Alloy broadening effect on optical properties of ... - Semantic Scholar
Microelectronics Journal 35 (2004) 207–212 www.elsevier.com/locate/mejo
Alloy broadening effect on optical properties of InGaAs grown by MOCVD with TMAs precursor J. Hellaraa,*, F. Hassena, H. Maarefa, H. Dumontb, V. Souliereb, Y. Monteilb a
Laboratoire de Physique des Semiconducteurs et des Composants Electroniques (LA-MA-06), Faculte´ des Sciences de Monastir, Avenue de l’Environnement, Monastir 5000, Tunisia b LMI-Universite´ Claude Bernard de Lyon-I, 43 Bd du 11 Novembre 1918, Villeurbanne Cedex 69622, France Received 1 July 2003; accepted 3 September 2003
Abstract We investigate the MOCVD growth characteristic of the In0.47Ga0.53As layers lattice matched to InP using TMAs as an alternative source of arsine. Improvement of the InGaAs quality was studied by means of PL lines, the origin of photoluminescence (PL), atomic force microscopy. Low temperature PL spectra exhibit a broad this broadening was analyzed using quantitative models for the linewidth of band exciton based on compositional fluctuations within the crystal volume. This statistical fluctuation of the composition affects not only the PL line width but also structural properties of the InGaAs epilayer. Furthermore, by increasing V/III ratio, a degradation of the InGaAs optical and structural quality was observed. q 2004 Elsevier Ltd. All rights reserved. Keywords: Photoluminescence; Trimethylarsine; InGaAs
1. Introduction The ternary semiconductor GaInAs lattice matched to InP has many important applications for novel microwave and optoelectronic devices [1] due to their excellent electron transport properties and his 1.3 –1.5 mm emitting photodetectors [2]. An increased interest has been paid to control the surface roughness in InGaAs alloys lattice matched to InP system by finding an alternative precursor to arsine [3 – 6]. The use of arsine as a precursor remains for growing InGaAs/InP but due to this high toxicity, safety considerations have oriented laboratories and epitaxial production to find an alternative precursor to arsine. In that way trimethylarsine (TMAs) has been used for this purpose. Until now less work has been done using TMAs precursor for the growth of InGaAs epilayers [4,6,7]. The random distribution of In and Ga atoms on the group III sites in InGaAs strongly influenced the electrical and optical properties. In fact, it induces statistical potential fluctuations, which affects the line width of the photoluminescence (PL) emission by causing a broadening in * Corresponding author. Tel.: þ 216-73500-274; fax: þ 216-73500-278. E-mail address: [email protected] (J. Hellara). 0026-2692/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.mejo.2003.09.013
the low temperature PL spectra of the InGaAs alloy. The purpose of this paper is to study the line broadening of the PL spectra of InGaAs. This important intrinsic broadening effect due to statistical fluctuation has been analyzed using Schubert’s model. Besides, we have analyzed the effect of V/III ratio on the InGaAs optical and structural quality.
2. Experimental procedure The experiments were performed on lattice matched InGaAs/InP he´te´rostructures labeled as R3079 and R3084. These samples have been grown at 760 Torr pressure under H2 gas flow by MOCVD on (100) InP. Substrate material was Fe doped semi-insolating InP with an orientation off (0.28) toward (111)A. Precursors were trimethylinduim and trimethylgallium for group III, TMAs and phosphine for group V. The InGaAs epilayer with a thickness of 1 mm is grown with a buffer layer of 50 nm. The growth temperature is 620 8C. The V/III ratio is 10 for sample R3084 and 20 for sample R3079. Low temperature (10 K) PL spectroscopy have been performed in an optical flow cryostat with the He exchange gas. The samples were excited with the 514 nm provided by an Arþ laser. The luminescence was analyzed
208
J. Hellara et al. / Microelectronics Journal 35 (2004) 207–212
with InGaAs detector. Atomic force microscopy (AFM) was also done for this samples.
3. Results and discussions In this section, we discuss optical and structural results of the In0.47Ga0.53As layers grown under H2 carrier gas using TMAs as precursor. 3.1. Photoluminescence study Low temperature (10 K) PL measurements of undoped InGaAs layer grown by MOCVD on InP substrate have been
performed. Fig. 1a shows a PL spectrum for the sample R3084 with V/III ¼ 10. The spectrum exhibits a single symmetric PL band at 0.776 eV with a full width at half maximum (FWHM ¼ 9.4 meV) attributed to InGaAs material. It should be noted that this FWHM is much broader compared to other works which typically have line widths of 1 –3 meV [8,9]. Furthermore, for a sample grown with AsH3 at the same reactor, the PL spectra has shown a less broad PL band (FWHM ¼ 7 meV) followed by an increase in InGaAs PL intensity. Since the FWHM is a good indicative of the sample quality, the best value is obtained for InGaAs grown with AsH3. The excitation density dependence of the PL energy and intensity of this transition has shown a linear increase of
Fig. 1. (a) Low temperature PL spectra of TMAs based InGaAs epilayers grown at V/III ¼ 10. Theoretical PL spectra was also shown for xGa ¼ 0:47: (b) Excitation density dependence of the InGaAs transition PL intensity and PL energy.
J. Hellara et al. / Microelectronics Journal 35 (2004) 207–212
the PL intensity and no variation of its PL energy (Fig. 1b). So this transition can be identified as a bound exciton peak (BE). Such behavior of this peak line shape with increasing density was observed by Battacharya et al. [8] and Schubert and Tsang [9], but in our case the PL line width is largely higher. From this, we can suggest that this PL transition can be identified as a superposition of several excitonic lines, originating from statistical potential fluctuation, which causes a broadening of the PL spectra. This broadening of the spectrum induces surface roughness and this alter the InGaAs optical and structural quality. In order to clarify this point, we have analyzed this line broadening of band exciton using Schubert’s model [9] based on composition fluctuations within the crystal volumes characteristic of this transition. Therefore, we propose to compare the experimental spectra to the theoretical ones calculated from Schubert’s model [9]. This model assumes that the In and Ga atoms are distributed randomly on the group III sites in an ideal
209
crystal. The probability of finding a Ga atom on any group III site is given by the alloy composition x: The Schubert’s model calculates the minimal theoretical line width of an ideal alloy semiconductor. The excitonic line width is given by Ref. [9] 2 DEexc ¼ 2:36sE ¼ 2:36
dEg 6 6 dx 4
31=2 7 xð1 2 xÞ 7 5 4 23 3 4a0 paexc 3
ð1Þ
where dEg =dx is the change of the energy gap with the alloy ˚) composition xGa ; a0 is the lattice parameter (a0 ¼ 5:869 A and aexc is the excitonic radius. The excitonic radius can be determined using hydrogenic model ðDEexc ¼ 1:6 meV for x ¼ 0:47Þ: A simple calculation using Eq. (1), yields to an ˚. excitonic radius of aexc ¼ 194 A The homogeneous carrier distribution within the random alloy potential yields an exciton line shape of
Fig. 2. (b) Comparison between experimental (solid curve) and theoretical (dots) line shape of the InGaAs PL transition. (b) Temperature dependence of the InGaAs PL peak position at V/III ¼ 10.
210
J. Hellara et al. / Microelectronics Journal 35 (2004) 207–212
Gaussian form [9]: "
f ðEÞ ¼
1 1 E 2 Eexc pffiffiffiffi exp 2 2 sE ðsE 2pÞ
2 #
ð2Þ
where sE is given by Eq. (1). Using the nominal composition xGa ¼ 0:47; we have plotted the theoretical line shape of the band exciton (Fig. 1a). We observe a large difference between the experimental and theoretical PL line widths. This suggest that the BE peak is due not to a dominant type of band exciton as mentioned in Ref. [2], but to a superposition of several excitonic lines due to composition fluctuation. Taking into account this effect (compositionP fluctuation), we have defined an average composition kxl ¼ ni¼0 xi ; xi ¼ x0 ^ i £ 0:01; where 0.01 is the step. In the next, we have calculated the theoretical line width using this average composition kxl in the Eqs. (1) and (2). Fig. 2a shows good correlation with experimental PL line and this for n ¼ 3:
From the above results, we can deduce that the compositional disorder is much stronger when the InGaAs layer is grown with TMAs relatively to that grown with arsine (AsH3). In addition, we have studied the temperature dependence of the PL emission at 0.776 eV of the InGaAs epilayer. The PL band exhibit a strong dependence on temperature (Fig. 2b), an anomalous S shape behavior is shown which is often explained in terms of exciton localization at potential fluctuation caused by random distribution of the group III elements on the cation sites of the crystal. This Sshape phenomenon has already been observed in alloys such as InGaP [10] and In0.52AL0.48As [11]. In the following, much attention will be given to study the effect of V/III ratio on the optical quality of TMAs based InGaAs (Fig. 3a). Two principal near band edge luminescence features was observed (sample R3079). The highest one located at 0.796 eV (8 meV) dominating the spectrum is attributed to the InGaAs BE transition. This peak is shift to
Fig. 3. (a) Low temperature PL spectra of InGaAs grown with TMAs as a function of V/III ratio. (b) Incident laser power dependency of the PL line intensity and energy associated to the transition involving the donor–acceptor recombination.
J. Hellara et al. / Microelectronics Journal 35 (2004) 207–212
211
Fig. 4. AFM surface topography (1000 nm £ 1000 nm) of InGaAs grown at 620 8C with TMAs under hydrogen flow. (a) V/III ¼ 10 and (b) V/III ¼ 20.
lower energy as the V/III is decreased which is related to an increase in In content in the InGaAs alloy and thus leads to a rise in the lattice mismatch ðDa=aÞ: The lower one which has a FWHM ¼ 33 meV, is a weaker transition, clearly shown at lower excitation density (insert Fig. 3a), is attributed to acceptor impurity unintentionally formed in InGaAs material. The presence of this impurity leads to an increase in the surface roughness. Furthermore, the exact nature of this defect was determined by excitation density dependence. Fig. 3b shows the excitation power dependence of the impurity PL energy and intensity. A tendency to a saturation of the PL intensity is observed when increasing the excitation density and a slight blue-shift (3 –4 meV) of the energy was observed. Based on these two instances of the excitation power densities this lower peak was assigned to a donor – acceptor pair recombination (DAP) [12]. The DPA peak occurs at approximately 17 meV below the exciton peak. From line shape analysis on material with similar doping densities, Goetz et al. [1] found that the acceptor binding energy was about 5 meV larger than the DPA and exciton peak separation. This would correspond to acceptor binding energy of 22 meV in our sample. Based on PL result from several authors on intentionally Zn doped GaInAs [1,8], the deep acceptor in our material appears to be Zn which has an activation energy of Ea ¼ 22 meV [1]. In addition, we note the non-appearance of the PL band related to the impurity for V/III ¼ 10, which indicates the lower surface roughness. In Section 3.2, we discuss the effect of V/III ratio on the structural quality of the InGaAs grown with TMAs under Hydrogen flow was taking into account in order to control the surface roughness. 3.2. Structural properties At V/III ¼ 10, the surface morphology shows an irregular step edge (Fig. 4a) with a surface roughness (RMS ¼ 0.23). Since the step flow growth mode is controlled by surface diffusion length of group III
elements adatoms (In, Ga). This irregular step edge can be interpreted as the result of methyl group desorption during the growth, which interrupt the diffusion length of group III adatom. To evaluate the V/III ratio dependence of the residual impurity, we have studied the effect of V/ III ratio on the structural properties of TMAs based InGaAs layers, by comparing two samples (R3079 – R3084). Fig. 4b shows the top view AFM image of InGaAs/InP grown with TMAs at V/III ¼ 20. For this V/ III ratio the surface roughness increases to RMS ¼ 0.57, as a result the surface morphology deteriorates rapidly with appearance of 2D islands resulting from saturation of the growth surface by methyl group. In fact, at high TMAs/III ratio, the strong methyl concentration on the step will induce a growth mode no longer controlled by adatoms (III elements) surface diffusion but by methyl desorption at the steps. As a consequence, a non-atomic steps was observed, and this suggests the presence of 2D nucleation growth mode. This is in accordance with the recent work of Dumont et al. [6] in which it has been demonstrated that the surface morphology of InGaAs grown with TMAs is very sensitive to V/III.
4. Conclusion InGaAs layers lattice matched to InP were grown by atmospheric pressure MOCVD using timethylarsine at growth temperature 620 8C under hydrogen carrier gas, and the effect of V/III ratio on layers properties was studied by Hall, AFM and PL measurements. We have studied the intrinsic effect of alloy broadening owning to the random distribution of AL and Ga in InGaAs and the influence of thermal broadening upon the luminescence line width, we have compared the experimental results with Schubert’s model. It was found that statistical fluctuation of the alloy composition are responsible for the observed line broadening at low temperatures,
212
J. Hellara et al. / Microelectronics Journal 35 (2004) 207–212
and thus also alter the structural properties of the InGaAs alloy.
References [1] K.H. Goetz, D. Bimberg, A.V. Solomonov, G.F. Glinskii, M. Razeghi, J. Appl. Phys. 54 (1983) 4543. [2] C. Charreaux, G. Guillot, A. Nouailhat, J. Appl. Phys. 60 (1986) 768. [3] J. Knauf, D. Schmitzm, G. Strauch, H. Jurgensen, M. Heyen, A. Melas, J. Cryst. Growth 93 (1988) 34. [4] H. Dumont, L. Auvray, J. Dazord, V. Souliere, Y. Monteil, J. Bouix, A. Ougazzaden, J. Cryst. Growth 197 (1999) 755.
[5] R. Hidaka, H. Asahi, R. Yamamoto, K. Asami, J. Satoh, S. Gonda, J. Cryst. Growth 181 (1997) 193. [6] H. Dumont, L. Auvray, J. Dazord, Y. Monteil, J. Bouix, A. Ougazzaden, J. Cryst. Growth 204 (1999) 1. [7] C. Pautet, P. Abraham, J. Dazord, R. Favre, F. Cauwet, Y. Monteil, J. Bouix, A. Ougazzaden, E.V.K. Rao, A. Mircea, Appl. Phys. Lett. 69 (1996) 209. [8] P.K. Battacharya, M.V. Rao, M.V. Rao, M.J. Tsai, J. Appl. Phys. 54 (1983) 5096. [9] E.F. Schubert, W.T. Tsang, Phys. Rev. B 34 (1986) 2991. [10] M. Kondow, S. Mingawa, Y. Inoue, T. Nishino, Y. Hamakawa, Appl. phys. Lett. 54 (1989) 1760. [11] S.M. Olsthoorn, F.A.J.M. Driessen, J. Gilling, J. Appl. Phys. 73 (1993) 7804. [12] M. Kamada, H. Ishikawa, J. Cryst. Growth 94 (1989) 849.
Alloy broadening effect on optical properties of InGaAs grown by MOCVD with TMAs precursor J. Hellaraa,*, F. Hassena, H. Maarefa, H. Dumontb, V. Souliereb, Y. Monteilb a
Laboratoire de Physique des Semiconducteurs et des Composants Electroniques (LA-MA-06), Faculte´ des Sciences de Monastir, Avenue de l’Environnement, Monastir 5000, Tunisia b LMI-Universite´ Claude Bernard de Lyon-I, 43 Bd du 11 Novembre 1918, Villeurbanne Cedex 69622, France Received 1 July 2003; accepted 3 September 2003
Abstract We investigate the MOCVD growth characteristic of the In0.47Ga0.53As layers lattice matched to InP using TMAs as an alternative source of arsine. Improvement of the InGaAs quality was studied by means of PL lines, the origin of photoluminescence (PL), atomic force microscopy. Low temperature PL spectra exhibit a broad this broadening was analyzed using quantitative models for the linewidth of band exciton based on compositional fluctuations within the crystal volume. This statistical fluctuation of the composition affects not only the PL line width but also structural properties of the InGaAs epilayer. Furthermore, by increasing V/III ratio, a degradation of the InGaAs optical and structural quality was observed. q 2004 Elsevier Ltd. All rights reserved. Keywords: Photoluminescence; Trimethylarsine; InGaAs
1. Introduction The ternary semiconductor GaInAs lattice matched to InP has many important applications for novel microwave and optoelectronic devices [1] due to their excellent electron transport properties and his 1.3 –1.5 mm emitting photodetectors [2]. An increased interest has been paid to control the surface roughness in InGaAs alloys lattice matched to InP system by finding an alternative precursor to arsine [3 – 6]. The use of arsine as a precursor remains for growing InGaAs/InP but due to this high toxicity, safety considerations have oriented laboratories and epitaxial production to find an alternative precursor to arsine. In that way trimethylarsine (TMAs) has been used for this purpose. Until now less work has been done using TMAs precursor for the growth of InGaAs epilayers [4,6,7]. The random distribution of In and Ga atoms on the group III sites in InGaAs strongly influenced the electrical and optical properties. In fact, it induces statistical potential fluctuations, which affects the line width of the photoluminescence (PL) emission by causing a broadening in * Corresponding author. Tel.: þ 216-73500-274; fax: þ 216-73500-278. E-mail address: [email protected] (J. Hellara). 0026-2692/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.mejo.2003.09.013
the low temperature PL spectra of the InGaAs alloy. The purpose of this paper is to study the line broadening of the PL spectra of InGaAs. This important intrinsic broadening effect due to statistical fluctuation has been analyzed using Schubert’s model. Besides, we have analyzed the effect of V/III ratio on the InGaAs optical and structural quality.
2. Experimental procedure The experiments were performed on lattice matched InGaAs/InP he´te´rostructures labeled as R3079 and R3084. These samples have been grown at 760 Torr pressure under H2 gas flow by MOCVD on (100) InP. Substrate material was Fe doped semi-insolating InP with an orientation off (0.28) toward (111)A. Precursors were trimethylinduim and trimethylgallium for group III, TMAs and phosphine for group V. The InGaAs epilayer with a thickness of 1 mm is grown with a buffer layer of 50 nm. The growth temperature is 620 8C. The V/III ratio is 10 for sample R3084 and 20 for sample R3079. Low temperature (10 K) PL spectroscopy have been performed in an optical flow cryostat with the He exchange gas. The samples were excited with the 514 nm provided by an Arþ laser. The luminescence was analyzed
208
J. Hellara et al. / Microelectronics Journal 35 (2004) 207–212
with InGaAs detector. Atomic force microscopy (AFM) was also done for this samples.
3. Results and discussions In this section, we discuss optical and structural results of the In0.47Ga0.53As layers grown under H2 carrier gas using TMAs as precursor. 3.1. Photoluminescence study Low temperature (10 K) PL measurements of undoped InGaAs layer grown by MOCVD on InP substrate have been
performed. Fig. 1a shows a PL spectrum for the sample R3084 with V/III ¼ 10. The spectrum exhibits a single symmetric PL band at 0.776 eV with a full width at half maximum (FWHM ¼ 9.4 meV) attributed to InGaAs material. It should be noted that this FWHM is much broader compared to other works which typically have line widths of 1 –3 meV [8,9]. Furthermore, for a sample grown with AsH3 at the same reactor, the PL spectra has shown a less broad PL band (FWHM ¼ 7 meV) followed by an increase in InGaAs PL intensity. Since the FWHM is a good indicative of the sample quality, the best value is obtained for InGaAs grown with AsH3. The excitation density dependence of the PL energy and intensity of this transition has shown a linear increase of
Fig. 1. (a) Low temperature PL spectra of TMAs based InGaAs epilayers grown at V/III ¼ 10. Theoretical PL spectra was also shown for xGa ¼ 0:47: (b) Excitation density dependence of the InGaAs transition PL intensity and PL energy.
J. Hellara et al. / Microelectronics Journal 35 (2004) 207–212
the PL intensity and no variation of its PL energy (Fig. 1b). So this transition can be identified as a bound exciton peak (BE). Such behavior of this peak line shape with increasing density was observed by Battacharya et al. [8] and Schubert and Tsang [9], but in our case the PL line width is largely higher. From this, we can suggest that this PL transition can be identified as a superposition of several excitonic lines, originating from statistical potential fluctuation, which causes a broadening of the PL spectra. This broadening of the spectrum induces surface roughness and this alter the InGaAs optical and structural quality. In order to clarify this point, we have analyzed this line broadening of band exciton using Schubert’s model [9] based on composition fluctuations within the crystal volumes characteristic of this transition. Therefore, we propose to compare the experimental spectra to the theoretical ones calculated from Schubert’s model [9]. This model assumes that the In and Ga atoms are distributed randomly on the group III sites in an ideal
209
crystal. The probability of finding a Ga atom on any group III site is given by the alloy composition x: The Schubert’s model calculates the minimal theoretical line width of an ideal alloy semiconductor. The excitonic line width is given by Ref. [9] 2 DEexc ¼ 2:36sE ¼ 2:36
dEg 6 6 dx 4
31=2 7 xð1 2 xÞ 7 5 4 23 3 4a0 paexc 3
ð1Þ
where dEg =dx is the change of the energy gap with the alloy ˚) composition xGa ; a0 is the lattice parameter (a0 ¼ 5:869 A and aexc is the excitonic radius. The excitonic radius can be determined using hydrogenic model ðDEexc ¼ 1:6 meV for x ¼ 0:47Þ: A simple calculation using Eq. (1), yields to an ˚. excitonic radius of aexc ¼ 194 A The homogeneous carrier distribution within the random alloy potential yields an exciton line shape of
Fig. 2. (b) Comparison between experimental (solid curve) and theoretical (dots) line shape of the InGaAs PL transition. (b) Temperature dependence of the InGaAs PL peak position at V/III ¼ 10.
210
J. Hellara et al. / Microelectronics Journal 35 (2004) 207–212
Gaussian form [9]: "
f ðEÞ ¼
1 1 E 2 Eexc pffiffiffiffi exp 2 2 sE ðsE 2pÞ
2 #
ð2Þ
where sE is given by Eq. (1). Using the nominal composition xGa ¼ 0:47; we have plotted the theoretical line shape of the band exciton (Fig. 1a). We observe a large difference between the experimental and theoretical PL line widths. This suggest that the BE peak is due not to a dominant type of band exciton as mentioned in Ref. [2], but to a superposition of several excitonic lines due to composition fluctuation. Taking into account this effect (compositionP fluctuation), we have defined an average composition kxl ¼ ni¼0 xi ; xi ¼ x0 ^ i £ 0:01; where 0.01 is the step. In the next, we have calculated the theoretical line width using this average composition kxl in the Eqs. (1) and (2). Fig. 2a shows good correlation with experimental PL line and this for n ¼ 3:
From the above results, we can deduce that the compositional disorder is much stronger when the InGaAs layer is grown with TMAs relatively to that grown with arsine (AsH3). In addition, we have studied the temperature dependence of the PL emission at 0.776 eV of the InGaAs epilayer. The PL band exhibit a strong dependence on temperature (Fig. 2b), an anomalous S shape behavior is shown which is often explained in terms of exciton localization at potential fluctuation caused by random distribution of the group III elements on the cation sites of the crystal. This Sshape phenomenon has already been observed in alloys such as InGaP [10] and In0.52AL0.48As [11]. In the following, much attention will be given to study the effect of V/III ratio on the optical quality of TMAs based InGaAs (Fig. 3a). Two principal near band edge luminescence features was observed (sample R3079). The highest one located at 0.796 eV (8 meV) dominating the spectrum is attributed to the InGaAs BE transition. This peak is shift to
Fig. 3. (a) Low temperature PL spectra of InGaAs grown with TMAs as a function of V/III ratio. (b) Incident laser power dependency of the PL line intensity and energy associated to the transition involving the donor–acceptor recombination.
J. Hellara et al. / Microelectronics Journal 35 (2004) 207–212
211
Fig. 4. AFM surface topography (1000 nm £ 1000 nm) of InGaAs grown at 620 8C with TMAs under hydrogen flow. (a) V/III ¼ 10 and (b) V/III ¼ 20.
lower energy as the V/III is decreased which is related to an increase in In content in the InGaAs alloy and thus leads to a rise in the lattice mismatch ðDa=aÞ: The lower one which has a FWHM ¼ 33 meV, is a weaker transition, clearly shown at lower excitation density (insert Fig. 3a), is attributed to acceptor impurity unintentionally formed in InGaAs material. The presence of this impurity leads to an increase in the surface roughness. Furthermore, the exact nature of this defect was determined by excitation density dependence. Fig. 3b shows the excitation power dependence of the impurity PL energy and intensity. A tendency to a saturation of the PL intensity is observed when increasing the excitation density and a slight blue-shift (3 –4 meV) of the energy was observed. Based on these two instances of the excitation power densities this lower peak was assigned to a donor – acceptor pair recombination (DAP) [12]. The DPA peak occurs at approximately 17 meV below the exciton peak. From line shape analysis on material with similar doping densities, Goetz et al. [1] found that the acceptor binding energy was about 5 meV larger than the DPA and exciton peak separation. This would correspond to acceptor binding energy of 22 meV in our sample. Based on PL result from several authors on intentionally Zn doped GaInAs [1,8], the deep acceptor in our material appears to be Zn which has an activation energy of Ea ¼ 22 meV [1]. In addition, we note the non-appearance of the PL band related to the impurity for V/III ¼ 10, which indicates the lower surface roughness. In Section 3.2, we discuss the effect of V/III ratio on the structural quality of the InGaAs grown with TMAs under Hydrogen flow was taking into account in order to control the surface roughness. 3.2. Structural properties At V/III ¼ 10, the surface morphology shows an irregular step edge (Fig. 4a) with a surface roughness (RMS ¼ 0.23). Since the step flow growth mode is controlled by surface diffusion length of group III
elements adatoms (In, Ga). This irregular step edge can be interpreted as the result of methyl group desorption during the growth, which interrupt the diffusion length of group III adatom. To evaluate the V/III ratio dependence of the residual impurity, we have studied the effect of V/ III ratio on the structural properties of TMAs based InGaAs layers, by comparing two samples (R3079 – R3084). Fig. 4b shows the top view AFM image of InGaAs/InP grown with TMAs at V/III ¼ 20. For this V/ III ratio the surface roughness increases to RMS ¼ 0.57, as a result the surface morphology deteriorates rapidly with appearance of 2D islands resulting from saturation of the growth surface by methyl group. In fact, at high TMAs/III ratio, the strong methyl concentration on the step will induce a growth mode no longer controlled by adatoms (III elements) surface diffusion but by methyl desorption at the steps. As a consequence, a non-atomic steps was observed, and this suggests the presence of 2D nucleation growth mode. This is in accordance with the recent work of Dumont et al. [6] in which it has been demonstrated that the surface morphology of InGaAs grown with TMAs is very sensitive to V/III.
4. Conclusion InGaAs layers lattice matched to InP were grown by atmospheric pressure MOCVD using timethylarsine at growth temperature 620 8C under hydrogen carrier gas, and the effect of V/III ratio on layers properties was studied by Hall, AFM and PL measurements. We have studied the intrinsic effect of alloy broadening owning to the random distribution of AL and Ga in InGaAs and the influence of thermal broadening upon the luminescence line width, we have compared the experimental results with Schubert’s model. It was found that statistical fluctuation of the alloy composition are responsible for the observed line broadening at low temperatures,
212
J. Hellara et al. / Microelectronics Journal 35 (2004) 207–212
and thus also alter the structural properties of the InGaAs alloy.
References [1] K.H. Goetz, D. Bimberg, A.V. Solomonov, G.F. Glinskii, M. Razeghi, J. Appl. Phys. 54 (1983) 4543. [2] C. Charreaux, G. Guillot, A. Nouailhat, J. Appl. Phys. 60 (1986) 768. [3] J. Knauf, D. Schmitzm, G. Strauch, H. Jurgensen, M. Heyen, A. Melas, J. Cryst. Growth 93 (1988) 34. [4] H. Dumont, L. Auvray, J. Dazord, V. Souliere, Y. Monteil, J. Bouix, A. Ougazzaden, J. Cryst. Growth 197 (1999) 755.
[5] R. Hidaka, H. Asahi, R. Yamamoto, K. Asami, J. Satoh, S. Gonda, J. Cryst. Growth 181 (1997) 193. [6] H. Dumont, L. Auvray, J. Dazord, Y. Monteil, J. Bouix, A. Ougazzaden, J. Cryst. Growth 204 (1999) 1. [7] C. Pautet, P. Abraham, J. Dazord, R. Favre, F. Cauwet, Y. Monteil, J. Bouix, A. Ougazzaden, E.V.K. Rao, A. Mircea, Appl. Phys. Lett. 69 (1996) 209. [8] P.K. Battacharya, M.V. Rao, M.V. Rao, M.J. Tsai, J. Appl. Phys. 54 (1983) 5096. [9] E.F. Schubert, W.T. Tsang, Phys. Rev. B 34 (1986) 2991. [10] M. Kondow, S. Mingawa, Y. Inoue, T. Nishino, Y. Hamakawa, Appl. phys. Lett. 54 (1989) 1760. [11] S.M. Olsthoorn, F.A.J.M. Driessen, J. Gilling, J. Appl. Phys. 73 (1993) 7804. [12] M. Kamada, H. Ishikawa, J. Cryst. Growth 94 (1989) 849.
