Upload
kj-hong
View
213
Download
0
Embed Size (px)
Citation preview
Journal of Crystal Growth 240 (2002) 57–63
Point defects of ZnSe epilayers grown by hot wall epitaxy
K.J. Honga,*, S.H. Youa, T.S. Jeongb, C.J. Younb, M.S. Hongc,J.S. Parkd, C.S. Parkd
aDepartment of Physics, Chosun University, Kwangju 501-759, South KoreabDepartment of Physics and Semiconductor Physics Research Center (SPRC), Jeonbuk National University,
Jeonbuk 561-756, South KoreacDepartment of Mechanical Engineering College of Engineering, Chosun University, Kwangju 501-759, South KoreadDivision of Metallurgical and Material Science Engineering, Chosun University, Kwangju 501-759, South Korea
Received 10 December 2001; accepted 30 December 2001
Communicated by M. Schieber
Abstract
The ZnSe epilayers were grown on the GaAs substrate by hot wall epitaxy. After the ZnSe epilayers were treated in
the vacuum, Zn- and Se-atmosphere, respectively, the defects of the epilayer were investigated by means of low-
temperature photoluminescence measurement. The dominant peaks at 2.7988 and 2.7937 eV obtained from the PL
spectrum of the as-grown ZnSe epilayer were found to be consistent with the upper and the lower polariton peak of the
exciton. I2 (D1, X), bounded to the neutral donor associated with the Se-vacancy. This donor-impurity binding energy
was calculated to be 25.3meV. The exciton peak, Id1 ; at 2.7812 eV was confirmed to be bound to the neutral acceptor
which corresponded with the Zn-vacancy. The Id1 peak was dominantly observed in the ZnSe/GaAs:Se epilayer treated
in the Se-atmosphere. This Se-atmosphere treatment may convert the ZnSe/GaAs:Se epilayer into the p-type. The SA
peak was found to be related to a complex donor like a (VSe–VZn)–VZn. r 2002 Elsevier Science B.V. All rights
reserved.
PACS: 61.72.Ji; 78.55.Et; 81.15.�z
Keywords: A1. Annealing treatment; A1. Defects; A1. Photoluminescence; A3. Hot wall epitaxy; B1. Zinc compounds
1. Introduction
ZnSe has been recently tried to be grown as p-type ZnSe for fabricating blue laser diode andlight-emitting diode [1–8]. Generally, the grownZnSe epilayer is known to be of n-type. Therefore,a major problem for obtaining better device
performance is to grow the p-type ZnSe layer withlow electrical resistivity. The difficulty in the p-type growth and the conductivity control for ZnSeare well known to be strongly related to the nativedefects and self-compensation of the ZnSe due tothe stoichiometric deviation generated duringgrowth or the additional thermal treatment[9,10]. The stoichiometric deviation is mainlycaused by the partial vapor pressure of seleniumwhich is higher than that of zinc during growth.
*Corresponding author.
E-mail address: [email protected] (K.J. Hong).
0022-0248/02/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 0 2 2 - 0 2 4 8 ( 0 2 ) 0 0 8 5 4 - 0
These native defects consist of Se-vacancy (VSe),Zn-vacancy (VZn), Se-interstitial (Seint), Zn-inter-stitial (Znint) and complexes of these single defects.Among the defects, the VSe and Znint acted as adonor. Other defects such as VZn and Seint couldwork as deep levels and/or as acceptor. The low-temperature crystal growth method for ZnSe hasbeen preferred to reduce native defects. Hot wallepitaxy (HWE) [11] is one of the low-temperaturecrystal growth technologies, hence it can grow ahigh-purity ZnSe epilayer at low temperature.HWE has been especially designed to growepilayers under the condition of near thermody-namics equilibrium [12].In this paper, the ZnSe epilayer was grown using
HWE and its crystal quality was investigated by
means of the double-crystal X-ray diffractiontechnique. The ZnSe epilayers treated in thevarious atmospheres were investigated using thePL spectra. Based on these results, we will discussthe origin of point defects formed in the ZnSeepilayer.
2. Growth and experimental procedure
A HWE apparatus used for growing the ZnSeepilayers (ZnSe/GaAs) on the semi-insulating(1 0 0) GaAs is shown in Fig. 1. Prior to growth,the GaAs substrate was cleaned ultrasonically for1min in successive baths of trichloroethylene,acetone, methanol, and 2-propanol and etched
Fig. 1. Schematic diagram of the HWE apparatus.
K.J. Hong et al. / Journal of Crystal Growth 240 (2002) 57–6358
for 1min in a solution of H2SO4:H2O2:H2O(5:1:1). The substrate was degreased in organicsolvents, and rinsed with deionized water(18.2MO). After the substrate was dried, thesubstrate was immediately loaded onto the sub-strate holder in Fig. 1 and was annealed at 5801Cfor 20min to remove the residual oxide on thesurface of the substrate. The grown ZnSe/GaAsepilayers were analyzed by the double-crystalX-ray diffraction (Bede Scientific Co. FR 590) toobtain the optimum growth condition. However,to grow the undoped ZnSe/GaAs epilayers, themost suitable temperatures of the substrate andthe source containing ZnSe powder turned out tobe 4001C and 6701C, respectively. The minimumvalue of a full width at half maximum (FWHM) ofthe ZnSe/GaAs epilayer obtained from the X-rayrocking curves was 195 arcsec, as shown in Fig. 2.This value is better than the value obtained fromthe ZnSe/GaAs epilayer grown with MOVPEby Fujita et al. [13]. Fig. 3 shows the surfacemorphology and the cross section of the ZnSe/GaAs epilayer observed by scanning electronmicroscopy (SEM). The ZnSe/GaAs epilayer grew
with a very smooth surface like a mirror, as shownin Fig. 3(a). Also, the thickness and the growthrate of epilayer were 1.8 mm and 0.03 mm/min,respectively.After growing, the as-grown ZnSe/GaAs epi-
layers were prepared under the following condi-tions: (1) After the epilayer and the powder of Znwere sealed in a quartz ampoule at B10�5 Torr,the ampoule was annealed for 1 h at 6001C, ZnSe/GaAs:Zn. (2) After the epilayer and the powder ofSe were sealed in a quartz ampoule atB10�5 Torr,the ampoule was annealed for 1 h at 2301C, ZnSe/GaAs:Se. (3) The as-grown epilayer was annealedin vacuum for 1 h at 6001C, ZnSe/GaAs:vac.The PL measurements at 10K were carried out
using a cryogenic helium refrigerator (AP, CSA-202B). The samples mounted on the cold finger of
Fig. 2. The X-ray rocking curves of the as-grown ZnSe/GaAs
epilayer.
Fig. 3. The photographs of SEM for the as-grown ZnSe/GaAs
epilayer: (a) the surface morphology and (b) the cross section of
epilayer.
K.J. Hong et al. / Journal of Crystal Growth 240 (2002) 57–63 59
a cryostat were focused using the 325 nm line ofHe–Cd laser (1K 545IR, Kimmon Electric Co)with a power of 10mW and the emitted light wasdetected by a photomultiplier tube (RCA, C31034) through the monochromator. The detectedsignal was amplified by a lock-in amplifier(EG&G, 5208) and recorded in an x–y plotter.
3. Results and discussion
3.1. As-grown ZnSe/GaAs epilayer
Fig. 4 shows a typical PL spectra of the as-grown ZnSe/GaAs epilayer measured at 10K. Thefree exciton peak, Ex; at 442.4 nm appears onthe shoulder toward the short-wavelength region.The energy of the Ex is 2.802 eV, which is equal tothe values obtained from the undoped ZnSe/GaAsepilayers grown with MOMBE by Migita et al. [14]and with MBE by Akimoto et al. [15], respectively.As shown in the sub-figure of Fig. 4, the verystrong intensity peaks, I2 (D1,X) were observed to
be at 443 nm (2.7988 eV) and 443.8 nm (2.7937 eV),which are believed to be the peaks bounded to theneutral donor. Each of the peaks represents theupper polariton, IU2 ; and the lower polariton, IL2 ;respectively [16,17]. The splitting energy betweenthe upper and the lower polariton was 5.1meV.This polariton is known to be caused by the straindue to the lattice mismatch between substrate andepilayer in the heteroepilayer growth. The FWHMvalue of the IL2 peak was 5.7meV. This value hasbeen reported to be the 6.35meV by Hingerl et al.[18], which was obtained from the 17K cathodo-luminescence spectrum. Also, the bound excitonpeak, I2; was observed at 443.4 nm (2.7963 eV) inthe 4.2K PL of sample grown with MBE by Yao[19]. The observance of the IL2 suggests that theundoped ZnSe epilayers grown in this experimenthave a very high optical quality. And the IL2 isgenerally known to be the bound exciton, I2[20,21]. Therefore, the binding energy [22] of thedonor-impurity, ED; can be calculated using
EðDo;XÞ ¼ Eg � Ebex � 0:15ED; ð1Þ
where Ebex is the binding energy of the free exciton.
ED was determined to be 25.3meV. This value isclose to the ionization energies of donors such asAl, Cl, and Liint, which have been reported to be25.3, 25.9 and 26meV, respectively [23]. A neutralacceptor bound exciton Id1 (Ao,X), of the sharpintensity peak at 445.8 nm (2.7812 eV) and its LOphonon replicas at 451 nm (2.7491 eV) and456.3 nm (2.7172 eV) appear on the right regionof the wavelength. A Iov emission and its LOphonon replicas were observed at 447 nm(2.7737 eV) and 452.5 nm (2.7400 eV), respectively.The thermal stability of the binding force of the Iovcould be characterized by observing a regular peakposition irrespective of the epilayer growth condi-tion. The origin of the Iov emission is related to thedislocation or the complex defects which acted asthe dislocation [24]. The peaks at 476.5 nm(2.6020 eV) and 482.2 nm (2.5712 eV) are coinci-dent with Yo emission [25] and its LO phononreplica associated with the dislocation generateddue to the lattice mismatch. The observance of theYo emission in the epilayer indicates that thegrown sample is a high-quality crystal [26]. A566 nm (2.1906 eV) peak of flat slope with a low
Fig. 4. Photoluminescence spectrum at 10K for the as-grown
ZnSe/GaAs epilayer.
K.J. Hong et al. / Journal of Crystal Growth 240 (2002) 57–6360
intensity at the long-wavelength region corre-sponds to a self-activated (SA) emission.
3.2. Annealing effect of the ZnSe/GaAs epilayers
In order to know the origins of the several peaksof the as-grown ZnSe/GaAs, we measured the PLspectra for samples annealed in vacuum, Zn- andthe Se-atmosphere. The obtained PL spectra areshown in Figs. 5 and 6. First, when the ZnSe/GaAs epilayer annealed in vacuum for 1 h at6001C (the ZnSe/GaAs:vac epilayer), the epilayerbecame non-stoichiometry because the Zn and theSe atoms vapored out, leaving the vacancies suchas VZn and VSe. Therefore, we can observe allpeaks related to the VZn and VSe. Fig. 5 shows thatthe I2 emission at 443.4 nm (2.7962 eV) and the I1-like emission at 446.5 nm (2.7768 eV) dominantlyappeared in the PL spectrum of ZnSe/GaAs:vacepilayer. The I2 is consistent with the donor and
the I1-like emission corresponds to the acceptor.Also the peaks related to LO phonon replicas ofthe I1-like are seen at the wavelength range from451.5 nm (2.7461 eV) to 462.1 nm (2.6831 eV). Onthe other hand, the Iov and the Yo peaks observedin the as-grown epilayer disappeared and theintensity of the SA spectrum at 620 nm(1.9998 eV) increased. The new peak associatedwith the Cu-green emission was observed to be at535.4 nm (2.3158 eV). This Cu-green emission isknown to be associated with the residual Cuimpurities in ZnSe powder.Second, to know the role of Zn, we prepared
ZnSe/GaAs samples annealed in Zn-atmospherefor 1 h at 6001C (ZnSe/GaAs:Zn epilayer). TheZn-atmosphere annealing could diffuse the addi-tional Zn into the ZnSe/GaAs epilayer. Conse-quently, the Zn-vacancies in the ZnSe/GaAsepilayer are filled with the diffused Zn. Bycomparing the peaks in Fig. 4 with those inFig. 5, we found that the peaks related to the I1and the SA emission completely disappeared in theZnSe/GaAs:Zn epilayer and the I2 emission at443.4 nm (2.7962 eV) became the dominant peak.This I2 peak was generally observed in the typical
Fig. 5. Photoluminescence spectra at 10K for the ZnSe/
GaAs:vac and ZnSe/GaAs:Zn.
Fig. 6. Photoluminescence spectrum at 10K for the ZnSe/
GaAs:Se.
K.J. Hong et al. / Journal of Crystal Growth 240 (2002) 57–63 61
n-type. The FWHM value of the I2 peak was takento be 7.5meV. The disappearance of the I1 and theSA emission indicates that these peaks arecertainly associated with VZn. This disappearancealso implies that the I2 peak is not related to Zn-vacancy, VZn, because the sites of VZn should besubstituted with the diffused Zn. As the intensityof the I2 peak increases, the intensity of the SAemission decreases. This may mean that the ZnSe/GaAs:Zn epilayers are purified by annealing in Zn-atmosphere. Among the samples prepared in thisexperiment, the donor–acceptor pair (DAP) emis-sion and its LO phonon replicas were observedonly in the sample annealed in the Zn-atmosphere.Such DAP emission is caused by an interactionbetween donors (such as Al and Cl) and shallowacceptors.Third, to know the role of Se, the ZnSe/GaAs
samples were annealed in Se-atmosphere for 1 h at2301C (ZnSe/GaAs:Se). Due to Se-atmosphereannealing the Se-vacancies in the ZnSe/GaAsepilayer are filled with the diffused Se. As shownin Fig. 6, the Id1 peak at 445.8 nm is very sharp andits PL intensity is high. However, the Ex; IU2 and IL2are observed to be relatively very weak emissionsat 442.4, 443 and 443.5 nm, respectively. With acomparison of the peaks in Fig. 4, the dominantIU2 emission significantly decreases and the inten-sity of the Id1 peak increases. The FWHM value ofthe Id1 peak was 2.5meV and very sharp. The LOphonon replicas of the Id1 are the dominant peaksin the PL spectra of the ZnSe/GaAs:Se. The Id1 hasbeen known to be associated with a deep acceptorlevel originated from the VZn due to stoichiometricdeviation [27]. The origin of the Id1 is also reportedby other authors [28,29]. Additionally, the as-grown ZnSe epilayers were annealed at 4001C and6001C in the Se-atmosphere, respectively, to studythe heat-treatment effect of Se. However, only theSA emission was seen and the emission peaks onthe short-wavelength region could not be observedin these samples. This can be caused by theabsorption of light emitted from the sample dueto the recrystallized Se on the sample surface. Thedominant Id1 peak and its replicas could beobserved in the sample annealed in the Se-atmo-sphere at 2301C. This suggests that only the Se-atmosphere annealing at 2301C could convert the
as-grown ZnSe into p-type. The conversion of thep-type ZnSe crystal is well known to be difficultdue to the self-compensation, although the ZnSecrystal is intentionally doped with impurities suchas Li and P.From results of the Id1 emission, the binding
energy [22] of the acceptor-impurity, EA; isobtained using
EðAo;XÞ ¼ Eg � Ebex � 0:08EA; ð2Þ
where EA was calculated to be 268meV. This valueis larger than 200meV reported by Yao et al. [30].The origin of acceptor-impurity has been reportedby Bhargava [31] to be Cu-related. The intensity ofthe I2 peak enhanced after the Zn-atmosphereannealing. On the other hand, the intensity of theI2 decreased after the Se-atmosphere annealing,but did not disappear completely. Therefore, theorigin of the I2 is related to VSe and the present I2emission is originated to a residual donor impuritysuch as Al and Cl. Though the Id1 peak dis-appeared after the Zn-atmosphere annealing, theId1 peak became the dominant peak after the Se-atmosphere treatment. This indicates that theorigin of the Id1 emission is related to VZn. TheSA emission disappeared after the Se-atmosphereannealing. This means that the origin of the SAemisison is related to VSe. And the ZnSe/GaAsepilayer after the Se-atmosphere annealing is alsopurified like a Zn-atmosphere annealing. The S-band [32] at 516.4 nm (2.4010 eV) was seen, evenorigin of this peak was not known and the Y0
emission at 476.5 nm was also observed in the Se-atmosphere-treated samples.
4. Conclusions
The ZnSe/GaAs epilayers were grown on thesemi-insulating (1 0 0) GaAs by HWE method. Theoptimum growth temperatures of the substrateand the source containing ZnSe powder werefound to be 4001C and 6701C, respectively.FWHM from the X-ray rocking curves andthickness were obtained to be 195 arcsec and1.8 mm, respectively. The PL measurement showedthat the dominant peaks at 2.7988 and 2.7937 eVobtained from the as-grown ZnSe epilayer
K.J. Hong et al. / Journal of Crystal Growth 240 (2002) 57–6362
corresponded to the upper and the lower polaritonpeak of the exciton, I2 (D
o,X). This polariton peakis associated with the strain due to the latticemismatch between substrate and epilayer. Whenthe samples were treated in the vacuum, Zn- andSe-atmosphere, respectively, the I2 peak wasobserved and its origin was not related to VZn
but VSe. The donor-impurity binding energy wascalculated to be 25.3meV. The exciton bounded toa neutral acceptor, Id1 ; was also seen. However, theId1 emission and its LO phonon replicas weredominant peaks in the spectrum of ZnSe/GaAs:Se.The PL measurement showed that the ZnSe/GaAs:Se epilayer was obviously converted intothe p-type and its origin of the Id1 is related to VZn.The acceptor-impurity binding energy of the Id1was estimated to be 268meV. The Id1 was related tothe Zn-site replaced by the residual Cu-impurity.The origin of the SA emission may be associatedwith a complex donor like a (VSe�VZn)�VZn.
Acknowledgements
This study was financed by research grants fromChosun University, 2001.
References
[1] J. Nishizawa, R. Suzuki, Y. Okuno, J. Appl. Phys. 59
(1986) 2256.
[2] R.M. Park, M.B. Troffer, C.M. Rouleau, J.M. Depuydt,
M.A. Haase, Appl. Phys. Lett. 57 (1990) 2127.
[3] M.A. Haase, J. Qiu, J.M. Depuydt, H. Cheng, J. Appl.
Lett. 59 (1991) 1272.
[4] H. Jeon, J. Ding, W. Patterson, A.V. Nurmikko, W. Xie,
D.C. Grillo, M. Kobayshi, R.N. Gunshor, Appl. Phys.
Lett. 59 (1991) 3619.
[5] S. Guha, H. Munekata, F.K. LeGoues, L.L. Chang, Appl.
Phys. Lett. 60 (1992) 3220.
[6] L.H. Kuo, L. Salamanca-Riba, J.M. Depuydt, H. Cheng,
J. Qui, Appl. Phys. Lett. 63 (1993) 3197.
[7] M. Drechster, B.K. Meyer, D.M. Hofmann, P. Ruppert,
D. Hommel, Appl. Phys. Lett. 71 (1997) 1116.
[8] K.K. Fung, N. Wang, I.K. Sou, Appl. Phys. Lett. 71
(1997) 1225.
[9] R.W. Jansen, O.F. Sankey, Solid State Commun. 64 (1987)
197.
[10] M. Karai, K. Kido, H. Nait, K. Kurosawa, M. Okuda, T.
Fujino, M. Kitagawa, Phys. Status Solidi A 117 (1990) 15.
[11] H. Yang, A. Ishida, H. Fujiyasu, H. Kuwabara, J. Appl.
Phys. 65 (1989) 2838.
[12] A. Lopez-Otero, Thin Solid Films 49 (1987) 3.
[13] Sz. Fujita, Sg. Fujita, J. Crystal Growth 117 (1992) 67.
[14] M. Migita, A. Taike, M. Shiik, H. Yamamoto, J. Crystal
Growth 101 (1990) 835.
[15] K. Akimoto, T. Miyajima, Y. Mori, Phys. Rev. B 39 (1989)
3138.
[16] D.D. Sell, S.E. Stokowski, R. Dingel, J.V. Dilorenzo,
Phys. Rev. B 7 (1973) 4568.
[17] H. Cheng, J.M. Depuydt, J.E. Potts, T.L. Smith, Appl.
Phys. Lett. 52 (1988) 148.
[18] K. Hingerl, H. Sitter, D.J. As, W. Rothemund, J. Crystal
Growth 101 (1990) 180.
[19] T. Yao, J. Crystal Growth 72 (1985) 31.
[20] Y. Shirakawa, H. Kukimoto, J. Appl. Phys. 51 (1980)
2014.
[21] J.L. Merz, H. Kukimoto, K. Nassau, J.W. Shiever, Phys.
Rev. B 6 (1972) 545.
[22] R.R. Sharma, S. Rodriguez, Phys. Rev. 159 (1967) 649.
[23] O. Madelung, in: Landolt-Br .onstein (Ed.), Numerical
Data and Functional Relationships in Science and
Technology, New Series, Group III, Vol. 17b, Springer,
Berlin, 1982.
[24] K. Shahzad, D.J. Olega, D.A. Cammaek, Phys. Rev. B 39
(1989) 13016.
[25] S. Myhailinko, J.L. Batsone, H.J. Hutchinson, J.W. Steel,
J. Phys. C 17 (1984) 6477.
[26] M. Isshiki, T. Kyotani, K. Masumoto, W. Uchida, S. Suto,
Phys. Rev. B 36 (1987) 2568.
[27] K. Mochizuki, K. Masumoto, Jpn. J. Appl. Phys. 27
(1988) 1669.
[28] K. Mochizuki, K. Masumoto, T. Yasuda, Y. Sgawa, K.
Kimoto, J. Crystal Growth 135 (1994) 318.
[29] Yu.V. Korestelin, V.I. Kozlovsky, A.S. Nasibov, P.V.
Shapkin, J. Crystal Growth 159 (1996) 181.
[30] T. Yao, Y. Makita, S. Maekawa, Jpn. J. Appl. Phys. 20
(1981) L741.
[31] R. Bhargava, J. Crystal Growth 59 (1982) 15.
[32] K. Morimoto, J. Appl. Phys. 64 (1988) 4951.
K.J. Hong et al. / Journal of Crystal Growth 240 (2002) 57–63 63