• PDF / 1,075,744 Bytes
• 6 Pages / 414.72 x 648 pts Page_size
• 105 Downloads / 286 Views

DOWNLOAD

REPORT

---

Mat. Res. Soc. Symp. Proc. Vol. 423 01996 Materials Research Society

phase epitaxy (HVPE), followed by the in-situ etch removal of the initial substrate to leave a thick free-standing GaN film. The availability of GaN homoepitaxial substrates would significantly reduce current problems in heteroepitaxial nucleation and eliminate defects arising from differences in the coefficient of thermal expansion between the epilayer and the substrate. EXPERIMENTAL PROCEDURE The HVPE technique has been discussed widely in the literature (4,5,6,7). Details concerning our HVPE growth of GaN have been presented elsewhere (8). MOVPE buffer layers and growth studies were carried out in a horizontal MOVPE system, operated at 76 Torr, utilizing trimethylaluminum, trimethylgallium, and ammonia as the reactants. Typical gas velocities were over 30 cm/sec, with cation precursor partial pressures on the order of 0.011 Torr, and a V/Il ratio of 1500. Low temperature AIN buffer layers were utilized for MOVPE growths of GaN on sapphire, in accordance with the techniques reported by other researchers (1). For growths of AIN on Si(l 11) substrates, the Si substrates were initially oxidized, with the oxide removed by HF etch just prior to growth. MOVPE AIN layers were grown on Si(l 11) using trimethylaluminum and ammonia as the precursors at I 100'C, similar to the method reported by Watanabe et al (9). Both 0)-scan (transverse) and 0-20 scan (longitudinal) x-ray diffraction geometries were used to characterize the GaN films. The x-ray apparatus was equipped with a four-reflection monochrometer and a three-reflection Si analyzer crystal at the detector. In the co-scan geometry, the detector was held at a fixed position, and the sample was tilted during the scan. In the 0-20 geometry both the sample and detector were moved, with the detector moving at twice the angular velocity of the sample. RESULTS AND DISCUSSION The growth rate of GaN, based on the apparent activation energy for growth and the dependence on GaC1 partial pressure, previously reported (8), appears to be limited by mass transport of GaCl to the growth front rather than by surface reactions. The growth thickness uniformity was ±8% over a 2.5 cm sample, with a smooth specular surface. Attention to surface cleaning and higher growth rates were found to improve the growth morphology. Higher GaCl partial pressures may increase the rate of nucleation, leading to a higher density of GaN nuclei on sapphire, and transition to the 2-D growth mode at a lower thickness. The theoretical relationship between the heterogeneous nucleation rate and the supersaturation for the general case is widely discussed in the literature (10). HVPE GaN layers on sapphire were analyzed using X-ray scans in the (o-rocking (transverse) and 0-20 rocking geometries (longitudinal). The co-scan values always exhibit a much broader FWHM to the peak than the 0-20 scan (i.e., 636 arcsec versus 16 arcsec for a 110 jtm thick HVPE GaN/sapphire sample). Broadening in the co-scan is due to composite broadening from both