Hydroxymethylcyclopropane on oxygen-covered Mo(110): A radical clock on a surface

Full text of DOI:10.1021/ja002769h

J. Am. Chem. Soc. 2000, 122, 12395-12396
12395
Scheme 1
Hydroxymethylcyclopropane on Oxygen-Covered
Mo(110): A Radical Clock on a Surface
Ilona Kretzschmar, Joshua A. Levinson, and
Cynthia M. Friend*
Department of Chemistry, HarVard UniVersity
Hydrocarbon products are formed between 480 and 600 K
during temperature-programmed reaction of hydroxymethylcy-
clopropane on O-covered Mo(110). The products are identified
as 1-butene, 1,3-butadiene, and ethene based on analysis of the
mass spectrometer fragmentation patterns (Figure 2, Table 1,
Supporting Information). The product distribution is estimated
to be in a ratio of 22:60:18 for 1-butene:1,3-butadiene:ethene.
The only other products formed are H2, water, and CO. Carbon
monoxide is formed at ∼950 K via reaction of carbon and oxygen,
indicating that deposition of carbon competes with elimination
of hydrocarbons.
The detection of linear C4-carbon products at elevated tem-
peratures suggests that ring opening has occurred; however, the
mass spectra of 1-butene and methylcyclopropane16 as well as
those of 1,3-butadiene and methylenecyclopropane17 are very
similar. Furthermore, the linear C4 carbons could be evolved either
via direct desorption of the ring-opened radical or via decomposi-
tion of alkyl or alkoxy intermediates bound to the surface.
Infrared reflection absorption spectroscopy is used to probe
for ring opening and to distinguish between the two paths. The
similarity in the spectrum obtained for hydroxymethylcyclopro-
pane adsorbed on O-covered Mo(110) at 100 K (Figure 1a) to
that for gas-phase hydroxymethylcyclopropane17 (Table 2, Sup-
porting Information) indicates that intact alcohol is present on
the surface at low temperature. At high coverage, the broad OH
absorption band of hydroxymethylcyclopropane is visible at
∼3280 cm-1. Heating to 220 K induces sublimation of condensed
hydroxymethylcyclopropane and some cleavage of the O-H bond
(not shown).
Heating to 350 K leads to completion of OH bond breaking.
Two new peaks at 905 and 936 cm-1, attributed to the formation
of a Mo-O-C bond,13 develop in the infrared spectrum. There
are no shifts in the 800-3010 region when an 18O-labeled surface
is used, indicating that the C-O bond is retained at 350 K. The
presence of the ν(C-O) mode at 1029 cm-1 further supports this
assertion. Hydroxy is identified by the appearance of a sharp
ν(OH) band at 3572 cm-1 (Figure 1b),13,18 that shifts to 3561 cm-1
when the reaction is carried out on 18O-covered Mo(110). The
labeling experiments demonstrate that hydrogen is transferred to
surface oxygen, leaving the C-O bond intact. The remainder of
the spectrum is attributed to methylcyclopropyloxide (c-C3H5-
CH2O-, Table 2, Supporting Information). There are two sharp
features at 1393 and 1434 cm-1, in the region expected for the
ring-CH2 scissors mode19 and one dominant peak in the C-H
stretch region at 3010 cm-1. Notably, there is no peak detected
in the region expected for a ν(CdC) mode, 1645 cm-1, indicating
that ring opening has not occurred at 350 K.
Cambridge, Massachusetts 02138
ReceiVed July 27, 2000
Radicals are proposed as intermediates for a range of hetero-
geneous reactions, but little is known of their lifetimes because
of their transient nature. This study describes the first observation
of the methylcyclopropyl-3-butenyl radical clock rearrangement
(Scheme 1) on a surface under ultrahigh-vacuum conditions. We
use the fact that alkyl radicals are transiently formed from reaction
of alcohols on Mo(110) established in previous studies.1,2 Because
both unrearranged and rearranged species are trapped on oxygen-
covered Mo(110) during the reaction, the lifetime of the radical
on the surface is on the order of 10-8 s. Since their introduction
by Giller and Ingold3 in 1980, free radical clocks3,4 have become
a well-established technique in organic chemistry to determine
radical lifetimes.5,6 The radical rearrangement shown in Scheme
1 is a well-calibrated example of a fast-reacting radical clock.7
In practice, a radical clock reacts with a radical of unknown
lifetime. The lifetime of the unknown radical is subsequently
determined by combining information on product distribution,
reactant concentration, and known reaction rates.8 For example,
the viability of a discrete substrate radical species during
hydrocarbon oxidation by soluble methane monooxygenase was
assessed using substituted cyclopropane probes.9 Hanzlik et al.10
used substituent effects on radical clocks to distinguish between
single-electron and hydrogen-atom transfer processes. Curtis and
Drucker11 used the rearrangement of the methylcyclopropyl radical
formed via cleavage of the thiolate C-S bond to verify the
formation of a radical intermediate during hydrodesulfurization
of thiols by a Co/Mo/S cluster. Ethyl- and isopropylcyclopropane
radical clocks were employed as mechanistic probes in the
catalytic oxidation of alkanes and alkenes by titanium silicates
in solution.12 In this study, we probe the relative time scales for
competing processes on a metal surface.
There are three requirements for radical clocks on a surface:
(i) the molecule must be anchored to the surface, (ii) the radical
rearrangement must be controlled via temperature, and (iii)
products must be trapped on the surface for detection. Our probe
molecule, hydroxymethylcyclopropane, was selected because
alcohols are known to react by O-H bond cleavage, yielding an
adsorbed alkoxide, on O-covered Mo(110).13-15
(1) Serafin, J. G.; Friend, C. M. J. Am. Chem. Soc. 1989, 111, 8967.
(2) Wiegand, B. C.; Uvdal, P.; Serafin, J. G.; Friend, C. M. J. Phys. Chem.
1992, 96, 5063.
(3) Griller, D.; Ingold, K. U. Acc. Chem. Res. 1980, 13, 317.
(4) Newcomb, M.; Glen, A. G. J. Am. Chem. Soc. 1989, 111, 275.
(5) Tian, F.; Battiste, M. A.; Dolbier, W. R. Org. Lett. 1999, 1, 193.
(6) Jiaang, W. T.; Lin, H. C.; Tang, K. H.; Chang, L. B.; Tsai, Y. M. J.
Org. Chem. 1999, 64, 618.
(7) Halgren, T. A.; Roberts, J. D.; Horner, J. H.; Martinez, F. N.; Tronche,
C.; Newcomb, M. J. Am. Chem. Soc. 2000, 122, 2988.
(8) Newcomb, M. Tetrahedron 1993, 49, 1151.
(9) Valentine, A. M.; LeTadic-Biadatti, M.-H.; Toy, P. H.; Newcomb, M.;
Lippard, S. J. J. Bio. Chem. 1999, 274, 10771.
(10) Chen, H.; de Groot, M. J.; Vermeulen, N. P. E.; Hanzlik, R. P. J.
Org. Chem. 1997, 62, 8227.
Heating to 460 K induces ring opening and disproportionation
of adsorbed OH. Two new vibrational peaks appear at 1243 and
1645 cm-1, while the ring modes at 1393 and 1434 cm-1 and the
ν(O-H) at 3572 cm-1 disappear (Figure 1c). The appearance of
the ν(CdC) peak20 at 1645 cm-1 is strong evidence that ring
opening has occurred and that the product is trapped on the
(11) Curtis, M. D.; Druker, S. H. J. Am. Chem. Soc. 1997, 119, 1027.
(12) Khouw, C. B.; Dartt, C. B.; Labinger, J. A.; Davis, M. E. J. Catal.
1994, 149, 195.
(16) Wiegand, B. C.; Napier, M. E.; Friend, C. M.; Uvdal, P. J. Am. Chem.
Soc. 1996, 118, 2962.
(17) Stein, S. E. In IR and Mass Spectra; Stein, S. E., Ed.; National Institute
of Standards and Technology: Gaithersburg MD, February 2000.
(18) Kretzschmar, I.; Friend, C. M. unpublished results.
(19) Martel, R.; Rochefort, A.; McBreen, P. H. J. Am. Chem. Soc. 1998,
120, 2421.
(13) Chen, D. A.; Friend, C. M. J. Am. Chem. Soc. 1998, 120, 5017.
(14) Uvdal, P.; MacKerell, A. D., Jr.; Wiegand, B. C. J. Electron Spectrosc.
Relat. Phenom. 1993, 64/65, 193.
(15) Chen, D. A.; Friend, C. M. Langmuir 1997, 14, 11451.
10.1021/ja002769h CCC: $19.00 © 2000 American Chemical Society
Published on Web 11/23/2000

12396 J. Am. Chem. Soc., Vol. 122, No. 49, 2000
Communications to the Editor
Scheme 2
keep the radical in the vicinity of the surface for a sufficiently
long time for ring opening. Once the butenyl radical is formed,
it may either form a carbon-metal bond or add to adsorbed
oxygen to form the 3-butenyl-1-oxide that is identified by the
1645 cm-1 peak. The C-O bond strength of the 3-butenyl-1-
oxide is estimated to be 1.2 eV26,27 greater than the methylcy-
clopropyloxide providing a thermodynamic driving force for
reformation of the C-O bond.
Figure 1. FT-IR spectra of adsorbed products of hydroxymethylcyclo-
propane reaction on oxygen-covered Mo(110) upon heating to: (a) 100
K, (b) 350 K, and (c) 460 K. Trace (d) shows the spectrum obtained for
3-buten-1-ol under the same conditions as (c). The crystal is allowed to
cool to 130 K prior to the IR-measurement.
surface. The disappearance of the ν(O-H) peak coincides with
the evolution of water.
A ring-opening mechanism mediated by the surface28 that does
not involve cleavage of the C-O bond is ruled out by experiments
performed recently in our laboratory using 1-cyclopropylethanol.18
Neither ring opening nor O-containing products are observed for
1-cyclopropylethanol when heated to elevated temperatures; only
C5 species desorbsall below 350 K. Further, the formation of a
metallacyclobutane species is excluded, because the characteristic
band at 2775 cm-1 is not detected.29 The mechanism depicted
implies that the radical rearrangement is faster than the trapping
of the radical by the surface. The latter, in turn, is faster than
hydrogen abstraction from the surface by either the unrearranged
or the rearranged radical, which rules out the direct desorption
mentioned above.
Changes in the C-O stretch region indicate that one of the
ring-opened products is the alkoxide, 3-butenyl-1-oxide. Assign-
ments are facilitated by experiments performed on an 18O-labeled
surface (Table 2). Notably, the two new peaks at 1243 and 1645
cm-1 do not shift, whereas shifts are observed in the 850-1100
cm-1 region. The shifts are strongest (10 and 8 cm-1) for the
900 and 939 cm-1 peaks, indicating some component of ν(Mo-
O-C).
Further evidence for the formation of the ring-opened alkoxide,
3-butenyl-1-oxide, from reaction of hydroxymethylcyclopropane
is obtained via comparison to the temperature-dependent spectrum
of 3-buten-1-ol adsorbed on O-covered Mo(110) (Figure 1d). All
peaks present in the spectrum obtained after heating 3-buten-1-
ol to 460 K are contained in Figure 1c. There are additional peaks
at 900, 915, 1243, 1342, and 2958 cm-1 in Figure 1c, which are
not present in Figure 1d. Modes in the 1200-1250 cm-1 region
have been attributed to the δ (CH2) mode of alkyl groups bound
to metal atoms,21,22 thus the 1243 cm-1 peak is assigned to a metal-
bound butenyl species.23
The reaction scheme shown for hydroxymethylcyclopropane
on O-covered Mo(110) is formulated based on all of our results
(Scheme 2). The ring-closed alkoxide forms at 350 K via O-H
bond scission. Some of the hydroxy hydrogens are transferred to
oxygen on the surface to form adsorbed OH. Above 350 K,2,13
the surface hydroxyl decomposes to gaseous water, the C-O bond
of the ring alkoxide species breaks, and the methylcyclopropyl
radical forms in the vicinity of the surface. It remains there long
enough to rearrange and subsequently form either new C-O or
C-Mo bonds. The time-scale for ring opening is on the order of
10-8 s in solution. We suggest that dipole-image dipole24 and
van der Waals interactions25 between the ring and the surface
On the basis of our study, the following relative ordering of
rates on O-covered Mo(110) is indicated: krearrange ) 4.7 × 108
s-1 > ktrap > kH-abstraction, assuming that the rates of rearrangement
are similar on the surface and in solution. Notably, hydrogen
abstraction from clean Mo(110) is faster, based on formation of
methylcyclopropane from methylcyclopropanethiol.16 The differ-
ence may be due to differing hydrogen reactivity, because the
desorption temperature for hydrogen is ∼125 K higher on
O-covered Mo, indicating that radical trapping rates depend on
the nature of the surface. This work provides a general approach
to studying radical lifetimes and reactions on surfaces.
Acknowledgment. We acknowledge support for this research by the
U.S. Department of Energy, Basic Energy Sciences, under Grant No.
DE-FG02-84-ER13289. I.K. thanks the Alexander von Humboldt founda-
tion for a Feodor-Lynen fellowship. Professor J. G. Chen is thanked for
helpful discussions.
Supporting Information Available: Tables and Figure of product
distributions for temperature dependent reaction spectroscopy and IR peak
assignments (PDF). This material is availabe free of charge via the Internet
at http://pubs.acs.org.
(20) Bent, B. E. Chem. ReV. 1996, 96, 1361.
(21) Eng, J.; Bent, B. E.; Fru¨hberger, B.; Chen, J. G. Langmuir 1998, 14,
1301.
JA002769H
(22) Abdelrehim, I. M.; Pelhos, K.; Madey, T. E.; Eng, J., Jr.; Chen, J. G.
J. Phys. Chem. B 1998, 102, 9697.
(25) Stinnett, J. A.; Madix, R. J. J. Chem. Phys. 1996, 105, 1609.
(26) Preliminary results of calculations using the method described in ref
27.
(27) Uvdal, P.; MacKerell, A. D., Jr.; Wiegand, B. C.; Friend, C. M. Phys.
ReV. B 1995, 51, 7844.
(28) Touvelle, M. S.; Stair, P. C. J. Catal. 1993, 139, 93.
(29) Ephritikhine, M.; Francis, B. R.; Green, M. L. H.; Mackenzie, R. E.;
Smith, M. J. J. Chem. Soc., Dalton Trans. 1977, 1131.
(23) Experiments involving an additional submonolayer coverage of NO
on top of the O-overlayer show that the 1645 cm-1 peak is retained, while
the 1243 cm-1 peak is no longer present in the IR spectrum. From this
observation it can be concluded that the 1645 cm-1 peak is an independent
species assigned as the alkoxide species.
(24) Coltrin, M. E.; Kay, B. D. J. Chem. Phys. 1988, 89, 551.