J. Phys. Chem. 1980, 84, 1445-1448 (101) (102) (103) (104) (105)
F. 0. Goodman, Surf. Sci., 27, 157 (1971). F. 0. Goodman, J. Chem. Phys., 5 5 , 5742 (1971). N. Garcia and J. Ibanez, J . Chem. Phys., 64, 4803 (1976). T. R. Knowles and H. Suhl, Phys. Rev. Lett., 39, 1417 (1977). F. 0. Goodman and N. Garcia, Phys. Rev. B, 20, 813 (1979).
(106) (107) (108) (109) (1 10)
1445
W. S. Liu, J. Chem. Phys., 59, 2490 (1973). C . Strachan, Proc. R . SOC.London, Ser. A , 158, 591 (1937). R. T. Allen and P. Feuer, ref 5, p 101. J. R. Manson and J. Tompkins, ref 20, part 1, p 603. L. G. Carpenter, private communication.
Silicon Production from the Interactfon between a Si0 Molecular Beam and a Tungsten Carbide Surface Owen K.
1.Wut
and Richard P. Burns*
Department of Chemistry. University of Iiiinois at Chlcago Circle, Chicago, Iiiinols 60680 (Received October 1, 1979)
The chemical reaction between a Si0 molecular beam and a tungsten carbide surface is investigated by Auger electron spectroscopy (AES) and flash desorption mass spectrometry (FDMS). For Si0 adsorbed on a carburized tungsten surface, AES shows that the silicon peaks at 64, 78, and 92 eV dominate. (The 92-eV peak is characteristic of elemental silicon, the 64- and 78-eV peaks are characteristicof higher oxidation states of silicon.) FDMS shows that there are two desorption peaks for mass 28. The low temperature peak is identified as CO (T, 1100 K, AW 37 K, P = 30.3 K s-'), which is unlike the flash desorption spectrum of CO from a polycrystalline clean W surface. Desorption from this state is described by a firstcorder process, but the activation energy increases with increasing coverage. This is interpreted as due to complex (CO + SiO) formation on the surface, and is confirmed by a separate coadsorption experiment. The other peak at higher temperature is identified as Si ( T , = 2020 K, AW = 155 K, /3 = 30.3 K d), and these data indicate a simple first-order desorption process with Ed = 123 kcal mol-l and u = 1.2 X 1013s-l. If the surface temperature is increased to 1500 K, AES shows only a 92-eV peak for silicon and its peak-to-peakheight is comparable to that for a pure silicon crystal. Thus, we conclude that silicon monoxide molecules were reduced by the tungsten carbide surface to elemental silicon.
-
-
Introduction The reduction of silicon oxides has long been an interesting subject not only in the semiconductor industry but also in the solar photovoltaic program.lV2 In this study, the surface chemical reactions between chemisorbed silicon monoxide and a tungsten carbide surface are investigated by Auger electron spectroscopy (AES) and by flash desorption mass spectrometry (FDMS). On one hand, AES provides (a) information on the elemental surface composition and (b) information on the oxidation state of s i l i ~ o n . On ~ ~ ~the other hand, FDMS provides (c) an identification of the gaseous species desorbing from the surface and (d) a determination of the kinetics of the desorption process. Thus, in this study, FDMS is used in combination with AES and the use of the two methods leads to a more complete description of chemical reactions on the surface. Experimental Section The ultrahigh vacuum system and sample preparation have been described e l ~ e w h e r e .Briefly, ~ ~ ~ the apparatus includes an Extranuclear 270-9 quadrupole mass spectrometer and a P:HI 15-120 hemispherical four grid LEED/Auger system. Silicon monoxide molecules were produced from a resistively heated oven containing Si(s) + Si02(s),315The beam of SiO(g) molecules could be turned on and off by means of a molecular beam flag3 which was inserted between the oven and the surface. The carburized tungsten surface was deposited by cracking ethylene molecules6Jon a clean tungsten surface. The ethylene was introduced onto the W surface though a stainless-steel dosing syringe directed at the front face of the surface.8
'
Gould Inc., Gould Laboratories, Electrical and Electronics Research, 40 Gould Center, Rolling Meadows, Ill. 60008. 0022-3654/80/2084- 1445$0 1.OO/O
An additional flag (mass spectrometer shutter) opens and closes the ionizer entrance aperture. When the mass spectrometer shutter is closed, condensable gaseous species (e.g., Si and SiO) cannot enter the mass spectrometer ionization region and are not detected. On the other hand, noncondensable gases (e.g., CO and C02) can still enter the ionizer and are detected with the mass spectrometer shutter closed. Thus, the use of the mass spectrometer shutter allows the differentiation between the condensable and noncondensable gaseous species at mass 28 (Si and CO) and at mass 44 (Si0 and C 0 2 ) . In addition the species identification included appearance potential measurements. In these experiments the flux scale was calibrated as follows. First, the tungsten surface was replaced with an A1203(1012) surface by means of a sample manipulator. Then the S i 0 molecular beam was allowed to strike the A1203surface. The angular distribution of the S i 0 molecules desorbed from the high temperature surface was cosine for both the tungsten and A1203surfaces. Next, the desorption rate of Si0 was compared with the known rate of evaporation of the A1203 (1012) surface at its melting point in order to calibrate the S i 0 molecular beam fluxa3 Finally, the tungsten surface was repositioned into the path of the Si0 molecular beam and the surface coverage of Si0 on the tungsten surface was determined by sticking probability measurements and by integrating the flash desorption peaks. This procedure has been described in detail previ~usly.~
Results and Discussion Auger Electron Spectra of the SiO(g) + W(carburized) System. Figure 1shows the Auger electron spectrum of a carburized W surface. This spectrum indicates that the carbon exists in the form of a surface metal ~arbide.~JO For SiO(O=l) adsorbed on a carburized tungsten surface 0 1980 American Chemical Society
1440
The Journal of Physical Chemistry, Vol. 84, No. 12, 1980
Wu and Burns
AUGER ELECTRON SPECTRA O F A CARBURIZED TUNGSTEN SURFACE
SILICON
TUNGSTEN
FLASH DESORPTION SPECTRA OF MASS28 FROM SiO(e=l)+W(CARBURlZEc
-
CARBON
OXYGEN
252eV
co
3
Tm=11450K
a: > a:
t a:
Q
c
MODULATION VOLTAGE 163eV
t
w 4v c 1ov 0 1ov
168eV
---I
272eV
ENERGY (eV)
50
70
90
150
170
190
240
280
260
3 0 0 490
TEMPERATURE('K) I
510
530
Flgure 1. Auger electron spectra of a carburized tungsten surface. Primary electron beam energy, E, = 3 keV. AUGER ELECTRON SPECTRA OF SILICON MONOXIDE ( O = i ) ADSORBED ON A CARBURIZED TUNGSTEN SURFACE
I
I
1000
1400
1200
I
I
1600
1800
I
2200
2000
+
Flgure 3. Flash desorption spectra of mass 28 from the SiO(8=1) W(carburized) system. Heating rate p = 30.3 K si. The ionizing electron energy (E,) in the quadrupole mass spectrometer ion source is 16 eV. FLASH DESORPTION SPECTRA OF MASS 28
SILICON
TUNGSTEN
CARBON
FROM SlO(~=l)+W(CARBURlZED)
OXYGEN
ELECTRON ENERGY =10eV IONIZATION POTENTIAL
505eV 272eV MODULATION VOLTAGE SI 4 v w 4v
168eV
92eV
5clLI
E;
---ENERGY(eV)
150
170
190
240
260
260
3 0 0 480
500
1
520
+
Flgure 2. Auger electron spectra of the SiO(S=I) W(carburized) system. Primary electron beam energy, E, = 3 keV.
(Figure 21, AES shows that the silicon peaks at 64,78, and 92 eV dominate. (The 92-eV peak is characteristic of elemental silicon, the 64- and 78-eV peaks are characteristic of higher oxidation states of ~ i l i c o n . )The ~ ~ ~carbon peak at 272 eV is similar to the Auger spectrum of CO adsorbed on a tungsten surface? This spectrum indicates that the carbon on the surface has been oxidized by the silicon monoxide molecules. Flash Desorption Spectra of Mass 28 (CO and Si) and Mass 44 (SiO)from the SiO(g) W(carburized)System. Figure 3 shows the flash desorption spectrum of mass 28 from the SiO(d=l) W(carburized) system. This spectrum indicates that there are at least two desorption peaks for mass 28. By using the mass spectrometer shutter8and the appearance potential (see Figure 4) of the molecule, we identified the low temperature peak as CO and the high temperature peak as Si. Figure 5, curves A-C, shows the flash desorption spectrum of mass 44 for different Si0 coverages on a carburized W surface. Again, by using the ionization potential and mass spectrometer shutter (shown in Figure 5, curve D), we can identify the desorption peak as SiO. This desorption peak arises from the unreacted Si0 on the surface. Figure 6 shows the flash desorption spectrum of CO and Si from different coverages of Si0 adsorbed on a carburized W surface. The desorption of CO ( T , 1100 K, AW 37 K, p = 30.3 K s-l) from the carburized surface described here is unlike the flash desorption spectrum of CO from a polycrystalline clean W surface (see Figure 7). De-
I
C 1000
I
1200
I
1400
1600
I
1800
I
2000
I
2200
I
240(
TEMPERATURE (OK)
Flgure 4. Flash desorption spectra of mass 28 from the SiO(8=1) 4W(carburized) system. E, = 10 eV.
,
FLASH DESORPTION SPECTRA 0
+
+
-
-
(D) 1
1000
OFF
I
1200 1400 1600 TEMPERATURE (OK)
1800
I
Flgure 5. Flash desorption spectra of SiO(mass 44) from the SiO(g) W(carburized) system: (A) SO(& 1.0); (B) Si0(8=0.57); (C) SiO(8=0.26);(D) SiO(O=l.O). E, = 16 eV.
+
sorption from this state is described by a first-order process, but the activation energy increases with increasing coverage. The flash desorption spectra of Si (T,= 2020 K, A W = 155 K, p = 30.3 K s-l) shown in Figure 6 indicate that the desorption is a first-order process with activation energy Ed = 123 kcal mol-1 and preexponential factor v =
The Journal of Physical Chemistry, Vol. 84, No. 12, 1980
SI0 Beam-Tungsten Carbide Interaction
1447
FLASH DESORPTION SPECTRA OF CO AND St0 FROM THE COADSORPTION
FLASH DESORPTION SPECTRA OF CO AND SI FROM SIO+W(CARBURIZED)
RR-SEC EXPERIMENT S i 0 ( ~ = O . ~ ) ~ C O ( ~ X ~ O ~ ~ T ODOSE)+W(CLEAN)
si T,=2020°K
LY: 1400
-
1800
2000
I
I
I
1400
800
1000
I
1000
to0
1200
I
1200
1400
TEMPERATURE (OK)
2200
TEMPERATURE (OK)
+
Flgure 6. Flash desorption spectra of CO and Si from the SiO(g) W(carburized) system: (A) SiO(6= 1-0); (B) Si0(6'=0.57); (C) SiO(6 =0.26). E, = 16 eV. FLASH DESORPTION SPECTRUM OF CO FROM COtW(CLEAN)
Figure 8. Flash desorption spectra of CO(mass 28) and SiO(mass 44) from the coadsorption experiment. First dose with CO(2 X lo-' torr s), second dose with Si0(8=0.5). E , = 16 eV.
n
AUGER ELECTRON SPECTRA OF SiOtW(CARBURIZED), SPECTRA ARE TAKEN AFTER CO DESORPTION
4Iv fi TUNGSTEN
'LICON (A)2x10-7TORR-SEC DOSE CO AT T = 8 W K Tm=1296'K
OF
CARBON
OXYGEN
163eV
178eV
MODULATION VOLTAGE
168eV
t
Si:3V
w 4v c:5v 0:5V
-- - ENERGY (eV)
70
1000
1200
1400
1600
90
110
150
170
190
250
270
290 480
500
520
Figure 9. Auger electron spectra of the Siqg) 4- W(carburized) system, spectra are taken after CO and unreacted Si0 desorption. Primary electron beam energy, E, = 3 keV.
1800
TEMPERATURE (KO)
+
Flgure 7. Flash desorption spectra of CO from the CO(g) W(clean) system: (A) 2 X torr s dose of CO at T = 818 K; (B) 6 X lo-' torr s dose of 60 at 7'= 818 K. E, = 16 eV.
1.2 X 1013s-l, The data determined here for desorption of silicon from tungsten may be compared with that determined previously.11J2 It should be noted that at the same coverage the peak temperature and peak shape for CO in Figure 6 are, within experimental error, identical with that for S i 0 in Figure 5. This comparison suggests that a surface complex is formed between CO and SiO. Moreover, the narrow CO and Si0 peak widths indicate a rather high preexponential factor for desorption of CO and SiO. This also supports the existence of a airface complex. Flash Desorption, Spectra of CO and S i 0 from the Coadsorption of CO and Si0 Experiment. In order to further confirm the existence of the CO and Si0 complex on the W surface, a separate coadsorption experiment was carried out. Figure 8 shows the flash desorption spectra of CO and Si0 from the sequential coadsorption experiment [C0(2X lo-' torr s) + Si0(6=0.5)]. This spectrum indicates that the peak shape for both Si0 and CO is identical with that observed in Figures 5 and 6. In addition, the peak temperature ( T , = 1145 K) of CO is,
within experimental error, equal to that (T, = 1145 K) of SiO. Thus, the coadsorption experiment supports the existence of a surface complex. Silicon Production on a Tungsten Surface. Based on the above description of the surface processes, one can predict that, if the surface temperature is increased to 1500 K, the CO and unreacted Si0 will be desorbed from the SiO(g) + W(carburized) system leaving elemental silicon on the surface. The Auger electron spectrum shown in Figure 9 confirms this prediction since only the 92-eV peak characteristic of pure silicon is observed. The 92-eV peak-to-peak height is comparable to that observed for a pure silicon crystal. Reaction Mechanism. From the above results, we propose the following reaction steps (mechanism):
--
SiO(g) SiO(ads) (1) SiO(ads) + C(ads) CO(ads) + Si(ads) (2) SiO(ads) CO(ads) (Si0 + CO) (surface complex) (3)
+
-
(Si0 + CO) (surface complex)
flash
CWg) + SiO(g) (4) (5)
1448
J. Phys. Chem. 1980, 84, 1448-1453
Conclusion Thus, we conclude that silicon monoxide molecules were reduced by the carbide surface to elemental silicon and that the carbide reduction of Si0 offers a possible method for producing.,ultrathin films of silicon on various surfaces. Acknowledgment'This research was in part by the National Science Foundation Grant No. GP-34320. 0. K. T. Wu acknowledges - a 1978-1979 American Vacuum Society-Scholarship. References and Notes (1) H. Kelly, Science, 199, 634 (1978). (2) L. P. Hunt in "Semiconductor Silicon", H. R. Huff and E. Sirti, Ed.,
The Electrochemical Society, London, 1977, p 803. (3) 0. K. T. Wu and R. P. Burns, Surf. Sci., 77, 626 (1978). (4) C. C. Chang in "Characterizatlon of Solid Surfaces", P. F. Kane and G. B. Larrabee, Ed., Plenum Press, New York, 1974, pp 509-575. (5) K. F. Zmbov, L. L. Ames, and J. L. Margrave, H@hTemp. Sci., 5, 235 (1973). . . (6) R. R. Rye and R. S. Hansen, J . Chem. Phys., 50, 3585 (1969). (7) B. D. Barford and R. R. Rye, J . Chem. Phys., 60, 1046 (1974). (8) 0. K.T. Wu, Ph.D. Dissertation, University of Illinois at Chicago Circle, 1979. (9) T. W. Haas, J. T. Orant, and G. J. Dooiery in "Adsorption-Desorption Phenomena", Ricca, Ed., Academic Press, New York, 1972, p 359. (10) J. B. Benziger, E. I. KO, and R. J. Madix, J. Catal., 54, 414 (1978). (11) 0. F. Swenson and M. K. Sinha, J. Vac. Sci. Techno/., 9, 942 (1972). (12) H. Newmann, Ann. Phys. (Leiprig), 18, 145 (1966).
Interaction of Gaseous Hydrogen Chloride and Water with Oxide Surfaces. 2. Quartz' Yoonok Kang, Jean Ann Skiles, and J.
P. Wightman"
Chemistry Department, Virginia Polytechnic Institute & State University, Blacksburg, Virginia 2406 1 (Received August 16, 1979)
The influence of temperature, pressure, and outgas conditions on the adsorption of hydrogen chloride and water vapor on quartz has been studied. Characterization of the quartz adsorbent was performed using X-ray powder diffraction,differentialscanning calorimetry, scanning electron microscopy (SEM),infrared spectroscopy, BET nitrogen surface area measurements, and electron spectroscopy for chemical analysis (ESCA). Water vapor adsorption isotherms at 30,40, and 50 "C were measured after outgassing at 100 and 400 "C. Both outgas temperature and adsorption temperature influenced the adsorption of water vapor on quartz. The water vapor adsorption was completely reversible on reoutgassing at 100 "C. Hydrogen chloride adsorption isotherms at 30 "C were measured on quartz after outgassing at 100, 200, and 400 "C. The adsorption was only partially reversible on reoutgassing at 100 "C indicative of a moderate adsorption process. The heats of immersion of the quartz, outgassed from 100 to 400 "C were determined in both water and hydrochlorfc acid solutions of varying concentrations to complement the vapor phase measurements. ESCA spectra of quartz were obtained before and after exposure to hydrogen chloride to elucidate the nature of the adsorbed chlorine species.
Introduction The NASA Space Shuttle uses a solid rocket propellant composed of an ammonium perchlorate oxidizer,powdered aluminum, and other additives contained in a polybutadiene binderO2*The major exhaust components with percentages are A1203(30%), CO (24%), and HC1 (21%) amounting to about 233 tonsaZb During launch at the NASA-Kennedy Space Center it is estimated3that perhaps as much as 50 tons of soil principally silica is swept upwards into the atmosphere. Water vapor from the atmosphere and gaseous exhaust products including hydrogen chloride could adsorb onto the silica surface. The question arises as to the extent to which silica can act as a sink for hydrogen chloride. The objective of this research was to study the adsorption of hydrogen chloride and water on quartz which was chosen as the adsorbent since it constitutes about 90% of the soil at NASA-KSC. We have previously reported4 on the adsorption of hydrogen chloride and water on a- and y-alumina. Experimental Section Matheson reagent grade (purity 99%) anhydrous hydrogen chloride and Matheson ultrahigh purity (99%) nitrogen were used as adsorbates. The nonporous quartz adsorbent, Min-U-Sil5, was obtained from the Pennsylvania Glass Sand Corp. According to the manufacturer, the production of Min-U-Si1 involves dry grinding of the source material to fine particle size followed by air separation. Min-U-Si1 is not heated above 300 "C during manufacture and is not exposed to water or oiljhydro0022-3654/80/2084-1448$0 1.OO/O
carbon vapors during and after grinding. The sample after receipt from the manufacturer was given only the pretreatment(s) described below. X-ray powder diffraction patterns were obtained with a Diano-XRD 8000 diffractometer using Cu Ka radiation. Calculated d spacings of 0.429 nm for the (100) plane and 0.335 nm for the (101)plane agree well with the reported values for cu-q~artz.~ Edmonds has reported6 a similar result for Min-U-Sil. Thermograms of Min-U-Si1 5 were obtained using a DuPont 990 thermal analyzer. An exothermic peak a t approximately 535 "C observed in the DSC (differential scanning calorimetry) thermogram of MinLU-Si1 is characteristic of a-quartz.' Infrared spectra were obtained on dry powders, on KBr pellets, and on concentrated aqueous slurries of Min-U-Si1 using a Perkin-Elmer 283 grating spectrophotometer. The IR spectra matched very well with the spectrum of quartz reported in the literature.8 The doublet at 800 and 780 cm-l characteristic of a-quartz was particularly well resolved. Surface areas of the quartz were determined by the BET method as described by Gregg and Singg using a Micromeritics Model 2100 D Orr surface area pore volume analyzer. Prior to the surface area measurements, the quartz was outgassed for 2 h at 100 and 400 "C at 10" torr, which corresponded to the range of thermal pretreatment used in the hydrogen chloride and water vapor adsorption studies. Particle agglomeration in quartz was noted using an Advanced Metals Research Corp. Model 900 scanning electron microscope (SEM) operating at 10 kV. Energy 0 1980 American
Chemical Society
