cap layer was deposited and a thermal annealing at 750 °C for 30 min is used. The use of nitrogen over hydrogen was motivated by the increased nitrogen incorporation efficiency in GaAs under N2 carrier gas. The precursors were tri-

methylgallium, arsine and 1,1-dimethylhydrazine. The lattice-mismatch a

a

∆

and the N content in

GaAsN were determined by X-ray diffraction from separation angle between the GaAsN layer and sub-strate (400) diffraction peaks, assuming coherent tensile strain and a Poisson ratio of 0.313. In Table 1 we have summarized all the characteristics of our samples studied in this work Photoluminescence measurements were carried out between 10 and 300 K while keeping the samples in a closed-cycle helium circulation cryostat. The excitation wavelength used is the 514.5-nm line of the cw argon ion laser.

3 Results and discussion

In order to compare the optical properties of the GaAsN epilayer with those of GaAsN/GaAs single quantum well, the 10-K PL spectra of both the GaAsN/GaAs epilayer and QW structures are plotted

0.9 1.0 1.1 1.2 1.3 1.4

Nominal layer

Interfacial layer

PL

Inte

nsity

(ar

b.un

its)

Energy (eV)

(a)

x=3,5%

x=2% (b)

PL

Inte

nsity

(ar

b.un

its)

1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35

Energy (eV)

Fig. 1 Low-temperature photoluminescence (10 K) spectra of GaAs1–xNx/GaAs: a) Epilayers structures with x = 2%; b) SQW structures: (�) 2% nitrogen content and 25-nm well width. (ο) 3.5% nitrogen content and 10-nm well width.

430 F. Saidi et al: Optical investigation of GaAs1–xNx/GaAs heterostructure properties


together in Fig. 1. In the spectrum of Fig. 1a, which is associated to the epilayer, shows two lumines-cence bands. The main line, at 1.26 eV, is associated to the radiative emission of the bound excitons in the epilayer of the nominal nitrogen concentration. The additional one at the low energy side, located at 1.14 eV, is attributed to the radiative carrier recombination induced by the presence of a native interfa-cial layer formed between GaAs substrate and GaAsN epilayer [19]. It has been shown recently, by Dumont et al. [19], by using photoluminescence spectroscopy at 300 K, that this interfacial layer does not originate from the growth processes such as the commutation of the precursor flux when depositing the GaAsN layer on GaAs buffer layer. But, it is expected that it is related to an intrinsic phenomenon induced by the GaAsN growth conditions. Its origin is not yet clearly demonstrated [19]. However, in the QW structure, we observe a single luminescence band for each nitrogen concentration (x = 2% and 3.5%) Fig. 1b. This PL band emission shows a red shift with increasing the nitrogen content. This phe-nomena originates from the bandgap reduction of the GaAsN alloys with the increasing nitrogen compo-sition reported elsewhere [7, 8, 12, 20]. In the QW structures, the red shifts, caused by the bandgap reduction with increasing nitrogen content, dominates the blue shift induced by the confinement effect related to the quantum well width. So an increase in the nitrogen content of 1% reduces the bandgap to about 110 meV. However, the confine-ment energy for the small well width ∼15 nm is about 50 meV. The competition between the bandgap reduction and the confinement effect explain the observed PL band red shift when the nitrogen composi-tion increases (Fig. 1b). These results agree well with the calculation, described bellow, based on the band anticrossing model (BAC). Due to its small width compared to the GaAsN quantum well region, the interfacial layer PL band is not observed in the QW structures (Fig. 1). Thus, the electronic level of the GaAsN quantum well domi-nates the energy level in the conduction band. In fact, the quantum luminescence efficiency of the inter-face layer is lower than that of the nominal GaAsN QW.

0 10 20 30 401.10

1.12

1.14

1.16

1.24

1.26

1.28

1.30(a)x=2%

Pea

k E

nerg

y (e

V)

Excitation power density (W/cm–2)

1 10

(b)Nominal layerInterfacial layer

PL

Inte

nsity

(ar

b.un

its)


(c)x=2%x=3.5%

0.1 1 10


1.18

1.19

1.20

1.21

1.22

1.23

1.24

1.25

Pea

k E

nerg

y (e

V)

1 .17

Fig. 2 Peak energy and integrated PL intensity versus excitation power density for GaAsN epilayer a) and b). c) Peak energy versus excitation power density of the GaAsN/GaAs QW structures.



0 20 40 60 80 100 120 140

1.24

1.25

1.26

1.27 (a) GaAsN/GaAs epilayer

x=2%

Pea

k en

ergy

(eV

)

T (K)

Fig. 3 Temperature dependence of the peak energy for the both structures: a) Epilayer; b) SQW. In order to identify the PL transition mechanism associated to the emission lines, excitation power and temperature dependencies of the optical properties were studied. In Fig. 2a and 2b, we report the inte-grated PL peak intensity and the peak energy versus the excitation power of the GaAsN epilayer struc-tures. We can note a saturation behavior of the peak intensity (Fig. 2b), which is more remarkable for the interfacial peak. Since an increase in the excitation density will cause a gradual filling of the energy states localized at the band tails. Moreover the interface roughness causes the saturation of the PL inten-sity. The blue shift observed in Fig. 2a is about 10 meV for the interfacial PL peak and 20 meV for the intrinsic one, is due to the gradual saturation of lower energy states [9, 20]. Similarly to the epilayers, in the QW structures, the peak energy shows a blue shift of about 12 meV/decade by increasing excitation power (Fig. 2c). Moreover, we have not observed a saturation behavior in the integrated PL intensity with increasing excitation power. These properties indicate that the luminescence emission results from the recombination between spatially separated electrons and holes. So it has been shown by Sun et al. [13] that nitrogen incorporated in GaAs induces a sufficiently small perturbation to the valence band. In fact the electron is localised in the GaAsN conduction band but the hole is localised in the GaAs valence band. Furthermore, the annealed sample shows a strong improvement of the PL intensity and a remarkable reduction of the FWHM (from 80 meV for unan-nealed to 42 meV for the annealed sample). The same properties are shown by Buyanova et al. [11], where the optical properties are clearly enhanced after annealing samples. Figure 3a shows the temperature dependence of the PL band energy of the GaAsN/GaAs epilayers. It exhibits an anomalous behavior, the so-called S-shape. From 10 to 45 K the emission energy decreases as expected, but an increase is seen between 70 K and 90 K. This S-shape behavior is attributed to the recombination of photogenerated carriers trapped by localised states within the GaAsN as previously observed [12, 17, 20]. The S-shape behavior results from the carrier localisation on the modulated poten-tial structure caused by the nitrogen composition fluctuation [20]. To further identify the PL mechanism in these SQWs, one can now compare the temperature behavior of the GaAsN epilayer and the two GaAs1–xNx /GaAs SQW (Fig. 3). In contrast to the bulk epilayer, the PL energy peak decreases monotonically in the whole temperature range investigated as T increases, this typical behavior can be fitted by empirical relation ship proposed by Varshni:

Eg(T) = Eg(0) – αT2/(β + T), (1)

where α is an empirical parameter related to the joint density of states and β an effective temperature. By fitting the experimental results, using Eq. (1), we have deduced the α and β parameter values.

α = 5.2 × 10–4 eV/K and β = 300 K for x = 2% and

α = 3.8 × 10–4 eV/K and β = 125 K for x = 3.5% .

This behavior suggests that the PL emission observed in the SQW structure results from the bandgap transition.

(b) x=2%Varshni a=5.2x10–4 eV/Kb=320 K

0 50 100 150 200 250 300T (K)

1.25

1.20

1.15

1.10

Pea

k en

ergy

(eV

)

x=3.5%Varshni a=3.8x10–4 eV/Kb=125 K

432 F. Saidi et al: Optical investigation of GaAs1–xNx/GaAs heterostructure properties


The main difference between the GaAsN epilayer and SQW structures is observed in the temperature behavior. In fact, the S-shape observed for the GaAsN epilayer explains the carrier localization on the modulated potential structure due to the non-uniform insertion of nitrogen atoms [20]. However, in GaAsN/GaAs SQW the variation of the PL peak energy versus temperature indicates that we have a band-to-band transition. This difference is due to the fact that the confinement effects screen the poten-tial modulation. The localization phenomenon produced by the potential fluctuation in the GaAsN/GaAs epilayer can be screened by a GaAsN/GaAs SQW structure. A theoretical calculation, of the bandgap and the confinement energy level in the GaAs1–xNx/GaAs structures, based on the anticrossing model theory [18] and the envelope function model [15] was devel-oped to identify the PL band observed experimentally and to deduce the nitrogen concentration in the interfacial layer. The bandgap reduction observed is a result of the conduction band edge shift with increasing of the nitrogen composition. The repulsive interaction between the nitrogen level and the GaAs conduction band minimum splits the conduction band of (III-V)-N alloy into two subbands denoted, respectively, E+ and E–. The lower conduction subband E– describes the bandgap of GaAs1–xNx, which is given by:

2 21M N M N NM2( ) [( ( ) ) ( ( ) ) 4 ]E k E k E E k E C x

−

= + − + + ⋅ ⋅ , (2)

where CNM is a constant that describes the interaction strength between the N level and the conduction band of GaAs and it is equal to 2.7 eV [14]. EM(k) is the conduction band of GaAs. It can be represented by a standard parabolic dispersion function. EN is the nitrogen level introduced in the forbidden GaAs band. It is given, relative to the valence band edge of GaAs, by the relation EN(meV) = 1675–2520x [14]. The bandgap of GaAs1–xNx given by Eq. (2) for E–(0), agrees well with the experimental data [15]. In our simple theoretical calculation, we have considered that the perturbation induced by the nitrogen introduction is reflected through the conduction band effective mass. Which is given by [18]

GaA

M N

2 2M N NM

*2*

(0)1

( (0) ) 4

smm

E E

E E C x

=

−

−

− ⋅ ⋅

. (3)

This approximation indicates that the electron effective mass m* in GaAsN increases with increasing nitrogen composition in the low composition range compared to the GaAs effective mass. This behavior is unusual and in fact it is opposite to conventional semiconductor behavior, where the value of the effec-tive mass decreases with decreasing bandgap energy [18]. Our model is based on a simple approach but there is a more complicated model to treat the optical and electronics processes in the Ga(In)AsN such as the microscopic model developed by Hader et al. [21].

0 5 10 15 20 25 30

(a)GaAs

1-xN

N/GaAs SQW

x=2%Modelised Transition energyModelised gap energyOur experimental workBuyanova et al[9]

Well width (nm)

1.45

1.40

1.35

1.30

1.25

1.20

Ene

rgy

(eV

)

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.01.051.101.151.201.251.301.351.401.451.50

(b) GaAsN/GaAs SQWModelised Transition energyModelised gap energyOur experimental workSun et al [13]

Ene

rgy

(eV

)

Nitrogen contetx (%)

Fig. 4 Theoretical calculation and experimental values of the transition dependencies on the nitrogen composition and well width.



Using our model, we have seen that experimental PL band positions of our SQW are well predicted: see Fig. 4a and b. Based on this result, we have tried to estimate the nitrogen content in the interfacial layer formed in the GaAsN epilayers structures. Figure 4a, shows the transition dependency of the well width, calculated for 2% nitrogen composition. In Fig. 4b, we plot the variation of the bandgap energy of epilayers structures and the transition energy, in quantum well structures, versus the nitrogen composition up to 4%. Transition energies values are calculated for 10-nm well widths. This plot allows us to estimate the nitrogen content in the interfacial layer observed experimentally in the GaAsN/GaAs epilayer. The triangle shows the experimental PL band energy of the interfacial epilayer observed in sample R3-443. Assuming that the interfacial layer can be considered as a rectangular quantum well with a 10-nm width extracted from the TEM measure-ments, the nitrogen concentration is found to be about 3%. From this result, one can deduce that the N concentration is higher than 4% in order to get a 1-eV energy peak.

4 Conclusion

In conclusion, we have studied the optical properties of GaAs1–xNx /GaAs epilayers and single quantum well structures, grown by MOCVD. We have shown that the increase in the nitrogen content causes a reduction of the bandgap in GaAsN. The new PL band on the low-energy side in the epilayer sample is attributed to the luminescence within the GaAsN interfacial layer formed with a high nitrogen content. In the SQW structure, we have observed a single luminescence band for each nitrogen concentration used. From the temperature and the excitation density dependence on the PL line, we have shown that the confinements effects in GaAsN SQW structures screens the localization effects induced by the potential modulation, observed in GaAsN epilayer. The anticrossing model has been used to estimate the nitrogen concentration in the interfacial layer and it is found to be about 3%.

References

[1] L. Bellaiche, S. Wie, and A. Zunger, Phys. Rev. B 54, 17568 (1996). [2] M. Weyers, M. Sato, and H. Ando, Jpn. J. Appl. Phys. 31, L 853 (1992). [3] D. Perkins, A. Mascurenhas, Y. Zhang, J. F. Geisz, D. J. Friedman, J. M. Chen, and S. R. Kurtz, Phys. Rev.

Lett. 82, 312 (1999. [4] L. Bellaiche, S. H. Wei, and A. Zunger, Appl. Phys. Lett. 70, 3558 (1997). [5] G. Pozina, I. Ivanov, and B. Monemar, J. Appl. Phys. 84, 3830 (1998). [6] K. Onabe, D. Aoki, J. Wu, H. Yaguchi, and Y. Suiraki, phys. stat. sol. (a) 176, 231 (1999). [7] U. Tisch, E. Finkman, and J. Salzman, Appl. Phys. Lett. 81, 463 (2002). [8] S. Francoeur, G. Sivaraman, Y. Qlu, S. Nikishin, and H. Temkin. Appl. Phys. Lett. 72, 1857 (1998). [9] T. A. Bayanova, W. M. Chen, G. Posina, J. P. Bergman, and B. Monemar, Appl. Phys. Lett. 75, 501 (1999). [10] Y. Zhang, A. Mascurenhas, H. P. Xin, and C. W. Tu, Phys. Rev. B 61, 443(2000). [11] I. A. Buyanova, W. M. Chen, and B. Monemar, MRS J. Nitride Semicond. Rev. 62, (2001). [12] M. Takahashi, A. Moto, S. Tanaka, T. Tanabe, S. Takaggishi, K. Karatani, M. Nakayama, K. Matsuda, and

T. Saiki, J. Cryst. Growth 221, 461 (2000). [13] B. Q. Sun, D. S. Jiang, X. D. Luo, Z. Y. Xu, Z. Pan, L. H. Li, and R. H. Wu, Appl. Phys. Lett. 76, 2862 (2000). [14] B. Gil, Solid State Commun. 114, 623 (2000). [15] A. Lindsay and E. P. O’Reilly, Solid State Commun. 112, 443 (1999). [16] P. N. Hai, W. M. Chen, I. A. Buyanova, B. Monemar, H. P. Xin, and C. W. Tu, Mater. Sci. Eng. 882, 218

(2001). [17] A. Kaschner, T. Luttgert, H. Born, and A. Hoffmann, Appl. Phys. Lett. 78, 1391 (2001). [18] J. Wu, W. Shan, W. Walukiewiez, K. M. Yu, J. W. Ager, E. E. Haller, H. P. Xin, and C. W. Tu, Phys. Rev. B

64, 0885320 (2001). [19] H. Dumont, L. Auvray, Y. Monteil, C. Bondoux, L. Largeau, and G. Patiriarche, Appl. Phys. Lett. 80, 2460

(2002). [20] F. Saidi, F. Hassen, H. Maaref, L. Auvray, H. Dumont, and Y. Monteil, Mater. Sci. Eng. C 21, 245 (2002). [21] J. Hader, S. W. Koch, and J. V. Moloncy, Solid State Electron. 47, 513 (2003).

Documents

Optical investigation of GaAs1−xNx/GaAs heterostructure properties