9425
J. Phys. Chem. 1993,97, 9425-9433
Interactions of Alcohols with a NiO( 100) Surface Studied by High-Resolution Electron Energy Loss Spectroscopy and Temperature-Programmed Desorption Spectroscopy Ming-Cheng Wu, Charles M. Truong, and D. Wayne Goodman. Department of Chemistry, Texas A&M University, College Station, Texas 77843-3255 Received: January 20, 1993; In Final Form: June 18, 1993"
Interactions of alcohols with well-defined NiO( 100) films prepared on Mo(100)have been studied using combined high-resolution electron energy loss spectroscopy (HREELS) /temperature-programmed desorption spectroscopy (TPD). The results show that alcohols adsorb associatively on NiO(100) a t 90 K and reversibly desorb upon heating under ultrahigh-vacuum conditions. Both methanol and ethanol are bonded to the cation sites of NiO( 100) via their oxygen atom with the methyl and ethyl group, respectively, directed away from the surface. At low coverages, there is a repulsive interaction between adsorbed alcohols, resulting in a decrease in the apparent activation energy of desorption. Below monolayer coverages, the change in the relative mode intensities in the HREELS spectra with respect to the methanol coverage correlates excellently with two desorption states evident in the TPD studies, indicating reorientation of a portion of the adsorbed methanol at high coverages. In the case of ethanol adsorbed on NiO( loo), however, only a single desorption state above 200 K is observed, consistent with the HREELS data which show that the relative mode intensities remain essentially unchanged within the monolayer coverage range. The absence of the u(0H) loss feature in the region of OH vibrations below monolayer coverages indicates the formation of strong hydrogen bonds between the hydroxyl hydrogen and the lattice oxygen. This feature, along with consideration of the surface symmetry of Ni0(100), leads to the conclusion that the hydroxyl proton of the molecules likely fluctuates about the neighboring anion sites, acting as a mobile proton. This result is important to our understanding of the nature of the precursor to the dissociated state in an acid-base reaction of the following type: A H B F? A- HB+.
+
I. Introduction Interactions of alcohols with well-defined solid surfaces are a subject of both fundamental and practical interest. For example, methanol synthesis from carbon monoxide/carbon dioxide and hydrogen over oxide supported catalysts is a process of major industrial importance. Accordingly, the kinetics and mechanism of this reaction have been studied extensively for the past decade.'-' The key step in methanol synthesis is presumed to involve the formation of adsorbedsurface intermediatesduring the reaction.'-7 Thus, in the reverse process, studies of methanol decomposition on well-defined solid surfaces under ultrahigh-vacuum (UHV) conditions should provide useful information regarding adsorbed CH,O species that may be of importance in this reaction. Adsorption and decomposition of alcohols on a variety of oxide catalysts have been studied extensivelyusing infrared techniques and other physical chemical methods.5-u A general trend revealed in these studies is that methanol decomposition proceeds via sequentiallyformed methoxy and formate intermediates."-22 On methanol synthesis catalysts, for example, methanol adsorbs dissociativelyto form a methoxy species at relatively low reaction temperatures (400 K) in the presence of gas-phase methanol.12 The methoxy species decomposes rapidly to form formate and carbonate intermediates upon removal of methanol from the gas phase and upon heating.I2 Heating of magnesia in methanol vapor in the 373-673 K temperature range, however,was reported to give rise exclusively to methoxide, and only in the presence of gas-phase oxygen was a formate species 0bserved.1~This differs from the early work of Kagel and GreenlerI4 in which the production of a formate-like surface compound was found upon heating of magnesia samples at 438 K in pure methanol vapor. The presence of surface carbonates on MgO was reported after heating the sample to 700 K.14 Desorption products of methanol adsorbed onto powdered oxide catalysts at elevated pressures were found to be primarily H2, To whom correspondence should be addressed. Abstract published in Advance ACS Absrracts, August 15, 1993.
+
CO, and C02.23924 In some cases, the production of methane and water was also reported.19 Combined infrared/TPD studies of methanol adsorbed onto ZnO powdered samplesz3showed that the desorption of H2 found within the 450-550 K temperature range is due to the decomposition of the methoxy groups as they convert to formate species. Also, concurrent desorption of CO and COZ at higher temperatures is a consequence of the decomposition of the formate intermediate. The formation of carbonates on the stoichiometric spinel, ZnCrz04, was also reported24and was observed to lead to the desorption of COZat 720 K. In all of the above cases, it has been shown*1-*4that the participation of substrate oxygen is necessary to form formate and carbonate species on the catalyst surfaces. Following the above studies, it is clear that although methanol chemistry on oxide materials has been explored extensively,these studies have been carried out primarily on powdered samples. Accordingly, the information obtained necessarily represents an average of the various crystal faces exposed plus a variety of defects present on the polycrystalline powdered samples. In particular, the reaction of methanol over powdered oxide catalysts may well be controlled by various defect sites whose respective contribution to the decomposition mechanism is difficult to identify. The investigation of methanol interaction with a welldefined solid surface is thus of great importance to elucidate the nature of active sites and the microscopies of methanol decomposition. Accordingly, methanol adsorption and decomposition on single-crystalmetal surfaces under UHV conditions have been a focus of intensive research for the past decade.2s34 Several studies using ultraviolet photoelectron spectroscopy (UPS)z5,2*and high-resolutionelectron energy-lossspectroscopy (HREELS)28-34have shown that methanol is bonded via the hydroxyl oxygen of the molecule with the methyl group directed away from the surface. The decomposition mechanism of methanol on most metal surfaces involves OH bond cleavage to form methoxy, followed by its decomposition to carbon monoxide and additional hydrogen ad atom^.^^^^ No CHzO or CHO intermediate in the decomposition of methoxy to CO could be
0022-365419312097-9425%04.00/0 0 1993 American Chemical Society
9426
The Journal of Physical Chemistry, Vo1. 97, No. 37, 1993
identified on these surfa~es.~’j~~9~4 Theone exception to this trend appears to be the Pd( 110) surface,33 where after annealing a saturated methanol layer to 300 K, loss features a t 1555 and 1640 cm-l were observed and were tentatively assigned as a v( C 4 ) mode of a HCO surface species adsorbed onto two different sites. Of particular relevance to the present study are UHV HREELS studies of alcohols adsorbed onto metals. The HREELS spectra of molecular methanol chemisorbed onto several transition-metal surfaces at monolayer coverages exhibit a number of loss features in the 200-4000-cm-1 spectral region.28-34 A predominant loss feature found between 1000 and 1035 cm-1 is due to excitation of the CO stretching mode (v(C0)). Weaker loss features observed in the frequencyranges 1120-1 140and 1430-1470cm-l arise from the CH3 rock (p(CH3)) and deformation (6(CH3)), respectively. The CH3 symmetric (v,(CH3)) and asymmetric (degenerate)stretch (v,(CH3)) modes were found in the frequency ranges 2800-2845 and 2945-2975 cm-l, respectively. The frequency of the OH stretch (u(0H)) depends upon the methanol coverage and exhibits a red shift by more than 300 cm-I from the gas-phase value of 3681 cm-1. The OH bending (?r(OH)) mode and the oxygen-metal stretching (v(0-M)) mode have been observed in the lower frequency range 235-770 cm-I in several of these studies.28,30,313 An intriguing feature revealed in these EELS studies is the change of the relative mode intensities of molecularly adsorbed methanol with respect to the methanol co~erage.~89~0 For example, HREELS spectra of low coverages of adsorbed methanol on Rh(IOO)3O resemble closely those of methoxy with v(0-M) and v(CO) modes dominating the spectra, while spectra of high coverages of methanol exhibit much more intense n(OH), 6(CH3), v(CH), and v(0H) modes, similar to the spectra of methanol multilayers. These changes imply that the interaction between the neighboring molecules increases at the higher coverages. Such an increase likely results in reorientation of some of the adsorbed methanol such that the vibrational motions associated with these hydrogenic features have dipole moment components perpendicular to the surface.30 As part of a continuing study of relatively complex organic species chemisorbed on insulating metal oxides, we report here the results of a combined HREELS/temperature-programmed desorption (TPD) investigation of the adsorption of alcohols on NiO( 100) films under UHV conditions. These studies focus on the adsorption geometry of alcohols on NiO(100) and on the nature of bonding between alcohols and the oxide surface. The reaction intermediates formed upon adsorption of alcohols on NiO( 100) and the desorption products during annealing experiments will be examined. Our new strategy to study oxides involves the following procedure: (1) preparation of ultrathin metal oxide films on a single-crystal refractory metal surface. Utilization of a refractory metal as a substrate minimizes potential interfacial reactions and interdiffusion at the interface and provides a high melting point conducting substrate; (2) deposition of oxide precursor metals onto the refractory metal surface in a controlled oxygen environment; and (3) spectroscopiccharacterization of adsorbates chemisorbed on these well-characterized oxide films utilizing combined HREELS/TPD. Such an approach allows us to employ various surface-sensitive techniques such as HREELS, Auger electron spectroscopy (AES), and low-energy electron diffraction (LEED) without difficulties associated with surface charging. Combined HREELS/TPD spectroscopies together with the thinfilm preparation technique facilitate the acquisition of a variety of data on oxide materials which have been heretofore precluded. Until recently, very little work has been done using HREELS to study adsorbates on ionic substrates because of the great difficulty associated with lossesdue to excitation of surface optical phonons of the oxide substrates. Since the intense multiple phonon
Wu et al.
KINETIC ENERGY (eV)
Figure 1. Auger spectrum of a -20 monolayer (ML) NiO(100)film grown on Mo(100) at 300 K.
losses generally extend over a wide vibrational frequency range of the HREELS spectra, it is not practical to observe directly adsorbate losses (which are several orders of magnitude smaller in intensity than the phonon losses) in the 0-4000-cm-1 spectral range. Recent progress has been made in circumventing the difficulties associated with these phonon losses. For example, deconvolution techniques have been implemented to remove or suppress surface phonon combination peaks from the HREELS ~pectra.3~936Clearly, the Fourier deconvolution methods are superior to simply acquiring difference spectra between clean and adsorbate-covered ~ u r f a c e s nevertheless, ; ~ ~ , ~ ~ it is still basically an “indirect” technique. Of particular concern are the spurious features that may be introduced in the spectrum by the Fourier deconvolution procedure itself due to difficulties inherent in deconvolution schemes.35~36In the recent st~dies,3’-3~we have developed a new approach to acquiring HREELS data in order to circumvent the difficulties associated with these phonon losses. By utilizing a high-energy incident electron beam, this new approach enables the direct observation of weak loss features due to the excitation of adsorbates without serious interference from intense multiple surface optical phonon losses of oxide materials. In the present study, we exploit these recent successes to study the interaction of alcohols with NiO films. 11. Experimental Details
These studies were carried out utilizing an ultrahigh-vacuum (UHV) system containing a two-tiered chamber, described elsewhere,37J8 with capabilities for HREELS, Auger electron spectroscopy (AES), low-energy electron diffraction (LEED), and TPD and for sample heating and cooling. The refractory metal Mo( 100) was chosen as a substrate for preparing the thin NiO films. NiO possesses a rock salt lattice, with the anions and cations situated at the corners of two interpenetrating facecentered-cubic sublattices. Since the lattice mismatch between the (100) face of nickel oxide and the Mo( 100) structure is only a few percent, NiO films are anticipated to grow epitaxially on the (100) face of Mo. Thin NiO films were prepared by depositing Ni onto Mo( 100) in a controlled oxygen atmosphere, followed by annealing in oxygen ambient. Nickel deposition was performed via thermal evaporation of a high-purity ribbon tightly wrapped around a tungsten filament. The film stoichiometry was adjusted by tuning the flux on the Ni evaporation, which was monitored by a mass spectrometer mounted in line with the metal source, a t an oxygen pressure of 7 X lO-7Torr. The growth of the films was controlled a t a rate of approximately one monolayer (ML) per minute, and the films were typically 20 M L thick. Displayed in Figure 1 is the Auger spectrum acquired following the film synthesis which shows features essentially identical to those of single-crystal NiO( Structural studies using LEED show that the NiO films
1The Journal of Physical Chemistry, Vol. 97, No. 37, I993
Interactions of Alcohols with a NiO( 100) Surface 2.50 r
I
TPD CH,OH/NIO(l OO)/Mo(iOO)
> k
1.25
v) 1.00
55
2
0.75 0.50
ll
0.25
0.00
100
140
180
220
280
300
340
380
420
460
500
(K) Figure 2. TPD spectra of CH3OH adsorbed on a NiO( 100) surface at 90 K as a function of coverage (ML): (a) 0.09, (b) 0.14, (c) 0.23,(d) 0.35,(e) 0.42,( f ) 0.52,(g) 0.65, (h) 0.94, (i) 1.09, and 1.40. The desorption states designated as a,41, and 4 2 are indicated. TEMPERATURE
u)
prepared using the above approach exhibited an excellent (1 X 1) LEED pattern indicative of long-range order in the films. The detailed studies regarding film synthesis and characterization are described elsewhere.4l Spectroscopicgrade methyl alcohol, 99.8%, methyl-d alcohol, 99.5+ atom % D, and ethyl alcohol, 99.8%, were used in the adsorption study and were further purified in the manifold via freeze-pump-thawcycles prior to introduction of the adsorbates into the vacuum chamber. Alcohols were introduced via a gas nozzle doser, which was directed to the crystal surface and at a distance of approximately 0.5 cm from the crystal face, for the thermal desorption experiments; however, the exposure of the specimen to alcohols was performed via back-filling the UHV chamber for the HREELS measurements. Alcohol surface coverages are reported with respect to the saturation or monolayer coverages. The monolayer coverages (ML) of alcohols on NiO(100) deduced from the TPD spectra are assumed to be 0 = 1. This value is used to calibrate the alcohol coverages deduced by measuring the area of the alcohol TPD feature. The HREELS spectra were acquired in the scattering compartment of the two-tiered chamber using an electron beam of the spectrometer (LK-2000, Larry Kesmodel Technologies)with a primary energy of Ep 50 eV and at an incident angle of 60’ with respect to the surface normal. The other parameters used in obtaining the EELS spectra presented here were as follows: spectral resolution (full width at half-maximum of the elastically reflected electron beam), 70-90 cm-l; typical count rate in the elastic peak of the clean surfaces, 50 000 Hz. Adsorption of alcohols resulted in approximatelya 10-fold decreasein the elastic peak intensity. Because of the low signal levels encountered in our HREELS measurements, an optimal Fourier filtering technique42has been utilized to remove high-frequencyrandom noise and spikes in the HREELS spectra. All HREELS spectra reported below have been smoothed using this Fourier filtering technique.42 The temperature-programmeddesorption spectrawere obtained at a linear heating rate of 5 K/s using a quadrupole mass spectrometer (Model lOOC, UTI Instruments).
III. Results A. TPD Studies. The adsorption of alcohols on NiO(100) films has been examined using TPD. Thermal desorption products from methanol adsorbed on NiO( 100) films at 90 K have been followed by monitoring several masses ( m l e = 2,18,28,3 1, and 32). Displayed in Figure 2 are thermal desorption signals for m / e = 3 1, the strongest fragment in the cracking pattern of CH3OH.43 Since during the thermal desorption process the intensity ratio of m l e = 3 1 to m / e = 32 remained essentially constant, the possibility that the formation of the mass 31 fragment is a
9421
consequence of the direct desorption of a C H 3 0 species can be ruled out. No desorption signal was observed for masses other than the parent peak and its fragments, indicating reversible adsorption of methanol on NiO(100) surfaces under UHV conditions. This behavior contrasts with that found for methanol chemisorbed on most transition-metal surfaces under UHV conditions where dissociative adsorption leading to further decomposition of methanol to form adsorbed CO and H is generally found.30J1s4 TPD spectra of methanol chemisorbedon Ru(001),3I Rh(001),30 and Pt( 11l),34 for example, show desorption products of H2 and CO, arising from CH30H decomposition, in the temperature ranges 300-400 and 450-550 K, respectively. Desorption of the parent molecule occurs at a lower temperature range of 150-300 K.30,3I ,34 At low exposures (e < 0.5), TPD spectra ( m / e = 31) are characterized by the presence of a single desorption state, designated as a 82 state, shown in Figure 2. Further exposure results in the concurrent appearance of two new states at lower temperatures. The desorption state (81 state) at 247 K remains unchanged in its peak maximum as the CH3OH exposure is increased, whereas the peak maximum of the j32 state shifts from 298 K at 0 = 0.09 to 271 K at 6 N 1.00. These j31 and 82 states can be interpreted as arising from the desorption of monolayer methanol. The sharp desorption feature (astate) found at 150 K, whose peak intensity increases continuously with increasing CH30Hexposure, apparently arises from condensed multilayers of methanol. Physically condensed multilayers of methanol on severalmetal have been found to desorb at a similar temperature of 150 K. It is evident in Figure 2 that population of the multilayers of adsorbed methanol begins prior to completion of the first monolayer. This behavior is likely due to the high sticking probability (by hydrogen bonding) and low mobility of methanol molecules at the low adsorption temperature. It is particularly noteworthy that the a-state peak in Figure 2 does not exhibit a common “leading edge” on the lowtemperature side, behavior frequently observed for many physisorbed systems.@ It has been argued28that the desorption of multilayers of methanol should not be described by zero-order kinetics. Instead, the observed shifts in the leading edge and peak maximum indicate a fractional order rate process, which demonstrates the existence of hydrogen bonding within the condensed methanol phase.28 The formation of hydrogen bonding in the multilayers of methanol condensed on NiO(100) is confirmed by our HREELS studies which follow. The evolution of various desorption features as a function of ethanol exposure has also been followed by monitoring several masses ( m / e = 2, 18, 28, 31,45, and 46). Displayed in Figure 3 are thermal desorption spectra of the strongest fragment ( m / e = 31) of the parent m0lecule.~3 There were no detectable desorption signals for masses other than the parent molecule and its fragments, indicating reversibleadsorption of ethanol on NiO(100) under UHV conditions. In contrast to methanol adsorbed on Ni0(100), only a single desorption peak (j3 state) was found for ethanol above 200 K. Similar to the j32 state of methanol adsorbed on Ni0(100), this j3 state peak maximum shifts from 336 K at 8 = 0.05 to 276 K at a coverage approaching saturation of the first layer, as shown in Figure 3. The signal from desorbing ethanol, which levels in the 180-220 K temperature range relative to that of monolayer desorption, was considerably higher in the ethanol TPD spectra compared to the methanol TPD spectra. Likewise, a lowtemperature desorptionfeature at 160 K is assigned to condensed multilayers of ethanol since its peak intensity increases continuously with increasing CzHsOH exposure. B. HREELS Studies. Adsorption of alcohols on thin NiO(100) films has been further investigated using HREELS.
-
3000.0
TPD C,H,OH/NiO(i 00)/Mo(i 00)
c
HREELS CH,OH/NiO(lOO)/Mo(lOO)
2700.0 -
I 921.00
* Wu et al.
9428 The Journal of Physical Chemistry, Vol. 97, No. 37, 1993
2400.0
k 1.91,
1150
71
-
f
17.50
xi00
2100.0
-
2 1800.0 A
7.00
*
b L
v
II
0.00 100
5 140
180
220
260
300
340
380
420
460
500
TEMPERATURE (K)
1500.0
2690
-
z w I-
Figure 3. TPD spectra of C2HsOH adsorbed on a NiO( 100) surface at 90 K as a function of coverage (ML): (a) 0.05, (b) 0.10, (c) 0.14, (d) 0.25, (e) 0.38, ( f ) 0.51, (9) 0.60, (h) 0.91, (i) 1.17,d) 1.53,and (k) 1.91. The desorption states designated as CY and @ are indicated.
r-1475
z 1200.0
-
900.0 -
600.0 -
7000.0
HREELS CH,OH/NiO(i OO)/Mo(lOO) 1475
6000.0
1140
300.0
-
I
L
I A
0.0 -2
0
1000
2000
3000
a
4000
5000
E.4000.0
WAVENUMBER (cm-l) F i p e 5. HREELS spectra of CH3OH adsorbed on a -20 ML NiO(100) film at 90 K at different annealing temperatures: (a) 90 K and (b) 175 K. The surface coverage of methanol is 3.33 ML. The spectra werecollectdatE0 N 5OeVandat thespecularlyreflactedbeamdirection.
5
TABLE I: Vibrational Frequencies (in cm-l) and Mode
5000.0
LI
N
z
Assignments for Methanol Chemisorbed on NiO(100) and for Methanol on Various Single-CrystalMetal Surfaces mode assignt NiO(100) Rh(100) Pd(100) Pd(ll0) Pt(ll0) Ni(ll1) 3335 3345 3400 3280 3215 2920 2960 2945 2975 2930 2955 2845 2800 2810 1450 1455 1470 1465 1430 1455 1155 1135 1130 1120 1080 1005 1025 1015 1000 1035 745 770 730 700 685 260 330 235
3000.0
z 2000.0
1000.0
0.0
J ,
0
1000
2000
3000
WAVENUMBER
4000
5000
(em-‘)
Figure 4. HREELS spectra of methanol adsorbed on a -20 ML NiO-
(100) film at 90 K as a function of CH3OH coverage (ML): (a) 0.23, (b) 0.67, (c) 1.67, (d) 2.33, and (e) 3.33. The spectra were collected at EO Y 50 eV and at the specularly reflected beam direction. Displayed in Figures 4 and 5 are two sets of HREELS spectra of methanol adsorbed onto a -20 monolayer NiO film a t 90 K as a function of exposure and annealing temperature, respectively. Exposing the sample to CH30H at 90 K gives rise to a number of distinct loss features in the 7504000-cm-l region. By analogy to the results obtained from infraredl l-24 and HREELSZ8-” studies of methanol adsorbed onto a variety of surfaces, the assignments of these losses a t monolayer coverages are straightforward. The losses observed at 1080, 1155, 1450, and 2920 cm-l are due to excitations of the CO stretch (v(CO)), the CH3 rock (p(CH3)), the CH3 deformation (6(CH3)), and the CH3 stretch (v(CH3)), respectively.28-34 Compiled in Table I are HREELS vibrational frequencies and mode assignments for chemisorbed methanol on NiO( 100) and on various metal s~rfaces.~8-’~ It is noteworthy that the v(C0) mode is blue-shifted by -50 cm-1 relative to the gas-phase value of 1033 whereas the frequency of this mode shifts slightly downward when adsorbed onto metal surfaces (see Table I). Infrared studies performed on
powdered oxide materials12-22also show a higher CO stretch frequency of methanol than the corresponding gas-phase value. The origin of this blue shift is probably analogous to that giving rise to blue-shifted CO adsorbed onto NiO( loo), that is, a “wall e f f e ~ t ” . The ~ ~ ,positive ~ ~ shift of the CO stretch frequency from the CO gas-phase value is interpreted to be due to the repulsion arising from C - 0 stretching against the rigid surface without moving its center of mass from the surface.46~47Consequently, this “wall effect” is predicted to vary strongly with the distance of CO from the ~ u r f a c e . ~ , ~ ’ There are differences in the relative mode intensities between the spectra at low coverages and those a t high coverages. The spectra at low coverages (8 < 0.5) resemble closely those at low coverages of methanol on single-crystal metal ~urfaces,28-3~ with a predominant loss peak at 1080 cm-l and some weaker features found a t 1150, 1450, and 2920 cm-l. On the other hand, the spectra a t high coverages (0.5 < 0 < 1.0) exhibit much more intense p(CH3), 6(CH3), and v(CH3) modes, as shown in Figure 4. Thep(CH3)and b(CH3) modes, for example, appear as intense as the v(C0) mode as shown in Figure 4b. The origin of these changes in the relative mode intensities will be discussed in the following section. At coverages exceeding one monolayer, the p(CH3) and v(C0) modes merge into one and become indistinguishable.
The Journal of Physical Chemistry, Vol. 97, No. 37, 1993 9429
Interactions of Alcohols with a NiO(100) Surface 2500.0
7000.0
HREELS
2000.0
-E.
XIGO
a t
\
03
4.1
1375 1035 I
-
o
i
l
ii4pi465
6000.0
s
COVERAGE (ML)
5000.0
N
1500.0
* bz E z
E
Iw z
HREELS CH,OD/NiO(l OO)/Mo(lOO)
1000.0 XI0
4000.0
3000.0
2000.0 X40
500.0
1000.0
0.0
b
- ,
1
0.0
WAVENUMBER (crn-l) Figure 6. HREELS spectra of methanol adsorbed at 90 K on (a) NiO(loo), 0 = 1.0, and (b) Mo(100), 0 = 1.0. Spectrum c was acquired following adsorption of formic acid (0 = 2.0) on a NiO(100) surface at 90 K and subsequently heating to 220 K. The insert shows the intensity ratio of the OH stretch to the CH stretch plotted as a function of coverage
upon adsorption on NiO( 100) surfaces. The spectra were collected at EO 50 eV and at the specularly reflected beam direction.
It is noteworthy that Figure 4 shows no loss feature in the vibrational frequency region of the OH stretch; only at coverages exceeding one monolayer does the u(0H) loss appear. The frequency of this OH stretching vibration is red-shifted by approximately 430 cm-* with respect to the methanol gas-phase value of 368 1cm-1945 and its bandwidth is considerablybroadened. These are clear indications2’” that hydrogen bonding is occurring in the multilayers of methanol adsorbed on NiO(100). The absence of the u(0H) loss is also clearly evident in the HREELS spectra acquired following the annealing experiments. Heating of a methanol multilayer ( 6 = 3.33) to 175 K results in the desorption of physisorbed multilayers of methanol, the disappearance of the u(0H) feature, and the appearance of the u(C0) and p(CH3) loss features, as shown in Figure 5 . The reappearance of the u(C0) and p(CH3) loss features is consistent with the resultsshown in Figure 4 in which these two losses become indistinguishableas the CH30H coverage exceeds one monolayer. In all of the above HREELS spectra, no loss feature associated with free hydroxyl groups arising from the abstraction of the hydroxyl hydrogen of the molecule could be identified. Such free hydroxyl groups should give rise to a distinct peak at 3650 cm-1.37-39.48 Displayed in Figure 6c is the HREELS spectrum acquired followingadsorption of formic acid on NiO( 100)surfaces and subsequent heating to 220 K under UHV conditions. Our previous studies48have shown that formic acid adsorbs associatively on NiO( 100) surfaces at 90 K and undergoes heterolytic dissociation upon heating to >200 K to form a formate intermediate. The OH stretch at -3550 cm-l associated with dissociated hydrogen adsorbed on surface lattice oxygen sites is clearly evident in the spectrum of Figure 6c. On the basis of this, together with the TPD results presented in the preceding section, we conclude that under UHV conditions methanol adsorbs associatively on NiO(100) surfaces at 90 K and undergoes reversibledesorption upon annealing. This result is not surprising considering that the NiO( 100) surface is relatively unreactive
-
0
,
a,-,
I000
2000
3000
4000
5000
WAVE NUMBER (crn-l) Figure 7. HREELS spectra of methanol-d, 99.5+ atom 96, adsorbed on a -20 ML NiO( 100) film at 90 K as a function of CH3OD coverage (ML): (a) 0.33, (b) 0.67, (c) 1.00, (d) 1.33, (e) 1.67, and (f) 3.33. The spectra were collected at Eo 50 eV and at the specularly reflected
beam direction. with respect to chemis~rption.~~ In fact, nearly perfect UHVcleaved NiO( 100) has been found to be completely inert to the chemisorption of 0 2 , HzO, SO*,and many other organic molecules.49 The associative adsorption of methanol on NiO(100) is also consistent with the trend revealed in the study of the acid/base properties of NiO( 100) films.48.50,51 Our previous studies‘s have shown that formic acid dissociates on NiO( 100) surfaces only at temperatures greater than 200 K. Given that formic acid is a much stronger acid than methanol, it is not likely that methanol dissociates on NiO(100) at 90 K. The absence of the u(0H) feature of methanol adsorbed on NiO( 100) thus contrasts sharply with the results observed for methanol adsorbed on several transition-metal ~urfaces.28*30-3~ These results indicate a distinct u(0H) feature at -3300 cm-1, whose intensity is comparable to that of the CH3 stretch, upon methanol adsorption at 6 5 1.O. Figure6,a and b, shows HREELS spectra of methanol adsorbed on NiO( 100)and Mo( 100)surfaces, respectively. Exposing Mo( 100)surfaces to CH3OH at monolayer coverages gives rise to a distinct loss feature at 3325 cm-l, which arises from the OH stretch of molecular methanol on Mo(100). The insert in Figure 6 shows the intensity ratio of the OH stretch to the CH stretch plotted against methanol coverage. For methanol adsorbed on NiO( loo), this intensity ratio increases only at coverages exceeding monolayer. The absence of an OH stretch following methanol adsorption in the HREELS spectra of methanol-d on NiO( 100) (shown in Figures 7 and 8) is likewise consistent with an unusually low cross section for the OH stretch. Exposing the sample to CH3OD at 90 K gives rise to a number of distinct loss features essentially identical to those of CH3OH on NiO( 100). The losses observedat 1080,1165,1455,and2930cm-1 areduetoexcitations of the CO stretch (u(CO)), the CH3 rock (p(CH3)), the CH3 deformation (6(CH3)). and the CH3 stretch (u(CH3)), respe~tively.2”3~The u(0D) feature was observed only at coverages exceeding one monolayer. The above results also eliminatethe ambiguity that the u(0H) frequency might be shifted downward to the region of the OH stretch such that the two
9430 The Journal of Physical Chemistry, Vol. 97, No. 37, 1993
Wu et al. 12000.0
7000.0
HREELS CH3OD/NiO(1OO)/Mo(10O)
~
11000.0
6000.0 -
10000.0
9000.0
5000.0 -
-
x 1 m hI i 11170 450
77
N
+
*
=4000.0 -
bz
*ooo.o~ 7000.0
5z
W 6000.0
3000.0 -
t
z
(I xlooMQ 2955
1
5000.0
I l l
f
x100
f
\vu e
xloo
d
\
\
I
C
2000.0
-
1000.0
-
b
2940 I
1000.0
0.0
A , 0
1000
1 11
\
2955 l a
I
2000
3000
4000
5000
WAVENUMBER (cm-l) Figure 8. HREELS spectra of CHsOD adsorbed on a -20 ML NiO(100) film at 90 Kat different annealing temperatures: (a) 90, (b) 175, (c) 175, and (d) 250 K. Spectrum c was acquired at an angle of 8 O off
the specular direction. Note a decrease of the v(C0) mode relative to the p(CH3) mode in spectrum c. The surface coverage of methanol-d is 3.33 ML. The remaining spectra were collected at Eo = 50 eV and at the specularly reflected beam direction.
3500.0 1
HREELS C,H,OH/NiO(I OO)/Mo(lOO)
bands are superimposed. The changes of relative mode intensities as a function of methaonol exposure and annealing temperature in Figures 7 and 8, respectively, were found to be nearly identical to those found in Figues 4 and 5 . Spectrumc of Figure 8, acquired a t an angle 8 O off the specularly reflected beam direction, clearly shows a decrease of the v(C0) mode relative to the p(CH3) mode, indicating the dipole nature of this vibrational mode. Previous HREELS studies33have shown that the v(C0) mode is strongly dipole allowed, whereas hydrogen-associated modes of the molecule are predominantly excited by impact scattering. Adsorption of ethanol on thin NiO(100) films has also been examined using HREELS. Figures 9 and 10 display HREELS spectra of C2H5OH adsorbed on a -20 monolayer NiO film at 90 K as a function of exposure and annealing temperature, respectively. Exposing the sample to C2HsOH at 90 K gives rise to several distinct loss features in the 750-4000-~m-~ region. The losses observed at 830, 1090, 1415, and 2955 cm-* are due to excitations of the CC stretch (v(CC)), the CO stretch (u(CO)), the CH2,3 deformation (6(CH2,3)), and the CH2,3 stretch (v(CH2,3)), respectively.29 OH stretching vibrational modes at methanol coverages below one monolayer were also not observed for ethanol adsorbed on NiO(100) (Figures 9 and 10). The u(OH) feature appeared only at coverages greater than one monolayer. IV. Discussion A. Chemisorbed Alcohol Species and Their Adsorption Geometries on NiO(100). In the preceding section, the conclusion was reached that methanol and ethanol adsorb associatively on NiO(100) at 90 K and undergo reversible desorption upon annealing. The following section will address the adsorption geometry and bonding configuration of these molecules on NiO(100). Since the oxygen lone-pair electrons of the alcohols will likely play a dominant role in adsorption and reaction, the surface
2500.0 -
WAVEN uMBER (cm-' Figure 10. HREELS spectra of CzHsOH adsorbed on a -20 ML NiO(100) film at 90 K at different annealing temperatures: (a) 90, (b) 175, and (c) 250 K. The surfacecoverage of ethanolis 2.86 ML. The spectra werecollectedatE0 = 50eVandat thespecularlyreflectedbeamdirection.
chemistry for methanol and ethanol on NiO( 100) is expected to be similar. The following discussion will focus primarily on methanol adsorption. As shown in Figures 4-8, the p(CH3), 6(CH3), and v(CH3) loss features at 8 < 1.0 exhibit relatively small shifts from their corresponding gas-phase frequencies, indicating a largely un-
Interactions of Alcohols with a NiO( 100) Surface perturbed methyl group of methanol following adsorption onto NiO( 100). Accordingly, we propose that methanol adsorbs onto the cation sites of NiO(100) via the hydroxyl oxygen with the methyl group directed away from the surface. A bonding configurationof this type is anticipated for the following reason: the lone pairs of electrons on the oxygen atom of methanol constitute a donor ligand, whereas the metal cation is electron deficient, acting as an electron acceptor. In the classic picture of Lewis acid/Lewis base interactions, the adsorptionof methanol on nickel oxide can be described by methanol acting as a Lewis base and the nickel cation as a Lewis acid. The above bonding configurationis also proposed for methanol adsorbed On most surfaces* Work function measurements carried out on several metal s ~ r f a c e s ~ * ~ have ~ ~ ~ 5shown ~ - 5 ~a significant decrease of the work function upon methanol adsorption, clearly demonstrating that methanol adsorbs with the oxygen end of the molecule pointing toward the surface. This bonding model is also supported by HREELS studies.28-34 HREELS spectra of methanol adsorbed on a number of metal surfaces indicate that the v(0H) and ?r(OH) vibrational modes shift dramatically from their corresponding gas-phase values, whereas the frequencies of the losses associated with methyl vibrations remain essentially unchanged. Information relevant to the bonding strength of methanol to NiO(100) can be deduced from the thermal desorption studies. The binding energy of methanol on NiO( 100) should correspond approximately to the activation energy of desorption deduced at a coverage approaching zero. Assuming first-order kinetics and a preexponential factor v = 1 X 1014Hz, the activation energy of desorption of the 82 state at 8 = 0.09 is deduced to be 19.5 kcal/mol, using Redhead’s peak maximum method.55 This value is significantly higher than the correspondingvalues (- 11 kcal/ mol) estimated for methanol desorption from transition metals.28J 1.34 The adsorption of molecular methanol on transition-metal surfaces is, in fact, considered to be chemicallyweak.28 Thermal desorption studies performed on several metal surfaces with varying crystallographicstructure have shown the binding energies of methanol to be essentially the same.28v56 The sign and magnitude of the methanol-induced work function change are also very similar for methanol adsorbed on varying substrates: a decrease in the work function was found in all cases, with an average decrease of approximately 1.5 eV.28929352-54 On the basis of these observations, Christmann and Demuth28suggested that localized d orbitals play only a minor role; that is, bonding to the surface arises primarily from an interaction between the oxygen lone-pair electrons and the delocalized metal electrons with sp character. On the other hand, the present study exhibits a quite strong binding of methanol to Ni0(100), implying that the localized d levels of the Ni cations probably participate in bonding with methanol. Further studies utilizing techniques which probe electronic structure are necessary to clarify this point. The TPD spectra shown in Figure 2 reveal two distinct desorption states at monolayercoverages. At 8 < 0.5, the apparent activation energy of desorption of the 82 state (estimated using the Redhead’s peak maximum method) decreasesfrom 19.5 kcal/ mol at 8 < 0.1 to 17.7 kcal/mol at 8 = 0.5. The decrease in the activation energy of desorption with an increase in CH30H coverage clearly indicates a repulsive interaction between the adsorbatesin the 82state. Given that methanol is a polar molecule with a dipole moment in the gas phase of 1.70 D,57lateral repulsion between the neighboring dipoles, which are probably oriented in an ordered fashion at low coverages, is expected. Although the exact orientationof methanol cannot be determined in the present study, the CO axis of the molecule in the 82 state is likely oriented with its dipole component perpendicular to the surface. Likewise, it is proposed that the dipole axes of the adsorbates in the 81 state are inclined toward the surface with their dipole
The Journal of Physical Chemistry, Vol. 97, No. 37, 1993 9431
W
Figure 11. A ball model for methanol adsorbed on NiO( 100). The arrows represent the hydroxyl proton fluctuating around the four neighboring oxygen anions.
moments oriented in random directions. This adsorption geometry of the molecules in the 81 state is consistent with the TPD results (Figure 2), which show essentially a constant peak desorption maximum of the 81 desorption state with an increase in CH3OH coverage. The constant desorptionpeak temperature likely arises from weaker dipole-dipoleinteractions (due to randomlyoriented dipoles) between CH30H molecules in the 81 state. Further support for the adsorption geometries of methanol in the 81 and 8 2 states comes from the HREELS studies which follow. In Figures 4, 5 , 7, and 8, HREELS spectra show the change of relative mode intensities with respect to the methanol coverage. This feature correlates excellently with the presence of the two distinct desorption states, i.e., the 81 and 8 2 states. At 8 < 0.5, a distinct mode intensity pattern of HREELS spectra was observed, as describedin the preceding section, which corresponds to the 82 state. A close resemblance of these spectra with those of methoxy on metal surfaces supports an adsorption geometry of the methanol molecules in the 82 state in which the dipole axes of the molecules are predominately perpendicular to the surface. The HREELS spectra of high coverages of methanol (8 > O S ) , on the other hand, exhibit much more intense modes associated with vibrations of the methyl group. This mode intensity change relates closely to the change in the TPD spectra where an additional desorption state, i.e. the 81 state, is observed. The change of the mode intensity then arises from a reorientation of the molecules in the state such that the vibrational modes of the methyl group have their most significant dipole moment component perpendicular to the surface. In the case of ethanol adsorbed on NiO( loo), only a single desorption peak above 200 K is apparent as shown in Figure 3. This feature correlates well with that found in the HREELS spectra in which the relative mode intensities remain unchanged as a function of ethanol coverage, as shown in Figures 9 and 10. In view of the complexityof ethanol,the orientationand desorption state of the molecules are not as discrete as those of methanol. This also accounts for the rather broad desorption peak of the ethtinol monolayer and the higher desorption signal levels in the 180-220 K temperature range. B. Delocalization of the Hydroxyl Proton of Alcohols. Adsorption of alcohols on NiO( 100) gives rise to a number of distinct loss features which can be unambiguously attributed to the correspondingvibrational modes of alcoholsin the gas phase4s or on the s u r f a ~ e . ~ However, ~”~ in all the cases, no loss peak in the region of v(0H) (or v(0D) vibrations was found at 8 I 1.O. The absenceof the v(0H) (or v(0D)) mode thus contrasts sharply with the results found for methanol adsorbed on several metal surfaces. The origin of this “invisibility” of the v(0H) (or v(OD)) mode will be discussed in this section. In the preceding section, we have shown that alcoholmolecules in the first layer are bonded to the surface cation sites of NiO(100) via the oxygen atom, with the methyl/ethyl group directed away from the surface. Depicted in Figure 11 is a ball model for methanol adsorbed on NiO( 100). In this bonding configuration,
9432
The Journal of Physical Chemistry, Vol. 97, No. 37, 1993
the hydroxyl hydrogen of methanol is located at a distance with respect to the surface anion sites such that formation of hydrogen bonds to the surface oxygen anions is expected. Due largely to the development of modern X-ray crystallographic and infrared/Raman spectroscopictechniques,the nature of hydrogen bonding in solids is now well understood. On the basis of data collected for hundreds of solid compounds, Novak5* found a remarkable correlation between the OH stretching frequency and the oxygen-to-oxygendistance; that is, all data lie on a universal curve.58 Three distinct regions were revealed based on the slope of this universal curve. Accordingly, the hydrogen bond can be classified into three categories: weak, strong, and intermediate. The hydrogen bond is said to be weak if the 0 - a 0 distance exceeds 2.7 A. Typical examples of this type are solid alcohols where the v(0H) frequency is higher than 3200 cm-1 with a bandwidth of typically 10 cm-l. The hydrogen bond is termed to be strong if the 0-0 distance falls between 2.60 and 2.50 A. Typical examples of this type are acid salts, such as NaH(CH3C00)2,which exhibit extremely low v(0H) frequencies, somewhere between 2700 and 700 cm-l, and broad v(0H) bands which can be as wide as 1000 Between the above two extremes, there are some compounds having hydrogen bonds of intermediate strength. In the present study, assuming that the hydroxyl oxygen of the adsorbate is located at 1.5-1.7 A above a surface cation site, the distance between the hydroxyl oxygen and the surface lattice oxygen falls into the categories for the strong and intermediate hydrogen bonds. Although we are not aware of any theoretical studies having been undertaken for the adsorption of methanol on NiO( loo), calculationsusing an extended Hiickel tight-binding approach showed that the total energy of ammonia adsorbed on NiO( 100) reaches a minimum when the nitrogen-nickel distance is around 1.6 A.59 The absence of the v(0H) loss feature in the present study may well be explained by forming a strong hydrogen bond between methanol and NiO(100). More interestingly, the hydroxyl proton in such an adsorption geometry likely fluctuates or “whizzes”around the four adjacent surface oxygen anions, acting as a mobile proton, as indicated by arrows in Figure 11. This is a consequence of the symmetry of the surface geometry of NiO(100): the four nearest anion neighbors are in equivalent positions and thus have an equal probability to form a hydrogen bond with the hydroxyl proton. From the point of view of the surface potential energy, fluctuation of the proton will occur if the rotational energy barrier along the azimuthal direction is relatively low. In the preceding section,it has been concluded that the methanol and ethanol absorb associatively on NiO( 100). This contrasts with methanol adsorbed on MgO surfaces where dissociative adsorption leading to a stable methoxy species occurs as low as 90 K.3’938 This difference in reactivity between NiO( 100) and MgO(100) surfaces can be readily explained in terms of their surface basicity. Magnesium oxide is a typical basic oxide whose surface basicity is significantlystronger than that of nickel oxide. In fact, methanol dissociates on most powdered oxide catalysts under catalytic reaction conditions. The initial step for these reactions all involves OH bond cleavage to form methoxy. It is presumed that there is a short-lived transition state involving a proton transfer subsequent to the formation of a hydrogen bond between the hydroxyl hydrogen and the lattice oxygen. This transition state then leads to the dissociated state, forming a conjugated base anion adsorbed on a cation site and a dissociated hydrogen adsorbed on an adjacent anion site. In the present study, this transition state is uniquely demonstrated due to the relatively weak surface basicity of nickel oxide. This observation significantlyenhances our understanding regarding the nature of the precursor to the dissociated state in an acid/base reaction.
Wu et al. V. Conclusions Interactions of alcohols with well-defined NiO( 100) films prepared on Mo( 100) under UHV conditions have been studied using combined HREELS/TPD. The results are summarized as follows: 1. Alcohols adsorb associatively on NiO( 100) surfaces and undergo reversible adsorption upon annealing under UHV conditions. Both methanol and ethanol are bonded to the cation sites of NiO(100) via their oxygen atom with the methyl and ethyl group, respectively, directed away from the surface. 2. At low coverages, a repulsive interaction between adsorbed alcohols is observed, which results in a decrease in the apparent activation energy of desorption. Molecularly adsorbed alcohols are found to bond stronger to NiO(100) than to most metal surfaces. The change of relative mode intensities in HREELS spectra with respect to the methanol coverage, below monolayer coverages, correlates excellently with two desorption states revealed in TPD studies, indicating reorientation of some of the adsorbed methanol at high coverages. In the case of ethanol adsorbed on NiO( loo), however, only a single desorption state above 200 K is evident, consistent with the HREELS results which show a constant mode intensity pattern with respect to the ethanol coverage. 3. The absence of the v(0H) loss feature in the region of OH vibrations below monolayer coverages suggests the formation of strong hydrogen bonds between the hydroxyl hydrogen and the lattice oxygen. This feature, together with consideration of the surface symmetry of Ni0(100), leads to the conclusion that the hydroxyl proton fluctuates or “whizzes” around the neighboring anion sites, acting as a mobile proton. This observation greatly enhances our understanding of the nature of the precursor to the dissociated state in an acid/base reaction. Acknowledgment. We acknowledge with pleasure the support of this work by the Department of Energy, Officeof Basic Energy Sciences, Division of Chemical Science, and the Gas Research Institute. References and Notes (1) Waugh, K. C. Cafal. Today 1992,15, 51. (2) Klier, K.Adv. Caral. 1982, 31, 243. (3) Saussey, J.; Lavalley, J. C.; Rais, T.; Chakor, A. A.; Hindermann, J. P.;Kiennemann, A. J . Mol. Cafal. 1984, 26, 159. (4) Bowker, M.; Waugh, K. C.; Hadden, R. A.; Hyland, J. N. K.; Houghton, J . Caral. 1988, 109, 263. ( 5 ) Edwards, J. F.; Schrader, G. L. J. Catal. 1985, 94, 175. (6) Edwards, J. F.; Schrader, G. L. J . Phys. Chem. 1984,88, 5620. (7) Visser, L. G.; Matulewicz, E. R. A.; Hart, J.; Mol, J. C. J . Phys. Chem. 1983,87, 1470. (8) Idriss, H.; Barteau, M. A. J . Phys. Chem. 1992, 96, 3382. (9) Ishikawa, T.; Kondo, S.;Sakaiya, H. J . Chem. SOC.,Faraday Trans. I 1988,84, 1941. (10) Chung, J. S.; Bennett, C. U. J. Caral. 1985, 92, 173. (11) Au, C. T.; Hirsch, W.; Hirschwald, W. Surf.Sci. 1989, 221, 113. (12) Edwards, J. F.; Schrader, G. L. J . Phys. Chem. 1985,89, 782. (13) Mi1lar.G. J.;Rochester,C.H.; Waugh, K.C.J. Chem.Soc.,Faraday Trans. 1992,88, 1033. (14) Kapel, R. U.;Greenler, R. G. J . Chem. Phys. 1968, 49, 1638. (15) Kondo, J.; Sakata, Y.; Maruya, K.;Tamaru, K.; Onishi, T. Appl. Surf. Sci. 1981, 28, 415. (16) Wadayama, T.; Saito, T.; Suetaka, W. Appl. Surf.Sci. 1984, 20, 199. (17) Chung, J. S.; Miranda, R.; Bennett, C. U. J . Chem. Soc., Faraday Trans. 1 1985,81, 19. (18) Li, C.; Domen, K.;Maruya, K.-I.; Onishi, T. J . Catal. 1990, 125, 445. (19) Hussein,G.A. M.;Sheppard,N.;Zaki, M. I.; Fahim, R. B.J. Chem. SOC.,Faraday Trans. 1991,87, 2655. (20) Yamashita, K.;Naito, S.;Tamaru, K.J. Caral. 1985, 94, 353. (21) Malinksi, R.; Gallus, 0. J. Reacr. Kiner. Cafal.Leu. 1979,12,447. (22) Morrow, B. A. J. Chem. SOC.,Faraday Trans. I 1914, 70, 1527. (23) Roberts, D.L.; Griffin, G. L. J . Cafal. 1985, 95, 617. (24) Riva, A.; Busca, G.; Vaccari, A.; Trifiro, F.; Mintchev, L. J. Chem. Soc., Faraday Trans. 1 1988, 84, 1423. (25) Bowker, M.; Madix, R. J. Surf. Sci. 1980, 95, 190. (26) Baudais, F. L.; Borschke, A. J.; Fedyk, J. D.; Dignam, M. J. Surf. Sci. 1980, 100, 210.
Interactions of Alcohols with a NiO( 100) Surface (27) Goodman, D. W.; Yates, Jr., J. T.; Madey, T. E. Surf. Sci. 1980,93, L135. (28) Christmann, K.; Demuth, J. E. J. Chem. Phys. 1982, 76, 6308. (29) Sexton, B. A. Surf. Sci. 1979, 88, 299. (30) Parmeter, J. E.; Jiang, X.; Goodman, D. W. Surf. Sci. 1990,240,85. (31) Hrbek, J.; dePaola, R. A,; Hoffmann, F. M. J . Chem. Phys. 1984, 81, 2818. (32) Demuth, J. E.; Ibach, H. Chem. Phys. Lett. 1979,60, 395. (33) Bhattacharya, A. K.; Chesters, M. A.; Pemble, M. E.; Sheppard, N. Surf. Sci. 1988, 206, L845. (34) Sexton, B. A. Surf.Sci. 1981, 102, 271. (35) Cox, P. A.; Falvell, W. R.; Williams, A. A.; Egdell, R. G. Surf. Sei. 1985,152/153, 784. (36) Petrie, W. T.; Vohs, J. M. Surf. Sci. 1991, 245, 315. (37) Wu, M.-C.; Estrada, C. A.; Goodman, D. W. Phys. Rev. Lett. 1991, 67, 2910. J.S.;Goodman,D. W.J.Phys. (38) Wu,M.-C.;Estrada,C.A.;Comeille, Chem. 1992,96, 3892. (39) Wu, M.-C.; Goodman. D. W. Catal. Lett. 1992, 15, 1. (40) Furstenau, R. P.; McDougall, G.; Langell, M. A. Surf. Sci. 1985, 150, 55. (41) Wu, M.-C.; Truong, C. M.; Goodman, D. W. Manuscript in preparation. (42) Press, W. H.; Flannery, B. P.; Teukolsky, S. A.; Vetterling, W. T. Numerical Recipes in Pascal: The Art of Scienti’c Computing, Cambridge University Press: New York, 1989; p 459. (43) Stenhagen, F.; Abrahamsson, S.;McLafferty, F. W. Registry ofMass Spectral Data; John Wiley & Sons: New York, 1974; Vol. 1.
The Journal of Physical Chemistry, Vol. 97, No. 37, 1993 9433 (44) Desorption of metal multilayers on top of a diimilar metal, for example, exhibits typically a common “leading edge” on the low-temperature side. (45) Ibach, H.; Mills, D. L. In Electron Energy Loss Spectroscopy and Surface Vibrutionr; Academic: New York, 1982; p 349. (46) Pacchioni, G.; Cogliandro, G.; Bagus, P. S. Surf. Sci. 1991,255,344. (47) Pacchioni, G.; Bagus, P. S. Surf.Sci., in press. (48) Truong, C. M.; Wu, M.-C.; Goodman, D.W . J. Chem. Phys. 1992, 97, 9447. (49) Henrich, V. E. Rep. Prog. Phys. 1985, 48, 1481. (50) Wu, M.-C.; Truong, C. M.; Goodman, D. W. J. Phys. Chem. 1993, 97, 4182. (51) Truong, C. M.; Wu, M.-C.; Goodman, D. W. J. Am. Chem. Soc. 1993, 115, 3647. ( 5 2 ) Rubloff, G. W.; Demuth, J. E. J. Vac. Sci. Technol. 1977,14,419. (53) Behm, R. J.; Christmann, K.; Ertl, G.; van Hove, M. A. J. Chem. Phys. 1980, 73, 2984. (54) LBth, H.; Rubloff, G. W.; Grobman, W. D. Surf. Sci. 1977,65325. (55) de Jone, A. M.; Niemantsverdriet, J. M. Surf. Sci. 1990,233,355. (56) Rogers, Jr., J. W.; Hance, R. L.; White, J. M. Surf. Sci. 1981, 100, 388. (57) Weast, R. C., Astle, M. J., Beyer, W. H., Eds. CRC Hand6ook of Chemistry andPhysics, 67th ed.; Chemical Rubber: Boca Raton, FL, 1986; p E-59. (58) Novak, A. Structure and Bonding, Springer-Verlag: Berlin, 1973; Vol. 18, p 177. (59) Cain, S. R.; Matienzo, L. J.; Emmi, F. J. Phys. Chem. 1990, 94, 4985.
