Industrial applications of low-temperature plasma physics* Francis F. Chent University of California, Los Angeles, California 900241594 (Received 14 November 1994; accepted 6 February 1995) The application of plasma physics to the manufacturing and processing of materials may be the new frontier of our discipline. Already partially ionized discharges are used in industry, and the performance of plasmas has a large commercial and technological impact. However, the science of low-temperature plasmas is not as well developed as that of high-temperature, collisionless plasmas. In this paper several major areas of application are described and examples of forefront problems in each are given. The underlying thesis is that gas discharges have evolved beyond a black art, and that intellectually challenging problems with elegant solutions can be found. 0 I995 American institute of Physics. 1. INTRODUCTION During the past four decades, the science of high- temperature and collisionless plasmas has grown explosively, fueled by the challenging problems in magnetic fusion, iner- tial fusion, and space plasma physics. As funds for basic research in fusion and space plasmas dwindle, it is fortunate that a new application of plasma physics has loomed large within the past five years-that of the use of low- temperature, partially ionized plasmas in manufacturing and materials processing. Indeed, this aspect of plasma physics may ultimately be the one with the greatest impact on our everyday lives. Although industrial applications have drawn great interest in plasma physics-the number of papers pub- lished monthly on plasma-related topics in semiconductor processing alone far exceeds the number in fusion research at its peak-the field has not benefited from the expertise of the cadre of physicists who have honed their skills in the classi- cal areas of plasma physics. Gas discharges are viewed by them as being an empirical discipline, devoid of elegance and beset with unnecessary complications. The purpose of this paper is to show that intellectually challenging problems can be found in low-temperature plasma physics, and that the complications of high collisionality and multiple species may be no more complicated and resistant to treatment than, say, instabilities in toroidal magnetic fields. The subject is very broad; and with due apologies to all the scientists working in this field, we must limit our coverage to a few representative examples in each case. The succeeding sections wilI discuss semiconductor processing, flat panel displays, ion implanta- tion, plasma polymerization and coating, thermal plasmas, and basic physics of low-temperature plasmas. II. SEMICONDUCTOR PROCESSING A. Physical mechanisms in etching The production of integrated circuits consists of repeated steps of deposition, masking, etching, and stripping to form and connect circuit elements like transistors and capacitors. Hundreds of chips can be made simultaneously on a silicon *Paper 9RV. Bull. Am. Phys. Sot. 39, 1749 (1994). %&ted speaker. wafer, which is typically 4-8 in. in diameter now and lo-12 in. in the near future. To put some five million transistors on a Pentium chip, for instance, the individual elements have to be below 0.5 ,um in size and moving toward 0.25 pm. Such resolution cannot be achieved without a plasma. All comput- ers and other electronic devices of the future will depend on plasma processing; yet, at the moment, very few plasma physicists have been involved. The plasma is needed for etching in at least three ways: (1) it produces the atomic species, usually Cl or F, which does the etching; (2) it prepares the substrate surface so that the etchant species can be more effective: and (3) it provides the directionality that allows the etching to proceed in a straight line. The plasma does not always have to touch the surface to perform its functions. The symbiosis between a plasma and an etching gas was demonstrated in the classic 1979 experiment of Coburn and Winters’ (Fig. 11, in which they showed that the etch rate of fluorine in an argon plasma was over an order of magnitude larger than with either the gas or the plasma alone. In addition to the etch rate, the plasma also provides profile control-the ability to etch a trench with straight sidewalls. Purely chemical etching would undercut the mask and produce a trench with rounded corners. By accelerating ions through a sheath, one can make them impinge on the mask and substrate at right angles, therefore affecting only the surface at the bottom of the trench, not the sides. This is known as anisotropic etching. However, isotropic chemical etching is still present to de- grade the trench profile (Fig. 2). By a fortunate accident, some of the etch products form a plastic polymer that depos- its on the trench walls and protects them from the chemical etchant, unless they are cleaned by a flux of energetic ions. Only by carefully balancing this “passivation” mechanism and the plasma-enhanced etch rate can one produce a square trench profile2q3 (Pig. 3). Four types of materials need to be etched in Ultra Large- Scale Integration (ULSI) processing: silicon (monocrystal- line or polycrystalline, doped or undoped), dielectrics (usu- ally Si02 or SiN,), metals (usually aluminum, tungsten, or molybdenum), and photoresist. Each of these involves differ- ent chemistries and different groups of experts. The pro- cesses that follow are not necessarily those used in any actual production line but will serve to give the flavor of what is 2164 Phys. Plasmas 2 (6), June 1995 1070-664)(/95/2(6)/2164/V/$6.00 0 1995 American Institute of Physics Downloaded 27 Nov 2001 to 128.97.88.10. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp? 4 l 5 I3 3 : 8 3 3 2 I 0 h+Ion Beam + XeF, Gas 4 1 1 L / RESIST 0 100 200 300 400 500 6CKL 700 800 900 L Time (set) FIG. 1. Evidence for the catalytic effect of plasma on chemical etching [reprinted with the permission of the American Institute of Physics (Ref. I)]. involved.4 Silicon can be etched by either fluorine or chlo- rine. In chlorine etching, the plasma first dissociates Cl, into Cl atoms. These react with Si to form SiCl, and SiCl,: Si+2 Cl-SiCl,, Sic&+2 C1--+SiC14. Here SiCl, is a gas and can be pumped out. In addition,-SiClz polymerizes to form [SiC12], , the