Crystallization of Silicon Films on Glass: A Comparison of Methods

  • PDF / 3,913,470 Bytes
  • 12 Pages / 417.6 x 639 pts Page_size
  • 106 Downloads / 247 Views

DOWNLOAD

REPORT


ROSS A. LEMONS*, MARTIN A. BOSCH**, AND DIETER HERBST** *Los Alamos National Laboratory, MS D429, Los Alamos, NM 87545 **Bell Laboratories, Holmdel, NJ 07733

ABSTRACT The lure of flat panel displays has stimulated much research on the crystallization of silicon films deposited on large-area transparent substrates. In most respects, fused quartz is ideal. It has high purity, thermal shock resistance, and a softening point above the silicon melting temperature. Unfortunately, fused quartz has such a small thermal expansion that the silicon film cracks as it cools. This problem has been attacked by patterning with islands or moats before and after crystallization, by capping, and by using silicate glass substrates that match the thermal expansion of silicon. The relative merits of these methods are compared. Melting of the silicon film to achieve high mobility has been accomplished by a variety of methods including lasers, electron beams, and strip heaters. For low melting temperature glasses, surface heating with a laser or electron beam is essential. Larger grains are obtained with the high bias temperature, strip heater techniques. The low-angle grain boundaries characteristic of these films may be caused by constitutional undercooling. A model is developed to predict the boundary spacing as a function of scan rate and temperature gradient.

INTRODUCTION Two primary applications for crystallized silicon on amorphous substrates are dielectrically isolated devices and matrix displays. For dielectric isolation, most of the work has been done on thermally oxidized silicon wafers. This substrate is fully compatible with integrated circuit processing and eliminates thermal expansion problems. However, for the display application, silicon wafers are too small, too opaque, and too expensive. For a display medium such as a liquid crystal twist cell, a large area, transparent, inexpensive substrate is needed. Transistors fabricated on this substrate could provide threshold switching for each display element and decoding, multiplexing, and line driver circuitry. Thin film transistors have been studied for over 20 years. Various semi2 3 6 conductors, such as CdS,l CdSe, amorphous hydrogenated silicon, - and poly7 9 crystalline silicon, have been investigated. For display element switching, the mobility of these materials is adequate, but for peripheral circuitry, much higher mobility is needed. The necessary mobility can be obtained by melting a *Research performed by the authors was done at Bell Laboratories, Mat. Res.

Soc. Symp. Proc. Vol. 13 (1983) 0Elsevier Science Publishing Co.,

Inc.

Holmdel,

NJ.

582

deposited silicon film to produce large crystal grains. With this material, both switching and control circuitry potentially could be fabricated on a single substrate. The work on this problem has concentrated on transparent fused quartz substrates. In most respects, fused quartz is ideal. It is available with high purity and in arbitrarily large areas. It can be heated to the silicon melting point without deformatio