Growth of Silicon-On-Insulator Films using a Line-Source Electron Beam
- PDF / 882,030 Bytes
- 6 Pages / 417.6 x 639 pts Page_size
- 39 Downloads / 147 Views
J. A. Knapp and S. T. Picraux, Sandia National Laboratories, Albuquerque,
New Mexico 87185
ABSTRACT A swept line-source electron beam has been used to study unseeded Si-oninsulator crystallization at beam scan speeds of 150-1500 cm/s. For a particular sample configuration a maximum linear crystallization velocity of - 350 cm/s was observed. At higher sweep speeds, competing nucleation occurred at intervals across the film. Both the limit in crystallization velocity and the intervals between nucleation are tentatively explained by a simple model. INTRODUCTION A variety of approaches to the fabrication of Silicon-on-Insulator (SO0) films are being explored (1-3]. Our studies have centered on the use of a line-source electron beam [4,5], both exploring the unique capabilities of such an energy source and examining questions of relevance to all approaches. The results reported here deal with the dependence of the quality of SOI formation on the speed of the sweeping e-beam. This applies to the practical question of just how fast an SOI layer can be successfully formed, and addresses such issues as minimizing the time for contamination and the substrate temperature. These questions are also important to other approaches to SOI formation. Our results indicate that, for the particular sample and beam configuration used here, the advancing recrystallization front breaks into multiple fronts above a sweep speed of - 350 cm/s. This phenomonem can be understood by the use of a simple model of the heat flow in the layered structure. EXPERIMENTAL A schematic of the line-source electron beam system is shown in Fig. 1. The beam sweeping and beam diagnostics have been reconfigured since our earlier SOI results were reported [4,5]. The system uses a sheet beam focussed to a line measuring 2 cm x 1 mm at the sample table, with a peak current density 2 of up to - 9 A/cm and energy of 10-50 keV. Samples are swept under the beam on a rotating table at up to 5000 cm/sec. The table also includes a slit used for determining the beam profile and power density. The sample configuration used in this study was similar to that used at other laboratories in SOI studies (3]. Specifically, Si wafers were covered with 1 pm of SiO2 , 0.5 ýimof poly Si, and a capping layer of 1 ýIm SiO2 and 500 A of Si 3 N4 . No seeding was provided for the SOI growth. This configuration is not expected to be ideal for our system, but was used to provide a reference point to previous work. After treatment, samples were examined with Nomarski optical microscopy, both before and after etching the cap off, and after delineating the grain structure with a Secco etch [6]. The orientation of the recrystallized films was determined by electron channeling patterns formed with secondary electrons in a scanning electron microscope. RESULTS Figure 2 shows the relationship between beam power density and sweep speed *This work performed at Sandia National Laboratories supported by the U. S. Department of Energy under contract number DE-ACO4-76DP00789. Mat.
Res.
Soc. Symp.
P
Data Loading...
