ELSEVIER Thin Solid Films 254 ( 1995) 194- 199 A new ellipsometric model of wet and dry etched GaAs surfaces Anna M. Kaminska, Marek Guziewicz Institute of Electron Technology. AI.Lotnikox 32146, 02-668 Warszawa, Poland Received 15 February 1994; accepted 21 June 1994 Abstract In this paper we describe the application of ellipsometry to study (lOO)GaAs surfaces. Investigations concern the surfaces after polishing, cleaning, wet and dry etching of GaAs substrates and the surfaces of liquid phase epitaxy- and molecular beam epitaxy-grown GaAs epilayers. A new model is proposed to interpret the ellipsometric data. It assumes the presence of a transitional layer characterized by spherical disturbances situated between the bulk GaAs and the native oxide layer. The ellipsometric parameters are analysed as a function of the size and density of the spherical disturbances and the transitional and native oxide layers. Different angles (70:75”) of the incident light beam are used at a wavelength of 546.1 nm. A refractive index of 1.81 for the oxide film and (4.05 - iO.304) for the GaAs substrate are assumed. Our results show that the thickness of the native oxide layer changes from 11 A to 35 8, depending on the surface treatment. The thickness of the transitional layer reaches the radius of the spherical disturbances. Their radius ranges from 10 A to 30 A, while the complex refractive index changes from (4.05 - i0.34) to (4.70 ~ il.30). Kqvwords: Ellipsometry; Etching; Optical properties; Surface roughness 1. Introduction Ellipsometry, using the polarization of monochro- matic light reflected from a surface, is a non-destructive and very sensitive technique, ideally suited for studying semiconductor surfaces. In the case of substrate wafers, the parameters influencing the change of reflected light polarization are the following: crystallographic defects formed during single crystal growth; and factors dam- aging the surface and the near-surface layer caused by mechanical and chemical processing. As the result of processing, microcracks are formed in the near-surface layer and the surface becomes rough, covered by a very thin layer of native oxides. Recent surface investigations using ellipsometry have again become of interest. Dinges [ 11 described the study of the native oxides and the transitional layer on GaAs surfaces. He introduced an intermediate layer with a thickness of lo-20 A. Luke: [2] investigated a gas absorbed layer on top of the native oxide. Ohlidal and Libezny [3] adopted the immersion and multiple angle of incidence ellipsometry method. They obtained native oxide thicknesses of 22-36 A on sulfuric acid-etched GaAs. The spectroscopic ellipsometric method has been demonstrated by Xiong and Snycer [4] on Si and GaAs 0040-6090/95/$9.50 Cl 1995 ~ Elsevier Science S.A. All rights reserved SSDl0040-6090(94)06258-M surfaces. In their study, a similar model was used, but without spherical disturbances. They used Bruggeman’s effective medium approximation and found the exis- tence of a damaged layer with 2% voids and the native oxide replaced by 4-7 A microroughness. Scanning tunneling microscopy (STM) is also very helpful in the surface investigations. Dagata and coworkers [5,6], Vaterlaus et al. [7] and Moriarty et al. [8] used the STM technique to study GaAs surfaces. The STM images of GaAs surfaces showed that the apparent root-mean square surface roughness Rapp varied from 4 A for passivated P2S5/( NH&S , to 16 A [5,6] or greater (8OA) [8] for unpassivated samples. Vaterlaus et a1.[7] observed arsenic precipitates in GaAs epitaxial material cleaved in ultrahigh vacuum. The typical dimensions of the precipitates were a diameter of 50 8, and a height of 15 A. Gwo et a1.[9] investigated the ( 110) cross-section surface structure of GaAs multi- ple p-n junctions grown by molecular beam epitaxy (MBE). The STM image showed lo-20 A roughness. In our paper the influence of the native oxides, transitional damaged layer and surface roughness is taken into account. Due to the complexity of the prob- lem, a rigorous analysis of the influence of all factors on the polarization of reflected light is not possible.