J. zyxwvutsrqp Phys. zyxwvutsrqp Chem. zyxwvut 1981, zyxwvut 85, zyxwvu 3529-3532 3529 Chemiluminescence and the Reaction of Molecular Fluorine with Silicon J. A. Mucha,* V. M. Donnelly, D. L. Flamm, and L. M. Webb Bell Laboratories, Murray Hiii, New Jersey 07974 (Received: December 16, 1980; zyxwvuts In Flnal Form: July 17, 1981) Molecular fluorine etches silicon with a rate zyxwvu = [(3.94 f 0.65) X 10-12]T1/2nFle-0.397eVlkT A/min, a process that is accompanied by gas-phase chemiluminescence which exhibits the same activation energy as the etch process. The observed temperature and pressure dependencies of these phenomena are consistent with a mechanism in which SiFz is an etch product that is involved in a chemiluminescent gas-phase reaction with Fz. The results extrend similar studies of silicon etching by atomic fluorine. There also is evidence of desorption products other than SiFz. The reaction between Fz and SiOzis measurable at elevated temperatures and pressure (-500 torr, 100 “C) and the Si:SiOzetch ratio is greater than 1001. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPON Introduction The reaction between molecular fluorine and single- crystal silicon has been studied previously in two labora- tories. Kuriakose and Margravel measured an etch rate of 0.042 mg/(cm2 min torr) (-2000 A/(min torr)) at 373 K and an Arrhenius activation energy of 0.52 eV (12 kcal/mol). However, Chen et a1.2 obtained etch rates of 160 A/(min torr) at this temperature and an activation energy of 0.35 eV (8 kcal/mol). The magnitude of the etch rate reported by Kuriakose and Margrave is large enough to make a measurable contribution in our fluorine-atom studies3and in some fluorine-containingplasmas presently employed in the processing of silicon electronic devices, especially at higher temperatures. We have therefore reexamined F2 etching of silicon and its oxide. The results reported here do not support either of the discordant previous works but do compare more favorably with those of Chen et. al. indicating that the F2 contribution to etching in fluorine plasmas is negligible. We also report the observation of a broad, visible chemiluminescence accompanying the etching of silicon by F2, which has a spectrum nearly identical with that observed during F-atom et~hing.~ Etch rates and the intensity of the chemiluminescence were measured as a function of temperature and fluorine pressure. The origin of the luminescence and the mechanism of the etch process are consistent with the interpretation of the F-atom re- sults; however, in the present study, there was evidence for the formation of other desorption products in addition to SiFP Experimental Section The experimental a p p a r a t u ~ , ~ , ~ ~ , ~ sample preparation, and handling procedures3 have been detailed previously. (1) A. K. Kuriakose and J. L. Margrave, J. Phys. Chem., 68, 2671 (1964). (2) M. Chen, V. J. Minkiewicz, and K. Lee, J. Electrochem. SOC., 26, 1946 (1979). (3) D. L. Flamm, V. M. Donnelly, and J. A. Mucha, J. Appl. Phys., 52, 3633 (1981). (4) (a) V. M. Donnelly and D. L. Flamm, J. Appl. Phys., 51, 5274 (1980); (b) V. M. Donnelly, D. L. Flamm, and J. A. Mucha, “Optical Emission from Transient Species in Halocarbon and Fluorosilicon Plasmas”, Extended Abstracts, 157th Meeting of the Electrochemical Society, St. Louis, MO, May 1980, Vol. 80-1, p 323; (c) V. M. Donnelly, D. L. Flamm, and J. A. Mucha, “Studies of Chemiluminescence Accom- panying Silicon Etching by F Atoms”, Proceedings of the 88th National Meeting of the American Institute of Chemical Engineers, paper 47C, Philadelphia, PA, June, 1980; (d) C. I. M. Beenakker, J. H. J. van Dommelen, and J. Dieleman, “Origin of the Luminescence Produced by the Reaction of Fluorine Atoms with Silicon”, Extended Abstracts, 157th Meeting of the Electrochemical Society, St. Louis, MO, May 1980, Vol. (5) D. L. Flamm, C. J. Mogab, and E. R. Sklaver, J. Appl. Phys., 50, 624 (1979). 80-1, p 330. 0022-365418112085-3529$01.25/0 Briefly, single-crystal silicon (100) samples were patterned with steam-grown thermal oxide, bonded to the end of a temperature-controlled aluminum rod, and positioned in- line with the wall of an insulated aluminum reaction cell. The F2 (Air Products, Technical Grade) was passed through a sodium bifluoride scrubber to remove any traces of HF. Silicon etch depths were measured by using a Sloan Technology Model 90050 Dektak stylus thickness monitor after dissolution of the oxide mask in HF. Oxide thickness was measured with a Nanospec AFT Model 174 micro- spectrophotometer. Chemiluminescence, originating in the gas phase above the Si(100) samples, was monitored through a 1-in. diam- eter sapphire window in the reaction-cell body. A cooled photomultiplier tube (RCA C31034) equipped with a Corning CS 2-61 long-pass red filter was used to measure the emission intensity. Spectra were obtained by using the same photomultiplier tube and a 0.3-m scanning monochromator (Heath Model EU-700). Optical collection efficiency was improved by using a pair of fused quartz lenses (f/1.7) to collimate the emission and focus it onto the entrance slits (2 mm) of the monochromator. A 450-Hz chopper between the two lenses and synchronous detection with a lock-in amplifier (Ithaco Model 39730) were em- ployed to minimize interference from background radia- tion. Because of the extremely weak emission levels, high pressures (5-15 torr) and elevated temperatures (>373 K) were required to obtain useful spectra. The spectra were digitized and filtered by using a statistical procedure de- scribed by Cleveland.6 Atomic fluorine, generated in a radio-frequency dis- charge upstream of the reaction cell,3was used to produce a reference spectrum48 for comparison with the chemilu- minescence during F2 etching. Fluorine atoms were also used in selected experiments to clean the silicon sample surface and thereby test for possible effects of surface contamination. Results and Discussion Chemiluminescence Spectra. Figure 1 shows spectra (uncorrected for spectrometer response) of the chemilu- minescence emanating from the gas phase above an un- masked silicon sample during etching with F atoms and F, at 473 K. Since emission levels are extremely low with the Fz etchant (a factor of - 10-3-104 of that observed with F atoms), a higher pressure (10.8 torr) was necessary to increase the intensity to a level permitting detection after dispersion. (6) W. S. Cleveland, J. Am. zyxwv Stat. Assoc., 74, 829 (1979). 0 1981 American Chemical Society