Real-Time Pseudodielectric Function of Low-Temperature-Grown GaAs
- PDF / 1,119,954 Bytes
- 6 Pages / 415.8 x 637.2 pts Page_size
- 105 Downloads / 173 Views
National Research Council Postdoctoral Research Associate. SSemiconductor Electronics Division, Electronics and Electrical Engineering
Laboratory, Technology
Administration, U.S. Department of Commerce. Official contribution of NIST; not subject to copyright in the United States. 243
Mat. Res. Soc. Symp. Proc. Vol. 618 ©2000 Materials Research Society
(lower) values of Tg [1, 3-5, 9]. The value of hepi decreases with increasing A[As] due to increasing strain [1, 3-5, 9]. This high point-defect density gives LT-GaAs its unusual properties that make it attractive for certain device applications. For example, LT-GaAs displays a high electrical resistivity (- 108 n-cm), which makes it useful for high-resistance buffer layers to prevent sidegating and backgating in MESFETs. LT-GaAs also displays a short nonradiative recombination lifetime (< 1 ps), which makes it attractive for fast photodetector and THz spectroscopy applications. EXPERIMENT We grew the LT-GaAs layers on semi-insulating GaAs (001) "epi-ready" substrates, in an EPI [10] GEN-I MBE system with an As 4 - As 2 cracking zone. We thermally de-oxidized the substrates and then deposited a 1.0 gtm GaAs buffer layer at approximately 600 'C under an As2 :Ga beam equivalent pressure ratio RBEP of (28 + 2) to obtain a smooth, oxide-free substrate. We stopped the buffer layer growth by closing the Ga shutter and then ramped the substrate down to Tg, closing the As 2 shutter below 400 'C. We initiated the LT-GaAs growth by opening the As 2 shutter at Tg and then opening the Ga shutter 5 s later. We extracted the LT-GaAs real-time as follows. We used a J.A. Woollam Co., Inc. [10], 44-wavelength, rotating analyzer system to measure in situ SE data (data acquisition frequency better than 1 Hz) before and throughout the growth. We established the in situ SE incidence angle by modeling the data measured on the substrate at Tg before the growth. The model parameters consisted of the GaAs temperature-dependent [11] and a surface roughness layer. The modeling indicated that the substrate starting surfaces displayed submonolayer surface roughness. We did not rotate the substrates during the growth in order to maintain a constant incidence angle. Finally, we extracted the LT-GaAs real-time and growth rate values by fitting the growth data by using the virtual substrate approximation [ 12]. We actively controlled the substrate temperature during the growth by using a Thermionics DRS system supplying feedback to a commercial temperature controller [10, 13, 14]. The DRS system determines the substrate temperature by measuring the substrate absorption edge wavelength and comparing it to an empirical calibration table. This technique yields a demonstrated precision of better than 1 'C and a measurement speed of better than 1 s. RESULTS Figure 1 presents the real-time data extracted after the growth for three samples of 100 nm LT-GaAs (thinner than hepi) grown at 300 'C, 250 'C, and 210 'C and for a GaAs substrate at 300 'C. This data shows that the real-time of epitaxial LT-GaAs c
Data Loading...
