Incorporation behaviors of group V elements in GaAsSbN grown by gas-source molecular-beam epitaxy

(1)

Journal of Crystal Growth 310 (2008) 2854–2858

Incorporation behaviors of group V elements in GaAsSbN grown by

gas-source molecular-beam epitaxy

Ta-Chun Ma, Yan-Ting Lin, Hao-Hsiung Lin

Graduate Institute of Electronics Engineering and Department of Electrical Engineering, National Taiwan University, Taipei 10617, Taiwan Received 16 January 2008; received in revised form 14 February 2008; accepted 14 February 2008

Communicated by H. Asahi Available online 23 February 2008

Abstract

We report the incorporation behaviors of As, Sb, and N atoms in GaAsSbN grown by gas-source molecular-beam epitaxy. We found that N atom is more reactive and competitive than Sb atom at the growth temperature ranging from 420 to 450 1C. The increment in Sb beam flux hardly changes the N composition. However, the increment in N flux retards the incorporation of Sb. In addition, the increment in As2flux makes the Sb and N compositions decrease at the same rate. Based on these results, we have successfully grown GaAsSbN epilayers lattice-matched to GaAs substrates. The energy gap at room temperature is as low as 0.803 eV. Negative deviation from Vegard’s law in lattice constant is observed in these layers.

Keywords: A3. Molecular beam epitaxy; B2. Semiconducting quarternary alloys

1. Introduction

GaAsSbN is a promising dilute nitride with only one group-III element and has drawn a lot of attention recently because of the potential applications to GaAs-based long-wavelength detectors [1–4]. Since GaAsSbN has only one group-III element, the chemical change between the nitrogen and group-III atoms, observed in thermal-annealed InGaAsN [5], can be eliminated. Less blue shift in energy gap of thermal-annealed GaAsSbN has been reported [6], which is beneﬁcial to long wavelength applications. However, the growth of this alloy is quite crucial because it contains three group-V elements. Under-standing the incorporation behaviors of the three group-V elements is very essential for precise composition control, especially for the high-quality epilayers lattice-matched to GaAs. Previous studies on molecular-beam epitaxy (MBE) showed that adding Sb could enhance the incorporation of N [7–10]. In this paper, we report the behaviors of this

promising dilute nitride grown by gas-source molecular-beam epitaxy (GSMBE). We found that N incorporation is virtually invariant of Sb ﬂux. Based on these simpler behaviors, GaAsSbN epilayers lattice-matched to GaAs have been successfully deposited.

2. Experimental procedures

A VG-V80H gas-source MBE system was used to grow all the GaAsSbN samples on (1 0 0) semi-insulating GaAs substrates. Pure arsine (AsH3), precursor of As source, was leaked into the gas cell at 1000 1C to provide the As2beams for the epitaxy process. Sb source was supplied by an EPI cracking cell. The Sb4 flux from the reservoir zone was cracked at 1050 1C to supply a mixed beam of Sb2dimer and Sb monomer. The Ga beam was supplied by EPI SUMO cell. Its flux was calibrated using an ion gauge to keep the growth rate at 1 mm/h. The N species was generated by an EPI uni-bulb RF plasma K-cell operating at a radio frequency of 13.56 MHz. A PBN shutter was placed in front of the K-cell to reduce the N flux and ionized species. All the bulk films are 1 mm in thickness.

www.elsevier.com/locate/jcrysgro

Corresponding author. Tel.: +886 2 33663670; fax: +886 2 23632442. E-mail address:[email protected] (H.-H. Lin).

(2)

The composition of the GaAsSbN was quantiﬁed by JEOL JXA-8200 electron probe X-ray microanalyzer (EPMA). GaAs, GaN, and GaSb were used as standards for ZAF correction. To ensure that the probe region is indeed within the epitaxial layer, we intentionally grew a GaAsSbN layer on InP substrate and lowered the electron gun voltage until the In and P signals were undetectable. The lattice mismatch between epitaxial layer and substrate was conﬁrmed by a Bede D1 high-resolution X-ray diffract-ometer. The absorption measurement was carried out using a tungsten–halogen lamp dispersed by a SPEX 500 M spectrometer as the light source. The transmitting light was detected by an InGaAs photodiode and the transmittance was retrieved by standard lock-in techniques.

3. Results and discussion

Fig. 1 shows the Sb and N compositions of GaAsSbN epilayers grown at 490 1C as functions of N flow rate. The AsH3 flow rate and Sb to Ga beam equilibrium pressure (BEP) ratio for these samples were fixed at 2.79 SCCM and 0.4, respectively. When N flow rate is less than 0.2 SCCM, nitrogen is not detectable in GaAsSbN by EPMA. For larger flow rate, N composition increases at the expense of Sb incorporation. Since the reactive N species in the plasma is a complicated function of plasma power, gas pressure, and flow rate, such a threshold behavior is not uncommon. Next, we consider the effect of Sb flux on N incorporation. In the experiments, AsH3and N2flow rates were fixed at 2.79 and 0.76 SCCM, respectively. Three growth

tempera-tures, 420, 450 and 490 1C, were chosen. The power of the plasma cell was set at 200 W. As can be seen inFig. 2, Sb composition increases linearly with the Sb flux. However, the N composition only slightly declines as the Sb flux increases. Recall the reduction in Sb composition resulting from the increase of N2flow rate; as evidenced inFig. 1, we conclude that N atoms tend to supplant the Sb atoms in the incorporation process. This finding, however, is in conflict with the observation of previous reports that adding Sb will enhance the incorporation rate of N[7–10]. Harmand et al. attributed their Sb-dependent behavior to the dominance of metastable molecular N2* in their N source[7]. In our case, atomic N, another N species with high chemical potential, could dominate our nitrogen incorporation process. The high chemical potential makes the N adatoms more competitive in the incorporation process than Sb adatoms, and results in the aforementioned behavior.

The temperature-independent behavior of the nitrogen incorporation coincides with previous observations on InGaAsN and GaAsN [11]. A thermodynamic study for MBE grown GaAsN and InGaAsN showed that the unchanged nitrogen composition below 480 1C can be attributed to a near-unity sticking coefﬁcient for nitrogen adatoms. In order to study the dominant species in the nitrogen plasma, we measured the dissociation fraction by using a quadrupole mass spectrometer to detect and compare the N2+ signals with and without the plasma discharge [12]. The dissociation fraction is deﬁned as the difference between the two signals divided by the signal with the discharge off. Fig. 3 shows the relationship

0.0 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Sb N

Mole fraction of group

V element

Nitrogen flow rate (SCCM) T_G= 490°C

0.2 0.4 0.6 0.8

Fig. 1. Sb and N mole fractions as functions of N ﬂow rate in GaAsSbN epitaxial layers grown at 490 1C. The AsH3ﬂow rate and Sb to Ga BEP

ratio were 2.79 SCCM and 0.4, respectively.

0.0 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22

V element Sb to Ga BEP ratio Sb (420°C) Sb (450°C) Sb (490°C) N (420°C) N (450°C) N (490°C) 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Fig. 2. Sb and N mole fractions as functions of Sb to Ga BEP ratio at three different growth temperatures, 420, 450 and 490 1C. AsH3and N2

ﬂow rates were 2.79 and 0.76 SCCM, respectively. The RF power of the N plasma cell is 200 W.

(3)

between the N mole fraction in GaAsSbN and the activated nitrogen flow rate, which is the product of nitrogen flow rate and the dissociation fraction. As can be seen, the N composition becomes linearly dependent on the atomic N flow rate as the flow rate is over its threshold value. Note that metastable N2is related to the N2+signal with the plasma on. We found that the relationship between the N2+ signal and the nitrogen composition is not a monotonic increasing function. These findings allow us to infer that the dominant species in our source is atomic nitrogen.

As shown inFig. 2, the Sb composition increases linearly with the Sb/Ga BEP ratio. However, lower growth temperature gives rise to higher Sb composition. This behavior has been observed in MBE-grown GaAsSb and was attributed to the differences in the sublimation energy and atomization energy of Sb and As [13]. The higher sublimation energy for Sb leads to the decrement of Sb incorporation at high temperature. Since the Sb composi-tion in Fig. 2 is less than 0.2, saturation behavior in incorporation as reported in Ref.[13]is not observed.

Effect of AsH3flow rate on Sb and N compositions was also investigated. N2flow rate was fixed at 0.58 SCCM. Sb to Ga BEP ratio for these samples was 0.41. All the quaternary layers were grown at 490 1C. As can be seen in

Fig. 4, the increment in AsH3ﬂow rate leads to the slight decrement in Sb and N composition simultaneously. Due to the smallest sublimation energy, As cannot compete with Sb and N. Most of the As2impinged on the growth surface are desorbed. The slight decrements shown inFig. 4

are simply due to the crowding effect of the enormous As adatoms. It is interesting that the Sb to N composition ratio approximately keeps a constant of 6 through the

whole AsH3 flow rate range. The nearly constant ratio indicates that the lattice mismatch to GaAs of the layers is almost unchanged. Therefore, the V to III ratio of the epitaxial layers can be varied without significantly affecting the lattice mismatch of the layers. Furthermore, this finding also implies that the strain and surface free energy may play a role in the incorporations of Sb and N in these nearly lattice-matched GaAsSbN layers. RecallFig. 1, the increase of N incorporation leads to the decrease of Sb. Since the sum of N and Sb compositions is approximately a constant, the N incorporation does not affect As composi-tion. In other words, the highly reactive N atom replaces Sb atom in the incorporation process. This is the most efficient way to reduce the strain energy and is certainly supported by thermodynamics. This finding, again, in-dicates the important role of strain-related surface free energy in the growth.

Three nearly lattice-matched samples with high nitrogen composition were grown on GaAs substrates at three different temperatures, 420, 450, and 490 1C, with the same nitrogen conditions, 300 W in RF power and 0.76 SCCM in nitrogen flow rate. The growth parameters and the properties are summarized in Table 1. As can be seen, the nitrogen compositions of these three samples are within 3.0–3.3%, which is consistent with the finding that the incorporation of nitrogen is nearly independent of tem-perature and Sb flux. The lattice mismatches, determined from the high-resolution XRD spectra shown inFig. 5, are all within 10 3. The bumps on the low angle shoulder of the epilayer peaks, as evidenced in the spectra, indicate the existence of compositional fluctuation in the alloys. It is worth to note that the measured compositions of these

0.00 0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035

Mole fraction of N in GaAsSbN

Activated N flow rate (SCCM) T_G = 490°C

AsH3 flow rate:2.79 SCCM

0.05 0.10 0.15 0.20

Fig. 3. Mole fraction of N in GaAsSbN as a function of activated N ﬂow rate.

0.4 0.01

0.1

V element

Sb N TG = 490°C

AsH3 flow rate (SCCM)

0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0

Fig. 4. Sb and N mole fractions as functions of AsH3ﬂow rate. N2ﬂow

rate and Sb to Ga BEP ratio were 0.58 SCCM and 0.41, respectively. The growth temperature was 490 1C.

(4)

samples deviate from the lattice-match compositions predicted by Vegard’s law. If the lattice constant obeys Vegard’s law, GaAs1 x ySbyNx alloy requires an Sb/N ratio (y/x) of 2.56 to achieve the lattice match to GaAs. The binary lattice constants of GaAs, GaSb and GaN taken for this calculation are 5.6532, 6.0959 and 4.52 A˚, respectively. As can be seen in Table 1, the Sb/N ratio is within 3.2–3.8, indicating a negative deviation from Vegard’s law in these GaAsSbN samples. Negative deviation from Vegard’s law has been observed in GaAsN epilayers grown by GSMBE[14]and MOCVD[15]and has been attributed to high substitutional nitrogen density and to vacancy-containing defect complexes stabilized by hydrogen by a recent study based on a plane-wave pseudopotential density function theory[16].Fig. 6shows the absorption spectra of the as-grown GaAsSbN samples at room temperature. The energy gaps are extracted from the absorption coefﬁcient formula for direct semiconduc-tors, ahnp(hn Eg)1/2, where a is the absorption coefﬁcient,

hn the photon energy, and Egthe energy gap. As shown in the figure, the fitting on the GaAsSbN sample grown at 420 1C gives the lowest energy gap of only 0.803 eV. The high N incorporation enhances the negative deviation in lattice constant, which requires more Sb to meet the lattice match condition and thus results in lower energy gap. However, the absorption spectra in Fig. 6 also show a strong below-band-edge absorption. Since the XRD spectra of these samples indicate the existence of the composition fluctuation, we ascribe the below-band-edge absorption to the band tail states resulting from the fluctuation.

4. Conclusion

We have studied the incorporation behaviors of the three group V elements, As, Sb and N of GaAsSbN alloys grown by GSMBE. In response to the increment in N2ﬂow rate, N composition increases at the expense of Sb composition. On

Table 1

Growth conditions, compositions determined from EPMA, lattice mismatch determined from XRD, deviation from Vegard’s law, and absorption edge data of three GaAsSbN samples

Sample ID

Growth temperature (1C)

N Rf plasma condition EPMA data Sb/N ratio (0 0 4)XRD mismatch Deviation from Vegard’s law Absorption edge (eV) Power (W) N2ﬂow rate (SCCM) Sb (%) N (%) C2106 490 300 0.76 9.8_{70.1 3.070.1 3.27} 4.8 10 4 _{2.7 10} 3 _0.847 C2107 450 300 0.76 11.070.1 3.370.3 3.33 7.9 10 4 _{3.0 10} 3 _0.816 C2221 420 300 0.76 11.770.1 3.170.3 3.77 5.7 10 4 _{5.7 10} 3 _0.803 -1000 100 101 102 103 104 105 106 107 108 109 -80'' -105'' C2221 C2107 Intensity (a.u.) C2106 -65'' GaAs substrate ω/2θ (arc sec) -750 -500 -250 0 250 500

Fig. 5. High-resolution XRD spectra of three GaAsSbN samples grown at 420, 450 and 490 1C, respectively. The N compositions of these samples are within 3.0–3.3%. The mismatches are all within 10 3_.

0.75 0.0 2.0x107 4.0x107 6.0x107 8.0x107 1.0x108 C2107: Eg = 0.816eV C2221: Eg = 0.803eV (α hν ) 2 (a.u.) C2106: Eg = 0.847eV GaAsSbN bulk RT absorption (αhν)2_{= A(h}_ν-E g)

Photon energy hν, (eV)

0.80 0.85 0.90 0.95 1.00

Fig. 6. Absorption spectra of three GaAsSbN samples. Solid curves are the experimental results, while the dotted curves represent the ﬁtting curves based on the formula for direct semiconductor, ahnp(hn Eg)1/2

where a is the absorption coefﬁcient, hn the photon energy, and Egthe

(5)

the contrary, as the Sb flux increases, the N composition hardly changes in the temperature range from 420 to 490 1C. In addition, when AsH3flow rate increases, Sb and N decrease at similar rates, which is beneficial for keeping the lattice constant unchanged. Based on these findings, we successfully grew GaAsSbN on GaAs with mismatch less than 10 3. The lowest energy gap of these GaAsSbN alloys is 0.803 eV at room temperature. The lattice constants are smaller than those predicted by Vegard’s law, i.e., negatively deviates from Vegard’s law.

Acknowledgments

This work was supported by the National Science Council of ROC and Institute of Nuclear Energy Research, Atomic Energy Council of ROC under contract No. NSC 95-NU-7-002-001. T.C. Ma is thankful to Mr. C.Y. Kao of MSE, NTU for his assistance and valuable discussions in EPMA measurement.

References

[1] G. Ungaro, G. Le Roux, R. Teissier, J.C. Harmand, Electron. Lett. 35 (1999) 1246.

[2] W. Ha, V. Gambin, M. Wistey, S. Bank, S. Kim, J.S. Harris, Proc. SPIE 4651 (2002) 42.

[3] S. Wicaksono, S.F. Yoon, W.K. Loke, K.H. tan, B.K. Ng, J. Appl. Phys. 99 (2006) 104502.

[4] H. Luo, J.A. Gupta, H. C Liu, Appl. Phys. Lett. 86 (2005) 211121.

[5] S. Kurtz, J. Webb, L. Gedvilas, D. Friedman, J. Geisz, J. Olson, R. King, D. Joslin, N. Karam, Appl. Phys. Lett. 78 (2001) 748. [6] L.F. Bian, D.S. Jiang, P.H. Tan, S.L. Lu, B.Q. Sun, L.H. Li,

J.C. Harmand, Solid State Comm. 132 (2004) 707.

[7] J.C. Harmand, G. Ungaro, L. Largeau, G. Le Roux, Appl. Phys. Lett. 77 (2000) 2482.

[8] K. Volz, V. Gambin, W. Ha, M.A. Wistey, H. Yuen, S. Bank, J.S. Harris, J. of Crystal Growth 251 (2003) 360.

[9] H.B. Yuen, S.R. Bank, M.A. Wistey, J.S. Harris, J. Appl. Phys. 96 (2004) 6375.

[10] S. Wicaksono, S.F. Yoon, K.H. Tan, W.K. Cheah, J. of Crystal Growth 274 (2005) 355.

[11] V.A. Odnoblyudov, A. Yu. Egorov, A.R. Kovsh, A.E. Zhukov, N.A. Maleev, E.S. Seminova, V.M. Ustinov, Semicond. Sci. Technol. 16 (2001) 831.

[12] R.W. McCullough, J. Geddes, J.A. Croucher, J.M. Woolsey, D.P. Higgins, M. Schlapp, Schlapp, J. Vac. Sci. Technol. A 14 (1996) 152.

[13] C.A. Chang, R. Ludeke, L.L. Chang, L. Esaki, Appl. Phys. Lett. 31 (1977) 759.

[14] W. Li, M. Pessa, J. Likonen, Appl. Phys. Lett. 78 (2001) 2864. [15] J.N. Beaudry, R.A. Masut, P. Desjardins, P. Wei, M. Chicoine,

G. Bentoumi, R. Leonelli, F. Schiettekatte, S. Guillon, J. Vac. Sci. Technol. A 22 (2004) 771.

[16] N. Shtinkov, P. Desjardins, R.A. Masut, Phys. Rev. B 74 (2006) 035211.