J. Phys. Chem. 1994, 98, 259-263
259
IRAS Study of CO Adsorption on S-Modified Mo(ll0) Surfaces W. Kevin Kubn,+ J.-W. He,*and D. Wayne Goodman’ Department of Chemistry, Texas A&M University, College Station, Texas 77845 Received: July 12, 1993; In Final Form: October 25, 19930
Infrared reflection absorption Spectroscopy (IRAS)was used to study the adsorption of CO on a S-modified Mo( 110) surface. As the surface S coverage is increased to 0.5 ML, the sulfur overlayer structure changes from ~ ( 2 x 2 )to ~ ( 2 x 2 ) . These changes can be observed in the IRAS spectra of adsorbed CO. In addition, it was noted that in the ~ ( 2 x 2 ) overlayer s sulfur actedmainly as a site blocker, whereas in the ~ ( 2 x 2 ) overlayer s electron-withdrawing effects became apparent. Finally, the dissociative adsorption of CO was seen to occur on both clean Mo(l10) and p(ZXZ)S/Mo(llO) but not on c(2X2)S/Mo(llO).
Introduction
COISIMo(l10) ~
The interactions between sulfur and transition-metal surfaces have been the object of many studies over the past two decades. This interest is due in part to the pronounced effect that surface sulfur has on molecular adsorption and reaction processes on An additional reason for this interest is the utility of molybdenum sulfides as hydrodesulfurizationcatalysts during hydrocarbon conversion reactions.*g9 In an attempt to understand these effects, numerous studies have been conducted on the structure of surface S overlayers.l*l6 These studies have sought to determine the S adsorption sites as a function of S coverage, which would allow insight into the observed inhibition of adsorption (poisoning) or reaction selectivity changes observed on S-modified surfaces. In thisstudy, the adsorptionof carbon monoxideon a presulfided Mo(110) surface was examined using infrared reflectionabsorption sptctroscopy(IRAS) and thermal desorption methods. The results obtained in this study will be discussed in terms of the site-blocking (poisoning) and electron-withdrawing nature of the S adatoms. Experimental Section
2045
~~
ANNEALED TO 1400K
0.50 ML
8 z
3 8m
a , 0.32 ML
a .
2200
2100
2000
1900
1800
WAVENUMBER (cm-’) Figure 1. IR spcctra of CO on a Mo(ll0) surface at 90 K with the indicated sulfur coverages.
The experiments were performed in an ultrahigh-vacuum chamber (base pressure 55 X 1&lo Torr) equipped for IRAS, Auger electron spectroscopy, (AES), low-energy electron diffraction (LEED), and temperature-programmed desorption (TPD).This apparatus has been described in detail e1~ewhere.l~ The sample was spot-welded to two Ta wires on the back face of thecrystal, allowing resistive heating to 1500K and liquid nitrogen cooling to 90 K. An electron beam assembly allowed heating to 2000 K with the sample temperature being monitored with a W-5% Re/W-26% Re thermocouple. The crystal surface was cleaned using a relatively simple procedure: oxidation at 1500 K in 2 X 10-7 Torr of oxygen followed by annealing in vacuum at a temperature of 2000 K. After this procedure, C, 0,and S surface impurities were determined to be less than 1%. The cleanliness and long-rangeorder of the surface were verified with AES and LEED, respectively. The IRASspectra were acquired for 256 scans in the single reflection mode at a resolution of 4 cm-’ and an incidence angle of 8 5 O from the surface normal. The spectra shown are raw data, corrected only for the base line. To deposit sulfur onto the surface, the sample was dosed with HIS. Heating to >500 K caused the H2Sto decompose with the + Present address: Frank J. Sciler Research Laboratory (FJSRL/NC) 2354 Vandenkrg Dr.,Suite 2A35, USAF Academy, CO 80840-6272. t Prawnt address: DuPont Chemicals, Research & Development, Sabine River Laboratory, P.O. Box 1089 (FM-1006), Orange, TX. * To whom wrreapndence should be addressed. .Abstract published in Advunce ACS Abstracts, December IS, 1993.
hydrogen desorbing, leaving only S on the surface. The H2S used was cleaned by freezing the H2S in liquid nitrogen and then vacuum-distilling off the dissolved gases. The CO used in this work was obtained from Matheson, was 99.99% pure, and was used without further purification. Gas exposures in this work are given in terms of langmuirs (1 langmuir = 1 X 10-6 Toms). All adsorbate coverages are referenced to the number of Mo surface atoms (1.428 X 1015atoms/cm2) with one sulfur atom per Mo atom corresponding to 0s = 1.00 monolayer (ML).
Results and Discussion IR spectra of CO adsorbed on various coverages of S on a Mo( 110) surface are shown in Figure 1. The sulfur was deposited by exposing the sample to H2S at 90 K. Subsequent heating to 1400 K causes the H2S to decompose into H2 (which desorbs at 400-500 K) and elemental S and forms a stable, ordered S overlayer. The IRAS spectra shown in this figure were then acquired at 90 K after a saturation CO exposure. Figure 2 shows the peak positions and total integrated peak area for the spectra in Figure 1. As the sulfur coverage is increased from 0.0 to -0.35 ML, the CO peak position shifts from 2035 to -2010 cm-’. At a sulfur coverage of 0.29 ML, a new peak appears at 2037 cm-I. The intensity of this peak increases with further increasesin S coverage. At a sulfur coverage of 0.5 ML, this new peak is the only peak observable with a frequency of 2045 cm-1.
0022-3654194 f 2098-0259%04.50/0 0 1994 American Chemical Society
Kuhn et al.
260 The Journal of Physical Chemistry, Vol. 98, No. 1 , 1994 COISIMo(1 I O )
COISIMo(1I O )
1
s i r
I
1
ai
A
a
s
I
J
. 0.0
0.1
0.2
0.3
0.4
2010 0.5
SULFUR COVERAGE (ML)
Figure 2. Peak positions and integrated intensities for the spectra shown in Figure 1.
0
a
0 0
0
0
0
0
0
0
2qrs
a
0 0
es = 0.20 ML
0
co EXPOSURE
-
10 L
0 0
6L 0
4L a
0
3L
Figure3, Reprcsentationsofthep(2X2)andc(2X2) LEED patternsand corresponding real space structures.
The sulfur coverages were determined from the ~ ( 2 x 2LEED ) pattern (see Figure 3) that was observed for the highest S coverage obtainableusing theabove S depositionmethod. A ~ ( 2 x 2LEED ) pattern has a coverage defined to be 0.50 ML. The S coverage assignment is further supported by a ~ ( 2 x 2 LEED ) pattern at 8s = 0.25 ML. These LEED patterns were then used to calibrate AES measurements. S/Mo AES peak intensity ratios were then used to determine S coverages in all subsequent experiments. Previous work on S/Mo( 110) supports the use of LEED patterns to establish S coverages on Mo( 110).14 ThecontinuousshiftoftheCGMOpeakfrom2035to -2010 cm-1 as the sulfur coverage is increased from 0.00 to -0.35 ML can be explained by a simple site-blocking mechanism. Figure 2 shows that as the S coverage is increased from 0.00 to -0.3 ML, the total integrated CO intensity decreases to approximately 75% of its original value. This implies a 25% reduction in CO coverage. If the adsorbed S is acting simply as a site blocker, a 25% reduction in CO coverage would indicate a S coverage of 0.25 ML. This is in reasonable agreement with the 0.3 ML S coverage measured and supports a site-blocking mechanism. To further explore CO coverage changes as a function of S coverage, CO TPD spectra were obtained as a function of S coverage (see Figure 4). This figure also shows that the amount of adsorbed Codecreases as the S coverageis increased. The CO TPD results indicate that, at a S coverage of 0.2 ML, the CO coverage is only 75% of the amount on clean Mo(ll0). These TPD results are
2L
Is196 1L L
2200
2100
2000
1000
1800
in good agreement with the IR results and further support the previous data indicating that, for sulfur coverages up to -0.3 ML on Mo( 1 lo), sulfur acts primarily as a site blocker (poison). It is well-known that on almost all surfaces CO peak frequencies shift as the CO coverage changes.’*-20 This shift in peak position is usually toward higher frequency as the CO coverage increases and is typically explained as being due to a reduction in 2r backdonation upon increasing CO coverage.1s-20 For CO adsorption on S/Mo(100) surfaces, this shift toward higher frequency was also seen. Figures 5-7 show the CO exposure dependence of the CO IRAS spectra at S coverages of 0.20, 0.35, and 0.50 ML, respectively. For these data, the Mo( 110) sample was exposed to H2S at 90 K and then flashed to 1400K to desorb the hydrogen and anneal the overlayer. CO exposureand IR spectral acquisition were at 90 K. For all three S coverages, there was an approximately 15-cm-1 increase in the CO stretching frequency as the CO coverage was increased to saturation (10 langmuirs). Thus, it is implicit that a decrease in the CO peak position would result from a decrease in the CO coverage. It is likely, therefore,
CO Adsorption on S-Modified Mo( 110) Surfaces COISIMo(ll0)
2014
2200
2100
EXPOSURE
2000
1900
1800
WAVENUMBER (cm-l) Figure 6. IR spectra of CO adsorbed on 0.35 ML of S on Mo(l10) at 90 K as a function of CO exposure. COISIMo(l10) 7
e,
20454
0.50 ML
2L
1L
2200
2100
2000
loo0
1800
WAVENUMBER (cm") Figure 7. IR spectra of CO adsorbed on 0.50 ML of S on Mo(110) at 90 K as a function of CO exposure.
that the red shift in peak frequency seen in Figure 2 as the S coverage is increased to -0.3 ML is due to a decrease in CO coverage. The data in Figures 5-7 indicate that CO adsorbed on ~ ( 2 x 2 ) S domains has a peak frequency of -2040 cm-I. This peak frequency is likely due to the electron-withdrawingnature of S. Assuming that the most likely position for S atom on the Mo(1 10) surface is in the hollow sites,14real space surface structures corresponding to the ~ ( 2 x 2 and ) ~ ( 2 x 2 LEED ) patterns can be deduced and are shown in Figure 3. It is apparent from the structure shown in Figure 3b for 0.5 ML of S on the Mo( 110) surface that the likely sites for CO adsorption would be on top between two S atoms or in a 4-fold hollow site with four adjacent S atoms. Due to the steric crowding at the a-top site, it is likely that the preferred site would be the hollow site. The stretching frequency for CO adsorbed on this surface, however, shows a peakat 2045 cm-l (see Figure 7). This is a rather high frequency to be associated with CO adsorbed in a hollow site. In a previous HREELS study of CO adsorption on S/Ni( loo), CO adsorbed on a ~ ( 2 x 2 )structure s was observed to have a frequency of 21 15 cm-l and was attributed to adsorption in a 4-fold Ni hollow between four S atoms with highly reduced back-donation into the 2rorbital due to theadsorbed sulfurss It is, therefore, plausible
The Journal of Physical Chemistry, Vol. 98, No. 1, 1994 261 that the peak at 2045 cm-l is due to CO adsorption in hollow sites with the unusually high frequency being due to the electronwithdrawing effect of the adjacent sulfur atoms. The spectra shown in Figure 6 for CO adsorption on a Mo(1 10) surface with 0.35 ML of S clearly have two distinct features One is at -2010 cm-l and the other at -2030 cm-1. The lowfrequency feature has been attributed to CO adsorbed on a p(2X2)S-covered surface while the high-frequency peak has been attributed to CO adsorbed on a c(2X2)S-coveredsurface. The fact that both features are present in Figure 6 clearly indicates that, at sulfur coverage of 0.35 ML, domains of both ~ ( 2 x 2 ) s and ~ ( 2 x 2 ) sare in coexistence on the surface. Due to the proximity of the two peaks, there is likely a strong transfer of intensity from the low-frequency peak to the high-frequency peak.21-23 As the CO coverage increases, this effect will increase due to the increased proximity of the CO molecules. "Intensitytransfer"'8J1.22 and " s ~ r e c n i n g " ~effects ~ - ~ ~ can significantly perturb observed IR peak intensities. The "intensity-transfer" effect is a strong dipoleciipole coupling between CO molecules adsorbed on a surface that results in the transfer of intensity from a low-frequency mode to a high-frequency mode.18121923 A good example of this effect is in the spectra of coadsorbed I Z C Wand 12C'80where the observed IRAS peaks are at equal intensity for a 5% I2Cl60/95%12C180 mixture.23 In addition, a suppression in the IRAS intensity of a CO-Pt( 111) peak has been observed when CO is coadsorbed with a highly polarizable species like CHpOH, H20, or Xe.26 The polarizable species can induce a considerable shielding of the local electric field around the CO molecule, "screening" its dipole moment. More importantly, this effect has also been observed for CO adsorbed on submonolayer Cu and Ag on Pt(l1 l)?4.2sCu on R u ( O O O ~ ) ?and ~ * ~Ni ~ on Mo(1 10).29,30Due to the polarizability of the overlayer adatoms upon CO adsorption, the CO-adatom ensembles are effective at screening the intensity of chemisorbed CO on the overlayer-free s~bstrate.~~-*s Thus, intensity-transfer effects make it likely that the low-frequency peak is the majority surface species, even at high CO coverages. One would intuitively expect the lowfrequency ( ~ ( 2 x 2 ) s peak ) to be larger since at a S coverage of 0.35 ML, 40% of the surface should be ~ ( 2 x 2 ) domains s while the remaining 60% would be ~ ( 2 x 2 ) domains. s However, these intuitive abundances do not account for the fact that, for the 1-langmuir exposure shown in Figure 6, a generous estimate of the amount of CO on ~ ( 2 x 2 ) domains s (high-frequency peak) would be 25% of the total surface CO. Thus, it would appear that the initial CO sticking probability is higher on the ~ ( 2 x 2 ) s domains than it is on the ~ ( 2 x 2 ) domains. s IR spectra of CO on S/Mo(llO) as a function of sample temperature are shown in Figures 8-10. For these figures, the Mo(110) surface was exposed to the HzS at 90 K and flashed to 1400 K to desorb the hydrogen and anneal the overlayer. CO exposure was at 90 K, and spectral acquisition was at the indicated temperatures. All three figures show the expected red shift in stretching frequency as the CO coverage is reduced due to desorption. The desorption temperatures indicated in the IR data are in excellent agreement with the TPD data shown in Figure 4. For S coverages of 0.35 ML or below, molecular desorptionoccurs at 330 K; however, for 0.5 ML of S,desorption occurs at 300 K. This decrease in desorption temperature for CO adsorbed on S/Mo( 110) surfaces with S coverages in excess of 0.35 ML is a further indication of a sulfur interaction withdrawing charge from the Mo surface. As discussed above, the electron-withdrawingeffect of the adsorbed S will reduce the charge donation from the Mo atoms to the adsorbed CO. This will lead to a weaker metal-CO bond (lower desorption temperature). The lower desorption temperature for CO bound to thec(2X2)S phase (high-frequency peak) comparedto CO bound to the ~ ( 2 x 2 phase ) (low-frequency peak) is clearly evident in Figure 9. In addition, by applying the Redhead approximation)'
Kuhn et al.
262 The Journal of Physical Chemistry, Vol. 98, No. 1, 1994 COISIMo(l10)
CO/S/Mo(lI O )
e,= 0.20 ML
IT
I
es = 0.50 ML TEMP.
W
o ^z U m
a
8m U
2WK
96K
00K
2200
2100
2000
le00
1800
WAVENUMBER (cm-')
Figure 8. IR spectra of CO on 0.20ML of S on Mo(110). CO exposure was at 90 K with spectral acquisition at the indicated temperatures. CO/SIMo(l I O )
e, = 0.35 ML
nr.
2200
2100
2000
le00
1800
WAVENUMBER (cm-') Figure 10. IR spectra of CO on 0.50ML of S on Mo(110). CO exposure was at 90 K with spectral acquisition at the indicated temperatures.
not only decreases the amount of dissociatively adsorbed CO but also facilitates the recombinative dissociation of the CO. In the previous study, the reduced dissociativeadsorption was explained as being due to an increase in desorption rate, not a decrease in the desorption rate c o n ~ t a n t .The ~ lower CO desorption temperatures at higher S coverages observed in this work are in qualitative agreement with this earlier study. The lack of CO dissociation on the ~ ( 2 x 2 ) soverlayer was observed in the IR spectra oftheCO adsorbed on this surface. Repeatedadsorptiondesorption cycles produced no IR features at -2100 cm-' due to C buildup on this surface. On the ~ ( 2 x 2 ) or s clean Mo( 1 10) surfaces, on the other hand, adsorption-desorptioncycles produced a pronounced peak at -2100 cm-I, due to carbon buildup.
Summary and Conclusions
05K
2200
2100
2000
1 m
1800
WAVENUMBER (cm") Figure9. IR spectra of CO on 0.25 ML of S on Mo(110). CO exposure was at 90 K with spectral acquisition at the indicated temperatures.
to the peak desorption maxima shown in Figure 4 and assuming a frequency factor of 1013. one can calculate adsorption energies of 20.1, 20.1, and 18.2 kcal/mol for CO adsorbed molecularly on clean Mo(llO), the ~ ( 2 x 2 ) soverlayer, and the ~ ( 2 x 2 ) s overlayer, respectively. These bond energies further reinforce earlier indications that, in the ~ ( 2 x 2 ) overlayer, s the S merely acts as a site blocker without noticeably affecting the bonding of the CO to the Mo surface. Whereas, for the ~ ( 2 x 2 ) overlayer, s the S electronically perturbs the surface changing the bonding OftheStotheMosurface. Thus, thec(2x2) 0.50MLSoverlayer both induces CO adsorption at 4-fold hollow sites (2045-cm-1 peak) and also decreases the binding energy of the CO adsorbed on these sites. In addition, it is possible that these bond energy differencesare a contributing factor to the preferential adsorption of CO on the ~ ( 2 x 2 ) sdomains that was observed for the 1-langmuir exposure spectrum in Figure 6. There is an additional effect that occurs when S is adsorbed onto a Mo( 1 10) surface. Dissociative CO adsorption is reduced. This result has been observed previously4 and is clearly evident in theCO TPD spectra shown in Figure 4. Increasing S coverage
Sulfur is a common surface contaminant and has been shown to have pronounced effect on the adsorption of C0.3,417 These effects are clearly evident in the IRAS spectra of CO adsorption on S/Mo( 110) presented earlier. The spectra shown in Figure 1 reflect the dual nature of adsorbed S. It can act as both a site blocker and/or an electron-withdrawing agent. The fact that S can affect CO adsorption in both a site-blocking' and electronwithdrawing3 fashion has, of course, been observed before. However, the fact that these effects can be observed in the CO IRAS spectra is another example of the utility of this technique. In Figures 1 and 2, it is clear that from a sulfur coverage of -0.25 ML up to a coverage of -0.35 ML that both ~ ( 2 x 2 ) s and ~ ( 2 x 2 ) domains s exist on the surface at the same time. The lack of an abrupt change as the S coverage is increased implies that this is not really a phase transition from the ~ ( 2 x 2structure ) to the 4 2 x 2 ) structure, but rather just the gradual filling in of the ~ ( 2 x 2 overlayer ) lattice structure with additional S atoms to give the ~ ( 2 x 2 structure. ) Finally, the results of this study clearly indicated that dissociative CO adsorption is strongly inhibited on the c(2x2)S/Mo( 110) surface.
Acknowledgment. We acknowledgewith pressure the support of this work by the Department of Energy, Office of Basic Sciences, Division of Chemical Science. References and Notes (1) Clarke, L. J. SurJ Sci. 1981, 102, 331. (2) Farias, M.H.; Gellman, A. J.; Somorjai, G. A,; Chianelli, Liang, K.S.Surf.Sci. 1984, 140. 181.
R. R.;
CO Adsorption on S-Modified Mo( 1 10) Surfaces (3) Gland, J. L.; Madix, R. J.; McCabe, R. W.; DeMaggio, C. Surf. Sci. 1984,143,46. (4) E r i h n , J. W.; Estrup, P. J. Surf. Sci. 1986,167,519. (5) Koestner, R. J.; Salmeron, M.; Kollin, E. B.; Gland, J. L. Surf. Sci. 1986. 172.668. (6) Roberts, J. T.;Friend, C. M. Surf. Sci. 1987,186,201. (7) Jiang, X.;Goodman, D. W. Caral. Lrrr. 1990, 4, 173. (8) Tauster, S.J.; Pecoraro, T. A.; Chianelli, R. R. J . Carol. 1980,63, 5.
(9) Harris, S.;Chianelli, R. R. J . Carol. 1986,98, 17. (IO) Peralta, L.; Berthier, Y.; Oudar, J. Surf. Sci. 1976,55, 199.
(11) Wilson, J. M. Surf. Sci. 1976,59,315. (12) Salmeron, M.; Somorjai, G. A.; Chianelli, R. R. Surf. Sci. 1983,127, 5.
(13) Maurice, V.;Peralta, L.; Berthier, Y.; Oudar, J. Surf. Sci. 1984,148, -I
~~ L J_ .
.
(14) Sanchez, A.; DeMiguel, J. J.; Martinez, E.; Miranda, R. Surf. Sci. 1986,171, 157. (15) Gellman, A.; Tysoc, W. T.; Zacra, F.; Somorjai, G. A. Surf. Sci. 1987,191,271. (16) Knight, C. C.; Somorjai, G. A. Surf. Scf. 1990,240, 101.
The Journal of Physical Chemistry, Vol. 98, No. 1, 1994 263 (17) Campbell, R. A.; Goodman, D. W. Reo. Sei. Instrum. 1992,63,172. (18) Hoffmann, F. M. Surf. Sci. Rep. 1983,3, 107. (19) VibrationalSpecrroscopyofMoleculesonSurjoces;J. T.,Yatea, Jr., T. E. Madey, Eds.; Plenum: New York, 1987. (20) Y. J. Chabal, Surf. Sei. Rep. 1988,8, 211. (21) Browne, V. M.; Fox, S.G.; Hollins. P. Caral. Today 1991, 9, 1. (22) Ryberg, R. In Advances in Chemical Physics; K. P., Lawley, Ed.; Wiley: New York, 1989. (23) Persson, B. N.J.; Ryberg, R. Phys. Reo. B 1981, 24,6954. (24) Rodriguez, J. A.; Truong, C. M.;Goodman, D . W. J. Chem. Phys. 1992,96,7814. (25) Rodriguez, J. A.; Truong, C. M.; Goodman, D. W. Surf. Sci. 1992, 271,L331. (26) Ehlcrs,D.H.;~r,A.P.;Spitzer,A.;Luth,H.Surf.Sci.1987,191, 466. (27) Hoffmann. F.~ M.: J.. J. ~ Chcm. 6 . 2290. ,~ ~ . ~ ~ .Paul. ___.,. . ~ ~ ~ Phvs. ~ . , .. . 1987. -.- . , 8 . . ,_ _ ... (28) Hoffmann, F. M.; Paul, J. J. Chem. Phys. 1987,87, 1857. (29) He, J.-W.; Kuhn, W. K.; Goodman, D. W. Surf. Sei. 1993,292,248. (30) He, J.-W.; Kuhn, W. K.; Goodman, D. W. J. Am. Chcm. Soc. 1991, 113,6416. (31) Redhead, P. A. Vacuum 1962,12,203.
.-
.~
