- Email: [email protected]
Lateral inhomogeneities in engineered Schottky barriers
Journal of Crystal Growth 201/202 (1999) 795}799
Lateral inhomogeneities in engineered Schottky barriers S. Heun *, T. Schmidt , J. Slezak , J. DmH az , K.C. Prince , B.H. MuK ller , A. Franciosi Sincrotrone Trieste, Basovizza, I-34012 Trieste, Italy Laboratorio Nazionale TASC-INFM, Area di Ricerca, Padriciano 99, I-34012 Trieste, Italy
Abstract We have measured the lateral homogeneity in Al/GaAs(0 0 1) Schottky barriers engineered for low barrier height through fabrication of a Si layer at the interface under As #ux. We used a spectroscopic photo emission and low-energy electron microscope (SPELEEM) which combines a low-energy electron microscope with an imaging band pass "lter to allow spatially resolved synchrotron radiation photoemission experiments. As-grown samples with a Si interlayer thickness of 0.5 and 3 monolayers appeared homogeneous within the spatial resolution of the SPELEEM (22 nm). Annealing an Al/Si(As)/GaAs(0 0 1) heterostructure for 10 min at 5003C, however, caused inhomogeneous As outdi!usion which is correlated with a local As 3d core level shift of 0.3 eV. We suggest that the reported degradation of such engineered Schottky barriers might be correlated with laterally inhomogeneous As out-di!usion upon annealing. 1999 Elsevier Science B.V. All rights reserved. PACS: 07.85.T; 68.35.C; 73.30 Keywords: Schottky barriers; Interface engineering; Spectromicroscopy; Photoemission spectroscopy
1. Introduction The band alignment across heterostructures determines carrier injection and con"nement in electronic and optoelectronic devices, and several
* Corresponding author. Tel.: #39-40-375-8685; fax: #3940-375-8565; e-mail: [email protected]. Present address: Institut fuK r Halbleitertechnologie, UniversitaK t Hannover, Appelstr. 11A, 30167 Hannover, Germany. Also with Department of Chemical Engineering and Materials Science, University of Minesota, Minneapolis, MN 55455, USA, and Dipartimento di Fisica, Universita` di Trieste, I-34127 Trieste, Italy.
authors have attempted to tune such parameters through local chemical or structural modi"cations of the interface [1]. For example, large changes in the Schottky barrier height have been observed in Al/n-GaAs(0 0 1) diodes following the fabrication of Si interface layers two monolayers thick under an excess #ux of group V or group III atoms. Schottky barriers as low as 0.2 eV and as high as 1.2 eV have been observed in the two cases [2]. The application of such heterostructures in practical devices hinge on the lateral homogeneity of the engineered interfaces. Surprisingly large lateral inhomogeneities in the band alignment have been observed in some metal}semiconductor junctions
0022-0248/99/$ } see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 0 2 4 8 ( 9 8 ) 0 1 4 7 2 - 9
796
S. Heun et al. / Journal of Crystal Growth 201/202 (1999) 795}799
as well as semiconductor heterojunctions [3] by photoemission spectromicroscopy, and might play an enhanced role in engineered interfaces containing highly strained interlayers. A second important issue for its technological implications relates to the stability of engineered interfaces. It was recently shown that diodes engineered for a low barrier degrade gradually above 2503C. The nature of this early degradation mechanism remains to be ascertained [4]. We measured the lateral homogeneity of Al/Si(As)/GaAs(0 0 1) Schottky barriers using the spectroscopic photo emission and low-energy electron microscope (SPELEEM) of the Technical University of Clausthal at the synchrotron light source ELETTRA [5]. The instrument combines a lowenergy electron microscope (LEEM) with an imaging band pass "lter, and has been operating successfully since the end of 1996 at the undulator beamline 6.2 at ELETTRA. The high brightness of the ELETTRA light source together with an optimized instrument setting result in a spatial resolution of 22 nm and an energy resolution better than 0.5 eV in X-ray photoemission electron microscope (XPEEM) operation mode. This is the highest lateral resolution ever reported for a spectromicroscope of this kind [6]. The SPELEEM is therefore ideally suited to the search for compositional and structural inhomogeneities which correlate with inhomogeneities in the electronic parameters (Schottky barriers).
2. Experimental details All epilayers were grown by solid-source MBE at the TASC-INFM facility [7]. In detail, n-GaAs bu!er layers 500 nm thick, doped with Si at 3;10 cm\, were initially grown at 6003C on n>-GaAs(0 0 1) wafers (N "3;10 cm\) using 1 the method described previously [7]. After growth of the bu!er, the samples were cooled to 3003C. Si epilayers 0.5 monolayers (ML) or 3 ML thick were subsequently grown under an As equivalent beam pressure of 4;10\ Torr. The typical Si growth rate employed was 20 As /h. The Si "lm thickness was calibrated as described in Ref. [7].
The heterostructures were protected by an amorphous As cap layer during transfer in air from the MBE facility to the beamline, where thermal desorption of the cap layer was performed. After this treatment, the sample showed a 1;1 LEED pattern. We veri"ed that similar treatments followed by in situ Al deposition yielded the same low Schottky barrier as that observed in junctions fabricated by sequential deposition of GaAs, Si (under As #ux), and Al without any capping/decapping cycle. In the same vacuum chamber, 13 As of Al were deposited in situ on the substrate kept at room temperature. The Al "lm thickness was determined from the attenuation of the As 3d and Ga 3d core levels, assuming a mean free path of the photoelectrons of 5 As at a photon energy of 156.1 eV [8]. To study thermal degradation of the engineered barriers, the samples were also annealed at 5003C for 10 min. The annealing temperature was measured with a thermocouple calibrated by means of an infrared pyrometer. During SPELEEM operation, secondary electrons or photoelectrons from the surface are accelerated to 18 keV, and pass through an electron microscope column and energy analyzer before striking a #ourescent screen to produce an image, which is acquired by a CCD camera [9]. All measurements discussed in this paper were performed with a photon energy of 156.1 eV. The spatial resolution of the SPELEEM microscope was found to be 22 nm in photoemission, while a 8 nm resolution has been achieved in low-energy electron microscopy [6]. The measured energy resolution of the instrument is better than 0.5 eV [6].
3. Results and discussion After decapping the 3 ML Si(As)/GaAs heterostructure in the microscope preparation chamber, the relevant core levels (Si 2p, As 3d, and Ga 3d) and the valence band were measured in XPEEM. The sample appeared to be very homogeneous both in LEEM and XPEEM studies. Also after 13 As of Al had been deposited at room temperature, the sample surface still appeared very homogeneous in LEEM and XPEEM. Fig. 1 shows a XPEEM image
S. Heun et al. / Journal of Crystal Growth 201/202 (1999) 795}799
Fig. 1. XPEEM image of an Al/n-GaAs junction incorporating a 3 ML-thick Si interface layer grown under As #ux. The Al overlayer was 13 As -thick. The image was recorded at a photon energy of 156.1 eV and a photoelectron kinetic energy of 138.4 eV corresponding to the Ga 3d peak. The "eld of view is 19 lm. The integration time for the image was 5 min. The sample appears to be very homogeneous, no inhomogeneities can be observed in the image. The bright spot in the upper part of the image is an artifact of the measurement and caused by a defect in the channelplate.
of the sample after Al deposition, recorded at a photoelectron kinetic energy of 138.4 eV, and corresponding to the peak intensity of the Ga 3d core emission. The "eld of view is 19 lm. Typical intensity variations from point to point in Fig. 1 are 10%, consistent with the signal to noise ratio observed by monitoring the same sample location as a function of time, and yielding an upper limit of 0.3 eV on the detectable Schottky barrier inhomogeneities from a single image. This upper limit can be decreased measuring a whole series of images at di!erent kinetic energies and "tting the so obtained spectra. Applying this method we calculated the lateral homogeneity of our Schottky barriers to be better than $0.10 eV. No lateral inhomogeneities were observed in the images obtained using the Si 2p, Al 2p, and As 3d core levels (not shown), as well. Analogous results were ob-
797
tained for the 0.5 ML Si(As)/GaAs heterostructure before and after RT-deposition of 13 As of Al. The absence of sample inhomogeneities within experimental uncertainty even for submonolayer Si coverages suggests that the scale of any lateral inhomogeneity in the Si distribution is small as compared to the experimental resolution. Assuming a homogeneous Schottky barrier height, the observed small #uctuations in core level intensities in Fig. 1 would be consistent with a maximum change in Si coverage of $0.2 ML. The results of annealing the 13 As Al/3 ML Si(As)/GaAs(001) heterostructure at 5003C for 10 min are shown in Fig. 2. An annealing of the same type of sample above 2503C reportedly yields a marked increase in the barrier height [4]. In fact, we observed sample inhomogeneities (see Fig. 2) as a result of annealing. We measured several images showing a domain (region A in Fig. 2) which appears bright at the energy of the As 3d core level, dark at the energy of the Al 2p core level, and invisible for other energies like the Ga 3d and Si 2p core level (not shown). This can be understood as an inhomogeneous As out-di!usion. Fig. 2 also shows spectra of the As 3d and Al 2p core levels which were taken from the regions indicated in the images in Fig. 2. The region A corresponds to the domain where As out-di!usion was observed, while region B corresponds to the homogeneous background. The spectra were obtained by areal integration in a series of images measured from the same region of the sample but at di!erent kinetic energies of the photoelectrons. In the region of the As outdi!usion, the As 3d core level shifts by 0.3 eV to higher kinetic energies, while the Al 2p core level remains energetically invariant. If we assume that the inhomogeneity observed in Fig. 2 is an early stage of the type of interdi!usion phenomena that would lead to barrier degradation, we can draw conclusions about the microscopic degradation mechanism. Several mechanism We caution the reader that the inhomogeneity in Fig. 2 still occupies a relatively small fraction of the junction area, and would not yet be detectable in spatially integrated photoemission and current}voltage studies. At this stage we have insu$cient data to unambiguosly identify a local Schottky barrier height variation corresponding to the inhomogeneity in Fig. 2.
798
S. Heun et al. / Journal of Crystal Growth 201/202 (1999) 795}799
Fig. 2. Left: XPEEM images from an Al/Si(As)/GaAs heterostructure annealed at 5003C for 10 min. The region A is bright at the energy of the As 3d core level, indicating preferential As out-di!usion, while it appears darker at the energy of the Al 2p core level. The image was recorded at a photon energy of 156.1 eV and a photoelectron kinetic energy of 113.4 and 79.9 eV, corresponding to the As 3d peak and Al 2p peak, respectively. The "eld of view is 5 lm;3 lm. The integration time for each image was 3 min. Right: Spectra of the As 3d and Al 2p core levels, taken from the regions indicated in the images. In the region of As out-di!usion, the As 3d core level shifts to higher kinetic energies, while the As 2p core level remains energetically invariant.
could in principle explain the reduced thermal stability of engineered Schottky barriers. Di!usion of Si into the Al overlayer is one possible mechanism. However, our experiment does not provide any evidence for an inhomogeneous out-di!usion of Si into the Al overlayer. The formation of an AlAs interface phase could also in principle yield a local increase in the Schottky barrier height. Although we observe As out-di!usion into the Al overlayer, we can exclude the formation of AlAs domains, because it would result in a shift of the Al 2p core level by more than 1 eV [10]. Instead, our data suggest a temperature-induced As depletion of the Si(As)/GaAs interface. Annealing of GaAs surfaces in vacuum is known to yield increasingly Ga-rich surfaces as a result of As desorption [11,12]. Also, Si interface interlayers de-
posited without an excess As #ux are known to yield a Schottky barrier similar to that observed in Al/GaAs junctions in the absence of any Si control layers. An As depletion of the interface is therefore likely to yield a local barrier which increases toward the &natural' Al/GaAs value.
Acknowledgements Discussions with E. Bauer and L. Sorba are greatfully acknowleged. This work was supported in part by INFM under the TUSBAR advanced research project and by NSF under Grant no. DMR-9525758. We thank the Technical University Clausthal for the loan of the SPELEEM.
S. Heun et al. / Journal of Crystal Growth 201/202 (1999) 795}799
References [1] A. Franciosi, C.G. Van de Walle, Surf. Sci. Rep. 25 (1996) 1. [2] M. Cantile, L. Sorba, S. Yildirim, P. Faraci, G. Biasiol, A. Franciosi, T.J. Miller, M.I. Nathan, Appl. Phys. Lett. 64 (1994) 988. [3] F. Gozzo, H. Berger, I.R. Collins, G. Margaritondo, W. Ng, A.K. Ray-Chaudhuri, S. Liang, S. Singh, F. Cerrina, Phys. Rev. B 51 (1995) 5024. [4] L. Sorba, S. Yildirim, M. Lazzarino, A. Franciosi, D. Chiola, F. Beltram, Appl. Phys. Lett. 69 (1996) 1927. [5] L.H. Veneklasen, Ultramicroscopy 36 (1991) 76.
799
[6] Th. Schmidt, S. Heun, J. Slezak, J. Diaz, K.C. Prince, G. Lilienkamp, E. Bauer, Surf. Rev. Lett. 5 (1998) 1287. [7] L. Sorba, G. Bratina, A. Antonini, A. Franciosi, L. Tapfer, A. Migliori, P. Merli, Phys. Rev. B 46 (1992) 6834. [8] M.P. Seah, W.A. Dench, Surf. Interface Anal. 1 (1979) 2. [9] E. Bauer, C. Koziol, G. Lilienkamp, T. Schmidt, J. Electron Spectrosc. Relat. Phenom. 84 (1997) 201. [10] L. Sorba, J.J. Paggel, A. Franciosi, J. Vac. Sci. Technol. B 14 (1996) 2994. [11] W. Chen, M. Dumas, D. Mao, A. Kahn, J. Vac. Sci. Technol. B 10 (1992) 1886. [12] D.K. Biegelsen, R.D. Bringans, J.E. Northrup, L.-E. Swartz, Phys. Rev. B 41 (1990) 5701.
Lateral inhomogeneities in engineered Schottky barriers S. Heun *, T. Schmidt , J. Slezak , J. DmH az , K.C. Prince , B.H. MuK ller , A. Franciosi Sincrotrone Trieste, Basovizza, I-34012 Trieste, Italy Laboratorio Nazionale TASC-INFM, Area di Ricerca, Padriciano 99, I-34012 Trieste, Italy
Abstract We have measured the lateral homogeneity in Al/GaAs(0 0 1) Schottky barriers engineered for low barrier height through fabrication of a Si layer at the interface under As #ux. We used a spectroscopic photo emission and low-energy electron microscope (SPELEEM) which combines a low-energy electron microscope with an imaging band pass "lter to allow spatially resolved synchrotron radiation photoemission experiments. As-grown samples with a Si interlayer thickness of 0.5 and 3 monolayers appeared homogeneous within the spatial resolution of the SPELEEM (22 nm). Annealing an Al/Si(As)/GaAs(0 0 1) heterostructure for 10 min at 5003C, however, caused inhomogeneous As outdi!usion which is correlated with a local As 3d core level shift of 0.3 eV. We suggest that the reported degradation of such engineered Schottky barriers might be correlated with laterally inhomogeneous As out-di!usion upon annealing. 1999 Elsevier Science B.V. All rights reserved. PACS: 07.85.T; 68.35.C; 73.30 Keywords: Schottky barriers; Interface engineering; Spectromicroscopy; Photoemission spectroscopy
1. Introduction The band alignment across heterostructures determines carrier injection and con"nement in electronic and optoelectronic devices, and several
* Corresponding author. Tel.: #39-40-375-8685; fax: #3940-375-8565; e-mail: [email protected]. Present address: Institut fuK r Halbleitertechnologie, UniversitaK t Hannover, Appelstr. 11A, 30167 Hannover, Germany. Also with Department of Chemical Engineering and Materials Science, University of Minesota, Minneapolis, MN 55455, USA, and Dipartimento di Fisica, Universita` di Trieste, I-34127 Trieste, Italy.
authors have attempted to tune such parameters through local chemical or structural modi"cations of the interface [1]. For example, large changes in the Schottky barrier height have been observed in Al/n-GaAs(0 0 1) diodes following the fabrication of Si interface layers two monolayers thick under an excess #ux of group V or group III atoms. Schottky barriers as low as 0.2 eV and as high as 1.2 eV have been observed in the two cases [2]. The application of such heterostructures in practical devices hinge on the lateral homogeneity of the engineered interfaces. Surprisingly large lateral inhomogeneities in the band alignment have been observed in some metal}semiconductor junctions
0022-0248/99/$ } see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 0 2 4 8 ( 9 8 ) 0 1 4 7 2 - 9
796
S. Heun et al. / Journal of Crystal Growth 201/202 (1999) 795}799
as well as semiconductor heterojunctions [3] by photoemission spectromicroscopy, and might play an enhanced role in engineered interfaces containing highly strained interlayers. A second important issue for its technological implications relates to the stability of engineered interfaces. It was recently shown that diodes engineered for a low barrier degrade gradually above 2503C. The nature of this early degradation mechanism remains to be ascertained [4]. We measured the lateral homogeneity of Al/Si(As)/GaAs(0 0 1) Schottky barriers using the spectroscopic photo emission and low-energy electron microscope (SPELEEM) of the Technical University of Clausthal at the synchrotron light source ELETTRA [5]. The instrument combines a lowenergy electron microscope (LEEM) with an imaging band pass "lter, and has been operating successfully since the end of 1996 at the undulator beamline 6.2 at ELETTRA. The high brightness of the ELETTRA light source together with an optimized instrument setting result in a spatial resolution of 22 nm and an energy resolution better than 0.5 eV in X-ray photoemission electron microscope (XPEEM) operation mode. This is the highest lateral resolution ever reported for a spectromicroscope of this kind [6]. The SPELEEM is therefore ideally suited to the search for compositional and structural inhomogeneities which correlate with inhomogeneities in the electronic parameters (Schottky barriers).
2. Experimental details All epilayers were grown by solid-source MBE at the TASC-INFM facility [7]. In detail, n-GaAs bu!er layers 500 nm thick, doped with Si at 3;10 cm\, were initially grown at 6003C on n>-GaAs(0 0 1) wafers (N "3;10 cm\) using 1 the method described previously [7]. After growth of the bu!er, the samples were cooled to 3003C. Si epilayers 0.5 monolayers (ML) or 3 ML thick were subsequently grown under an As equivalent beam pressure of 4;10\ Torr. The typical Si growth rate employed was 20 As /h. The Si "lm thickness was calibrated as described in Ref. [7].
The heterostructures were protected by an amorphous As cap layer during transfer in air from the MBE facility to the beamline, where thermal desorption of the cap layer was performed. After this treatment, the sample showed a 1;1 LEED pattern. We veri"ed that similar treatments followed by in situ Al deposition yielded the same low Schottky barrier as that observed in junctions fabricated by sequential deposition of GaAs, Si (under As #ux), and Al without any capping/decapping cycle. In the same vacuum chamber, 13 As of Al were deposited in situ on the substrate kept at room temperature. The Al "lm thickness was determined from the attenuation of the As 3d and Ga 3d core levels, assuming a mean free path of the photoelectrons of 5 As at a photon energy of 156.1 eV [8]. To study thermal degradation of the engineered barriers, the samples were also annealed at 5003C for 10 min. The annealing temperature was measured with a thermocouple calibrated by means of an infrared pyrometer. During SPELEEM operation, secondary electrons or photoelectrons from the surface are accelerated to 18 keV, and pass through an electron microscope column and energy analyzer before striking a #ourescent screen to produce an image, which is acquired by a CCD camera [9]. All measurements discussed in this paper were performed with a photon energy of 156.1 eV. The spatial resolution of the SPELEEM microscope was found to be 22 nm in photoemission, while a 8 nm resolution has been achieved in low-energy electron microscopy [6]. The measured energy resolution of the instrument is better than 0.5 eV [6].
3. Results and discussion After decapping the 3 ML Si(As)/GaAs heterostructure in the microscope preparation chamber, the relevant core levels (Si 2p, As 3d, and Ga 3d) and the valence band were measured in XPEEM. The sample appeared to be very homogeneous both in LEEM and XPEEM studies. Also after 13 As of Al had been deposited at room temperature, the sample surface still appeared very homogeneous in LEEM and XPEEM. Fig. 1 shows a XPEEM image
S. Heun et al. / Journal of Crystal Growth 201/202 (1999) 795}799
Fig. 1. XPEEM image of an Al/n-GaAs junction incorporating a 3 ML-thick Si interface layer grown under As #ux. The Al overlayer was 13 As -thick. The image was recorded at a photon energy of 156.1 eV and a photoelectron kinetic energy of 138.4 eV corresponding to the Ga 3d peak. The "eld of view is 19 lm. The integration time for the image was 5 min. The sample appears to be very homogeneous, no inhomogeneities can be observed in the image. The bright spot in the upper part of the image is an artifact of the measurement and caused by a defect in the channelplate.
of the sample after Al deposition, recorded at a photoelectron kinetic energy of 138.4 eV, and corresponding to the peak intensity of the Ga 3d core emission. The "eld of view is 19 lm. Typical intensity variations from point to point in Fig. 1 are 10%, consistent with the signal to noise ratio observed by monitoring the same sample location as a function of time, and yielding an upper limit of 0.3 eV on the detectable Schottky barrier inhomogeneities from a single image. This upper limit can be decreased measuring a whole series of images at di!erent kinetic energies and "tting the so obtained spectra. Applying this method we calculated the lateral homogeneity of our Schottky barriers to be better than $0.10 eV. No lateral inhomogeneities were observed in the images obtained using the Si 2p, Al 2p, and As 3d core levels (not shown), as well. Analogous results were ob-
797
tained for the 0.5 ML Si(As)/GaAs heterostructure before and after RT-deposition of 13 As of Al. The absence of sample inhomogeneities within experimental uncertainty even for submonolayer Si coverages suggests that the scale of any lateral inhomogeneity in the Si distribution is small as compared to the experimental resolution. Assuming a homogeneous Schottky barrier height, the observed small #uctuations in core level intensities in Fig. 1 would be consistent with a maximum change in Si coverage of $0.2 ML. The results of annealing the 13 As Al/3 ML Si(As)/GaAs(001) heterostructure at 5003C for 10 min are shown in Fig. 2. An annealing of the same type of sample above 2503C reportedly yields a marked increase in the barrier height [4]. In fact, we observed sample inhomogeneities (see Fig. 2) as a result of annealing. We measured several images showing a domain (region A in Fig. 2) which appears bright at the energy of the As 3d core level, dark at the energy of the Al 2p core level, and invisible for other energies like the Ga 3d and Si 2p core level (not shown). This can be understood as an inhomogeneous As out-di!usion. Fig. 2 also shows spectra of the As 3d and Al 2p core levels which were taken from the regions indicated in the images in Fig. 2. The region A corresponds to the domain where As out-di!usion was observed, while region B corresponds to the homogeneous background. The spectra were obtained by areal integration in a series of images measured from the same region of the sample but at di!erent kinetic energies of the photoelectrons. In the region of the As outdi!usion, the As 3d core level shifts by 0.3 eV to higher kinetic energies, while the Al 2p core level remains energetically invariant. If we assume that the inhomogeneity observed in Fig. 2 is an early stage of the type of interdi!usion phenomena that would lead to barrier degradation, we can draw conclusions about the microscopic degradation mechanism. Several mechanism We caution the reader that the inhomogeneity in Fig. 2 still occupies a relatively small fraction of the junction area, and would not yet be detectable in spatially integrated photoemission and current}voltage studies. At this stage we have insu$cient data to unambiguosly identify a local Schottky barrier height variation corresponding to the inhomogeneity in Fig. 2.
798
S. Heun et al. / Journal of Crystal Growth 201/202 (1999) 795}799
Fig. 2. Left: XPEEM images from an Al/Si(As)/GaAs heterostructure annealed at 5003C for 10 min. The region A is bright at the energy of the As 3d core level, indicating preferential As out-di!usion, while it appears darker at the energy of the Al 2p core level. The image was recorded at a photon energy of 156.1 eV and a photoelectron kinetic energy of 113.4 and 79.9 eV, corresponding to the As 3d peak and Al 2p peak, respectively. The "eld of view is 5 lm;3 lm. The integration time for each image was 3 min. Right: Spectra of the As 3d and Al 2p core levels, taken from the regions indicated in the images. In the region of As out-di!usion, the As 3d core level shifts to higher kinetic energies, while the As 2p core level remains energetically invariant.
could in principle explain the reduced thermal stability of engineered Schottky barriers. Di!usion of Si into the Al overlayer is one possible mechanism. However, our experiment does not provide any evidence for an inhomogeneous out-di!usion of Si into the Al overlayer. The formation of an AlAs interface phase could also in principle yield a local increase in the Schottky barrier height. Although we observe As out-di!usion into the Al overlayer, we can exclude the formation of AlAs domains, because it would result in a shift of the Al 2p core level by more than 1 eV [10]. Instead, our data suggest a temperature-induced As depletion of the Si(As)/GaAs interface. Annealing of GaAs surfaces in vacuum is known to yield increasingly Ga-rich surfaces as a result of As desorption [11,12]. Also, Si interface interlayers de-
posited without an excess As #ux are known to yield a Schottky barrier similar to that observed in Al/GaAs junctions in the absence of any Si control layers. An As depletion of the interface is therefore likely to yield a local barrier which increases toward the &natural' Al/GaAs value.
Acknowledgements Discussions with E. Bauer and L. Sorba are greatfully acknowleged. This work was supported in part by INFM under the TUSBAR advanced research project and by NSF under Grant no. DMR-9525758. We thank the Technical University Clausthal for the loan of the SPELEEM.
S. Heun et al. / Journal of Crystal Growth 201/202 (1999) 795}799
References [1] A. Franciosi, C.G. Van de Walle, Surf. Sci. Rep. 25 (1996) 1. [2] M. Cantile, L. Sorba, S. Yildirim, P. Faraci, G. Biasiol, A. Franciosi, T.J. Miller, M.I. Nathan, Appl. Phys. Lett. 64 (1994) 988. [3] F. Gozzo, H. Berger, I.R. Collins, G. Margaritondo, W. Ng, A.K. Ray-Chaudhuri, S. Liang, S. Singh, F. Cerrina, Phys. Rev. B 51 (1995) 5024. [4] L. Sorba, S. Yildirim, M. Lazzarino, A. Franciosi, D. Chiola, F. Beltram, Appl. Phys. Lett. 69 (1996) 1927. [5] L.H. Veneklasen, Ultramicroscopy 36 (1991) 76.
799
[6] Th. Schmidt, S. Heun, J. Slezak, J. Diaz, K.C. Prince, G. Lilienkamp, E. Bauer, Surf. Rev. Lett. 5 (1998) 1287. [7] L. Sorba, G. Bratina, A. Antonini, A. Franciosi, L. Tapfer, A. Migliori, P. Merli, Phys. Rev. B 46 (1992) 6834. [8] M.P. Seah, W.A. Dench, Surf. Interface Anal. 1 (1979) 2. [9] E. Bauer, C. Koziol, G. Lilienkamp, T. Schmidt, J. Electron Spectrosc. Relat. Phenom. 84 (1997) 201. [10] L. Sorba, J.J. Paggel, A. Franciosi, J. Vac. Sci. Technol. B 14 (1996) 2994. [11] W. Chen, M. Dumas, D. Mao, A. Kahn, J. Vac. Sci. Technol. B 10 (1992) 1886. [12] D.K. Biegelsen, R.D. Bringans, J.E. Northrup, L.-E. Swartz, Phys. Rev. B 41 (1990) 5701.