Extended Abstracts of the 1996 International Conference on Solid State Devices and Materials, Yokohama, 1996, pp. 428-430 PD‐ 4‐ 7 Effects of Thermal Stability of Si1-'-nGerCn Layers on Properties of Their Contacts with Aluminum Jian MI, Ashawant GUPTA, and Cary Y. YANG Microelectronics Laboratory, Santa Clara University, Santa Clara, CA 95053, USA Jintian ZHU and Paul K.L. YU Department of Electrical and Computer Engineering, University of California, San Diego La Jolla, CA 92093, USA Patricia WARREN and Michel DUTOIT Institute for Micro- and Optoelectronics, Department of Physics Swiss Federal lwtitute ofTechnologt, CH 1015 Lausanne, Switzerlond The eff€cts of themral stability of Sir-*ace.q hyers on the elecfrical properties of AVSiro-rGe,C, Schottky diodes have been investigated. I-V, C-V, and x-ray diftaction measurements were performed to examine tle electical properties and lattice stucture of the alloy layers. Nealy ideal I-V and C-V characteristics were obtained for Si,-*-"Ge*C" with strain reduced subsantially by carbon incorporation. High effective dopant concentration was observed for strained alloy layers, which had be€n subjected to thermal annealitg. This effective doping is attibuted to lattice-relaxation related defects. 1. Introduction Recently, Si1-*-rGe*CrlSi heterostructures have atfracted increased int-erest in Si-based technologyr), since, compared with the binary Sit-*Ge* alloy, Si1-*-rGe*C, alloy can provide more flexible bandgap engineering. Carbon incorporation reduces the strain induced by germanium in the alloy layers and thus improves their mechanical stability. In addition, the ternary alloy has an additional degree of freedom in bandgap design. One of the major issues related to device application is the thermal stability of SiGeC layers. Since there exists the possibility of silicon carbide precipitation, SiGeC layers are chemically metastable in addition to the inherent mechanical metastability that exists in SiGe layers. Previous studies of thermal stability have focused on physical and chemical changes of lattice structure such as relaxation and precipitation under ex-situ stress (high-temperature annealing)2-a). In this paper, we report results on electrical measurements of Al/SiGeC Schottky diodes that were fabricated using conventional silicon processing. Unlike previous work, our investigation is focused on the effects of carbon incorporation on the electrical propenies of the SiGeC alloy, using these Schottky diodes as test vehicles. 2. Experimental The alloy layers used to fabricate AVSiGeC Schottky diodes were grown by rapid thermal chemical vap-o.r deposition. The growth process was described previously'). Briefly, on n-type (100)-Si.substrates with phosphorus concentration of about 5x10r5 cffi-3, an undoped Si epi- layer layer of 200 nm was grown as a buffer layer followed by an undoped alloy layer of 100 to 150 nm depending on the composition. Si1-*Ge* layers were obtained with compressive strain up to 1.5% (x:0.2). The carbon concentration in Si1-*-rGe*C, layers varies from 0 to 2 at%o. Characterizations of the alloy layers show that they are free of structural defects and precipitation of silicon carbide. All the carbon atoms were confrmed to be in substitutional sites6). Two batches of Schottky diodes were fabricated. For the first batch, using a photoresist layer of about I pm as a mask, phosphorus was implanted into the alloy layer to ensure good electrical contact. The energy and dose of the implantation were 40 keV and 5x10t' ,--t, respectively. Donor activation was achieved with a furnace anneal at700 oC for 60 min in Nr. The second batch was produced without phosphorus implantation but with one sample annealed using the same condition of donor activation as for the first batch. Aluminum electrodes of 2.8x10-3 cm2 were made by Al sputtering and wet pafferning. While the n- zone under Al electrode did greatly enhance the rectiffing characteristics at high biases (> 0.3 V), it did not have much influence on results at low biases. The two different processes allowed us to isolate the effects of thermal annealing. C-V and I-V measurements were performed at room temperature. X-ray diffraction (XRD) was used to monitor lattice relaxation. 3. Results and Discussion Log (I) - V curves of the first batch of samples are shown in Fig. 1. From results in the low-bias regime, an ideality factor of 1.04 was exffacted for the Si epi-layer (reference sample). This implies that the current was a result of thermionic emission and/or diffusionT). Th. current and ideality factor of the SiGe sample were much higher than those of the reference sample, infening that other mechanisms than the thermionic emission and diffusion also contributed to the total current. However. 428