Reflectivity Difference Spectra of GaAs and ZnSe (100) Surfaces
- PDF / 455,166 Bytes
- 7 Pages / 414.72 x 648 pts Page_size
- 63 Downloads / 194 Views
Mat. Res. Soc. Symp. Proc. Vol. 406 ©1996 Materials Research Society
of GaAs (011) [3,12]. EXPERIMENT The heart of our SE setup is the photoelastic modulator, which operates at 50 kHz. The system has been improved to measure spectral data up to 6 eV, so that the E, structure of wide bandgap materials may be obtained. The experimental setup, procedure and method of data reduction have been reported previously [14]. The SE was converted into the RDS using the optical bridge configuration suggested by Aspnes et al. [11]. As a reference, the xs axis on the sample surface is taken to be perpendicular to the plane of incidence. GaAs (100) wafers, which were Si doped approximately lxl01 8 cn 3 , were purchased from the Sumitomo company, which provided crystallographic orientations for each wafer. ZnSe films were grown on a GaAs (100) substrate by molecular beam epitaxy (MBE) [13]. The Hall measurement showed that ZnSe is n-type doped approximately I x 1018 cm 3 . The surface of both types of samples was rinsed gently with methanol before the RD spectra were taken. The RD spectra Ar/roo -- (r011 -roTl)/rOol , Ar/rOTo - (r0 T1 -roTT)/roTo and Ar/roTi = (rol -roTo)/roTh were obtained by orienting the sample in such a way that the xs axis lay along the [001], [010] and [011] axes, respectively. BecauseAr must vary sinusoidally with a period of 1800 regardless of the origin of the RD spectra, correct spectra must satisfy the relation Ar/roT0 = - Ar/rooI. Our experimental data did not satisfy that relation, because of unwanted effects such as misalignments in 0u and Op, nonideal optical devices and optical activity [11]. However, because those effects induce the same error in every spectrum, by taking both the Ar/roo1 and the Ar/ro-o spectra, we have eliminated that error by subtracting
(Ar) err
1 22
A&r Ar rol
r0To)
from each of the measured spectra. THEORY Aspnes [3] has derived theoretical lineshapes for the RD spectra arising from five different physical origins. By comparing these lineshapes with his RD data on Si (111), Si (110), GaAs (110) and InP (110) surfaces, he found [3,12] that only three of these origins need to be considered: the intrinsic surface charge distribution treated either as a surface local field (SLF) or in the contact-exciton approximation (CEA), bulk spatial distribution of charge (BSD) and surface roughness (SR). The lineshapes he found for each of these contributions depend only on the bulk crystal symmetry and on the ambient and bulk dielectric functions Ea and Eb, The signs and amplitudes of the contributions depend on the crystal orientation with respect to the xs axis. Furthermore, the lineshapes for these three contributions to the reflectivity difference spectra Ar/r are easily found from Aspnes' derivation of ARIR F2Re{Ar/r}. The BSD contribution can be written: _roll
-rOT1
r°°l I BSD
flaEb d (E2d), 2 2me c b(Eb -Ea) dE
-CBsD
where E is the photon energy. The SLF contribution can be written:
320
(2)
(
I110111
ro 1 ) SLF
hc eb-E)a 47E n b(Ia
-iCL
where the complex conjuga
Data Loading...
