ii Liquid Phase Epitaxy (LPE). The one sample in this group, R137, is an exceptional example of the LPE growth technique since its central cell structure shows that there is only one major residual contaminant. However the electrical characteristics quoted are not as good as those of material produced routinely a
2 “ 1 — 1
number of years ago (which had 200,000 cm V s~ mobilities at 77K).
iii Vapour Phase Epitaxy (VPE). Both RR98B and SI are samples which represent the ’state of the art* for VPE growth. SI has very high mobility and low impurity concentration, while RR98B has a very low compensation ratio ('^0.05), Apparently the reduction in the residual acceptor concentration in RR98B was achieved by baking out the gallium source at high temperatures prior to the growth run. Prior to the use of the bake-out procedure photoluminescence studies showed that the dominant acceptor was
Table 4.1.
Electrical Characteristics of the GaAs Samples.
Sample .Growth Technique Source
cm"^ Mobility (77K) 1 0^ cm^v“^s“^ MBV20 MBE Phillips 7 X 10^* 49 MBV380 MBE Phillips 3 X 10^4 55 R137 LPE Mas-Planck 1 X 10^3 150 81 VPE Motorola 5 X 10^^ 160
RR98B VPE Wrig,ht Patterson AFB 2 X 1 0^^ 163
zinc but no zinc could be detected in runs using the bake out procedure. The impurity concentration profile for RR98B is shown in Figure 4.4. From this profile it is clear that RR98B has quite a thick layer of ~30]Jim.
The combination of very low compensation and high purity in RR98B was such that a second peak in the mobility versus temperature curve could be seen at a temperature of ~9K (shown in the inset to Figure 4.4 after Look and Colter 1983). This second mobility peak, which had not previously been observed in III-V materials, occurs as ionized impurity scattering is reduced as electrons freeze out onto the impurity sites.
All the material was undoped and contained only residual impurities and was grown on semi-insulating substrates. With the exception of the two MBE samples, electrical contacts to the samples were made by alloying indium dots to the epitaxial surface using the procedure described in Chapter 3. All the experiments were performed by measuring the photoconductivity as a function of magnetic field at temperatures of 4.2 or 1.8K.
In Figure 4.5 the spectrum of the ls-2p^^ transition for each of the samples is shown. The recordings were taken with a laser wavelength of 118.8iim and at 4.2K and the samples were illuminated with intrinsic radiation from a quartz halogen lamp.
For the two MBE samples (MBV20 and MBV380) the ls-2p^^ peak is rather broad and the central cell components are not well resolved. However it is clear that both samples contain two components. More unresolved components may also be present. The peak positions of MBV20 are at slightly higher magnetic field than those in MBV380 and both samples have an asymmetric lineshape with a longer tail on the high
CK LU CL < ce: o LJ QC LU OC cc < LJ RR98B î I T fK) 8 12 16 20 24 28 32 36 40
DISTANCE FROM SURFACE (MICRONS)
Figure 4.4: Carrier concentration against depth for sample RR98B (Colter, unpublished). The inset diagram shows the Hall mobility against temperature for RR98B. Note the second peak in the mobility at ~9K due to a decrease in the ionized impurity scattering as carriers freeze out and neutralize the donors (from Look and Colter, 1983).
X = 118-8 urn T= 4-2K 1s-2p,i MBV20 MBV380 R137 g < H- > I— LJ gz RR96B o LJ o*— ox: Q . 3 8 3 6 3 7 3-5
MAGNETIC FIELD (TESLA)
Figure 4.5: Spectra of the central cell structure of the ls-2p^^ transition for the five n-GaAs samples using an FIR laser wavelength of 118.8ixm. The sample temperature was 4.2K and the samples were illuminated with band gap excitation from a quartz halogen lamp. The
magnetic field side. As the transition linewidth in MBV20 is significantly greater than that in MBV380 this shift in peak position and the asymmetry is probably a consequence of the quadratic Stark broadening mechanism discussed earlier. The low magnetic field component can be positively identified as the donor which is present in the other three samples, and which has been associated with sulphur donors. Originally X^ was identified as silicon, but more recent work with back doped samples has resulted in the reassignment of Xg to sulphur while X^ has been identified as silicon (Low et al 1983a,b,c; Ozeki et al 1977). The identity of the high field donor is not clear, and it could be due to either of the components identified as Sn and X^ (silicon) which are clearly resolved in the LPE sample R137. Other donors, eg selenium, may also appear in this region (Armistead et al 1984). Low et al (1982a) have made a detailed study of high purity MBE material (with 77K mobilities up to 110000
2 -1 -1
cm V s ) and found that X^ (S), Sn, X^ (Si) and Pb were