Rotational Isomerization of Esters on Ni(111) - ACS Publications

formate chemisorbed on Ni(lll) are found to be consistent with surface ... interaction between the opposed O—R2 and C=0 dipoles.10 A ... SCHEME 1: (...
1 downloads 0 Views 1MB Size
17906

J. Phys. Chem. 1995,99, 17906-17916

Rotational Isomerization of Esters on Ni(ll1) E. Zahidi, M. Castonguay, and P. H. McBreen* Dtpartement de Chimie, Uniuersitt Laual, Qutbec, Canada G I K 7P4 Received: August 16, 1995@

Reflectance absorbance infrared (RAIRS) and thermal desorption (TPD) data for methyl formate and ethyl formate chemisorbed on Ni( 111) are found to be consistent with surface coverage dependent rotational isomerization. The most stable gas phase rotameric forms of these molecules are the (Z) or cis forms. However, the reflectance infrared data provide direct evidence for isomerization to the (E) or anti forms as their surface coverages are increased. This rotational isomerization, around the C-0 bond, is observed to occur at coverages less than half that required to saturate the chemisorbed layers. In combination with infrared integrated intensity data, the thermal desorption results indicate that steric repulsion is responsible for the observed rotational isomerization. It is argued that two factors contribute to a reduction of the rotational barrier. First, steric repulsion is reduced continuously along the (Z) to (E) rotational coordinate. Second, image dipole stabilization increases continuously along the (Z) to (E) rotational coordinate. A number of interesting parallels exist between the chemisorption properties of these three ester molecules on Ni( 111) and the properties of Lewis acid-carbonyl molecule adducts. These similarities will be discussed briefly with reference to use of Lewis acids to achieve stereoselective control in organic synthesis, and with reference to the issue of enantioselective heterogeneous catalysis.

Introduction The orientation of adsorbed species with respect to the surface plane is a subject of central interest in surface chemistry and physics. For example, the issue of whether molecules such as CO or N2 are perpendicular or parallel to a metal surface has obvious implications for their dissociation probability. A complicating factor is that the molecular orientation of adsorbed molecules may depend on the surface coverage of the species. A case in point is CO on Ni(l10).3 A variety of techniques including photoelectron d i f f r a ~ t i o n ,electron ~~ stimulated desorption ion angular distribution, and metastable quenching s p e c t r o ~ c o p yhave ~ ~ shown that the CO bond axis is normal to the surface for coverages below 0.66 and tilted from the normal at higher coverages. The photoelectron diffraction data establish the tilt angle as 17", and the origin of the tilting is attributed to intermolecular repulsive interactions within the adsorbed layer. Surface coverage dependent molecular reorientation has been observed for several more complex molecule^.^-^ For example, angle-resolved photoemission data for high coverages of cyanogen on Ni( 110) are consistent with a tilting of the molecule away from the surface plane. Rosch et aL4 used a molecular modeling approach, based on the Lennard-Jones equation, to evaluate the latter observation in terms of the packing energy per adsorbed cyanogen molecule. They found that for a densely packed layer of cyanogen, a tilt angle of 25" is required to achieve chemisorption bond energy compensation of repulsive lateral interactions. Given these variations in adsorption geometry, one might expect the interadsorbate reaction chemistry of such species to display a related dependence on surface coverage. A nice example of such an effect is given in a recent study by Xi and Bent.7 These authors used surface vibrational spectroscopy to show that iodobenzene lies flat on Cu( 111) at low coverages and is tilted away from the surface plane at high coverages. They then correlated the emergence of a lowtemperature pathway for biphenyl formation with the presence of the tilted species. @Abstract published in Advance ACS Absrmcts, November 15, 1995.

0022-365419512099-17906$09.00/0

The examples given above mostly deal with the reorientation of molecular axes with respect to a reference plane. To the best of our knowledge, there exists no direct spectroscopic information on surface coverage dependent rotational isomerization in adsorbed monolayers. It should be noted, however, that Hoffmann et aL4 have reported detailed RAIRS results which show that ethyl iodide adsorbs on Pt( 111) with the C-C bond parallel to the surface at low coverages and in a configuration with the C-C bond perpendicular to the surface at high surface coverages. This paper deals with RAIRS and TPD data related to rotational isomerization of ethyl formate and methyl formate on Ni( 111). As discussed below, rotational isomerization within adsorbed layers of ester molecules is relevant to the issue of enantioselective surface chemistry.* Different rotameric forms of carboxylic acid esters are illustrated in Scheme 1. The stable gas phase conformational forms of molecules such as methyl formate, ethyl formate, and methyl acetate are the (Z) or cis form^.^-'^ The greater stability of the (Z) rotameric forms is due, mainly, to the electrostatic interaction between the opposed 0-Rz and C=O dipoles.'O A list of calculated and experimentally determined equilibria parametersI2-l5 for (Z) to (E) interconversion is given in Table 1. The large energy difference between the (E) and (Z) rotamers of methyl acetate, as compared to that for methyl formate, is due to the steric repulsion between the two methyl groups in the (E) rotamer.I0 Ethyl formate may be found in two distinct (Z) conformations.I6 These are the (Z)-trans and (Z)-gauche conformations illustrated in Scheme 4. The values listed in Table 1 show that significant modifications occur on going from the gas phase to a highly polar solution phase and that complexation to a Lewis acid introduces further changes. For example, the energy difference between (Z)- and (@-methyl acetate is reduced from its gas phase value of 8.5 kcallmol by a factor of 2 in a highly polar solventi2and by a factor of 7 on complexation to a tungsten derived Lewis acid.'j The dipole moments of the (E) conformers are more than double those of the (Z) forms due to the parallel orientation of the 0-R2 and C=O bond dipoles.I2 Hence, the effect of a polar solvent 0 1995 American Chemical Society

Rotational Isomerization of Esters on Ni( 111)

J. Phys. Chem., Vol. 99, No. 51, 1995 17907

SCHEME 1: (Z) and (E) Rotameric Forms of Carboxylic Acid Esters

only one direction of the moving mirror, with the optical resolution set at 4 cm-I. The mirror speed was 6.33 cm/s for an acquisition time of 2.56 min.

9

Results

II

/I

Ethyl formate

H

C2H5

(E)

(Z)

Methyl acetate

CH3

CH~

reflects the induced dipole stabilization of the (E) form in solvents possessing a high dielectric constant. For example, NMR measurement^'^ show that the free energy difference between the (Z) and (E) conformers of rert-butyl formate varies linearly with the Onsager parameter, (E - 1 ) / ( 2 ~ l), where E is the dielectric constant. The role of complexation to Lewis acids in modifying the free energy difference is more complex and is less well understood. A detailed analysis of the origin of the barrier for rotation around the CO single bond in simple carboxylic acid esters is given by Wiberg and Wong.'la The significant rotational barriers are usually attributed to the loss of resonance stabilization which occurs on rotation of the 0-R2 group out of the RlCOO plane.'Ib An overall view of the adsorption geometry and the surface decomposition chemistry of the three ester molecules on Ni(111) is given elsewhere.18a The chemisorption of all three molecules occurs through the interaction of a carbonyl lone pair with the nickel surface. Application of the surface selection rule shows the molecular plane of the adsorbed molecules to be roughly perpendicular to the surface. A striking aspect of the chemisorption induced shifts in the infrared absorption spectra of these molecules is that they closely match those reported for ester-Lewis acid adduct^.^^,'^ This issue will be discussed below with reference to heterogeneously catalyzed enantioselective chemistry. A brief report on the data for ethyl formate is given elsewhere.20

+

Experimental Section The experiments were performed at a base pressure of 2 x 1O-Io Torr using the upper level of a UHV-high pressure reaction cell system.'8a The Ni( 111) sample, purchased from Monocrystals Inc., was cleaned by repeated cycles of sputtering, oxygen treatment, and annealing to 900 K. Surface cleanliness was verified using retarding field Auger, CO desorption, and RAIRS measurements on adsorbed CO. The adsorbate liquids were purified by repeated freeze-thaw cycles. The infrared spectra were collected using a Mattson Galaxy 4020 Fl'IR spectrometer with a 1 mm2active area narrow band MCT detector. The beam enters and exits the chamber through a set of differentially pumped Viton O-ring sealed NaCl windows. The MCT detector sets the experimental lowfrequency cutoff at around 750 cm-'. However, the signal to noise ratio below 920 cm-' was not high enough to extract information. In addition, the signal in the CH stretching region was, in most cases, not good enough to unambiguously identify bands in the submonolayer region. This problem was, however, intrinsic to the three molecules under study, because good quality data in the v(CH) region was readily obtained for methoxy and ethoxy in the related thermal decomposition studies. As a result, we concentrate on the absorption bands observed between 2100 and 920 cm-I. A 42 kHz low-pass filter is used for removing the higher frequency signal in the doublesided interferograms to avoid folding-in of noise from outside the spectral region. The reported spectra are expressed in absorbance units and represent the ratio of 1000 sample scans to 1000 scans of the clean surface. Spectra were recorded in

First, we will give a general summary of the results and this will be followed by a detailed presentation of the data. The TPD data for methyl formate, ethyl formate, and methyl acetate from Ni( 111) display a strong coverage dependence. Three distinct coverage regions may be discerned. Complete decomposition takes place at low coverages. Multipeak molecular desorption spectra are observed at intermediate coverages, and desorption from condensed multilayers takes place for high exposures. The coverage dependence of the RAIRS spectra is even more complex. The methyl formate spectra display changes in line widths as well as the appearance of a new carbonyl band at coverages well below those required to saturate the chemisorbed layer. The spectra for ethyl formate show changes in line widths, the disappearance of some bands, and the emergence of new bands as the chemisorption layer coverage is increased. The RAIRS spectra for chemisorbed methyl acetate, on the other hand, are qualitatively invariant as a function of coverage. Infrared spectra characteristic of the condensed ester molecules are observed for high exposures. Integrated infrared absorbance data for each of the three molecules display a similar surface coverage dependence. The latter dependence takes the form of a linear increase for low exposures, followed by an intermediate region where the absorbance levels off, and finally a third region where a further increase occurs. Overall, there is a strong correlation between the coverage dependent TPD and RAIRS data. Coverage-dependent TPD data for the molecular desorption of methyl formate, ethyl formate, and methyl acetate from Ni(111) are presented in Figures 1-3, respectively. Three different coverage regimes may be distinguished on the basis of these TPD spectra. Each of the three esters undergoes thermal decomposition in the low coverage regime. Thus, there is no molecular desorption of methyl formate and ethyl formate, at exposures below 2 langmuirs. Similarly, molecular desorption of methyl acetate only takes place at exposures above 0.6 langmuir. The thermal decomposition processes for each of the three molecules were described in a previous paper,lsa and a summary is given in Scheme 2. Multilayer or condensed layer desorption is observed for exposures of -7 langmuirs for methyl formate and -4-5 langmuirs for methyl acetate and ethyl formate. These features are observed as sharp peaks at 120, 136, and 128 K, respectively. Finally, there is an intermediate region, between the surface decomposition and the multilayer desorption regimes, for which distinct molecular desorption peaks are observed. These multipeak features correspond to desorption from the chemisorption layer. The methyl formate data display peaks at 160 and 208 K for an exposure of 3.0 langmuirs. On increasing the coverage, the low-temperature peak becomes much more intense than the one at 208 K and shifts to 142 K. In the case of ethyl formate, molecular desorption peaks appear at 170 and 224 K, respectively, for exposures of 3 langmuirs. As in the case of methyl formate, the low-temperature ethyl formate peak increases in intensity and shifts by -20 K to lower temperatures as the exposure is increased. For methyl acetate, a single molecular desorption peak at 207 K is observed for low exposures. On increasing the exposure, the molecular desorption peak shifts to the 165 K position shown in Figure 3. As described elsewhere,18aand as illustrated in Scheme 2, the low-coverage decomposition of the three ester molecules

Zahidi et al.

17908 J. Phys. Chem., Vol. 99, No. 51, 1995

TABLE 1: Approximate Relative Stabilities (AE,kcaVmo1) of (E) and (Z) Rotamers of Methyl Formate, Ethyl Formate, and Methyl Acetate. Approximate Rotational Barriers (Erot)for (Z) to (E) Interconversion in gas phase observed calculated comuound methyl formate ethyl formate methyl acetate

AE 4.7"

E,,, 10-15"

8.5"

AE 5.2b

13Sb

8.5b

13.5b

" Reference 13, AH". Reference 12a, A G m

K.

AE 2.1' 1.7'

E,,,

9.9d lod

observed

AE 1.7b

12.4b

5.2b

13.3b

E,,,

J \

2.4L (x200)

9 1.2L 200

Em

calculated

AE

Em

1.3'

13.5'

AE

E,,,

Reference 14, AGlw K. Reference 14, A G m K. e Reference 15, A G m K.

2.7L

65

complexed to Lewis acid

in a polar solvent observed calculated

A

2.5L

I

0.6L

4

300

200

00

Temperature (K)

300

4

Temperature (K)

Figure 1. Thermal desorption data (mle = 31) for methyl formate on Ni( 11 1) as a function of exposure. The heating rate was 0.5 IUS.

Figure 3. Thermal desorption data (m/e = 43) for methyl acetate on Ni( 111) as a function of exposure. The heating rate was 0.5 Ws. SCHEME 2: Thermal Decomposition Pathways for Methyl Formate, Ethyl Formate, and Methyl Acetate on Ni(ll1)"

k

gOOCH:

H COOCHi

Y

eo

CPjCOHC03C'~

u

CC

co!ll

AAJn SK AWL ZK h (1)

,7,,7,7,,

5L

c

-s

Y

4L

0 E

01 .VI

%

3.5L

c.

Ee

C'IICWCHj

;,;

&7,,,m

4-

K

M A r,2YIY A 7 A

COIY

Iz( fllbl

'+ CHjCOOCHiy

VI fn

s

"The data supporting these pathways is given in a separate publication. I R a

2.5L

2L 100

200

Temperature

300

(K)

Figure 2. Thermal desorption data (mle = 31) for ethyl formate Ni(l11) as a function of exposure. The heating rate was 0.5 Ws.

I

results in co,d,, Hadr,and surface methoxy or ethoxy species. The major desorption products of the decomposition processes are CO and Hz; however, methane and acetaldehyde desorption are also observed for ethyl formate. The latter two desorption products result from the decomposition of the ethoxy intermediate. On the basis of the work of Gates et a1.,'8bethoxy decomposition also leads to the deposition of a small amount of carbon

on the surface. There is reasonably good evidence for the formation of a stable surface acetyl species, in addition to methoxy species, in the decomposition of methyl acetate on Ni(1 l l).lSa The CO desorption peak areas, displayed in panel a of Figure 4, may be used to monitor the thermal decomposition of the adsorbed esters. The CO desorption peak area plots show that decomposition is completely saturated prior to the completion of the chemisorbed layers. The surface decomposition of methyl formate, ethyl formate, and methyl acetate saturate at exposures of 3, 3, and 1.2 langmuirs, respectively. Ethyl formate produces less CO than methyl formate due to the desorption of acetaldehyde. Panel b of Figure 4 shows the molecular desorption peak areas are linear as a function of exposure. In combination, the results shown in both panels a and b of Figure 4 indicate that the surface coverage of the esters

J. Phys. Chem., Vol. 99, No. 51, 1995 17909

Rotational Isomerization of Esters on Ni( 111)

TABLE 2: Vibrational Data (cm-I) for Esters on Ni(ll1) Methyl Formate: HCOOCH3

e-o

e-0.5

1659 1458 1262 4

0

12

8

4

0

Expe6Un (L)

8

12

Exposum (L)

Figure 4. (a) The integrated CO desorption signal as a function of coverage for methyl formate, ethyl formate, and methyl acetate on Ni(1 l l). (b) Molecular desorption peak areas plotted as a function of exposure.

1710 1690 1655 1460 1442 1266 1254 1182

Lewis multimode approx acid layer (sym species) descripn adduct

rotamers

Z-

E-

cis

trans

1730

v4a’)

v(C=O)

1653 1618

1745 1777

1220

vs(a’) ~.5(a’) vg(a’)

das(CH3) 1464 &(CH3) 1437 v(C-0) 1348 1337

1459 1483 1433 1464 1205 1239

1162

vg(a’)

1161 1099

Ethyl Formate: HCOOC2H5

0

- e0

Lewis rotamers multimode approx ZZacid 0.5 layer (sym species) descripn adduct trans gauche

1700 1655 1444 1394

1732

vs(a’)

v(C-0)

1635 1732

1732

1465

v20(a”)

1215

&s(CH3) &(CHd CHZtwist v(C-0)

1443 1373 1300 1260 1187

1443 1373 1300 1187

1OL

1267 1255 1242

v9(a’) vZl(a’’) vlda’)

7L

1013

1152 1007

1152 1007

1650 1444 1304 1260 1242

1162 Methyl Acetate: CH3COOCH3

I-”-I I“-I 1800

1700

1600 15001400

A

O.0L

e

1-01 .5LI 1300

1200

1100

Frequency (cm-’ )

Figure 5. RAIRS spectra for methyl formate (HCOOCH3) on Ni( 11 1) as a function of exposure at 86 K. The left-hand box highlights data for the v(C=O) stretching region of the spectra. Data for the d(CH3) region are displayed in panel b. Data for v(C-0) stretching and CH3 rocking regions are shown on the right-hand side (c).

BOO

1700

1000

rioo

Frequency (em-’

1200

)

Figure 6. RAIRS spectra for ethyl formate (HCOOCZH~) on Ni( 11 1) as a function of exposure at 86 K. The characteristic v(C=O) and v(C-0) regions are shown in panels a and c, respectively. The 6(CH3) region is displayed in panel b.

is approximately linear in exposure. This linear dependence is important when considering the coverage-dependent IR absorbance data presented in Figures 9- 11. Representative coverage-dependent RAIRS data for HCOOCH3 and HCOOC2H5 are shown in Figures 5 and 6, respectively. In order to facilitate the description of the coverage-dependent RAIRS data, a complete listing of the observed bands as well as their gas phase assignments is given in Table 2. Several papers in the literature deal with the analysis of the vibrational

- 0 8

1666 1456 1364 1302 1041

multimode 0.5 layer (sym species)

1749 1442 1423 1375 1050

v5(a’) vg(a‘) or v7(a’) vg(a’) vlo(a’) vjZ(a’)

approx descripn v(C-0) 6as(OCH3)or 6,(CCH3) G,(CCH3) v(C-0)

Lewis rotamers acid ZEadduct cis trans

1619 1761 1774

1360 1246 1255 1057 1116

spectra of methyl formate?’ ethyl formate,22 and methyl These studies show that the CH3 rocking, the C-0 stretching, the 0-R2 stretching and the OCO deformation modes are strongly coupled. Figures 5 and 6 concentrate on the characteristic regions of the IR spectra of both molecules. The carbonyl stretching region, a, the CH3 deformation region, b, and the C-0 stretch and CH3 rock regions, c, are shown for HCOOCH3. The Y E mode of methyl formate and the Y I I mode of ethyl formate may alternatively be described as the asymmetric COC stretching The low-coverage spectra of methyl formate (Figure 5) display a single peak in the carbonyl stretching region. As may be seen from Table 2, this absorption band, located at 1659 cm-I, is red-shifted by 71 cm-I with respect to the carbonyl frequency for the multilayer. In contrast, the Y(C-0) mode, located at 1262 cm-I, is blue-shifted by 42 cm-’ with respect to the multilayer value. Such shifts are consistent with what is known about the IR and Raman spectra of ester-Lewis acid adducts,I9 as well as the vibrational spectra of methyl formate on C ~ ( l l 0 and ) ~ ~on copper catalysts.24 Some literature data for ester-Lewis acid adducts is included in Table 2. Two new features appear in the carbonyl stretching region at 1710 and 1690 cm-l, respectively, as the exposure is increased to 1.2 langmuirs. These new features are accompanied by the emergence of two other bands at 1442 and 1182 cm-I. The pair of new features in the carbonyl stretching region are shifted by 51 and 31 cm-’, respectively, to higher frequency with respect to the v4 mode in the low-coverage spectrum. The new feature at 1690 cm-’ shifts down to 1683 cm-’ as the exposure

Zahidi et al.

17910 J. Phys. Chem., Vol. 99, No. 51, 1995

1OL

L

7L 6L

5L

4L JL 2L

?L

0.OL

: rdoo 0.a O.ZL

2000

1800

1600

1400

1200

1000

Frequency ( c m - 1 )

Figure 7. RAIRS spectra for ethyl formate (HCOOC2H5) on N i ( l l 1 ) as a function of exposure at 86 K.

approaches 7 langmuirs. At this level of exposure three new features are visible in the spectrum, at 1729, 1220, and 1162 cm-I, respectively. The latter features may readily be attributed to the vj, vg, and vg modes of condensed layer HCOOCH3. With increasing exposure, the peaks characteristic of the condensed layer (multilayer) simply increase in intensity and undergo a slight upward shift in frequency. The important point to be taken from the data in Figure 5 is that additional features in the carbonyl stretching region appear at exposures well below those required to saturate the chemisorption layer. That is, a new chemisorption state is populated as the coverage is increased. Surface decomposition species can be ruled out, since the low-coverage spectra are clearly characteristic of molecularly adsorbed methyl formate. There is a possibility that the features at 1690 and 1710 cm-' are due to a Fermi resonance enhancement of some overtone or combination features.25 Such resonances occur, for example, between 2vl0,2vl6, and v4 in the spectrum of (Z)-DCOOCH32'b and in the spectra of several ester species.25 There are no Fermi resonance features in the v(C=O) region for free HCOOCH3, but this cannot be generalized to chemisorbed methyl formate. Nevertheless, the simultaneous emergence of features at 1182 and 1442 cm-l suggests that the shifted carbonyl bands should be attributed to a distinct adsorption state, or states, rather than to a Fermi resonance enhancement. Surface coverage dependent RAIRS data for ethyl formate on Ni( 111) are shown in Figures 6 and 7. The emergence of the Characteristic multilayer peaks at 1732, 1215, and 1162 cm-I provides a clear demarcation between the sub- and supramonolayer regimes. They show that multilayer formation begins at 4-5 langmuirs. As may be seen from Figures 6 and 7, the coverage dependence of the RAIRS spectra of chemisorbed ethyl formate displays several striking features. The changes which are easily visible in the spectra are as follows. A relatively intense feature at 1304 cm-I, present from the lowest exposures, disappears at 4 langmuirs. The width of the v5 band increases significantly at 2 langmuirs. An additional set of changes takes place at 2 langmuirs. These include the emergence of new features at 1394 and 1013 cm-I. The v(C-0) region of the spectrum displays a quite complex surface coverage dependence. A single peak, at 1242 cm-', is detected at the lowest surface coverages studied. A second component, at 1260 cm-I, becomes visible at an exposure of 0.4 langmuir. The relative intensity of the latter band increases with increasing coverage. A third component is visible as a shoulder at 1267 cm-' for exposures in the range 2-3 langmuirs. A single band of line width 25 cm-I, at 1255 cm-', is observed for an exposure of 4

1600

1400

1200

1000

Frequency (cm-')

Figure 8. RAIRS spectra for methyl acetate on Ni( 11 1) as a function of exposure at 86 K.

m

0 00 0

4

8

I2

Exposure (L)

Figure 9. Integrated absorbance plots for the carbonyl stretching vibration for methyl formate on Ni( 11 1). Data for the sum over all observed carbonyl peaks (W) and for the peak at 1659 cm-' are shown.

langmuirs. An additional, albeit not so evident, change requires comment. A band at 1444 cm-' is visible in the low-coverage spectra. The emergence of the band at 1394 cm-' for exposures > 2 langmuirs coincides with a decrease in the intensity of the 1444 cm-' band. Furthermore, the latter feature is hardly visible for exposures in the multilayer region. In contrast, the multilayer spectra display a new band at 1465 cm-l. The latter band is present as a shoulder at an exposure of 3 langmuirs. The coverage dependence of the band at 1444 cm-' parallels that of the band at 1304 cm-I in the sense that both peaks fade out as the chemisorption layer nears saturation. Similarly, the features at 1394 and 1465 cm-' grow in together. In summary, the data display a complex surface coverage dependence. Some changes occur in the spectra at exposures well below those required to produce features characteristic of the condensed layer, while others occur close to the completion of the chemisorbed monolayer. Data for methyl acetate are displayed in Figure 8. The spectra display no coverage-dependent behavior other than the appearance of features characteristic of the condensed layer at high exposures. Figure 9 displays integrated absorbance data for the carbonyl stretching vibration of adsorbed methyl formate. The square points were obtained by integrating over the entire carbonyl stretching region. Thus, the increase at high exposures ('5 langmuir) is associated with the growth of the multilayer band at 1729 cm-I (see Table 2 and Figure 5). The round point plot is for the intensity of the carbonyl peak at 1659 cm-I and

Rotational Isomerization of Esters on Ni( 111)

J. Phys. Chem., Vol. 99, No. 51, 1995 17911 I

0.4 I

-

0.3

3

d3

I

8

A

MUhylFome EthylFome MnhylAcnate

1 A

:

. I '

h

A

4

s 0 B

-

e

02-

MethylFome EthylFomc MnhylAcetatc

e

e

8 8

A

8A'

1.'

O"

0.0

t i 0

4

0

12

8

4

- 0

I2

8

Exposure (L)

Exposure (L)

Figure 12. Line widths of the lowest frequency carbonyl band for

Figure 10. Integrated absorbance for all C-0 stretching bands (originating from both the chemisorbed layers and the multilayers) observed for ethyl formate, methyl formate, and methyl acetate on Ni(1 l l ) as a function of exposure at 86 K.

chemisorbed methyl formate, ethyl formate, and methyl acetate as a function of exposure at 86 K.

SCHEME 3: Illustration of Possible Adsorbed (Z) and (E) States of Methyl Formate

0.3

(Z)

(E)

>

Y3C

H-C ?

3

L'

H-C

$0

..............................

0.2

v 0

Y

.e

1

P

2

4

6

8

8

Ethyl Formate Exposure (L)

9

8

'CI

'

0

B 0.1

Q

m

I

e

e

e

e

e

8

8 8 8

0.0 0

2

4

6

8

10

12

Exposure (L)

Figure 11. Integrated absorbance data as a function of exposure for ethyl formate on Ni(ll1) at 86 K. Plots for the chemisorption layer carbonyl signal and the total carbonyl (chemisorption and multilayer) signal are shown. The insert displays the C(1s) signal for ethyl formate on Ni( 111) as a function of exposure.

therefore arises from chemisorbed methyl formate. The latter peak is present from the lowest exposures. Both plots display an initial linear increase followed by a saturation, or plateau region, in the integrated absorbance. The difference between the integrated absorbances in the plateau regions of the two plots indicates that 30-40% of the carbonyl signal for chemisorbed methyl formate falls on the high-frequency side of the 1659 cm-' band. That is, the contribution of the bands at 1690 and 1710 cm-' is somewhat greater than that indicated by a simple visual inspection of Figure 5 . Integrated absorbance data for the three ester molecules is displayed in Figure 10. These data show that all three molecules display a similar three-region behavior. Data specific to ethyl formate on Ni( 111) is shown in Figure 11. The insert to Figure 11 contains an important piece of corollary XPS data. It shows the C(1s) peak area as a function of exposure for ethyl formate on Ni(ll1). The C(1s) peak area, and hence the coverage of

ethyl formate, is seen to be linear in exposure over the region required to satura5e the chemisorbed layer. (This XPS analysis of the chemisorbed layer ignores any possible modulation of the C(1s) signal due to forward scattering.) Thus, the leveling off of the integrated absorbance, in the exposure range 2-4 langmuirs, does not arise from a saturation of available adsorption sites. The fact that the coverage is linear in exposure means that a good estimate of the coverages may be obtained by normalizing each exposure to that required to saturate the chemisorbed ethyl formate layer. We assume that saturation of the chemisorptionlayer coincides with the threshold exposure for multilayer formation. A threshold exposure of 4-5 langmuirs may be determined from an inspection of either the TPD or the RAIRS data. The molecular desorption peak area data shown in panel b of Figure 4 strongly suggest that the coverage of methyl formate and methyl acetate is also linear in exposure. Data for the surface coverage dependence of the line widths of the carbonyl stretching vibrations of the chemisorbed ester molecules is shown in Figure 12. The low-coverage value of the fwhm is approximately 10 cm-' in each case. The line widths increase to 20-25 cm-' for exposures of 2 langmuirs. The values for methyl formate and methyl acetate remain constant above 2 langmuirs. However, the line width for ethyl formate increases to around 33 cm-' at 5 langmuirs. The linewidth data is presented for completeness and will not be discussed. Many different effects could be at the origin of the increase in line widths at higher coverage^.*^^*^

Discussion A. Evidence for Rotational Isomerization. The data will be interpreted in terms of coverage-dependent rotational isomerization. An illustration of different adsorbed conformations of methyl formate and ethyl formate is illustrated in Schemes 3 and 4, respectively. These schemes do not specify any particular adsorption site. The surface is treated as a featureless mirror

Zahidi et al.

17912 J. Phys. Chem., Vol. 99, No. 51, 1995

SCHEME 4: Illustration of Possible Adsorbed Conformations of Ethyl Formate CHI

0

CHI

-

CHI

CHI

0

CHj

-CHi

CHI

0

- CHl

plane. However, as outlined in a separate publication,Igathe RAIRS data provide unequivocal evidence for chemisorption via the interaction of a carbonyl lone pair with the surface. The RAIRS spectra also indicate that each of the three esters are oriented perpendicular to the surface, as illustrated in Schemes 3 and 4. The Ni-0-C bond angle is not known and it may vary slightly from one surface conformation to the next. RAIRS measurements are particularly appropriate for distinguishing between adsorbed rotameric form^.^^-^^.^^^.^^.^^ Rotational isomerization sometimes leads to characteristic shifts in IR spectra, and the RAIRS technique possesses the resolution required to detect such changes. Furthermore, rotational isomerization within a molecule adsorbed on a metal will in general result in surface selection rule related changes in the spectra. The intensity of certain bands will be modified as internal molecular bonds are rotated relative to the surface plane. This leads to a cos2 6 dependence of the infrared intensity on the angle which the transition dipole moment makes with the normal to the surface.29 In many cases, RAIRS possesses sufficient sensitivity to detect at least some of such orientation induced changes in band intensity. In the discussion of the RAIRS data, we rely heavily on published IR spectra for different conformers of methyl formate, methyl acetate, and ethyl formate. For example, Blom and GunthardI3 used IR spectroscopy to detect the (E) forms of methyl formate and methyl acetate isolated in argon matrices. The (E) conformers were produced by using a Knudsen cell as a thermal molecular beam source. Similarly, Muller et aL2' used photoisomerization (hv = 5.0 eV) to produce (E) methyl formate in an argon matrix for IR studies. Rushkin and Bauer21eused a transient, temperature drift,IR absorption technique to detect (@-methyl formate. Charles et aL2*assigned features in the IR spectrum of ethyl formate to (Z)-trans and (Z)-gauche rotamers. They found no evidence for ethyl formate in an (E) conformation. Lenaerts et aLzs performed an IR study of (Z) to (E) photorotamerization of matrix isolated methyl acetate. First, we will consider the results for methyl formate. The TPD and RAIRS spectra, given in Figures 1 and 5, respectively, show that the onset for multilayer formation occurs in the exposure range 5-7 langmuirs. The characteristic IR bands of the multilayer, at 1729, 1220, and 1162 cm-I, are clearly visible in the spectra recorded for exposures of 7 langmuirs. The multilayer desorption peak, at 120 K, is also clearly visible for exposures of 7 langmuirs. Thus, both the TPD and RAIRS data show that only chemisorbed methyl formate is present at 4 langmuirs and that the onset of multilayer formation occurs at

5-7 langmuirs. As detailed in the Results section, the important aspect of the RAIRS data is that a number of changes occur in the spectra well below the exposure required for the onset of multilayer formation. At low exposures, less than 1 langmuir, the chemisorbed layer is characterized by a v(C=O) band at 1659 cm-l. However, a new feature develops at 1690 cm-' for exposures 2 1.2 langmuirs. The 1690 cm-' band is shifted from the low coverage band by Av = 3 1 cm-I. By comparison with literature data, the latter shift is immediately suggestive of the presence of (@-methyl formate on the surface. IR studies of t r a ~ p e d l ~or. ~transient28 '~ (E)-methyl formate report blue shifts of 32-34 cm-I, with respect to the carbonyl stretching vibration of the (Z)-conformer. This agreement with the literature data then raises the possibility that (Z) to (E) isomerization of methyl formate occurs on Ni( 111). Rotational isomerization of adsorbed methyl formate should lead to additional changes in the RAIRS spectra. Both Blom and GunthardI3 and Muller et aL2Ibreport that the IR spectra of (E)-methyl formate display an additional characteristic feature at 1099 cm-I, which is about half as intense as the shifted carbonyl band. Muller et al. identify the 1099 cm-' band as arising from the highly mixed v g mode. Our spectra do not display a band in the neighborhood of 1099 cm-I. However, a feature at 1182 cm-' grows in simultaneously to the emergence of the shifted carbonyl band. On the basis of the reference data listed in Table 2 we tentatively assign the 1182 cm-' band to the v9 mode of adsorbed (E)-methyl formate. The justification for such an assignment is as follows. Note from Table 2 that the adsorption of all of the three ester molecules on Ni( 111) leads to a substantial blueshift in the v(C-0) vibration. In particular, the v(C-0) mode for chemisorbed methyl formate is shifted by 42 cm-I to higher wavenumbers with respect to that for the multilayer. One would also expect the v g mode of chemisorbed (@-methyl formate to undergo a blue shift since it is principally a v(C-0) vibration.*lb Hence, an assignment of the feature at 1182 cm-] to the v9 mode is plausible, although it would involve a shift of up to 83 cm-l. Rotational isomerization obviously involves a reorientation of some molecular bonds with respect to the plane of the surface. Thus, the so-called surface selection rule may be invoked to account for the emergence of a new feature in the d(CH3) region at 1442 cm-I, as the chemisorbed layer coverage is increased. The 1458 cm-' band present at low coverage is assigned to an in plane asymmetric CH3 deformation mode. The new feature at 1442 cm-' could possibly be assigned to the symmetric CH3 deformation mode, since it would be RAIRS active in the (E)conformation due to the fact that the 0-CH3 bond axis is roughly perpendicular to the surface. However, we do not wish to place too much emphasis this assignment of the new band at 1442 cm-I, since the symmetric CH3 deformation mode appears in the IR spectrum of argon isolated (@-methyl formate at 1467 cm-I. In summary, there are a number of indications that adsorptioninduced rotational isomerization of methyl formate occurs as the surface coverage is increased. The principal evidence is, however, the 31 cm-' blue shift in the v(C=O) band. This evidence, in itself, is not sufficient to conclude that (Z) to (E) isomerization takes place. Rather, the data for methyl formate must be considered in parallel with those for ethyl formate. As outlined below, the RAIRS spectra for ethyl formate on Ni(111) display more unambiguous evidence for (Z) to (E) isomerization within the chemisorbed layer. The coverage dependence of the ethyl formate RAIRS spectrum differs markedly from the case of methyl formate in that there appears to be a collective transformation of the entire

Rotational Isomerization of Esters on Ni( 111)

J. Phys. Chem., Vol. 99, No. 51, 1995 17913

TABLE 3: Summary of Observed Changes in the RAIRS Spectrum of Chemisorbed Ethyl Formate as a Function of Coverage 8-0

8-1

1650 (fwhm = 11 cm-') 1443

1650 (fwhm = 30 cm-I)

1304 1260 weak 1242 intense (fwhm = 11 cm-I)

mode assignment

RAIRS activity consistent with

v(C-0)

all

1394

das(CH3),a" ddCH3)

1255 intense (fwhm = 25 cm-I)

CHI twist v(C0C) or (C-0,) st

(Z)-gauche (E)-trans, (2)-trans (Z)-gauche all

,1013

v(C-C)

(E)-trans, (Z)-trans

adsorption layer from one state to another. A summary of the coverage dependent variations for ethyl formate is given in Table 3. These coverage-dependent variations will be discussed in terms of the relative populations of the three rotameric forms illustrated in Scheme 4. The rotational barrier separating (2)and (@-ethyl formate in a polar solvent is approximately 10.0 kcdmol, and the (Z) form is more stable than the (E) form by approximately 1.7 kcaYm01.I~ The room temperature conformational composition of gas phase (a-ethyl formate is 41% trans and 59% gauche.I6 The C-C bond is directed 95" out of the HCOO plane in the gauche form, and the rotational barrier between the two (Z) forms is 1.1 kcal/mo1.l6 Although the gauche conformation is less stable than the trans conformation, it is doubly degenerate, and hence is the more dominant species in the vapor phase at room temperature. The ethyl formate RAIRS spectra shown in Figures 6 and 7 provide direct evidence for preferential adsorption into the (Z)gauche state at low surface coverages. The signature for the (Z)-gauche state is the relatively intense band at 1304 cm-I. The only band in the neighborhood of 1304 cm-I for the free molecule appears at 1300 cm-I . This band is a assigned to the CH;! twisting mode, 191.Since its symmetry species is a", it will be RAIRS inactive for the (Z)-trans and (E)-trans adsorbed states, both of which display Cs symmetry. The (Z)-gauche configuration,on the other hand, does not possess a C, molecular plane and the CH2 twisting mode becomes allowed. A further indication of the presence of a low-coverage (Z)-gauche state is the band at 1444 cm-'. The latter frequency is identical to that observed for the out-of-plane CH3 deformation mode of free ethyl formate. We are reasonably certain of this assignment, since no other band falls within 23 cm-I of the Y ~ mode O of vapour phase ethyl formate. The symmetry species of the ~ 2 0 mode is also art for (@-trans- and (z)-trans-ethyl formate; hence it would only be active for the (Z)-gauche state. Both the feature at 1304 cm-I and the feature at 1444 cm-I disappear as full monolayer coverage is obtained at 4-5 langmuirs exposure. This indicates that the (Z)-gauche state is depopulated as the surface coverage is increased. The two new features which appear at 1394 and 1007 cm-' for exposures greater than 2 langmuirs are then characteristic of a new surface conformation. By comparison to literature data,22 these two bands may be assigned to the symmetric CH3 deformation mode and the C-C stretching mode, respectively. An application of the surface selection rule shows that these two vibrations would be RAIRS inactive for the (Z)-gauche surface conformation. However, they would be RAIRS active for both the (Z)-trans and the (E)trans conformations. In summary, the coverage-dependent changes in the RAIRS spectra are consistent with rotational isomerization out of the (Z)-gauche conformation as the adsorbate layer packing density is increased. We have concluded that the RAIRS data show that the (Z)-

gauche conformation is the dominant form at low surface coverages. This is a reflection of the fact that the (Z)-gauche state maximizes the nonbonding interaction of the methyl group with the surface, since it places the methyl group closer to the surface. On the basis of the work of Madey and Yates3I and Sexton and Hughes32 on adsorption energies for alkanes, one can reasonably attribute a value of 2.0 kcal/mol to the van der Waals interaction of a CH3 group with a metal surface. Recall that the gas phase rotational barrier between the (Z)-trans and (Z)-gauche forms is 1.1 kcal/mol.I6 Hence, it is plausible that the nonbonding interaction of the CH3 group with the surface could make the adsorbed (Z)-gauche form more stable than (Z)trans form. It is difficult to unambiguously decide whether it is the (E)trans or the (Z)-trans conformer which is populated at full monolayer coverage, since we do not know the Ni-0-C bond angle. A knowledge of this angle might allow a surface selection rule analysis of the activity of the ethyl group CH3 deformation and rocking bands. Both the (E)-trans and the (Z)trans configurations, as illustrated in Scheme 4, are consistent with the observation, at 4 langmuirs, of the CH3 deformation band at 1394 cm-I as well as the in-plane CH3 asymmetric deformation and CH3 rocking bands at 1450 and 1150 cm-I, respectively. However, an inspection of the Y(C-0) band provides some indirect evidence for the formation of (E)-trans adsorbed ethyl formate. Note that only a single band at 1242 cm-I is observed at the lowest exposures where the (Z)-gauche form dominates. In contrast, three peaks in the Y(C-0) stretching region are observed at intermediate coverages. The three bands may reflect the presence of the three surface conformers illustrated in Scheme 4. Only a single band, located at 1255 cm-I, is observed at full monolayer coverage. This may reflect a dominance of the (E)-trans configuration at close to full coverage of the chemisorbed layer. We may then summarize the surface coverage dependent RAIRS data for methyl formate and ethyl formate by noting that they present different types of evidence for (Z) to (E) rotational isomerization. The methyl formate data display a shift in the carbonyl band characteristic of (Z) to (E) isomerization. The ethyl formate data display intensity changes, or RAIRS activity changes, consistent with rotation around the C-0 band. Taken together, the data for both molecules are most readily interpreted in terms of (Z) to (E) isomerization. As discussed next, the data for methyl acetate provide an interesting contrast to the methyl formate and ethyl formate data. In contrast to the data for methyl formate and ethyl formate, the coverage-dependent RAIRS spectra for methyl acetate on Ni( 111) simply display an increase in absorbance followed by the growth of features characteristic of multilayer formation. An assignment of the observed absorbance bands is given in Table 2. The photoisomerizationwork of Lenaerts et aL2*show that two new bands, due to shifts in YIO and Y I ~ ,should appear in the 1050-1300 cm-l region if rotational isomerization takes place. The absence of rotational isomerization for adsorbed methyl acetate is of importance in considering the origins of the isomerization observed for the other two ester molecules. The latter issue will be discussed next. B. Origins of Rotational Isomerization in Adsorbed Layers of Esters. As pointed out in the Introduction, coveragedependent changes in adsorbate orientation are usually ascribed to interadsorbate steric interactions: Both thermal desorption measurements and IR absorbance data are sensitive indicators of adsorbate-adsorbate interactions. In this section, we will attempt to correlate evidence for lateral interactions in the TPD and IR data, displayed in Figures 1-3 and 9- 11, respectively,

17914 J. Phys. Chem., Vol. 99, No. 51, 1995 with the observed onsets for rotation around the C-0 bond. This analysis will concentrate on the data for ethyl formate, since the coverages of ethyl formate are known and since the evidence for its rotational isomerization are clearest. However, the arguments given below for ethyl formate could be extended to the results for methyl formate on Ni( 111). The integrated intensities of the carbonyl stretching bands, of the three chemisorbed esters, display a coverage dependence (Figures 9- 11) indicative of dynamic dipole-dipole coupling effects.33 For example, consider the case of ethyl formate (Figure 11). The sum of the integrated intensities of the observed carbonyl stretching bands displays three distinct regions as a function of exposure. There is a roughly linear region at low exposures (0-2 langmuirs), a plateau or leveling off region at intermediate exposures (2-4 langmuirs), and a further increase at high exposures. The IR data displayed in Figures 6 and 7 show that the latter increase reflects the formation of the multilayer. That is, only the initial two regions are associated with the chemisorbed layer. The XPS C( 1s) data included in Figure 1 show that the amount of ethyl formate on the surface is linear in exposure to -4 langmuirs. From the IR measurements shown in Figures 6 and 7, we know that multilayer formation commences at exposures of 4-5 langmuirs. Hence, an exposure of 2 langmuirs corresponds to approximately half monolayer coverage. Given the linear dependence of the coverage on exposure, the plateau region in the integrated intensity data cannot arise from a saturation of available adsorption sites. Instead, it may, most readily, be attributed to depolarization resulting from dipole-dipole coupling within the oriented chemisorption layer.26.27.33As a result, the integrated IR absorbance is nonlinear in coverage. This causes the integrated signal to level off as the coverage is increased and the interadsorbate distance is decreased. Hence, the initial linear region in the plot for ethyl formate (Figure l l ) , for example, shows that island formation does not occur at low coverages. That is, the ethyl formate layer is progressively compressed as the coverage increases. The onset of the plateau region in the integrated intensity at between one-third and one-half coverage reflects the occupation of nearest-neighbor sites. For example, the integrated intensity for CO on Ru(001) levels off at about one-third of a monolayer.34 Nearest-neighbor sites begin to be filled on a hexagonal surface at coverages above I l 3 . Hence, the onset of the leveling-off region in the integrated absorbance data is a rough indicator for the onset of close packing within the adsorbed layer. Note that the onset for isomerization out of the (Z)-gauche state roughly coincides with the depolarization onset. This may be taken as an indication that the isomerization process is dependent on steric interactions between closely spaced adsorbate molecules. The thermal desorption data provide a number of indications for lateral interactions within the chemisorbed layer of ethyl formate, in the coverage region where the (Z)-gauche state is depopulated. As in the integrated absorbance data, the thermal desorption results shown in Figure 2 also display three distinct regions as a function of exposure. Desorption of multilayer ethyl formate takes place at high exposures, two molecular desorption peaks from the chemisorbed layer are observed at intermediate exposures, and total decomposition occurs at low exposures.18aThe majority desorption products of the thermal decomposition of ethyl formate on Ni( 11l), as illustrated in Scheme 2 , are CO and H2. Hence, thermal decomposition is directly indicated by the integrated CO desorption signal shown in Figure 4a. The latter data shows that decomposition saturates at exposures which fall within the plateau region in the integrated IR intensity plot. The saturation behavior may be

Zahidi et al. interpreted in terms of steric constraints, or crowding effects, within the adsorbed layer. Ethyl formate chemisorbs to Ni(111) via the interaction of a carbonyl lone pair with the surface.lBa As illustrated in Scheme 2 , low-temperature decomposition yields adsorbed ethoxy, CO, and atomic hydrogen species. Thus, decomposition replaces a monodentate species with three coadsorbed species. Furthermore, we must take into account the fact that the adsorption energy for ethyl formate on Ni( 111) is approximately equal to its activation energy for surface decomposition.18a Hence, molecular desorption becomes favored over decomposition as the surface packing density increases and the number of free sites available for the decomposition products decreases. The saturation of the decomposition process, at coverages within the plateau region for the IR absorbance data, is therefore an indirect manifestation of steric constraints within the adsorbed layer. Two aspects of the molecular desorption peaks from the chemisorbed layer are consistent with repulsive adsorbateadsorbate interactions. These are the fact that there are multipeak molecular desorption spectra and the fact that there is a downward shift in the desorption temperature with increasing coverage. Downward shifts in the desorption temperature with increasing coverage are the typical TPD signature for repulsive lateral interactions. It is also well-known that lateral interactions within adsorbed layers can give rise to multipeak spectra if sudden discontinuities occur in the coverage dependence of the adsorption energy.35 Such discontinuities result from coverage-dependent lateral interactions. An alternative, albeit parallel, interpretation may also be given for the multipeak methyl formate and ethyl formate desorption spectra. That is, the multipeak desorption spectra reflect the presence of different conformers on the surface. Such an interpretation is supported by the fact that chemisorbed methyl acetate displays only a single molecular desorption peak. Methyl acetate desorption does not undergo rotamerization on Ni( 11l), hence the single desorption peak. The vibrational and thermal desorption data clearly indicate that isomerization takes place in the surface coverage region where lateral interactions or steric constraints in the adsorbed layer become evident. An examination of the different surface conformations for methyl formate and ethyl formate illustrated in Schemes 3 and 4, respectively, gives some idea as to how steric repulsion within the chemisorbed layers may be reduced. If we assume that there is free rotation around the metalcarbonyl group chemisorption bond, then each adsorbed molecule will project a circular area onto the surface. It may be seen from Scheme 4 that the projection of the (Z)-trans form of ethyl formate is approximately 9 times larger than the projection for the (E)-trans form. This suggests that, as the coverage is increased, steric repulsion will destabilize the (Z) form relative to the (E) form. It also suggests that steric repulsion between the bulky ethyl groups may be significantly reduced by (Z) to (E) rotamerization. Such a transformation requires rotation around the C-0 single bond, for which there typically is a high rotational barrier (Table 1). For example, the rotational barrier for ethyl formate in a highly polar solvent is of the order of 10 kcaVm01.'~ However, an examination of Scheme 1 reveals two features which indicate that the rotational barrier may be sufficiently reduced in the chemisorbed system, First, the interconversion from the (Z)-gauche to the (E)-trans conformation involves a continuous motion of the ethyl group away from the metal surface. This implies that steric repulsion is reduced continuously during the isomerization process, since we are dealing with lateral interactions within a two-dimensional system. Thus, lowering of steric repulsion will provide a driving

Rotational Isomerization of Esters on Ni( 111) force along the entire rotational coordinate. With respect to the latter point, molecular modeling calculations by Fox and Rosch4 show that the packing energies within dense adsorbed layers can 'be several times greater than the reported rotational barrier for solution phase ethyl formate. Hence, steric interactions at moderately high coverages contribute to both the fact that (Z) adsorbed form is destabilized with respect to the (E) form and the fact that the rotational barrier is sufficiently reduced to pennit rotamerization. The proposal that the steric interactions are at the origin of the observed rotational isomerization is supported by additional data for methyl acetate on Ni( 111). Rotational isomerization is not observed in the latter case, and this may be explained by presence of repulsive interactions between the two methyl groups in the (E) form of methyl acetate (10). The second feature evident from Scheme 4 is that rotation of the C-0 bond 90' out of the HCOO plane increases the static dipole sum associated with the C=O and the O-C2H5 bonds relative to that for the (Z) forms. The dipole sum is even greater for the (E) form. For example, Wiberg et al.I2 calculate values of 1.90, 4.31, and 2.95 D for the dipole moments of (a-methyl formate, (E)-methyl formate, and the transition state (rotation by 90" out of the plane) between the two rotamers, respectively. Their calculations show a corresponding solvent effect resulting from dipole-induced dipole stabilization. These calculations show that in highly polar, E = 35.9, solvents, the (2)to (E) energy difference is reduced by approximately 70% and the rotational barrier is reduced by around 8%. As may be seen from Table 1,the calculated solvent effects are in line with those observed in experimental studies. Dipole-image dipole stabilization for molecules adsorbed on metals, materials with very high dielectric constants, is a well-appreciatedphenomenon in surface science. Hence, a qualitatively similar image dipole stabilization of the adsorbed esters is expected to hold and to depend on the torsion angle for rotation around the C-0 bond. Given the very high polarizabi€ity of the metal conduction electrons and the fact that the molecules are aligned, the reduction of both the energy difference and the rotational barrier is expected to be greater than for the esters in polar solvents. In summary, for an aligned array of ethyl formate or methyl formate on a metal, both dipole induction and steric effects serve to minimize the barrier for rotation around the C-0 bonds and the formation of the (E)-rotamers on the surface. Obviously, the metal to molecule chemisorption bond will add an extra level of complexity in that it will also modify the equilibria parameters. Model calculations are required to predict the latter effect. The occurrence of rotational isomerization of ethyl formate and methyl formate on Ni( 111) is certainly related to the fact that the lattice parameter of nickel is relatively small and the adsorption energy of ethyl formate on Ni( 111) is relatively large. Discussion of the influence of these two factors on crystal packing forces in chemisorbed layers is given by Gavezotti et al.36and by Fox and Rosch!b The low-coverage value of the adsorption energy of ethyl formate and methyl formate on Ni(111) is in the range of 10-15 kcavmol. Such substantial adsorption energies ensure a dense packing of the adsorbates onto the hexagonal surface lattice. The resulting coveragedependent steric strain within the adsorbate layer leads to rotational isomerization. C. Conclusions and Relevance to Heterogeneously Catalyzed Enantioselective Chemistry. We have obtained direct spectroscopic evidence for rotational isomerization of ethyl formate on Ni(ll1) at 86 K. Rotational isomerization of chemisorbed ethyl formate occurs as its coverage on Ni( 111) is increased toward the half monolayer level. This isomeriza-

J. Phys. Chem., Vol. 99, No. 51, 1995 17915 tion, involving rotation around the C - 0 bond, induces several changes in the relative intensities of absorption bands in the infrared reflectance spectra. Rotational isomerization of methyl formate, at even lower coverages, is evidenced by a characteristic shift in the carbonyl stretching frequency. Both the infrared spectroscopy and thermal desorption data provide indicators for the onset of interadsorbate interactions as the surface coverage is increased. The observed rotational isomerization is attributed to steric repulsion as the packing density of the adsorbed esters is increased beyond a critical level. The relatively low coverages required to produce isomerization would be expected to be easily achieved under typical conditions for heterogeneous catalysis. Hence, the results of this study show that rotational isomerization should be considered in developing mechanistic schemes for enantioselective reactions of esters on surfaces. To date, only a very limited number of highly selective heterogeneously catalyzed enantioselective reactions have been r e p ~ r t e d . ~Most J ~ of these reports concemed the enantioselective hydrogenation of esters on alkaloid-modified Pt or Ir, and on tartrate-promoted Ni-supported metal catalyst^.^,^^ A recent paper, by Wan and Davis,38reported the design and application of a highly effective and stable heterogenized chiral catalyst. It is plausible to assume that the enantioselective hydrogenation is related to the local surface-adsorbate geometry created by the combination of the adsorption geometry of the ester, the proximity of the chiral modifier, and the atomic structure of the metal ~urface.3~In this scenario, the reaction takes place on the accessible face of the electronically activated ester molecules. Electronic activation arises from the chemisorption interaction with the metal surface. We have pointed out, above and in a separate publication,lg that the shifts in the RAIRS spectra of chemisorbed methyl formate, ethyl formate, and methyl acetate are very similar to those observed for esterLewis acid complex (Table 2). The electronic activation of esters chemisorbed on Ni( 111) may then be similar to that for carbonyl compounds attached to Lewis acid centers.40 This parallel between nickel and Lewis acids is interesting in the context of the observed rotational isomerization in the chemisorbed layer and the widespread use of Lewis acids to catalyze asymmetric organic synthesis.41 There are numerous papers in the organic chemistry literature which consider the roles of steric and conformational factors in Lewis acid catalyzed nucleophilic attack of carbonyl functions.42 We note, for example, that Lewis et al.43report that the complexation of ethyl cinnamate to SnC4 results in a cis to trans rotation of the enone moiety. Similarly, Faller and Mal5 found that complexation of methyl acetate to a strong and bulky Lewis acid enhanced the fraction of (E)rotamer. Ab initio calculations by Loncharich et a1.& predict that Lewis acid complexed acrylates will be in the s-trans conformation, in contrast to the s-cis conformation favored for uncomplexed acrylates. The results of the present study indicate that in its chemisorption properties with respect to ester molecules, Ni( 111) may be treated as a Lewis acid analogue. Therefore, this first observation of rotational isomerization of adsorbed esters constitutes an important step in establishing a molecular basis for heterogeneously catalyzed stereoselective chemistry.

Acknowledgment. Partial funding for this research has been provided by the Networks of Centers of Excellence in Molecular and Interfacial Dynamics. Funding for the project was also made available through an NSERC research grant and grants from the FCAR (Quebec). M.C. gratefully acknowledges the receipt of an NSERC graduate student fellowship. We thank J-R. Roy and A. St-Laurent for their assistance with the XPS measurements.

17916 J. Phys. Chem., Vol. 99, No. 51, 1995

References and Notes (1) (a) Shinn, N. D.; Madey, T. E. Phys. Rev. Lett. 1985, 53, 2481.

(b) Hoffman, F. M.; de Paola, R. A. Phys. Rev. Lett. 1984, 52, 1697. (2) (a) Whitman, L. J.; Bartosch, C. E.; Ho, W. J . Chem. Phys. 1986, 85, 3688. (b) Grunze, M.; Golze, M.; Hirschwald, W.; J-Freund, H.; Pulm, H.; Seip, U.; Tsai, M. C.; Kuppers, J. Phys. Rev. Lett. 1984, 53, 850. (3) (a) Wesner, D. A,; Coenen, F. P.; Bonzel, H. P. Phys. Rev. Lett. 1988,60, 1045. (b) Lee, J.; Arias, J.; Hanrahan, C.; Martin, R.; Metiu, H.; Klauber, C.; Alvey, M. D.; Yates, J. T., Jr. Surf. Sci. 1985, 159, L460. (c) Haq, S.; Love, J. L.; King, D. A. Surf. Sci. 1992, 275, 170. (d) Kuhlenbeck, H.; Neumann, M.; J-Freund, H. Surf. Sci. 1986, 173, 194. (e) Pangher, N.; Haase, J. Surf. Sci. 1993, 293, 1908. (4) (a) Rosch, N.; Fox, Th.; Netzer, F. P.; Ramsey, M. G.; Steinmuller, D. J. Chem. Phys. 1991, 94, 3276. (b) Fox, T.: Rosch, N. Surf. Sci. 1991, 256, 159. ( 5 ) Hoffmann, H.: Grittiths, P. R.; Zaera, F. Surf. Sci. 1992, 262, 141. (6) Hubbard, A. T. Chem. Rev. 1988, 88, 633. (7) Xi, M.; Bent, B. E. J . Am. Chem. SOC. 1993. 115, 7426. (8) (a) Simons, K. E.; Ibbotson. A,; Johnson, P.; Plum, H.; Wells, P. B. J. Catal. 1994, 150, 32 1. (b) Griffiths, S. P.; Johnson, P.; Vermeer, W. A. H.; Wells, P. B. J. Chem. Soc., Chem. Commun. 1994, 2431. (c) Minder, B.; Mallat, T.; Skabal, P.; Baiker, A. Catal. Lett. 1994, 29, 115. (9) (a) Jones, G. I. L.; Owen, N. L. J . Mol. Struct. 1973, 18, 1. (b) Mark, H.; Baker, T.; Noe, E. A. J. Am. Chem. SOC. 1989, 111, 6551. (10) (a) Wiberg, K. B.; Laidig, K. E. J . Am. Chem. SOC. 1988, 110, 1872. (b) Wang, X.; Houk, K. N. J . Am. Chem. SOC. 1988, 110, 1870. (11) (a) Wiberg, K. B.; Laidig. K. E. J. Am. Chem. SOC. 1987, 109, 5936. (b) Nagy, P.; Kiss, A. I.; Lopata, A. J. Mol. Struct. 1981, 86, 41. (12) (a) Wiberg, K. B.; Wong, M. W. J . Am. Chem. SOC. 1993, 115, 1081. (b) Wennerstrom, H.; Forsen, S.; Roos, B. J . Phys. Chem. 1972, 76, 2430. (13) Blom, C. E.; Gunthard, Hs. H. Chem. Phys. Len. 1981, 84, 267. (14) Grindley, T. B. Tetrahedron Lett. 1982, 23, 1757. (15) Faller. J. W.; Ma, Y. J . Am. Chem. SOC. 1991, 113, 1579. (16) Riveros, J. M.; Wilson, E. B., Jr. J . Chem. Phys. 1967, 44, 4605. (17) Drakenberg, T.; Forsen, E. J . Phys. Chem. 1972, 76, 3582. (18) (a) Zahidi, E.; Castonguay, M.; McBreen, P. H. J . Am. Chem. SOC. 1994, 116, 5847. (b) Gates, S . M.; Russel. J. N., Jr.; Yates, J. T., Jr. Surf. Sci. 1986, 171, 111. (19) (a) Taillandier, M.; Ligier, J.; Taillandier, E. J. Mol. Struct. 1968, 2, 437. (b) Dembitskii, A. D.; Sumarokova, T. N. Opt. Specrrosc. 1962, 12. 202. (c) Paul. R. C.; Chandha, S . L.; Vashisht, J. L. Indian J . Chem. 1969, 7, 275. (20) Zahidi, E.; Castonguay, M.; McBreen, P. H. Chem. Phys. Lett. 1995, 236, 122. (21) (a) Hollenstein, H.; Giinthard, Hs. H. J . Mol. Spectrosc. 1980, 84, 457. (b) Muller, R. P.; Hollenstein, H.; Huber, J. R. J . Mol. Spectrosc. 1983, 100, 95. (c) Susi, H.; Scherer, J. R. Spectrochim. Acta 1969, 25A, 1243. (d) Rushkin, S.; Bauer, H. S. J . Phys. Chem. 1980, 84, 3061. (22) Charles, S . L.; Jones, G. I. L.; Owen, N. L.: Cyvin, S. J.; Cyvin, R. N. J . Mol. Struct. 1973, 16, 225.

Zahidi et al. (23) (a) Sexton, B. A,; Hughes, A. E.; Avery, N. R. Appl. Surf. Sci. 1985, 22/23, 404. (b) Sexton, B. A.; Hughes, A. E.; Avery, N. R. Sut$ Sci. 1985, 155, 366. (24) (a) Millar, G. J.; Rochester, C. H., Waugh, K. G. J . Chem. SOC., Faraday Trans. 1991,87,2784. (b) Monti, D. M.; Cant, N.W.; Trim, D. L.; Wainwright, M. S. J. Catal. 1986, 100, 17. (25) Moravie, R.-M.; Corset, J.; Burneau, A. J. Chim. P h p . 1982, 79. 119. (26) Persson, B. N. J.; Ryberg, R. Phys. Rev. B 1981, 24, 6954. (27) Ryberg, R. Adv. Chem. Phys. 1989, I26 1. (28) Lenaerts, S.; Daeyart, F.; Vanderveken, B. J.; Maes, G. Spectrosc. Lett. 1989, 23, 289. (29) (a) Fan, J.; Trenary, M. Langmuir 1994, IO. 3649. (b) Greenler, R. G. J . Chem. Phys. 1966, 44, 310, (30) Ruschin, S.; Bauer, S. H. J . Phys. Chem. 1980, 84, 3061. (31) Madey, T. E.; Yates, J. T., Jr. Surf. Sci. 1978, 76, 397. (32) Sexton, B. A,; Hughes, A. E. Sur$ Sci. 1984, 140, 227. (33) Borguet, E.; Dai, H-L. J . Chem. Phys. 1994, 101, 9080. (34) Pfnur, H.; Menzel, D.; Hoffman, M. F.; Onega, A,; Bradshaw, A. M. .Sui$ Sci. 1980, 93, 431. (35) (a) King, D. A. Surf. Sci. 1975, 47, 384. (b) Payne, S . H.; Zhang, J.; Kreuzer, H. J. S u ~ Sci. . 1992, 264, 185. (c) Adams, D. L. Sui$ Sci. 1974, 42, 12. (36) (a) Gavezotti, A.; Simonetta, M. Chem. Phys. Lett. 1982, 92, 16. (b) Gavezotti, A.; Simonetta, M. Surf. Sci. 1982, 116, L207. (37) (a) Orito, Y.; Imai, S.; Niwa, S.; Nguyen, G-H. Synrh. Org. Chem. Jpn. 1979,37, 173. (b) Garland, M.; Blaser, H-U. J . Am. Chem. SOC. 1990, 112, 7048. (c) Izumi, Y. In Advances in Catalysis; Eley, D. D.. Pines, H., Weisz, P. B., Eds.; Academic Press: San Diego, CA, 1983; Vol. 32, p 215. (d) Webb, G.; Wells, P. B. Catal. Today 1992. 12, 319. (38) Wan, K. T.; Davis, M. E. Nature 1994, 370, 450. (39) (a) Sutherland, I. M.; Ibbotson, A.; Moyes, R. B.; Wells, P. B. J . Catal. 1990, 125,77. (b) Augustine, R. L.; Tanielyan, S. K.; Doyle. L. K. Tetrahedron: Asymmetry 1993, 4 , 1803. (40) (a) Reetz, M. T.; Hullmann, M.; Massa, W.; Berger, S.; Rademacher, P.; Heymanns, P. J . Am. Chem. SOC. 1986, 108. 2405. (b) Lewis, F. D.; Oxman, J. D.; Gibson, L. L.; Hampsch, H. L.; Quillen, S . L. J . Am. Chem. SOC. 1986, 108, 3005. (41) (a) Evans. D. A. Science 1988, 240, 420. (b) Denmark, S . A,; Almstead, N. G. J. Am. Chem. SOC. 1993, 115, 3133. (c) Shambayati, S.; Crowe, S . W.; Schreiber, S . L. Angew. Chem., Int. Ed. Engl. 1990, 29. 256. (42) (a)Bimey, D. M.; Houk, K. N. J . Am. Chem. SOC.1990,112.4129. (b) Corey, E. J.; Loh, T-P.; Sarshar, S.; Azimioara, M. Tetrahedron Lett. 1992,33, 6945. (c) Curran, D. P.; Kim, B. P.; Piyasena, H. P.; Loncharich. R. J.; Houk, K. N. J . Org. Chem. 1987, 52, 2137. (43) Lewis, F. D.; Quillen, S. L.; Hale, P. D.; Oxman, J. D. J . Am. Chem. Soc. 1988, 110, 1261. (44) Loncharich, R. J.; Schartz, T. R.; Houk, K. H. J . Am. Chem. SOC. 1987, 109, 14. JP952417A