Properties of carbon-doped low-temperature GaAs and InP grown

Preparing to load PDF file. please wait...

0 of 0
100%
Properties of carbon-doped low-temperature GaAs and InP grown

Transcript Of Properties of carbon-doped low-temperature GaAs and InP grown

Properties of carbon-doped low-temperature GaAs and InP grown by solid-source molecular-beam epitaxy using CBr4
W. K. Liua) and D. I. Lubyshev Quantum Epitaxial Designs, Inc., Bethlehem, Pennsylvania 18015
P. Specht, R. Zhao, and E. R. Weber Department of Materials Science and Mineral Engineering, University of California, Berkeley, California 94720
J. Gebauer Fachbereich Physik, Martin-Luther-Universita¨t, Halle-Wittenberg, 06099 Halle, Germany
A. J. SpringThorpe and R. W. Streater Nortel Technology, Nortel Advanced Components, Ottawa, Ontario, Canada K2H 8E9
S. Vijarnwannaluk, W. Songprakob, and R. Zallen Department of Physics, Virginia Tech, Blacksburg, Virginia 24061-0435
͑Received 10 October 1999; accepted 27 January 2000͒
Carbon-doped GaAs and InP grown at low temperatures by molecular-beam epitaxy contain a high concentration of antisite defects which gives rise to ultrafast carrier trapping time and desirable radiation-hard properties. The use of CBr4 as the dopant source introduced significant bromine incorporation during low-temperature ͑LT͒ growth. Incomplete dissociation of the CBr4 molecules gives rise to the formation of C–Br complexes and results in a reduction of electrically active carbon concentration. In this work, we present our studies on the incorporation mechanism of C and Br in LT-GaAs and report on the effect of carbon and bromine incorporation on carrier lifetime and concentration of arsenic antisite defects. Preliminary results on LT-InP:C characterization are also presented. © 2000 American Vacuum Society. ͓S0734-211X͑00͒01803-5͔

I. INTRODUCTION
GaAs layers grown by molecular beam epitaxy ͑MBE͒ at low temperature ͑LT͒ GaAs have been extensively studied because of their ultrafast carrier trapping time ͑Ͻ1 ps͒ and, upon annealing, very high resistivity (ϳ107 ⍀ cmϪ2).1–4 These properties are attractive for applications in radiationhard electronics and ultrafast photodetection. Carrier trapping time and resistivity of the LT GaAs are both governed by the concentration of arsenic antisite defects ͑AsGa).5 Thermal annealing during integrated circuit ͑IC͒ fabrication can cause reduction in AsGa concentration and give rise to out-diffusion of defects into the active area of devices. Be doping in LT-GaAs ͑LT-GaAs:Be͒ has been studied by a number of groups6–8 and was recently found to thermally stabilize AsGa antisite defects.7,8 Earlier, we found that the replacement of diffusive Be dopants by C from a CBr4 source can also improve the thermal stability of LT-GaAs.9 Compared to Be doping, C doping in LT-GaAs is far more complicated due to significant incorporation of Br at low substrate temperatures.
In this work, we report on C and Br incorporation in LT-GaAs using CBr4 source. Properties of C-doped LT-InP, a potential alternative for III–V semiconductor-based radiation-hard applications, were also studied. Secondary ion mass spectroscopy ͑SIMS͒ and transmission electron spectroscopy ͑TEM͒ were used to investigate the incorporation mechanism of C and Br in LT-GaAs. The effect of carbon
a͒Electronic mail: [email protected]

doping on carrier lifetime and concentration of antisite defects were studied by time-resolved reflectivity transients ͑TRRT͒ measurement, local vibration mode ͑LVM͒ spectroscopy, near infrared absorption ͑NIRA͒, and magnetic circular dichroism absorption ͑MCDA͒ measurements.
II. EXPERIMENT
All samples discussed in this work were grown in a solidsource Varian GENII MBE system equipped with a customdesigned gas injector for CBr4. The experimental setup for C doping has been described elsewhere.9 Three-inch diameter LT-GaAs͑100͒ and LT-InP͑100͒ samples were grown using elemental Group-III metal sources and As4 and P2 from valved crackers. Substrate temperature was calibrated using diffuse reflectance spectroscopy.9 The As-to-Ga beam equivalent pressure ͑BEP͒ ratio was kept constant at a value of 20 and the P-to-In BEP ratio was fixed at 18. Typically, 0.5 ␮m thick LT-GaAs and LT-InP layers were deposited at a growth rate of 0.9 and 0.5 ␮m/h, respectively. Annealing of the LT samples was done in proximity with a sacrificial wafer at 600 or 700 °C in nitrogen ambient for 30 min.
The LT samples were characterized using various techniques. The structural properties of as-grown and annealed layers were studied by cross-sectional TEM using a JEOL 4000FX microscope operated at 400 kV. Lattice mismatch parallel to the growth direction (⌬c/c) was measured by a five-crystal Philips high-resolution x-ray diffraction system ͑HRXRD͒ using the ͑004͒ reflection. Electrical properties were determined by Hall-effect measurement with van der

1594 J. Vac. Sci. Technol. B 18„3…, MayÕJun 2000 0734-211XÕ2000Õ18„3…Õ1594Õ4Õ$17.00 ©2000 American Vacuum Society 1594

1595

Liu et al.: Carbon-doped low-temperature GaAs and InP

1595

Pauw geometry at 300 K. C and Br concentrations were mea-
sured by SIMS. Infrared absorption measurements were per-
formed to study the localized vibrational modes of substitutional CAs in LT-GaAs:C. AsGa concentrations, ͓AsG0 a] and ͓AsGϩa], were obtained by NIRA and MCDA measurements performed at 1.8 K. The samples were polished down to Ͻ100 ␮m and mounted on an optical cryostat. The incident light beam was chosen to be weak enough to avoid any pos-
sible quenching of the arsenic antisites. The magnetic field
was set at 2 T for MCDA measurements. Carrier trapping times ͑CTT͒ were determined from TRRT recorded in a pump–probe laser pulse experiment.10 The laser was oper-
ated at a wavelength of 800 nm with 100 fs short pulses
repeated at a frequency of 1000 MHz. The recorded data taken in a time window of ϳ20 ps could be modeled by at most two exponential decays, providing two time constants.
The shorter time constant is influenced by the experimental
resolution, but may also contain carrier thermalization effects.4 CTT given in this article refer to the longer time
constant whenever two decays were fitted. CTT above 0.1 ps
can be reliably resolved.

III. RESULTS AND DISCUSSIONS
A. Chemical properties of LT GaAs:C and LT InP:C
Temperature staircase structures were grown for SIMS analysis to study the chemical properties of LT-GaAs:C and LT-InP:C. An ion-implanted reference Br standard was used to quantify the Br incorporation level as a function of growth temperatures and study their thermal stability. 1000-Å-thick C-doped LT layers were grown at successively lower temperatures, each separated by 1000 Å undoped spacer. The descending temperature sequence was chosen to eliminate any in situ annealing effect.
SIMS profile for an as-grown LT-GaAs:C temperature staircase sample doped nominally at 1ϫ1020 cmϪ3 is shown in Fig. 1͑a͒. Note that the C concentration ͓͑C͔͒ exhibits virtually no growth temperature dependence between 225 and 410 °C. The situation is very different for Br. For growth temperatures (Tg) above 410 °C, the Br signal is below the SIMS detection limit. This indicates complete dissociation of CBr4 molecules into C and Br atoms with the latter desorbed from the growing surface. At temperatures around 330 °C, however, significant Br incorporation was observed. To track the change in Br concentration more closely, a second staircase sample doped nominally at 6ϫ1019 cmϪ3 was grown with a narrower temperature range ͓Fig. 1͑b͔͒. ͓Br͔ rises steadily and the ͓Br͔-to-͓C͔ ratio increases from 0 to ϳ4 where it saturates. Note that the ͓Br͔-to-͓C͔ ratio of 4 corresponds to the stoichiometry of CBr4. The Signoid relationship ͑rather than Arrhenius͒ between the ͓Br͔-to-͓C͔ ratio and Tg suggests that the dissociation of CBr4 does not proceed via a first order reaction but may involve CBr radicals. The samples were then subjected to proximity annealing at 700 °C for 30 min. Both C and Br exhibit good thermal stability with less than 10% change in concentration and no broadening in the width of the doped layers upon annealing.

FIG. 1. SIMS profiles for two LT-GaAs:C temperature staircase structures showing the variation of ͓Br͔ with growth temperature: ͑a͒ Tg ϭ225– 410 °C, ͓C͔ϭ6ϫ10 19 cmϪ3 and ͑b͒ Tgϭ290– 370 °C, ͓C͔ϭ1 ϫ1020 cmϪ3.
Br incorporation was also detected in LT-InP:C samples grown at TgϽ300 °C. SIMS spectrum of LT-InP:C grown at Tgϭ260 °C reveal a Br concentration of 3 – 5ϫ1017 cmϪ3 for ͓C͔ ϭ5ϫ1019 cmϪ3. The similar behavior of Br incorporation in LT-GaAs:C and LT-InP:C suggest that this process is predominantly determined by the energetics of CBr4 dissociation.
B. Structural properties of LT GaAs
In order to understand the role of Br in the incorporation mechanism of C in LT-GaAs, experimental techniques sensitive to the location of impurities and point defects in crystal lattice were employed. Reflection high-energy electron diffraction patterns during MBE growth and HRXRD spectra and TEM images of as-grown samples all confirm the singlecrystal nature of C-doped LT materials. Upon annealing at high temperature, As precipitates were observed in LTGaAs:C, similar to that observed in undoped and Be-doped LT-GaAs.11,12 Based on these observations, we conclude that the growth mode of LT-GaAs:C is similar to their undoped and Be-doped counterparts, and was not changed by the incorporation of Br. The size of CBr4 molecules is significantly larger than the lattice volume of the constituent atoms in GaAs and InP. The incorporation of large amount of undis-

JVST B - Microelectronics and Nanometer Structures

1596

Liu et al.: Carbon-doped low-temperature GaAs and InP

1596

FIG. 2. Local vibration mode spectra from GaAs:C layers grown at 225– 330 °C with ͓C͔ϭ6ϫ1019 cmϪ3. Inset: Integral absorption intensity and percentage of ͓CAs] as a function of growth temperature.
sociated CBr4 molecules would inevitably result in the generation of dislocations. No evidence of dislocation, however, was observed by TEM. It would also be inconsistent with the lattice contraction observed in LT-GaAs:C.9 On the other hand, the covalent radius of Br ͑1.14 Å͒ is fairly close to that of Ga ͑1.26 Å͒ and As ͑1.20 Å͒. We, therefore, speculate that dissociation of CBr4 still occurs during LT growth but that both Br and C are incorporated into the lattice via the formation of C–Br complexes.
The concentration of C atoms at As sites ͓͑CAs͔͒ has been measured by LVM spectroscopy. The raw transmittance data is shown in Fig. 2. GaAs:C samples grown at both low and standard temperatures exhibit clearly resolved LVM bands near 580 cmϪ3, indicative of C atoms residing predominantly in the As substitutional sites. The LVM band in the LTGaAs:C samples is broadened, consistent with lattice disorder due to the presence of large variety of nonstoichiometric defects. The dependences of the integrated absorption intensity and the percentages of C in As substitutional sites as a function of growth temperature are presented in the inset of Fig. 2. The rapid decrease of CAs from 100 to Ͻ35% as Tg changes from 340 to 260 °C correlates well with the increase of ͓Br͔ in SIMS and supports our model of C–Br complex formation.
The neutral and singly charged arsenic antisite concentrations in LT-GaAs as a function of ͓C͔ were measured by NIRA and MCDA, respectively. The undoped LT-GaAs sample exhibits strong absorption below the bandedge, while strong free carrier absorption was observed in the 600 °C GaAs:C reference sample. According to Martin,13 ͓AsG0 a] in undoped LT-GaAs is typically 6.0ϫ1019 cmϪ3. For our LTGaAs:C sample, ͓AsG0 a] was found to decrease dramatically to below 2.0ϫ1019 cmϪ3 for ͓C͔Ͼ3.5ϫ1019 cmϪ3. The subbandgap ͑ϳ1.2 eV͒ absorption also decreases slightly as ͓C͔ increases. It has to be pointed out that a Br concentration about four times higher than ͓C͔ was detected in samples grown at Tgϭ240 °C and that the interaction between Br and AsGϩa, if any, is unknown. Contrary to Be-doped LT-GaAs,14
J. Vac. Sci. Technol. B, Vol. 18, No. 3, MayÕJun 2000

FIG. 3. MCDA spectra of as-grown LT-GaAs:C samples doped from 3.5 ϫ1019 to 1ϫ1020 cmϪ3. All samples are grown at 240 °C unless otherwise stated.
no AsGϩa absorption band in LT-GaAs:C is detected in the energy range near 0.94 eV from MCDA ͑see Fig. 3͒. The broad peak at 1.2 eV is related to AsG0 a. For comparison, the undoped sample grown at the same Tg contains ϳ2.2 ϫ1018 cmϪ3 AsGϩa due to the compensation by Ga vacancies (VGa). We, therefore, speculate that VGa is suppressed by the presence of Br, which may occupy Ga sites. Positron annihilation experiments are currently underway to determine the concentration of VGa in these LT-GaAs:C samples. C. Comparison of carrier trapping time in LT GaAs:C and LT InP:C
Analogous to LT-GaAs, CTT in LT-InP is governed by the antisite defect concentration ͓PIn͔. The potential of LT-
FIG. 4. Normalized TRRT spectra for as-grown LT-GaAs:C doped at 6.6 ϫ1019 cmϪ3 (Tgϭ215 °C) and as-grown undoped and C-doped InP grown at 460 and 260 °C.

1597

Liu et al.: Carbon-doped low-temperature GaAs and InP

TABLE I. Growth parameters and characterization results of LT-In:C samples.

Layer
InP InP:C LT-InP:C LT-InP

Tg ͑°C͒
460 460 260 260

͓C͔ ͑cmϪ3͒
undoped 1 ϫ 1019 1 ϫ 1019 undoped

͓ nHall͔ (cmϪ3͒
1.0e16 3.1e16 2.2e18 3.9e18

␮Hall ͑cm2/V s͒
3450 2244 565 984

⌬d/d ͑%͒
••• Ϫ0.0161 Ϫ0.0462
0.0136

1597

InP as buffer layers for radiation-hard applications depends on the concentration and stability of these defects. Here, we present some preliminary data on CTT in undoped and C-doped LT-InP and compared them to those in InP grown under standard MBE conditions ͑Tgϭ460 °C͒ and LTGaAs:C. TRRT spectra for GaAs and InP are shown in Fig. 4. HRXRD studies reveal lattice contraction in the InP:C lattice qualitatively similar to those observed in GaAs:C. Hall results indicate n-type conductivity for all InP:C samples as a result of C incorporation in In sites as donors and due to native point defects in LT-grown material. The growth parameters and characterization results are summarized in Table I.
Both undoped and C-doped LT-InP exhibit reduction in CTT compared to materials grown at standard temperatures. For undoped InP, CTT was found to decrease from ϳ200 ps at 460 °C to ϳ1.6 ps at 260 °C. For the C-doped layer, the corresponding change was from ϳ40 to ϳ9 ps. Compared to LT-GaAs, TRRT spectra of LT-InP are considerably more complicated and exhibit multiple decay mechanism. The CTT values extracted from the spectra is therefore less accurate than the corresponding values for LT-GaAs. However, based on these initial results and the similarity between LT:GaAs:C and InP:C, we believe LT-InP buffers can potentially be used to improve reliability of radiation-hard electronics.
IV. CONCLUSIONS
The chemical properties of LT-GaAs:C and LT-InP:C were studied by SIMS and LVM spectroscopy. The use of CBr4 as the dopant source was found to introduce significant bromine incorporation during low-temperature growth. We report on the effect of C and Br incorporation on carrier lifetime and concentration of antisite defects. A model for C

and Br incorporation was proposed and we speculated that
incomplete dissociation of the CBr4 molecules give rise to the formation of C–Br complexes and results in a reduction
of electrically active carbon concentration. Preliminary char-
acterization data on LT-InP:C indicate that it can potentially
be used as an alternative III–V material to LT-GaAs for
radiation-hard IC applications.
ACKNOWLEDGMENTS
This work was supported by DSWA through SBIR Phase
I Contract No. DSWA01-98-M-302 and AFOSR under Grant
No. F49620-98-1-0135. TRRT measurements were spon-
sored by JSEP under Grant No. F49620-94-04640.
1F. W. Smith et al., Appl. Phys. Lett. 54, 890 ͑1989͒. 2E. S. Harmon, M. R. Melloch, J. M. Woodall, D. D. Nolte, N. Otsuka, and C. L. Chang, Appl. Phys. Lett. 63, 2248 ͑1993͒. 3D. C. Look, Thin Solid Films 231, 61 ͑1993͒. 4T. S. Sosnowski, T. B. Norris, H. H. Wang, P. Grenier, J. F. Whitaker, and C. Y. Sung, Appl. Phys. Lett. 70, 3245 ͑1997͒. 5Z. Liliental-Weber et al., Proceedings of the Eighth Conference on Semiinsulating III–V Materials, edited by M. Godlewski ͑1994͒, p. 305. 6D. E. Bliss, W. Walukiewicz, K. T. Chan, J. W. Ager III, S. Tanigawa, and E. E. Haller, Mater. Res. Soc. Symp. Proc. 241, 93 ͑1992͒. 7M. R. Melloch et al., J. Appl. Phys. 72, 3509 ͑1992͒. 8P. Specht, H. Sohn, M. Luysberg, A. Prasad, J. Gebauer, R. KrauseRehberg, and E. R. Weber, Mater. Sci. Forum 258–263, 951 ͑1997͒. 9X. Liu, A. Prasad, W. M. Chen, A. Kurpiewski, A. Stoschek, Z. LilientalWeber, and E. R. Weber, Appl. Phys. Lett. 65, 3002 ͑1994͒. 10W. K. Liu, K. Bacher, F. J. Towner, T. R. Stewart, C. Reed, P. Specht, R. C. Lutz, R. Zhao, and E. R. Weber, J. Cryst. Growth 201Õ202, 217 ͑1999͒. 11M. R. Melloch, N. Otsuka, K. Mahalingham, C. L. Chang, P. D. Kirchner, J. M. Woodall, and A. C. Warren, Appl. Phys. Lett. 61, 177 ͑1992͒. 12W. K. Liu, K. Bacher, F. J. Towner, T. R. Stewart, C. Reed, P. Specht, R. C. Lutz, R. Zhao, and E. R. Weber, J. Vac. Sci. Technol. B 17, 1200 ͑1999͒. 13G. M. Martin, Appl. Phys. Lett. 39, 747 ͑1981͒. 14P. Specht, R. Zhao, R. C. Lutz, and E. R. Weber, in Proceedings of the Symposium on Nonstoichiometric III–V Compounds, edited by P. Kiesel, S. Maltzes, and T. Marek ͑Verlag, Erlangen, 1998͒, p. 15.

JVST B - Microelectronics and Nanometer Structures
ConcentrationPropertiesPhysFigGaas