In order to construct a reliable parameter set for the physic al modeling of 4H-SiC, we are collecting and examining the physical parameters. The results of mobility measurement are presented and compared with the built-in model in the device simulator. The doping depe n nce of the electron mobility is in agreement with the built-in model, whereas that of the hole mobility is different from the built-in model in the higher doping region. Further, the anisotropy of the electron and hole mobility is investigated. The anisotropy of the electron mobility ) 0001 ( / ) 00 1 1 ( > < μ > < μ is about 0.83 and is in agreement with the built-in model. The anisotropy of the hole mobility is observed and it is estimated to be 1.15. To our knowledge, this is the first report of the anisotropy of the hole mobility in 4H-SiC. Introduction Silicon carbide devices have outstanding features, namely higher speed and lower loss than silicon devices. Among the many polytypes of SiC, 4H-SiC has attracted gre at att ntion as a candidate material for the next generation of power semiconductor devices, due t o the excellent physical properties such as the electric breakdown field and mobility. In order to r alize SiC devices that make the best use of the excellent physical properties, device simulati on technology of SiC is indispensable. However, the comprehensive and reliable parameter set for the physic al modeling of 4H-SiC for device simulators has not been reported. As a first step in the construction of a reliabl e par meter set for the physical modeling of 4H-SiC, we are collecting and examini ng the physical parameters systematically by fabricating test chips that consist of the el ments for physical property measurements. This paper is the first report on our ongoing research . The final goal of our research is the release of the comprehensive parameter set. In this paper, we present results of mobility measurement and compare them with the previous results. Experimental Figure 1 shows the top view of a test chip of the first lot. A prec ise patterning of contact, electrode and mesa by the mask process guarantees the precision of the physical property measurements. A test chip consists of elements (Hall bars and the square and clover shaped four terminal pattern) for mobility measurements and pin diodes for the impact ionization coefficient mea surements. Hall bars are tilted to the crystallographic axis every fifteenth degree in order to de ect the anisotropy of the mobility. Test chips were fabricated on 4H-SiC epitaxial wafers. For the measurements of the electron mobility, Materials Science Forum Online: 2003-09-15 ISSN: 1662-9752, Vols. 433-436, pp 443-446 doi:10.4028/www.scientific.net/MSF.433-436.443 © 2003 Trans Tech Publications Ltd, Switzerland All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans Tech Publications Ltd, www.scientific.net. (Semanticscholar.org-12/03/20,10:56:51) 2 Title of Publication (to be inserted by the publisher) 10μm-thick n-type nitrogen doped epitaxial layers were grown on p-type s ubstrates or a semi-insulating substrate. For the measurements of the hole mobilit y, 15μm-thick p-type aluminum doped epitaxial layers were grown on n-type substrates. The doping conce ntrations of the epitaxial layers were between 15 10 1× cm and 18 10 1× cm for both cases. Deep mesas were formed for the isolation and the termination of pin diodes using inductively coupled plasma (ICP) reactive ion etching in SF6 chemistries. Nickel was deposited for the contact metallizati on after the contact implantation and 1600C o activation annealing, and sintered at 1000 C o . After the fabrication process, the chips were diced into chips, 3mm square. Each chip was used for the mobility measureme nt . Results and Discussion First, we examined the effect of the epitaxial “quality” on the mobility. Figure 2 shows the comparison of the electron mobility as a function of the temperature of a commercial epitaxial wafer and the epitaxial wafer, grown in our group (UPR sample). The dopant of the epitaxial layer is nitrogen and the doping density is below 15 10 5× cm for both samples. It can be seen that the difference of the mobility between the two epitaxial wafers inc reases with decreasing the temperature. From the SIMS analysis, the contamination with boron in the epitaxial layer of the commercial sample was 15 10 1× cm, whereas the boron concentration in the epitaxial layer of the UPR sample was 14 10 3× cm. The concentrations of other elements (Ti, Fe, V, Al) were below 14 10 2× cm. Thus, we can attribute the mobility degradation at low temperatures in the commercial sample to the contamination of boron: The reduction of carrier density by compensation and the increase of the ionized impurity scattering reduce the electron mobility in the com mercial epitaxial wafer at low temperatures. It should be noted that the electron mobility of the comm ercial epitaxial wafer almost coincides with that of the UPR sample at room temperature, because phonon scattering limits the electron mobility at room temperature. We also examined the effect of the “quality” of the substrate on the mobility. The temperature dependences of the mobility of an n-type epitaxial layer on a p-type substrate, a semi-insulating substrate and a p-type epitaxial buffer layer on an n-type substrat e were compared. We could not observe a distinct difference among them. The doping dependent bulk mobility model commonly used in the device simulator is based on the model suggested by Caughey and Thomas [1]: 0 500