Observation of shallow residual donors in high purity epitaxial GaAs by means of photoluminescence spectroscopy

Solid State Communications, Vol. 38, pp. 1053-1056. Pergamon Press Ltd. 1981. Printed in Great Britain.

0038-1098/81/231053-04502.00/0

OBSERVATION OF SHALLOW RESIDUAL DONORS IN HIGH PURITY EPITAXIAL GaAs BY MEANS OF PHOTOLUMINESCENCE SPECTROSCOPY R.J. Almassy,* D.C. Reynolds, C.W. Litton, K.K. Bajaj*~ and G.L. McCoy Avionics Laboratory, Air Force Wright Aeronautical Laboratories, Wright-Patterson Air Force Base, OH 45433, U.S.A. (Received 7 January 1981 by M. Balkanski)

We report the first observation of three different residual donors in undoped high purity vapor phase epitaxial GaAs using the high resolution photoluminescence spectroscopy technique at temperatures ~ 2 K. The binding energies of these shallow donors were determined from the excited state transitions of excitons bound to neutral donors and they are found to be in very good agreement with corresponding values obtained from high-resolution far infrared Fourier transform spectroscopy, using the modulated photoconductivity technique. CONSIDERABLE EFFORT has been devoted to the study and identification of hydrogenic shallow donors in high purity epitaxial GaAs over the past decade [1 ]. This work has shown that high purity vapor phase and liquid phase epitaxial GaAs have several different residual donors which are characterized by very small differences in binding energy. From high-resolution, Fourier transform infrared (FTIR) magnetospectroscopic studies, which employed the modulated photoconductivity detection technique to monitor the ls ~ 2p_ 1 transition of isolated hydrogenic donors in a timed magnetic field, Stillman et aL [2] have detected three residual donors in high purity vapor phase epitaxial (VPE) GaAs, while Stradling et al. [3 ] have found the same number of residual donors in liquid phase epitaxial (LPE) material. Wolfe and coworkers [4, 5] have performed similar spectral measurements in conjuction with back-doping experiments in VPE GaAs and have identified one of the residual donors as Silicon on the Gallium site (hereafter denoted as X2 ). The other two residual donors (denoted as XI and 2"3) could not be given a definite chemical identification; however, it was suggested that )(3 was probably due to Carbon on a Gallium site [5] and speculated that XI might have arisen from a native defect, possibly a Ga-vacancy [4, 5]. Following a procedure similar to that of Wolfe and coworkers, Ozeki et al. [6] have attempted to identify the residual donors in VPE GaAs. Their assignments, Present address: Directorate of Laboratories, Air Force Systems Command, Andrews Air Force Base, Maryland, U.S.A. t Present address: Universal Energy Systems, Inc., 3195 ~ Plainfield Rd., Dayton, OH 45432, U.S.A. Work performed at the Avionics Laboratory, WrightPatterson Air Force Base, Ohio under Contract F33615-76-C-1166.

however, differ from those of Wolfe et al. [5] ; moreover, the binding energy of the Silicon donor they measure is smaller than that determined by Wolfe et al. [5]. Recently Afsar [1] and coworkers have initiated a systematic effort to identify the residual donors in VPE GaAs by means of FTIR magnetospectroscopy performed over a wide range of high magnetic fields. Their preliminary results indicate that X2 and X3 are due to silicon and germanium, respectively. In their measurements, the assignment of X3 to germanium was made by studying germanium-doped VPE GaAs samples in which the doping was achieved by neutron transmutation of gallium. At this point, despite considerable experimental effort on this subject, there appears to be no general undisputed agreement concerning the identity of residual donors in VPE GaAs, even though the assignment of X2 to silicon and )(3 to germanium appears plausible and reasonably reliable. In this letter we report the Ftrst observation of three different residual donors in undoped high purity VPE GaAs using the sharp-line, high-resolution photoluminescence spectroscopy technique at superfluid liquid helium temperatures. We have determined the binding energies of these donors from the excited state transitions of excitons bound to neutral donors and Fred that these values agree very well with those of others. 1. EXPERIMENTAL DETAILS The samples employed in this study were undoped, high purity epitaxial layers grown on semi-iusulating GaAs : Cr substrates by means of the H2 : AsC13 : Ga vapor deposition technique. Hall and electrical conductivity measurements were used to characterize the electrical parameters of the samples. Electron mobilities of the samples were typically on the order of 10s cm 2

1053

1054

SHALLOW RESIDUAL DONORS IN HIGH PURITY EPITAXIAL GaAs Di, (I.51410)

Vol. 38, No, 11

(I.51415) D4,(I.5146T

E] (X3),(I.51007)

D~, (I.51456)

E3 (Xz), (I.51014) E 3 (Xl), (I.51020) o3 Z I-Z I11

X' 2 /

-1 ~J

I

1.51005

I

'/I

1.51025

1.51420 ENERGY (eV)

1.5t440

1.51460

Fig. 1. Typical photoluminescence spectrum of high purity VPE GaAs in the near band edge region at ~ 2 K. The origin of the various displayed transitions is explained in the text. V -~ sec -~ at 77 K and their total concentration of electrically active impurities was ~ 1014 cm -3. Photoluminescence samples were mounted in a strain-free manner on one end of a sample holder which was, in turn, immersed in the quartz tip of a pyrex helium Dewar containing a superfluid liquid He bath whose temperature was maintained in the range of 1.2-2.1 K, as determined by vapor pressure thermometry. A krypton ion laser radiating some 200 mW of c.w. power at 6471 A was employed to pump the luminescence; spectral analysis of the photoluminescence was achieved with a modified B&L 4 m grating spectrograph, equipped with a large (10 cm square) high resolution diffraction grating ruled to 2160 grooves mm -t and blazed to 5000 A in f'trst order. This instrument was capable of producing a first order reciprocal dispersion of approximately 0.54 A mm -~ over the wavelength range of interest. Photoluminescence spectra were photographically recorded on Kodak type 1N spectroscopic plates. Wavelength calibration of the plates was achieved by non-linearly interpolating the luminescence spectral lines,with respect to well known interferometricaUy measured neon spectral lines, using the grating equation, the known geometric dispersion of the instrument and a nonlinear least squares fitting technique. 2. RESULTS AND DISCUSSION When GaAs is irradiated with light whose photon energy exceeds that of the bandgap, electron-hole pairs are produced. These electrons and holes thermalize very rapidly to form free excitons, and impurity exciton

complexes at low temperatures. A large number of these bound electrons and holes then radiatively recombine emitting spectral radiation which is energetically characteristic of the excitonic system they are associated with. In this paper we are interested in the radiative transitions associated with donor exciton complexes. A typical photoluminescence spectrum in this energy region is displayed in Fig. 1. The line 2)1 arises from unresolved radiative recombination of excitons bound to neutral donors in which the donors are left in their ground states. Since the differences between the binding energies of different donors in GaAs are very small (the maximum spread in the binding energies of donors is at most 0.2 meV) it is not possible to resolve such transitions which arise from the recombination of excitons bound to different donors. Thus, if only one donor were present, the peak marked Ol would represent the resolve resolved transition of that donor to its ground state. It has been suggested [7] that the lines marked D2, D3 and D4 arise from radiative recombination of excitons bound to one or more neutral donors for the ease where the resulting donor exciton complex is initially in excited states, but it is not possible to calculate accurately the energies of these states. Energies of such states have, however, been estimated [7] by invoking a nonrigid rotator model as an approximation to the donor exciton complex, a model in which these states are identified with the low-lying rotational states of such a system. The energies thus obtained agree fairly well with observed values. The origin of the line marked E3 (XI) on the lowerenergy-side of the D peaks, is attributed to the following mechanism. An exciton bound to a neutral donor

Vol. 38, No. 11

SHALLOW RESIDUAL DONORS IN HIGH PURITY EPITAXIAL GaAs

FINAL

INITIAL

(•)--

x3--

X2 x, J

1055

deduced from careful high resolution FTIR spectroscopic measurements of the ls ~ 2p_~ transition of an isolated donor [2]. Central cell correction to the 2s state is assumed to be 1/8 of its value for the ls state. In the present work, we propose that the lines marked E3 (X2) and E3 (X3) arise from similar transitions due to two other donors. From an analysis of our spectral data, the binding energies of all three donors have been calculated and are presented in Table I, where we have assigned the transitions E3 (X1), E3 (X2) and E3 (X3) to the previously designated X~, X2 and X3 donors, respectively. For comparison we also display the values of the binding energy of these donors as measured by two other groups using high-resolution FTIR magnetospectroscopy. Although slightly larger, the values we determine agree rather well with those obtained by other groups. If we assume that the final state of the donor is

(~)= "+"

2S _ I ~ V 2/ /___~/ ~----I.SlOI4eV ~-~ 1"51020 eV I S½ / ......

Fig. 2. Schematic representation of radiative recombination of an exciton in the exciton-donor complex where the rmal state is the donor in the ground or in the excited configuration.

Table 1. Binding energies o f residual donors in VPE GaAs (units o f meV)

Donor designation

Measurement: Wolfe et al. [5] (FTIR)

Measurement : Ozeki et al. [6] (FTIR)

Measurement Present results (photoluminenscence)

X1 X2 X3

5.801 5.854 5.937

5.795 5.845 5.949

5.804 5.870 5.978

recombines when the complex as a whole is in the D3 excited state, leaving the donor electron in the first excited state of an isolated neutral donor, as shown in Fig. 2. In this process, the emitted photon will have an energy which is equal to the energy of the D3 line minus the energy of the above excitation. One can show from symmetry considerations (group theoretic arguments, in particular) that both the 2s and 2p states are allowed ffmal states. However, the energies of the 2s and 2p states are so close to one another that it is not possible to resolve them experimentally. Excited state transitions corresponding to the complex in D 1 and D 2 excited configurations are also observed. But the associated excited state transitions due to different donors (analogous to the Da mechanism denoted above) are much more difficult to resolve in the case of the Dt and D2 excited configurations, owing to the small separation between the D~ and D2 lines. If the t'mal state of excitation of the donor electron is assumed to be 2s, the difference between D3 and Ea (X1) is then the difference in energy between the ls and 2s states of this donor. The binding energy of this donor can therefore be calculated using a value of the impurity Rydberg, Ro = 5.737 meV (which incidentally includes a contribution due to nonparabolicity ot the conduction band). This value was

2p rather than 2s, then the values of the binding energy we obtain for all three donors are identical to those of Wolfe et al. [5]. Recently Cooke et al. [81 have also determined the binding energies of Xt, X2 and X3 donors in VPE GaAs using a number of samples from a variety of sources following a technique similar to that of Wolfe e t a l . [5]. The values of Ozeki etaL [6] are quite close to those of the other groups, except that their chemical assignments for these donors are different. As mentioned earlier, Ozeki et al. assign XI to silicon whereas Wolfe et al. [5] and Cooke et al. [8] and recently Afsar et al. [1 ] assign X2 to silicon. In addition, Wolfe et al. and Cooke et al. assign X3 to carbon whereas Ozeki et al. and Afsar et al. assign X3 to germanium. The donor Xt has not been chemically identified by any other group except by Ozeki et al. Since the samples used in the present study were not intentionally doped, we are not in a position to propose chemical identification of these residual donors. It should be pointed out that due to the extreme sharpness of bound exciton transitions and the high resolution ( ~ 0.007 meV) of our spectrograph, the excited state transitions due to different residual donors are very well resolved. On the other hand, high-resolution Fourier transform spectroscopic measurements require the appli-

1056

SHALLOW RESIDUAL DONORS IN HIGH PURITY EPITAXIAL GaAs

cation of fairly high magnetic fields in order to resolve transitions due to different donors in VPE GaAs and further require a theoretical analysis of the data in order to derive the binding energies at zero field. To conclude, we have observed for the first time three different residual donors in undoped high purity VPE GaAs using the sharp-fine high-resolution photoluminescence technique. From the excited state transitions of exciton-donor complexes, we have derived the binding energies of these donors. Our values compare very well with those measured by high-resolution FTIR magneto-spectroscopy, using the modulated photoconductivity detection technique. REFERENCES .

See for instance, M.N. Afsar, K.J. Button & G.L. McCoy, Int. J. o f lnfrared and Milimeter Waves 1, 145 (1980)and references cited therein.

2. 3.

4. 5. 6. 7. 8.

Vol. 38, No. 11

G.E. Stillman, D.M. Larsen, C.M. Wolfe & R.C. Brandt, Solid State Commun. 9,245 (1971 ). R.A. Stradling, L. Eaves, R.A. Hoult, N. Miura, P.E. Simmonds & C.C. Bradley, Gallium Arsenide andRelated Compounds. Inst. Phys. Conf. Ser 17, p. 65 (1972). C.M.Wolfe, D.M. Korn & G.E. Stillman, AppL Phys. Lett. 24, 78 (1974). C.M. Wolfe, G.E. Stillman & D.M. Korn, Gallium ArsenMe and Related Compounds. Inst. Phys. Conf. Set 33b, p. 120 (1977). M. Ozeki, K. Kitahara, K. Nakai, A. Shibatomi, K. Dazai, S. Okawa & O. Ryuzan,Jap. J. Appl. Phys. 16, 1617 (1977). W. Rul-de & W. Kfingenstein, Phys. Rev. B12, 7011 (1978). R.A. Cooke, R.A. Hoult, R.F. Kirkman & R.A. Stradling, J. Phys. D. Appl. Phys. 11,945 (1978).