J. Phys. Chem. 1991, 95, 3155-3164 at 1530, 1449,1341, 1143,955,747, and 681 cm-I. These have been assigned as A, , B, AI ,B, B, Bl,, and A,, in symmetry, TRus, the kaman bands observed are not consistent with CuPc lying flat, or nearly so, on the metal surface. Rather, the molecular plane is significantly canted. Spectral results in the visible: Reflectance and absorption spectroscopy were performed on CuPc films of thickness ranging from 10 to 200 nm. In all cases the spectra obtained were clearly identifiable as due to the a phase of CuPc." LuciaQ reports that the Q-band maxima of a-CuPc occur at 626 and 691 nm, with the 626-nm band being significantly more intense. In our own spectra, Figure 7, for example, we observe similar intensities for bands positioned near 628 and 690 nm. In comparison, the intensities are reversed in 8-CuPc, and the bands occur at 644 and 725 r ~ m . ~Thus, * all three spectral methods (IR, Raman, and visible absorption) indicate that only a-CuPc is present in these deposited films. We note that Barrett4g has studied the development of CuPc films by ellipsometry. Conclusion
On the basis of the IR, Raman, and visible spectra reported here, it appears that thin films of copper phthalocyanine deposited onto amorphous substrates held at room temperature exist in the a-phase modification. Further, as the film thickness decreases, there is a preferential ordering of the unique axis of the phthalocyanine parallel to the metal surface. Given the constancy of the CuPc vibrational frequencies with coverage and the nearly linear dependence of absorption coefficient with film thickness, one can reasonably infer that the CuPcsupport interaction is a weak one. This is also consistent with chemical intuition. The air-grown oxides are densely covered with hydroxyl groups and partially covered with formate and other The surface (48) (49)
Lucia, E. A.; Verderame, F. D. J . Chem. Phys. 1968, 18, 2674. Barret, M.A,; Borkowska, 2.;Humphreys, M.W.;Parsons, R. Thin
Solid Films 1975, 28, 289.
3755
presented to the CuPc entity early in the deposition offers few sites for significant interaction with the CuPc r orbitals. Thus, the most energetically favorable configuration is one in which islands form wherein the CuPc moieties orient with their unique axes parallel to the surface. Within these islands, the K systems of adjacent CuPc's would overlap. As the coverage increases, CuPc is condensing on a multilayer of CuPc. At some point in this process (a surface film thinner than 100 nm) randomly oriented granules develop. DebeZ2has reported that 100-nm films of H2Pc can, under favorable circumstances, become highly oriented with the unique axis making an angle of 2 6 9 from the surface normal. He has also found2' that 100-nm films of CuPc growing on HzPc can develop this so-called standing b-axis configuration. The results reported here are highly reproducible so long as the deposition rate, substrate temperature during deposition, and the oxidation process are all performed as described. There is ample evidence both in the literature and from our own experience that variations in these parameters can produce quantitatively and qualitatively different results. The most critical may be the substrate temperature. For example, there exists a "critical substrate temperat~re"'~ of 0.33Tbil a t which film properties of nonmetallic films vary rapidly. Typically, this special region is less than 20 OC wide. Above and below this narrow window the film properties vary slowly and smoothly with the substrate temperature during deposition. Very recently, Debe2 has shown that milli-g vapor transport films can grow with very high directional preference.
Acknowledgment. We thank the United States Environmental Protection Agency for partial support of this work under Grant R-8 163290 10. R&s@ NO. CUPC,147-14-8. (SO) Tunneling Spectroscopy; Hansma, P. K.,Ed.; Plenum Press: New York, 1982.
Adsorbed States of Acetone and Their Reactions on Rh(ll1) and Rh(111)-(2 X 2)O Surfaces Carl Houtman and Mark A. Barteaus Center for Catalytic Science and Technology, Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716 (Received: August 14, 1990) The binding and decomposition of acetone on clean and modified Rh( 111) surfaces were studied by using TPD (temperature-programmed desorption) and HREELS (high-resolution electron energy loss spectroscopy). On the clean surface the $(C,O) configuration was the dominant form of acetone. This species was characterized by a v(C0) frequency of 1380 cm-I. This value was 430 cm-'below the v(C0) frequency of liquid acetone, indicating a substantial reduction in bond order arising from the binding of acetone to the metal surface via the ?r and ?r* orbitals of the carbonyl. The strength of this interaction was also indicated by the reactivity of the $(C,O)-acetone intermediate. All acetone molecules adsorbed in the first monolayer on the clean surface decomposed to CO, H2, and surface carbon. This decomposition exhibited a primary kinetic isotope effect upon deuterium substitution in both TPD and HREELS experiments. Thus C-H rather than C-C scission was the ratedetermining step in acetone decompositionon Rh(ll1). Three different methods of estimating the acetone decomposition kinetics from the hydrogen TPD were compared. Accurate modeling of the kinetics required inclusion of the effect of hydrogen atom recombination kinetics on the rate of hydrogen evolution from the surface. Modification of the surface by addition of a (2 X 2) overlayer of oxygen resulted in a change in the acetone binding configuration from q2(C,0) to ~ ~ ( 0 This ). shift was the result of electronic modification of the surface by the electronegative oxygen atoms. The v(C0) frequency of acetone observed on the oxygen-predosed surface, 1660 cm-I, was much higher than that observed on the clean surface. The reduced interaction of acetone with the modified surface was also reflected in the decreased reactivity of this adsorbate. ~'(0)-Acetonetended to desorb rather than decompose as observed for q2(C,0)-acetone. Introduction
The formation and binding of carbonyl-containing moieties are Divotal in the svntheses of oxveenated oroducts from carbon
.-
* To whom correspondence should be addressed. 0022-3654191 12095-3755$02.50/0
monoxide and hydrogen and by olefin hydroformylation. Acetone has proven to be an effective model for studying the binding and reactions of carbonyl compounds under ultrahigh-vacuum conditions. The carbonyl group of acetone has been observed to bind to metal surfaces in two configurations: a weak donor bond via 0 1991 American Chemical Society
3156 The Journal of Physical Chemistry, Vol. 95, No. 9. 1991
Houtman and Barteau
cleavage, since similar reactions may control the conversion of the lone-pair electrons on the oxygen, designated ql(0),and a related aldehydes and ketones. dihapto bond, q2(C,0), involving both the carbon and the oxygen of the carbonyl. These two configurations exhibit different reExperimental Section activities: q'(0)-bonded species tend to desorb without reaction, The experiments were conducted in an ion-pumped ultrawhereas q2(C,0)-bonded species decompose to carbon monoxide high-vacuum chamber. With exception of an added diffusion and hydrocarbon fragments in temperature-programmed depump, this chamber has been described previously.s It is a sorption experiments. Thus it is postulated that the selectivity two-level stainless steel vacuum chamber, base pressure 8 X lo-" of catalytic reactions involving carbonyl-containing surface inTorr, equipped with an HREEL spectrometer (McAllister termediates may be controlled by surface modifiers that change Technical Services) and four-grid optics (Physical Electronics) the binding configuration of the carbonyl. for LEED and AES. This instrument had a mass spectrometer The adsorption of acetone on three close-packed transition-metal (UTI 1OOC) multiplexed with an IBM XT. The ionizer of the surfaces has been studied under ultrahigh-vacuum conditions. mass spectrometer was enclosed in a quartz shroud with a 7Avery' clearly demonstrated the formation of an q'(0)-acetone mm-diameter hole a t the end. To maintain effective pumping, on the Pt( 1 11) surface by HREELS. This species was weakly two other holes were placed on the sides. When the crystal was bound to the surface with an activation energy for desorption of placed within 1 mm of this opening, the shroud enhanced the 48 kJ/mol. There was spectroscopic evidence that the P t 4 - C collection of the desorption products and reduced the sensitivity bond axis was inclined relative to the surface normal. The binding of hexafluoroacetone on the same surface2 indicated that the ~'(0) to products desorbed from the support hardware and the back of the crystal. Calibration experiments with CO showed that the bond involved primarily electrons donated by the carbonyl oxygen, crystal had to be placed within 1.0 crystal diameters (10 mm) since the withdrawal of electron density by the fluorine atoms of the opening for the shroud to be effective. reduced the strength of the surfaceadsorbate bond to 35 kJ/mol. The Rh( 111) crystal was polished by using standard metalloThis result can be contrasted with the well-known formation of graphic techniques and aligned by Laue X-ray back-reflection q2(C,0)-bonded hexafluoroacetone on mononuclear transitionto & O S 0 . The crystal was spot-welded on two 0.5-mm tantalum metal comple~es.~Thus in this one case the q2(C,0)-analogue wires that served as a support. Heating was achieved by passing from homogeneous organometallic chemistry is not an effective current (30 A maximum) directly through the support wires. The model for adsorption on surfaces. In contrast to the positive crystal could be cooled to 85 K by thermal conduction through identification of q'(0)-acetone, isolation of the q2(C,0)-acetone a I/& copper feedthrough that was immersed in liquid nitrogen. intermediate was more difficult on Pt( 111). Binding of acetone The temperature was monitored with a chromel/alumel therin the q2(C,0)-configuration resulted in loss of the dipole activity mocouple spot-welded to the back of the crystal. The Rh( 111) of many of the skeletal modes, leaving only methyl stretching and crystal was cleaned by cycles of ion bombardment, oxygen TPD, deformation modes in the HREEL spectrum. This form of acetone and annealing to 1350 K. A B , HREELS, and oxygen TPD were decarbonylated to give hydrocarbon fragments and carbon used to judge the cleanliness of the crystal. monoxide. Anton et a1.4 also identified both bonding configurations The acetone samples were stored in glass tubes connected to of acetone on Ru(0001). The $(C,O) configuration was preferred a stainless steel dosing line and purified by freeze (in liquid on the clean Ru(0001) surface. The addition of a p(2 X 2)O N,)/pump/thaw cycles. The extra dry hydrogen, deuterium, and overlayer stabilized the ~'(0) relative to the qZ(C,O)-acetone oxygen were used as supplied by Matheson. These reagents were configuration. These workers attributed the shift in the favored dosed to the crystal through a 1.5-mm stainless steel needle bonding configuration to an increase in the Lewis acidity of the positioned within 2 cm of the front face of the crystal for direct Ku(0001) surface, resulting in less back-bonding by electron doses or with the crystal rotated 90° from the needle for backdonation to the carbonyl ?r* orbital. Finally, Davis and Barteaus ground doses. A background oxygen dose of 1.2 langmuirs (1 have studied acetone adsorbed on the Pd( 111) surface. They also langmuir = lod Toms) was found to produce a consistent (2 X identified both bonding configurations of acetone. After the 2) LEED pattern corresponding to one-quarter monolayer of removal of a condensed acetone layer that exhibited a vibrational oxygen atoms on the surface. All exposures have been corrected spectrum resembling that of liquid acetone, q2(C,0)-acetone was for ionization gauge sensitivity. A TPD ramp rate of 4 K/s was the dominant species on the clean surface. This q2(C,0)-species used in all experiments. The temperature ramp was controlled was characterized by a u(C=O) frequency of 1435 cm-I. Since by computer, and the rate was constant (h0.2 K/s) to 1350 K. the liquid-phase frequency is 1710 cm-I, it is clear that the bond The electron beam energy for the HREELS experiments was 5 order of the carbonyl group was significantly reduced. With the eV; this produced an elastic peak height of 6 X lo5 counts/s and addition of preadsorbed oxygen atoms, the q ' ( 0 ) configuration a fwhm of 70 cm-' for reflection from the clean surface. For the was favored. This species was characterized by a v ( C 4 ) fretemperature-programmed steps between the HREELS experiquency of 1670 cm-I. While acetone coadsorbed with atomic ments a ramp rate of 4 K/s was used to heat the crystal to the oxygen primarily desorbed without reaction, acetone adsorbed on desired temperature. The power supply was then turned off, and the clean surface had two competing pathways available: dethe maximum temperature recorded. This maximum was used composition and desorption. as the indicated temperature in the HREELS sequences below. Although the adsorbed states of acetone are well-defined on The crystal was allowed to cool to the initial temperature before metal surfaces, little attention has been devoted to the reaction the HREEL spectrum was collected. mechanisms of its decarbonylation. The focus of the present study was the chemistry of acetone on the Rh( 111) surface. HighResults resolution electron energy loss spectroscopy (HREELS)was used Acetone on the Rh ( 1 1 1 ) Surface. Acetone decomposed on the to identify adsorbed species and to probe the nature of their clean Rh( 111) surface. Figure 1 shows the carbon monoxide interaction with the surface. Temperatureprogrammed desorption desorption spectra during TPD following various acetone doses (TPD) was used to identify the desorption and reaction products a t 90 K. Since both the temperature of the maximum rate of and to determine the energetics of surface reactions. Particular desorption and the peak width match those observed after adattention was devoted to the mechanism of acetone decomposition, sorption of carbon monoxide, these peaks were attributed to Le., whether the initial reaction involves C-C or C-H bond molecular carbon monoxide adsorbed on the surface. The slight downward temperature shift of the peak with increasing coverage (1) Avery, N. R. Surf. Sci. 1983, 125, 771. was due to repulsive lateral interactions between the carbon (2) Avery, N. R. hngmuir 1985.1. 162. monoxide molecules. This effect has been observed for the ad(3) Clark, B.; Green,M.; Osborn, R. B. L.; Stone, F. G. A. J . Chem. Soc. sorption of CO at high coverages on the Rh( 11 1) surfacee6 The A 1969, 20. (4) Anton, A. B.; Avery, N. R.; Toby, B. H.; Weinberg, W. H. J . Am. Chem. Soc. 1986, 108, 684. (5) Davis, J. L.; Barteau, M. A. S u r , Sci. 1989, 208, 383.
(6) Castner, D. G.; Sexton, B. A.; Somorjai, G. A. Sur, Sci. 1978,71,519.
Adsorbed States of Acetone and Their Reactions on Rh
100
200
300
400
500
600
700
The Journal of Physical Chemistry, Vol. 95, No. 9, 1991 3757
800
Temperature (K)
Temperahire (K)
Figure 1. Carbon monoxide TPD after acetone exposures at 90 K (yields corresponding to these spectra appear in Figure 7).
Figure 3. Comparison of the hydrogen and deuterium TPD observed
I
after the adsorption of acetone and acetone-& respectively (the dose for both cases was 0.9 langmuir).
m/e 2
3 0K
Temperature (K)
b
400 350
300
250
200
h
Y
v
3
100
200
300
400
500
600
700
800
0.0025
0.0035
0.0345
Temperature (K)
llT
Figure 2. Hydrogen TPD after dosing acetone at 90 K (yields corresponding to these spectra are shown in Figure 7). The inset shows a comparison of the hydrogen desorption after adsorbing hydrogen with that after adsorbing acetone.
Figure 4. Kinetic analysis of the leading edges of the hydrogen and deuterium desorption spectra shown in Figure 3.
amount of carbon monoxide desorbed after a 1.5-langmuir dose of acetone was equivalent to a coverage of 0.18 monolayer. One monolayer (ML)is defined as the number of Rh atoms in the topmost layer of the (111) surface (1.62 X 10lsatoms/cm2). Hydrogen desorption spectra following acetone exposures from 0.1 5 to 1.5 langmuirs are shown in Figure 2. For the three higher acetone doses shown in the figure, the hydrogen desorption peak can be divided into two parts. The larger peak exhibited a maximum at 303 K. A second peak was observed as a shoulder centered a t 400 K with a tail continuing to 600 K. The higher temperature reaction-limited peak was assigned to the dehydrogenation of hydrocarbon fragments on the surface, as similar hydrogen peaks at 400 K and above were observed after ethylene adsorption on Rh( 11l).' The hydrogen peak at 303 K was not governed solely by the recombination kinetics of hydrogen atoms, since it was not equivalent to the desorption spectrum observed after adsorption of H2 on the clean Rh( 11 1) surface. The inset of Figure 2 shows a comparison of the hydrogen desorption after hydrogen adsorption with that produced by acetone decomposition. The two have approximately the same peak area, implying equivalent coverages of hydrogen atoms on the surface. However, the temperature of the maximum desorption occurred at 267 K for the direct hydrogen dose and at 303 K for acetone decomposition. Hydrogen normally exhibits second-order desorption kinetics. Second-order desorption peaks shift downward in temperature with increasing coverage, but the hydrogen peak from (7) Bent, B. E.; Mate, C. M.; Crowell, J. E.; Koel, B.E.; Somorjai, G.A. J . Phys. Chcm. 1907. 91, 1493.
acetone TPD did not exhibit a comparable shift with increasing coverage as shown by Figure 2. Thus one would expect that a t least the leading edge of the hydrogen desorption peak from acetone would be controlled by the release of hydrogen atoms during a reaction occurring on the surface and might, therefore, be utilized to extract the kinetics of that reaction. A kinetic isotope effect (KIE)was observed for hydrogen evolution after adsorption of acetone. A comparison of the TPD spectra of hydrogen and deuterium after adsorption of acetone and acetone-& respectively, is shown in Figure 3. These spectra have been scaled to account for the mass spectrometer sensitivity. The peak temperatures of these two spectra were nearly the same; the maximum hydrogen desorption rate occurred a t 303 K,and the maximum for deuterium occurred at 3 15 K. This small shift in peak temperature would lead to an estimate of the kinetic isotope effect of acetone decomposition of 3.57 at 300 K by using the activation energies estimated by the method of Redhead* and assuming a first-order preexpnential factor in each case of 10') s-I. The ratio of kH to kD from this simple analysis was suggestive of a primary isotope effect, but quantitative determination of the kinetics is required. As will be shown below, more rigorous analysis of the kinetics of acetone decomposition leads to even larger values of the kH/kD ratio, providing clear evidence for a primary isotope effect. Although the peak temperatures for Hzand Dz evolution from acetone were similar, the positions of the leading edges of the two peaks were quite different. Figure 4 shows an analysis of the leading edges, based on the assumption of first-order kinetics. This figure is a plot of the natural logarithm of the desorption rate (8) Redhead, P. A. Vucuum 1962, 12, 203.
Houtman and Barteau
3158 The Journal of Physical Chemistry, Vol. 95, No. 9, 1991 divided by coverage versus reciprocal temperature. The coverage was determined by partially integrating the low-temperature desorption peak after the area under the high-temperature shoulder was removed. The use of a first-order analysis assumes that the reaction that liberates hydrogen was first order with respect to surface acetone concentration and that the hydrogen recombination kinetics did not influence the desorption peak. In the low-temperature regions, uncertainty in the rate measurements was introduced by the limited resolution of the analog-todigital converter used to record the mass spectrometer signal. At higher temperatures the second-order kinetics of hydrogen recombination made the analysis invalid. Thus only 30% of each peak was used in the linear regression. The temperature ranges used for the analysis were 214-294 K for hydrogen and 263-308 K for deuterium. The analysis shown in Figure 4 gave an activation energy of 56 kJ/mol and a preexponential factor of 4.7 X lo8 s-I for hydrogen production from acetone and an activation energy of 94 kJ/mol with a preexponential factor of 3.9 X 1014 s-l for acetone-d6. With these kinetic parameters an apparent KIE of 3.60 at 300 K was calculated. The large and unexpected difference in the apparent preexponential factors determined for H2and D2 desorption from the first-order decomposition of acetone suggests that this level of analysis is still insufficient. Although the hydrogen and deuterium desorption peaks observed after the adsorption of acetone were fit by first-order kinetics, the rate of evolution of hydrogen may not reflect the actual decomposition kinetics of acetone but may be modified by the kinetics of dihydrogen formation. This possibility was explored by modeling numerically a two-step process leading to hydrogen desorption. Hydrogen desorption from the clean surface exhibited second-order desorption kinetics with a peak temperature at saturation coverage of 267 K, ca. 40 K below the peak decomposition temperature of acetone. However, the peak temperature of a desorption process controlled by second-order kinetics shifts upward in temperature with decreasing initial coverage. Thus at the onset of acetone decomposition, when the hydrogen atom coverage is low, the evolution of dihydrogen may be delayed by the hydrogen atom recombination kinetics. As the reaction proceeds, however, the hydrogen atom coverage will increase and the decomposition reaction may become rate limiting. Thus modeling is required. The sequential reactions can be represented by the following equations: (CH3)&O(ad)
kl
CO(ad)
+ 2xH(ad) + 2CH3,(ad)
(1)
2H(ad) A H2(g) where k l is the first-order reaction rate constant describing the decomposition of acetone, and k2 is the second-order rate constant describing the formation of dihydrogen and its desorption into the gas phase. The recombination of hydrogen or deuterium atoms on the Rh( 1 11) surface has been thoroughly studied by Yates et a1.9 They reported that the activation energy and the preexponential factor were dependent on the hydrogen or deuterium coverage. The kinetic parameters they calculated for both isotopes spanned a wide range: from an activation energy of 78 kJ/mol and a corresponding preexponential factor of 1.2 X cm2/s a t the zero-coverage limit to an activation energy of 20 kJ/mol and a preexponential factor of 6.3 X cm2/s at saturation. They attributed the strong coverage dependence to the influence of a precursor state on the desorption kinetics. A rate equation with a linear coverage dependence of both the activation energy and the logarithm of the preexponential factor derived from the values reported by Yates et al. was used to calculate the hydrogen desorption profile assuming an initial hydrogen atom coverage of 0.6 ML,approximately 75% of saturation. A comparison of this calculated peak with the TPD spectrum obtained after hydrogen adsorption indicated that it did not exactly reflect our observations. Subsequent calculations indicated that for specified initial coverages a simple second-order model with constant kinetic ( 9 ) Yates Jr., J. T.: Thiel, P.A.; Weinberg, W.H.Swf. Sd. 1979,84,427.
Hydrogen Decomposition Prnmelen E = 63 W h o l e A I2 8 ~ 1 0 ~ ~
Simulation
1
Deuterium Decomposilirm Plnmetcn E = 68 Whole A = 2.8110''
Simulation
200
300
400
500
600
Temperature (K)
Figure 5. Comparison of the hydrogen and deuterium desorption spectra from acetone with those predicted by the sequential reaction model (see text) accounting for the kinetics of hydrogen/deuterium atom recombination.
parameters more accurately fit our data. The parameters that best fit the hydrogen atom recombination at the coverage of interest were an activation energy of 45 kJ/mol with a preexponential factor of 1.7 X lo-* cm2/s. These were obtained from a hydrogen desorption spectrum that had the same area as the low-temperature portion of the hydrogen desorption peak observed after acetone adsorption. Similar values were obtained by Yates et aL9 for a hydrogen coverage that was approximately 50% of saturation. Since the spectra obtained after the adsorption of hydrogen and deuterium were indistinguishable except for the mass spectrometer sensitivity factors, the activation energies and preexponential factors were assumed to be the same. The system of sequential reactions above was used to model hydrogen and deuterium desorption during TPD experiments following acetone adsorption. For the purposes of this model x was assumed to be 2, Le., the hydrocarbon fragments deposited on the surface were assumed to be CH. This value was obtained by integrating the high-temperature shoulder of the hydrogen desorption peak that corresponded to hydrocarbon dehydrogenation and comparing this to the amount of carbon left on the surface after the TPD experiment. The system of ordinary differential equations was solved by using a fourth-order Runge-Kutta method. Initially the kl values determined from the leading edge analysis described above were used in the model, but these gave peak temperatures for both hydrogen and deuterium desorption that were 15 K above those in the respective experimental data. The values for kl used in the calculations were adjusted to produce a match with the observed peak temperatures and leading edges. It was determined that an activation energy for acetone decomposition of 63 kJ/mol with a preexponential factor of 2.8 X 10" s-I produced the best fit of the hydrogen desorption peak; the corresponding values for the decomposition of acetone-& were an activation energy of 68 kJ/mol with a preexponential factor of 2.8 X 10" s-I. The value of the preexponential factor in each case was selected to fit the slope of the low-temperature side of the desorption peak. A comparison of the calculated and observed desorption spectra is shown in Figure 5 . For this figure the scale factors were chosen such that the peak areas of the simulation were the same as the low-temperature portion of the experimental data. The simulations for both hydrogen and deuterium fit the peak positions and peak shapes quite well. This analysis demonstrated that hydrogen and deuterium evolution was controlled by the kinetics of both acetone decomposition and hydrogen atom recombination and that a simple analysis of the leading edge could not give the correct kinetic parameters. The kinetic isotope effect of 300 K that corresponds to the 5 kJ/mol difference in activation
Adsorbed States of Acetone and Their Reactions on Rh
The Journal of Physical Chemistry, Vol. 95, No. 9, 1991 3759
I78 K
I d e 43
Carbon Monoxide
q2-Acetone
200
100
300
400
500
600
700
255 E
800
Temperature (K)
Figure 6. Acetone TPD after dosing acetone at 90 K. The yields corresponding to these spectra are shown in Figure 7. 1 200
Yield vs. Dose 0.5
600 loo0 1400 1800 2200 2600 3000 3 x)
Energy Loss(cm.l)
A
Figure 8. HREEL spectra after adsorption of approximately 3 langmuir of acetone at 90 K and after subsequently heating to 255 and 330 K. An equivalent dose resulted in the desorption of 0.20 ML of CO in TPD. 0
A
0.2
0
0.1
0.0
0
1
Do=
2 (L)
Figure 7. TPD yields after acetone dosing (m H2;0,C O A, (CH,),CO 0, C(ad)). the vertical axis has units of monolayers of molecules. A monolayer is defined as the number of Rh atoms in the top layer of the clsoe-packed surface, 1.62 X IOt5 atoms/cm*. The horizontal axis has units of langmuirs corrected for ionization gauge sensitivity.
energies is 9.0. This value is entirely consistent with a primary isotope effect, and thus scission of a C-H bond must be the rate-determining step in hydrogen evolution. However, we have yet to show that this reaction was that of acetone as suggested in the model rather than that of, for example, methyl groups dissociated from it. This issue will be addressed further during the discussion of the TPHREELS experiments. Figure 6 shows the temperature-programmed desorption of acetone at different intial exposures. The peak observed at 178 K in the spectrum after an acetone dose of 1.5 langmuirs corresponds to a coverage of approximately 0.22 ML. This peak temperature yields an activation energy of 45 kJ/mol if one ass u m e a preexponential factor of 10” &. This may be compared to the latent heat of vaporization of acetone from the liquid state,’O 31 kJ/mol. The discrepancy between the two values suggests that a t the relatively low coverage of condensed molecular species characteristic of the highest exposure curve in Figure 6, there is still an interaction with the surface. The absence of acetone desorption after lower doses of acetone and the large peak observed after the highest dose indicated that saturation of the first monolayer occurred between 0.95 and 1.5 langmuirs and that most of the acetone adsorbed in the first layer decomposed rather than desorbed. Doses of acetone above 1.5 langmuirs did not result in saturation of the condensed state a t 178 K. This exposure threshold and the subsequent monotonic rise of acetone desorption with increasing acetone doses is illustrated in Figure 7. The horizontal axis has units of langmuirs, and the vertical axis has units of monolayer equivalents of the molecules. As can be seen in this figure, acetone desorption was not observed for exposures below 0.8 langmuir, but above this threshold there was a linear rise in the amount of acetone desorbed; the other products satu(10) Cox, J. D.; Pilcher, 0 . Thermochemistry of Organic and O r g u m metallic Compounds; Academic Press: 1970.
rated at a dose of approximately 1 langmuir. The stoichiometry of acetone was reflected in the relative yields of the decomposition products. The average ratios for the coverages shown in Figure 7, after correcting for mass spectrometer sensitivity, were 1.0:2.6:1.9 for CO:H2:surface carbon. The amount of carbon remaining on the surface after the acetone TPD experiment was quantified by analyzing the amount of CO and C02desorbed by subsequent adsorption and TPD of 1.2 langmuirs of oxygen. The only other product observed after the adsorption of acetone was a small amount of methane at 275 K. The amount of methane was approximately 0.25% of the amount of carbon monoxide desorbed during the experiment. The preadsorption of 0.3 M L of hydrogen atoms did not increase the amount of methane evolved during the acetone TPD. Thus the reaction that produced methane was not limited by hydrogen availability. In summary, acetone adsorbed in the first monolayer began to decompose at 275 K on the clean Rh(ll1) surface. Hydrogen, carbon monoxide and surface carbon were the only products formed in significant amounts. A kinetic isotope effect was observed when the methyl hydrogens were replaced with deuterium. Acetone desorption was observed only for acetone doses above 0.8 langmuir. It was concluded that this acetone was evolved from a condensed state. The intermediates responsible for the products desorbed during the TPD experiments were identified by using HREEL spectroscopy. Figure 8 shows the spectra obtained after a 3-langmuir acetone dose at 90 K. The 90 K spectrum was dominated by the v(C-H) and the b(CH,) modes of acetone at 3000 and 1410 cm-I, respectively. The mode at 1710 cm-’ was assigned to the v ( M ) mode of the carbonyl of a weakly bound condensed state of acetone. As we have previously demonstrated on the Pd(ll1) surface, HREELS is capable of resolving differences between acetone weakly bound to the surface in an ql(0)configuration from liquidlike condensed species in higher layerse5 The assignment of the 90 K spectrum of Figure 8 to q’(0)-acetone was precluded by the low intensity of the carbonyl mode. If the C-O bond axis were approximately normal to the surface, one would observe a very strong dipole loss in this region, as was clearly evident in spectra obtained after acetone exposure of the oxygen-dosed surface, discussed below. The existence of a weakly bound state upon acetone adsorption at 90 K was also consistent with the changes in band intensities observed upon heating the surface to 255 K. The increase in intensity of the 1260- and 1380-cm-I peaks indicated that the weakly bound state converted to the q2(C,0)-configuration upon warming. A small amount of the acetone remained in a binding configuration that was char-
3760 The Journal of Physical Chemistry, Vol. 95, No. 9, 1991
Houtman and Barteau
TABLE I: $-Acetone HREELS Assignments
hvdronenated Ru4 Rh ,
mode u(CH~) UiCO ') h(CH3) v,(C-C-c) u,(C-CC) p(CHd if(C4) h(C-C-c) #Not resolved.
liquid'l 296412924 1710' 1410/1361 1220 785 109 1/995 530 488
2955 1300 1370/1170
3000 1380 1380 1260 800 990 600
980 655
nP'
r annealed to 255
deuterated Ru'
I
K
Pd5 2975 1435 1270 1025/990 710 495
liquid" 2255 f 2222 1707' 1092/1005 1245 690 8841764 480 408
Rh
Pd'
2220 1275 1075/880
2200 1360 1100 1230
2260 1445
820 610
800 600 470
490 695
Acetone TPHREELS
100
200
300
400
Temperature (K)
d, -Acetone TPHREELS 200
600 1000 1400 1800 2200 2600 3000 3,
I
Energy Loss(cm")
Figure 9. HREEL spectra after adsorption of approximately 3 langmuir of acetone and acetone-d6at 90 K and subsequently heating to 255 K. An equivalent dose resulted in the desorption of 0.20 ML of CO in TPD for both cases.
acterized by a u(C0) frequency of 1660 cm-I. The formation of an q2(C,0)-acetone intermediate on the Rh( 111) surface is consistent with results reported for the close-packed surfaces of the two metals adjacent in the periodic table, Pd and Ru. The vibrational assignments for q2(C,0)-acetone on Rh and on both Pd and Ru are summarized in Table I. The IR assignments for liquid acetone are also included for comparison." The u(C=O) frequencies of &C,O)-acetone exhibited a monotonic increase upon moving rightward in the periodic table. The reduction in the u(C=O) frequency from that of gas-phase acetone can be related to the strength of the interaction with the metal surface. With a u ( C 4 ) frequency of 1300 cm-I, acetone interacted most strongly with Ru4 and acetone interacted the least strongly with Pd5 exhibiting a vibrational frequency of 1435 cm-I. Another characteristic mode observed for the $(C,O)-acetone intermediate was the T(C-0) mode at 600 cm-l. This mode was clearly resolved for &C,O)-acetone adsorbed on all the metals studied to date. The $(C,O)-acetone assignments were confirmed by comparison of the 255 K spectrum with a spectrum obtained after dosing acetone-d6 under similar conditions. These two spectra are shown in Figure 9 with the corresponding modes connected. From the comparison with the acetone-d, spectrum it was clear that the 1380-cm-I loss in the acetone spectrum was due to both the u(C=O) and the 6(CH3) modes. Since there was still a loss in this region after the methyl mode was shifted by deuteration, the existence of a u(C=O) mode at 1360 cm-' was confirmed. The 6(CD3) mode was observed a t 1100 cm-' in the deuterated case. The slight downshift of the mode at 1260 to 1230 cm-' upon deuteration was consistent with the assignment of this mode to the skeletal u(C-C-C) vibration. The increase in intensity of the characteristic carbon monoxide modes, u(M-CO) at 470 cm-I and u(C0) at 2020 cm-', at higher temperatures in Figure 8 indicated that the decomposition reaction had begun by 255 K, and by 330 K only carbon monoxide and (1 1)
Dellepiane, G.;Overend, J. Specrrochim. Acto 1966, 22, 593.
100
200
300
400
Temperature (K)
Figure 10. Top: ( 0 )TPHREELS of the adsorbed CO production after 0.9 langmuir of acetone dosed at 90 K (W). Simulation of TPHREELS calculated using the rate parameters determined in Figure 5. Bottom: (0)TPHREELS of the CO production after 0.9 langmuir of acetone-d6
dosed at 90 K (W). Simulation of TPHREELS calculated using the rate parameters determined in Figure 5. surface hydrocarbon species remained. Low-intensity hydrocarbon modes were observed in the 330 K spectrum at 800 and 2960 cm-I. These losses were clearly b(CH) and u(CH) modes of a single hydrogen atom bound to a carbon atom. This assignment was based on the observation of similar losses after the adsorption of hydrocarbons on the Rh( 111) surface.' This assignment was also consistent with the stoichiometry obtained from the integration of the hydrogen TPD peaks. The intensity of the carbon monoxide u(C0) loss in the range of 2020-2060 cm-I was used as a measure of the progress of the decomposition of acetone. The intensities of the elastic peak and the carbon monoxide peak at 2020-2060 cm-' were determined by integration. Figure 1 0 , top, is a plot of the percent of the elastic peak area represented by the CO peak area as a function of temperature. The reaction giving C O began a t 220 K and was largely complete by 285 K. A similar plot for the CO production after acetone-d6 adsorption is shown in Figure 10, bottom. For the deuterated case the reaction began at 236 K and was largely complete by 305 K. Although the scatter in these data preclude a more complete analysis, there did appear to be a kinetic isotope effect delaying the decomposition of acetone-d6 relative to the decomposition of acetone. To estimate the KIE, a straight line was fit t o the rising portion of the TPHREEL spectra by using a linear least-squares analysis of six data points. These lines were used to estimate the temperature at which 50% of the final CO had been evolved. This temperature was assumed to correspond to the peak temperature during a TPD experiment with a heating
Adsorbed States of Acetone and Their Reactions on Rh Ink2
m 1 0.8 L
100
200
300
400
to^
500
600
700
800
Temperature (K)
Figure 11. TPD after dosing 1.2 langmuir of oxygen at 300 K and 1.9 langmuir of acetone at 90 K (yields corresponding to these spectra were 0.085 ML of CO and C02and 0.25 ML of acetone). The acetone TPD after a 0.8 langmuir dose of acetone is also shown.
rate of 0.4 K/s. Since we were interested only in the difference in activation energy, the errors introduced by these approximations were not critical. The temperature where the progress was 50% for acetone was 274 and was 29 1 K for acetone-&. The kH/kD ratio determined by estimating the activation energy difference (4.8 kJ/mol) and extrapolating to 300 K was 6.8. Thus the TPHREELS experiments also indicated a primary KIE. Also plotted in Figure 10 are the results of a simulation that was calculated with the kinetic parameters obtained from the analysis of hydrogen evolution kinetics illustrated in Figure 5. For this simulation the temperature program was specified to match the stopping temperatures used during the TPHREELS experiments. The temperature was ramped linearly (4 K/s) up to each stopping temperature and then ramped linearly (4 K/s) back to the starting point and then up to the next temperature. When the temperature reached each of the stopping points, the calculated progress of the reaction was recorded, and these results are plotted in the respective figures. The solution of the two coupled ordinary differential equations (temperature and coverage) was performed using a fourth-order Runge-Kutta method. The comparison of the observed and calculated CO production indicated that the TPHREELS experiment were consistent with the TPD results and that the reaction that released hydrogen atoms to the surface also released carbon monoxide. The correspondence of carbon monoxide and hydrogen formation indicated that the decomposition of acetone was the ratecontrolling reaction. Thus the kinetic parameters obtained from the analysis of the hydrogen evolution rate reflected the decomposition of acetone. In summary, HREEL spectra collected after the adsorption of acetone on the clean Rh( 1 1 1) surface showed the formation of an qz(C,O)-acetone intermediate, and this was the only species present on the surface between 200 and 250 K. This intermediate began to decompose at 255 K, leaving carbon monoxide and hydrocarbon fragments on the surface. Acetone on the Rh ( 1 11)-(2 X 2)O Surfoce. Two adsorbed states of acetone were observed on the oxygen-predosed Rh( 1 11) surface. Figure 11 illustrates TPD spectra after dosing 1.9 langmuirs of acetone on the Rh( 11 1)-(2 X 2)O surface. The acetone desorption spectrum after dosing only 0.8 langmuir of acetone is included in the figure to distinguish more clearly the two acetone desorption peaks. The higher temperature acetone peak, centered at 275 K, was attributed to q'(0)-acetone as confirmed by HREELS results discussed below. This peak temperature corresponds to an activation energy of 70 kJ/mol assuming a preexponential factor of 10" s-'. The corresponding peak temperature for the strongly bound q*(C,O)-acetone state on the clean Rh(ll1) surface could not be resolved, as 100% of the molecules adsorbed in this state on the clean surface decomposed. Comparison of the spectrum obtained after 1.Plangmuir dose of acetone with that obtained after a 0.8-langmuir dose on
The Journal of Physical Chemistry, Vol. 95, NO. 9, 1991 3761 the Rh(ll1)-(2 X 2)O surface cllarly showed the filling of the condensed state, characterized by a peak at 190 K. This peak temperature is 12 K higher than the peak temperature observed on the clean surface, consistent with the suggestion above that there is some interaction of condensed species with the surface at these coverages and that this interaction is modified slightly by the addition of oxygen to the surface. The position of the 190 K peak, which roughly exhibits the shape expected for first-order desorption kinetics, corresponds to an activation energy of 48 kl/mol assuming a preexponential factor of s-'. Comparison of the two acetone peaks at the two coverages shown also indicated that the q'(O)-acetone state was saturated by a 0.8-langmuir dose. Integration of the q'(0)-acetone portion of this peak gave a saturation coverage of approximately 0.25 ML for the ~'(0)acetone state. The condensed state did not saturate with higher acetone doses but continued to grow in area. The extent of decomposition was also observed to reach a plateau at approximately the same dose as that required for saturation of the q'(0)-acetone state. Integration of the CO and C02 peaks in Figure 11 showed that 0.085 ML of acetone decomposed in the course of TPD from the Rh(ll1)-(2 X 2)O surface. For this integration the CO evolved above 550 K was excluded, since it was the product of the reaction of oxygen atoms and surface carbon formed from the methyl groups. Higher acetone exposures did not result in any increase in the amounts of the decomposition products. Thus 25% of the acetone adsorbed in the first layer decomposed, while the remainder desorbed intact. As noted above, the entire first layer of acetone adsorbed on the clean Rh( 1 1 1) surface decomposed. On the clean surface 0.18 ML of acetone decomposed in TPD experiments following acetone adsorption at 90 K. Thus the addition of oxygen reduced the absolute amount of acetone decomposition by 47%. The carbonand hydrogen-containing products from acetone decomposition on the oxygen-dosed surface were evolved via three pathways. Molecular carbon monoxide was observed to desorb at 470 K. Carbon monoxide was also seen to react with surface oxygen to yield C 0 2 at 400 K. This reaction has been previously reported for coadsorbed oxygen atoms and CO molecules by Root et al.'* Finally, the small water peak at 410 K indicated that hydrogen atoms were released by a surface reaction at this temperature. This desorption temperature is consistent with the decomposition of a surface acetate intermediate, for which some evidence was also obtained by HREELS, as described below. Similar peaks have been observed after the adsorption of acetic acid on Rh(1 1 l).13 Because of the difficulty of defining a baseline for the water spectrum and the overlap of the expected C 0 2 peak from acetate decomposition with that for CO oxidation, quantification of the amount of decomposition via the acetate pathway was subject to greater error than for other products, but the amount appeared to be less than 10% of the acetone decomposed. No carbon was left on the surface after the TPD experiment. A small excess of oxygen (less than 0.06 ML of oxygen atoms) was observed to desorb at 1000 K. The desorption of hydrogen at 200 K was consistent with the recombination of hydrogen atoms on the oxygen-predosed Rh(ll1) surface. Similar observations were made by Thiel et al.I4 They reported that the presence of oxygen atoms on the surface modified the hydrogen kinetics and shifted the hydrogen desorption peak temperature from 300 K on the clean surface to 200 K on the oxygen-dosed surface. The presence of this low-temperature peak in Figure 11 indicated that some reaction of acetone to liberate hydrogen atoms must have occurred below 200 K. Integration of the H2 peak at 200 K indicated that 0.17 ML of hydrogen atoms was involved, which implies the decomposition 0.028 ML of acetone. The source of this hydrogen could not be identified from the TPD results. The water peak at 190 K appeared to be molecular water incorporated into the multilayer on the surface. It likely arose either by adsorption from ~
~
~~
(12) Root, T. W.; Schmidt, L. D.; Fisher, G. Surf. Sci. 1985, 160, 173. (13) Houtman, C.; Barteau, M.A., manuscript in preparation. (14) Thiel, P.A.; Yates Jr., J. T.; Weinberg, W . H. Surf. Sci. 1979, 90, 121.
3162 The Journal of Physical Chemistry, Vol. 95, No. 9, 1991
Houtman and Barteau
TABLE 11: n'-AWtoW HREELS Adirmments ~~
mode
296412924 1710 141011361 1220 785 10911995 530 488 a
hydrogenated Ru4 Rh
liquid"
2950 1690 144011395
nP 8251780 900 54015 15
deuterated Ru4 Rh
Pd'
liquid"
2980 1665 141011350 1240
2985 1685 142511355 1220
225512222 1707' 1092/1005 1245 690 8841764 480 408
nr
nr
930
10501965
nr nr
495
nr
2070 1665 1030 1260 7201715 920 5051495
2230 1660 1015 1270
Pd' 2245/2100 1670 1030 1250
nr
nr
715 340
890 335 695
nr
Not resolved. XI00
-2
200
600 loo0 1400 1800 2200 2600 3000 34 !Q
Energy Loss(cm")
Energy Loss(cm")
Figure 12. HREELS after dosing 1.2 langmuir of oxygen and approximately 3 langmuir of acetone at 170 K and HREELS after subsequently heating to 252 and 300 K. An equivalent dose resulted in the desorption of 0.10 ML of C O and C 0 2 in TPD.
the background or was present in the acetone sample, since this temperature was t60 low for it to be the product of a reaction. No other products were observed during the TPD experiment. In summary, the addition of a (2 X 2) overlayer of oxygen to the R h ( l l 1 ) surface resulted in the binding of acetone to the surface in a less reactive state. This state was characterized by molecular acetone desorption at 275 K. Decomposition was still observed, but it accounted for only 25% of the acetone adsorbed in the first monolayer. HREEL spectra obtained after adsorption of acetone on the oxygen-dosed surface indicated that the acetone was bound to the surface in a configuration different from that observed on the clean surface. Figure 12 shows the HREEL spectra recorded after a 3-langmuir dose of acetone on the Rh( 11 1)-(2 X 2)O surface. The 535-cm-l peak of the 170 K spectrum was the u ( M - 0 ) vibration of adsorbed oxygen atoms. The mode assignments for v'(O)-acetone on Ru, Pd, and Rh are summarized in Table 11. Comparison of the 170 K spectrum with the spectrum collected after acetone adsorption on the clean surface, Figure 8, showed a change in the relative intensities of different vibrational modes. Acetone adsorbed on the oxygen-dosed surface exhibited more intense methyl modes (6(CH3) at 1430 cm-I and u(CH3) at 2960 cm-l) and an intense u ( C 4 ) mode at 1665 cm-l, while acetone adsorbed on the clean surface had a stronger skeletal u(C-C-C) mode at 1260 cm-' and a weak u ( C 4 ) mode a t 1380 cm-l. These shifts in intensity can be related to changes in bonding orientation. If acetone were adsorbed via the lone pair electrons on the carbonyl oxygen with the C=O bond essentially perpendicular to the surface, the methyl modes and the u ( C 4 ) mode would be enhanced relative to those of a species with the CO axis parallel to the surface. For the strongly rehybridized carbonyl
Figure 13. HREELS after dosing 1.2 langmuir of oxygen and approximately 3 langmuir of acetone-d6at 170 K and HREELS after subsequently heating to 252 and 300 K. An equivalent dose resulted in the desorption of 0.10 ML of CO and C 0 2 in TPD.
group of v2-acetone bound in a roughly parallel orientation, the asymmetric v ( C 4 - C ) mode would be more intense. For both of these orientations, the p(CH3) loss at 990 cm-' would have nearly the same intensity. These differences were precisely those observed between acetone on the (2 X 2)O and the clean Rh( 111) surfaces. The acetone mode assignments were confirmed by collecting spectra after the adsorption of acetone-& The 170 K spectrum of Figure 13 showed the C-D mode frequencies were reduced by a factor of approximately 1.39. The u(CH3) mode at 2960 cm-' shifted to 2225 cm-I. The 6(CH3) mode at 1430 cm-' shifted to 1030 cm-I, and the p(CH3) mode at 990 cm-' shifted to 715 cm-'. The shift of the 6(CH3) mode revealed the v(C-C-C) mode at 1260 cm-' more clearly. These mode assignments are summarized in Table 11. The large intensity of the u(C=O) mode a t 1660 cm-I in both the deuterated and nondeuterated case indicated that this mode did not have any hydrogenic character, as expected. The liquid acetone u ( C 4 ) frequency is 17 IO cm-'. Since this mode was only shifted down in frequency by 50 cm-I, clearly the binding of acetone in this configuration does not significantly change the bond order. On the clean surface, however, the u( C 4 ) loss was observed at 1360 cm-I. In this case the bond order was significantly less than 2. Heating the surface to 252 K resulted in minor changes in the spectrum. The losses associated with v'(O)-acetone and atomic oxygen species still dominated the spectrum. The small 1400-cm-' loss in Figure 13 suggested that a small portion of the acetone was bound in the d(C,O)-acetone state. This mode was obscured in Figure 12 by the large peak due the methyl deformation of acetone. It is likely that this $(C,O)-species accounts for the small amount of acetone observed to decompose on the oxygen-dosed Rh(ll1) surface. The r(C=O) mode of $(O)-acetone was
The Journal of Physical Chemistry, Vol. 95, No. 9, 1991 3163
Adsorbed States of Acetone and Their Reactions on Rh observed at 345 cm-' for acetone-& This mode for unlabeled acetone was obscured by the v(M-0) mode of the oxygen adlayer a t 535 cm-l. Further heating of the surface resulted in the loss of intensity of the v ( M ) mode, indicating a loss of acetone from the surface. This is consistent with the TPD experiment that showed an acetone desorption peak at 275 K. Coincident with the loss of q'(0)-acetone modes was the enhancement of the carbon monoxide mode at 2070 cm-I. This increase in intensity was due to a small amount of decomposition occurring on the surface. In the transition from 252 to 300 K another reaction occurred. The modes at 700,975, and 1425 cm-I, which appear in the 300 K spectrum of Figure 13, were characteristic of an acetate intermediate. Davis and BarteauI5 have identified similar losses as the 6(W-0), v(C-C), and ~ ~ ( 0 - C - 0modes ) of acetate intermediates formed from acetic-d4 acid on the Pd( 111) surface. For unlabeled acetone the growth of the mode at 1410 cm-l between 252 and 300 K was obscured by the 6(CH3) mode at 1430 cm-I in Figure 12. The formation of acetate intermediates was also suggested by the evolution of water at 410 K during the TPD experiments described above. In summary, the HREELS of acetone on the oxygen-dosed Rh( 11 1) surface indicated the formation of an q'(0)-acetone intermediate bonded to the surface through the lone-pair electrons on the oxygen. This species desorbed at 275 K in both HREELS and TPD experiments. Small amounts of the acetone decomposed or reacted to form acetate intermediates between 252 and 300 K.
Discussion The binding configuration had a profound effect on the reactivity of acetone bound to the Rh(ll1) surface. On the clean surface acetone adsorbed in the first layer decomposed to CO and surface hydrocarbons. On the oxygen-dosed surface only 25% of the acetone decomposed, and there was spectroscopic evidence for q2( C,O)-acetone that may have been responsible for this decomposition, although q'(0)-acetone was the dominant species on the surface. Thus the conclusion drawn from the results on other metals (R', Ru4, and Pd5) that q2(C,0)-acetonetends to react while q'(0)-acetone tends to desorb can be extended to Rh. Davis and Barteaus used various oxygen predoses to determine the origin of the effect of oxygen atoms on the binding configuration of acetone. For oxygen coverages between 0.05 and 0.25 ML, no qualitative differences in the HREEL spectra were observed. Thus an explanation based on site blocking was eliminated, and electronic modification of the surface was suggested as the origin of the effect. As proposed by Anton et aL4 the addition of oxygen to these metal surfaces enhances surface-acceptor properties toward electrons from the lone pair on the carbonyl, favoring the formation of a ~'(0) binding configuration. Likewise the addition of oxygen reduces the donor properties of the metal, thus reducing the availability of electrons for back-bonding into the T* orbitals of the carbonyl bound in the q2(C,0) configuration. The decomposition of acetone on clean R h ( l l 1 ) to liberate hydrogen exhibited a kinetic isotope effect when the acetone was fully deuterated. The kinetic isotope effect can be modeled by using statistical mechanics and transition-state theory. This approach was used by Gates et a1.I6 to predict the magnitude and temperature dependence of the KIE observed for removal of the hydroxyl hydrogen from methanol on Ni( 11 1). Our adaptation of their analysis is summarized here. The following expression for the KIE is given by transition-state theory:
TABLE III: Kinetic Isotope Effect for Acetone Decomposition ~alcdkH f kD freq, cm-' mode H modes D modes 270 K 300 K 330K v(CH,) 3000 2200 8.4 6.8 5.7 6(CH,b 1380 1100 2.1 2.0 1.9 p(CHY) 990 800 1.7 1.6 1.5
mead kulkn method peak temp leading edge analysis sequential reaction model TPHREELS
270K 4.1 18.9 1 1 -5 8.4
300K 3.6 3.6 9.0 6.8
330 K 3.2 0.9 7.4 5.7
partition functions for acetone-d6 are indicated by the D subscripts. Since the acetone molecule is bound to the surface, translational and rotational degrees of freedom may be omitted from the partition functions, and since the temperature of our experiments was too low to excite electronic transitions, only vibrational and internal rotational degrees of freedom contributed to the partition functions. If one of the vibrational modes is assigned as the reaction coordinate and the other modes are not significantly perturbed, the following expression for the KIE is obtained:
i#j
i#j
where the o,represent the vibrational modes of acetone bound to the surface, and N is the total number of vibrations. Internal rotations are assumed to make only a minor contribution to the partition functions. If the j t h vibrational mode is chosen as the reaction coordinate, it does not contribute to the partition function of the transition state, and the contributions of the unperturbed vibrational modes are found in the product of the partition functions of both the molecule and the transition state. Thus calculation of the KIE is reduced to a determination of the partition functions associated with the vibrational modes chosen as the reaction coordinate. The partition function for a vibrational mode that can be approximated as a harmonic oscillator is given by the following expression: e - ( ~ /n2 QHWi = 1 - ((wP/n (5) where a = hc/k has a value of 1.4387 K/cm-'. Note that this expression for the vibrational partition function implicitly defines the zero of energy as the bottom of the potential energy well, Le., this reference point is the same for both H and D-substituted species.I7 Putting this partition function into the expression for KIE above, one obtains a prediction of the KIE based on the vibrational frequencies of the parent molecule:
where kH and kD are the decomposition rate constants for acetone and acetone-& respectively, QHis the total partition function of both the acetone and the associated surface atoms, and QHt represents the partition function of the transition state. The
where wD and oHare the vibrational frequencies of the acetone-& and the acetone modes chosen as the reaction coordinate. With the vibrational frequencies obtained from the HREEL spectra for q2(C,0)-acetone and acetone-d6, Table I, the KIE was predicted from this expression. Table 111contains the kinetic isotope effects calculated for three temperatures assuming three different vibrational modes as the reaction coordinate. The KIE values calculated by using the rate parameters determined by the three methods of analyzing the TPD data and the TPHREELS data are also included in this table for comparison. It is clear that, regardless of the method of analysis, the kinetic isotope effect at 300 K was significantly greater than 2 and that the reaction clearly exhibited a primary kinetic isotope effect, Le., C-H scission was
(IS) Davis, J. L.; Barteau, M.A. Lmgmuir 1!389, 5, 1299. (16)Gates, S.M.;Russell,J. N.;Yates Jr., J. T. Surf. Sci. 1984.116, 199.
(17) Hill,T.L.An Introduction to Statistical Thermodynamics; Addison-Wesley: Baton, 1960.
(3)
3764 The Journal of Physical Chemistry, Vol. 95, No. 9, 1991 the rate-limiting step in acetone decomposition. The values obtained from the detailed simulation of the TPD and HREELS data, considering the sequential recombination of hydrogen atoms, are in excellent agreement with the expected value of 7 from statistical mechanics, which reflects the zero-point energy difference of C-H vs C-D vibrations. The KIE observed with the sequential reaction model exceeded the expected value of 6.8. The slightly greater value, 9, obtained from the sequential reaction model is not a significant discrepancy, given the inaccuracies introduced by the hydrogen/deuterium recombination kinetics and the resolution of the data and simulations in Figure 5. Discrimination of a KIE between 7 and 9 would require one to carry one additional significant figure on the activation energies used in these simulations and is not warranted by these data. Potential contributions of quantum mechanical tunneling effects to the slightly greater value of the KIE cannot, of course, be excluded. The origin of the bond breaking sequence for acetone decomposition on Rh( 111) is unclear. While proton abstraction to form enolate anions from acetone is a common reaction in basic solution and on basic oxide we are unaware of other examples of the activation of acetone by hydrogen abstraction on a metal surface. On thermochemical grounds alone, one would expect C-H scission to be disfavored relative to C-C scission in acetone; the homolytic bond dissociation energies for acetone are DCH= 41 1 kJ/mol and Dcc = 340 kJ/mol.2' Even more surprising was our observation that acetaldehyde decomposition on Rh( 11 1)22 has little in common with acetone decomposition. Acetaldehyde decomposition on Rh( 111) did not exhibit a kinetic isotope effect and produced significant yields of methane, in sharp contrast with the behavior of acetone. Both of these observations suggested that the C-C cleavage step in acetaldehyde decomposition preceded removal of methyl hydrogens; this reaction liberated methyl groups, some of which subsequently formed methane. The methylidyne or methylene groups produced by acetone decomposition were not hydrogenated to methane under the conditions of our experiments. The explanation for the difference between acetone and acetaldehyde decomposition sequences on Rh( 111) may arise from the requirements for the formation and decomposition of acetyl ( C H 3 C 4 ) species. Such species are common in carbonylation catalysis by soluble rhodium complexes,23and we have successfully isolated and identified them by HREELS in the course of acetaldehyde decomposition on Pd( 11 It is likely that (18) Miyata, H.; Toda, Y.; Kubokawa, Y. J . Card. 1974, 32, 155. (19) Koga, 0.;Onishi, T.; Tamaru, K. J . Chem. SOC.,Faraday Trans I 1980, 76, 19. (20) Vohs, J. M.; Barteau, M. A. J . Phys. Chem. 1991.95, 297. (21) McMillen, D. F.; Golden, D. M. Annu. Rev. Phys. Chem. 1982.33, 493. (22) Houtman, C.; Barteau, M. A. J . Caral., in press. (23) Dekleva, T. W.; Forster. D. Adu. C a r d 1986, 34, 81.
Houtman and Barteau acetaldehyde decomposition on Rh( 11 1) proceeds via acetyls (although we have yet to isolate these as stable species) and that these species decompose via C-C scission to liberate methyl groups. In the case of acetone, however, acetyl species could not be formed without prior cleavage of a C-C bond. Apparently the interaction of the rhodium surface with the carbonyl and methyl moieties in even the strongly bound v2(C,0)-acetone is insufficient to permit this reaction, and acetone decomposes instead by an initial rate-determining C-H bond scission.
Conclusions Acetone adsorbed in an v2(C,0)-configuration on the clean Rh(111) surface and in an v'(0)-configuration on the Rh( 111)-(2 X 2)O surface. In addition to altering the bonding configuration, electronic perturbation of the surface by the electronegative oxygen atoms greatly suppressed the decomposition of acetone. All of the acetone molecules adsorbed in the first layer on Rh( 11 1) decomposed unselectively to CO, H2, and C(ad);only 25% of the adsorbed acetone reacted on Rh( 111)-(2 X 2)0, and the products included those of oxidation, H20, C02, and CH3COO(ad), as well as decomposition. The large kinetic isotope effect observed by TPD and TPHREELS for decomposition of acetone and acetone-d6 on Rh( 111) clearly demonstrated that the bond-breaking sequence involved an initial rate-determining dehydrogenation preceding decarbonylation. Quantitative determination of the magnitude of the kinetic isotope effect required inclusion of the rates both acetone dehydrogenation and hydrogen atom recombination, since both processes influence the rate of dihydrogen evolution in the temperature range of interest, 250-350 K. Corroboration of the conclusion that dehydrogenation of acetone preceded decarbonylation was provided by the absence of methane production from acetone, in keeping with the liberation of C H or CH2species in the decarbonylation step. Both the hydrogen partitioning between high- and low-temperature channels in TPD and HREEL spectra of the adsorbed layer were consistent with the production of CH(ad) species via acetone dehydrogenation and subsequent rapid decarbonylation of the dehydrogenated surface intermediate. Acetone and acetaldehyde decomposition on Rh( 1 11) exhibited surprising differences that were consistent with the inability of acetone to form acetyl species by an initial C-C scission. Acknowledgment. This work was supported by the Department of Energy, Division of Chemical Sciences, Office of Basic Energy Sciences (Grant FG02-84ER13290). We thank Nicole Brown for assistance in the collection of TPD data. RWtry NO. 0, 7782-44-7; H, 12385-13-6; (CH,),CO, 67-64-1; Rh, 7440-16-6; deuterium, 7782-39-0. (24) Davis, J. L.; Barteau, M. A. J . Am. Chem. Soc. 1989, 1 1 1 , 1782.
