Extrinsic precursor-assisted synthesis of 1,5-hexadiene on Cu(100)

Article abstract of DOI:10.1021/ja0011414

The reaction between allyl bromide and previously chemisorbed η³-allyl to form 1,5-hexadiene on Cu(100) at cryogenic temperatures has been examined using reflection absorption infrared spectroscopy and temperature-programmed desorption. Above 110 K, the 1,5-hexadiene formation rate decreases with increasing temperature and is controlled by the residence time of dosed allyl bromide, whereas below 100 K, the rate increases with temperature and is controlled by the reaction of weakly adsorbed allyl bromide with chemisorbed η³-allyl. Above 110 K, a precursor state model is used to interpret the data. The activation energy difference, E_d-E_r, between desorption and reaction of allyl bromide with η³-allyl is 12 kJ mol^-1.

Full text of DOI:10.1021/ja0011414

2990
J. Am. Chem. Soc. 2001, 123, 2990-2996
Extrinsic Precursor-Assisted Synthesis of 1,5-Hexadiene on Cu(100)
H. Celio, K. C. Scheer, and J. M. White*
Contribution from the Center for Materials Chemistry, Department of Chemistry and Biochemistry,
UniVersity of Texas at Austin, Austin, Texas 78712-1167
ReceiVed March 31, 2000
Abstract: The reaction between allyl bromide and previously chemisorbed η3-allyl to form 1,5-hexadiene on
Cu(100) at cryogenic temperatures has been examined using reflection absorption infrared spectroscopy and
temperature-programmed desorption. Above 110 K, the 1,5-hexadiene formation rate decreases with increasing
temperature and is controlled by the residence time of dosed allyl bromide, whereas below 100 K, the rate
increases with temperature and is controlled by the reaction of weakly adsorbed allyl bromide with chemisorbed
η3-allyl. Above 110 K, a precursor state model is used to interpret the data. The activation energy difference,
Ed - Er, between desorption and reaction of allyl bromide with η3-allyl is 12 kJ mol-1
.
I. Introduction
while dosing allyl bromide, C3H5Br, on Cu(100) covered with
η3-allyl species and atomic bromine. This cryogenic coupling
reaction contrasts with typical carbon-carbon coupling reactions
of adsorbed alkyl and vinyl groups on Cu and other transition
metals under ultrahigh-vacuum (UHV) conditions.10,11 The
Communication also included the first high-resolution reflection
absorption infrared spectrum (RAIRS) of chemisorbed η3-allyl,
an important C3 species.12,13
The study of transition metal-catalyzed carbon-carbon bond-
forming reactions constitutes a major area of fundamental
research and technological significance, e.g., Fischer-Tropsch
synthesis1 and oxidative coupling of methane.2 For these and
other catalyzed coupling processes, overall reaction paths have
been proposed, but mechanistic details are sparse.1,2 Typically,
Langmuir-Hinshelwood, Rideal-Eley, and precursor mecha-
nisms have used to describe adsorption and reaction of hydro-
carbons on clean transition metal and nonmetallic surfaces.3
While the concept of transient precursor states, proposed four
decades ago by Kisliuk,4 is often used to describe impinging
reactants, characterizing the reaction kinetics of extrinsic
precursors remains challenging.3 Two types of precursors,
intrinsic and extrinsic, are distinguished on the basis of whether
the weakly adsorbed precursor molecule, trapped (physisorbed)
by dispersion forces, is located over an empty substrate site
(intrinsic) or over an adsorbate layer site (extrinsic). Kisliuk
originally proposed an intrinsic precursor state to explain the
dissociative adsorption kinetics of nitrogen over tungsten.4 More
recently, intrinsic precursor states have been identified in the
process of forming chemisorbed benzene on Si(111)-7 × 75
and in the kinetics of ethane dissociation on Si(100)-(2 × 1)
and Ir(110)-(1 × 2).6,7 Theoretical studies of both extrinsic
and intrinsic precursor states are also available.8
Other substrates, e.g., Pt(111),15,24 Al(100),16 Ag(111),17
Ag(110),18 and Ni(100),19 have been used in surface science
investigations of allyl groups formed from allyl halides. For
example, in a recent study of C3H5Br and C3H5I on Pt(111),24
RAIRS and TPD indicated the formation of η1- and η3-C3H5
following cleavage of the carbon-halogen bond. The products
formed in TPD depended on coverage; at low coverages, no
C-containing species desorbed, but at high coverages some
propene desorbed. The latter occurred in two TPD peaks. The
lower (185 K) was attributed to prompt desorption upon H
addition to η1-C3H5 and the higher (250 K) to the hydrogenation
of η3-C3H5 to a di-σ form of propene that undergoes significant
D-for-H exchange prior to desorbing as propene. There was no
evidence in this study for C-C bond formation.
Here, we report further on the coupling reaction that forms
1,5-hexadiene, focusing on the mechanism and temperature
dependence of its kinetics. Dissociative adsorption of allyl
chloride, C3H5Cl, was used to prepare a saturated and well-
In a recent Communication,9 we reported the facile formation
of 1,5-hexadiene, C6H10, at cryogenic temperatures (110 K)
(9) Celio, H.; Smith, K. C.; White, J. M. J. Am. Chem. Soc. 1999, 121,
10422-10423.
(10) Bent, B. E. Chem. ReV. 1996, 96, 1361-1390.
(11) Zaera, F. Chem. ReV. 1995, 95, 2651-2693.
(12) (a) Biloen, P.; Helle, N. J.; Sachtler, W. M. H. J. Catal. 1979, 58,
95-107. (b) Brady, R. C., III; Pettit, R. J. Am. Chem. Soc. 1981, 103, 1287-
1289.
(13) Maitlis, P. M.; Long, H. C.; Quyoum, R.; Turner, M. L.; Wang, Z.
Q. Chem. Commun. 1996, 1-8.
(14) Collman, J. P.; Hegedus, L. S.; Norton, J. R.; Finke, R. G. Principles
and Applications of Organotransition Metal Chemistry; University Science
Books: Mill Valley, CA, 1987; Chapter 19.
(15) Scoggins, T. B.; White, J. M. J. Phys. Chem. 1997, 101, 7958-
7967.
(16) Bent, B. E.; Nuzzo, R. G.; Zegarski, B. R.; Dubois, L. H. J. Am.
Chem. Soc. 1991, 113, 1143.
(17) Kershen, K.; Celio, H.; Lee, I.; White, J. M. Langmuir 2001, 17,
323-328.
(18) Dvorak, J.; Dai, H. L. J. Chem. Phys. 2000, 112, 923-934.
(19) Tjandra, S.; Zaera F. J. Phys. Chem. B 1997, 101, 1006.
* Corresponding author. Tel.: 512-471-3704. Fax: 512-471-9495. E-
mail: jmwhite@mail.utexas.edu.
(1) Anderson, R. B. The Fischer-Tropsch Synthesis; Academic Press:
London, U.K., 1984.
(2) Somorjai, G. A. An Introduction to Surface Chemistry and Catalysis;
Wiley-Interscience: New York, 1994; Chapter 7.
(3) Masel, R. I. Principles of Adsorption and Reaction on Solid Surfaces;
Wiley: New York, 1996; Chapter 10.
(4) Kisliuk, P. J. Phys. Chem. Solids 1957, 3, 95-101.
(5) Brown, D. E.; Moffatt, D. J.; Wolkow, R. A. Science 1998, 279,
542-544.
(6) Reeves, C. T.; Ferguson, B. A.; Mullins, C. B.; Sitz, G. O.; Helmer,
B. A.; Graves, D. B. J. Chem. Phys. 1999, 111, 7567-7575.
(7) (a) Kang, H. C.; Mullins, C. B.; Weinberg, W. H. J. Chem. Phys.
1990, 92, 1397-1406. (b) Soulen, S. A.; Madix, R. J. Surf. Sci. 1995, 323,
1-5.
(8) (a) Doren, D. J.; Tully, J. C. Langmuir 1988, 4, 256-268. (b) Doren,
D. J.; Tulley, J. C. J. Chem. Phys. 1991, 94, 8428-8440.
10.1021/ja0011414 CCC: $20.00 © 2001 American Chemical Society
Published on Web 03/09/2001

Synthesis of 1,5-Hexadiene on Cu(100)
J. Am. Chem. Soc., Vol. 123, No. 13, 2001 2991
characterized 1:1 layer of chemisorbed η3-C3H5 and Cl with
which impinging C3H5Br forms a precursor state and reacts to
form physisorbed C6H10 as follows:
the variable leak valve, we preset it and left it open at all times. With
the sample holder and substrate at 300 K, the leak valve was set so
that the pressure inside the chamber rose to 4 × 10-9 Torr when 100
mTorr of Ar was applied to it. This pressure was used to calculate
doses in Langmuir (1 L ) 1 × 10-6 Torr‚s), e.g., a 0.6 L dose of
C3H5Br requires a 150 s exposure. To dose, 100 mTorr of gas was
first placed in a vessel behind a butterfly valve (closed) that was
connected by an evacuated tube to the preset leak valve. The butterfly
valve was then opened. The dose terminated by evacuating the gas
behind the leak with the butterfly valve open. This procedure improved
the reproducibility of the coverages compared to the typical procedure
of opening and closing the leak valve. It is noteworthy that, with the
sample holder and substrate at cryogenic temperatures and the same
doser settings, the pressure rose to no more than 6 × 10-10 Torr when
dosing the allyl-containing molecules.
To prepare the C3H5(c) and Cl(c) layer, 1.0 L of C3H5Cl was dosed
on clean Cu(100) at 110 K, annealed to 200 K to desorb excess C3H5Cl,
and cooled to the temperature of interest. This procedure yielded highly
reproducible infrared spectra of C3H5(c) and Cl(c) that were saturated
with respect to further dissociation of dosed C3H5Cl.
Cl-covered Cu(100) without C3H5(c) was prepared by dosing C3H5Cl
at 110 K and annealing to 550 K to desorb C3H5. After three dose/
anneal cycles, the Cl/Cu AES ratio saturates, and there is no RAIRS
evidence for dissociation upon dosing and annealing more C3H5Cl. To
prepare bromine-covered Cu(100), the same procedure was followed
using C3H5Br. The extremely weak bromine Auger signal (LMM
transition at 1393 eV) precludes a direct assessment of the number of
cycles required to reach saturation.
C3H5Br(p) + η3-C3H5(c) + Cl(c) f C6H10(p) + Br(c) + Cl(c) (1)
The subscripts (p) and (c) denote physisorbed and chemisorbed
states, respectively. The reactive sticking probabilities of incident
C3H5Br were extracted using an analogue of the King and Wells
method20-22 based on reflection absorption infrared spectroscopy
(RAIRS). In a related work to be published elsewhere,23 we
show that for a surface saturated with C3H5(c) and the ac-
companying halogen as in (1), C3H5 takes the exo-η3- form
between 77 and 300 K, i.e., all three carbons bound to Cu (η3)
and oriented so the C-C-C angle is pointed away from the
accompanying halogen (exo). Further, temperature-programmed
desorption (TPD) is dominated by C3H5 radical desorption at
∼450 K.
Using RAIRS data, complemented by TPD data, and the rich
literature that describes organometallic reaction chemistry of
allyl groups (e.g., coupling reactions between η3-C3H5 groups),14
we answer two fundamental questions. First, how does C3H5Br
adsorb on a 1:1 layer of η3-C3H5(c) and Cl(c)? Second, what
kinetics and mechanism describe the reaction between C3H5Br
and η3-C3H5(c) to form physisorbed 1,5-hexadiene, C6H10?
As noted in the Introduction, we used a method related to that
frequently used in molecular beam scattering to extract precursor
reaction kinetics.22 Using RAIRS, we monitored the intensity of C3H5(c)
for active and inactive surfaces and used the ratio to compute the
reaction probability (Sr). Sr is analogous to the sticking probability (S),
typically measured with mass spectrometry. However, unlike mass
spectrometry, which monitors desorbed species, RAIRS monitors
surface species. The details of this procedure will be described below
in the context of the RAIRS data.
II. Experimental Section
The experiments were performed in a two-level ultrahigh-vacuum
(UHV) chamber with a base pressure of 4 × 10-10 Torr. The lower
chamber is equipped with standard surface analysis tools, including a
single-pass cylindrical mirror analyzer for Auger electron spectroscopy
(AES) and a differentially pumped quadrupole mass spectrometer for
TPD. The upper chamber houses a residual gas analyzer (RGA) and is
coupled to a commercial Fourier transform infrared spectrometer for
RAIRS. Infrared spectra (peak-to-peak noise ∼0.004% ∆R/R units) were
recorded by co-adding 1500 scans at 4 cm-1 resolution. All spectra
are reported with respect to an appropriate reference that accounts for
differing reflectivities (e.g., electronic absorption). The reference was
either clean Cu(100) or halogen-covered Cu(100). The Cu(100)
temperature was varied between 77 and 1000 K using a power supply
controlled by feedback from a type K thermocouple inserted into the
edge of the crystal.
III. Results
Multilayer C3H5Cl dosed at 110 K, annealed to 200 K, and
recooled to 110 K leaves a surface covered with equal and
saturated concentrations of C3H5(c) and Cl(c) but no C3H5Cl.30
The RAIRS, Figure 1A, comprises a strong band at 886 cm-1
assigned to η3-allyl(c),9 and two peaks at 2089 and 2114 cm-1
.
The surface was cleaned by cycles of Ar+ ion grazing-incidence
sputtering at 750 K (1.5 kV, 8 µA, 15 min) and annealing (975 K for
15 min). Cycles continued until impurity (C, O, Si, and S) concentra-
tions were below AES detection limits. A more stringent quality test,
RAIRS of adsorbed CO,25 confirmed both cleanliness and surface
ordering with minimal defects.
The latter are inverted, indicating their presence in the reference
spectrum, and are assigned to CO adsorbed on step edges.25
This CO accumulates primarily during the collection of the
background RAIRS at 110 K, 65 degrees below the desorption
temperature of CO.25
Spectrum B, which uses the same reference as spectrum A,
was taken after dosing 2.0 L of CO on the saturated 1:1 η3-
C3H5(c) and Cl(c) overlayer. Differences between A and B are
negligible, indicating strong inhibition of CO adsorption. Even
after a 10 L CO dose, there is no RAIRS change (not shown).
For spectrum C, system B was annealed to 550 K in a vacuum,
recooled to 110 K, and dosed with 2.0 L of CO. As expected,
there is no η3-allyl(c) signal. There is, however, an intense,
The adsorbates, allyl bromide (3-bromopropene, C3H5Br, Aldrich
98+%; stabilized with propylene oxide), allyl chloride (3-chloropropene,
C3H5Cl, Aldrich 99+%; stabilized with propylene oxide), and 1,5-
hexadiene (Aldrich 99+%), were each purified by several freeze-
pump-thaw cycles prior to each experiment. Reproducible dosing,
through a cylindrical tube (3 mm i.d., 60 mm in length) that terminated
3 cm in front of the Cu(100) surface, was achieved using an altered
version of the common procedure. Rather than opening and closing
narrow band (full width at half-maximum (fwhm) ) 7.5 cm-1
)
(20) King, D. A.; Wells, M. G. Surf. Sci. 1972, 29, 454-482.
(21) (a) King, D. A. In Critical ReViews in Solid State and Materials
Sciences; Schuele, D. E., Hoffman, R. W., Eds.; CRC Press: Boca Raton,
FL, 1978; pp 167-208. (b) Arumainayagam, C. R.; Madix, R. J. In
Molecular Beam Studies of Gas-Phase Collision Dynamics; Davidson, S.
G., Ed.; Progress in Surface Science 38; Pergamon: New York, 1991; pp
1-102.
(22) Cassuto, A.; King, D. A. Surf. Sci. 1981, 102, 388-404.
(23) Celio, H.; White, J. M. J. Phys. Chem., in press.
(24) Chrysostomou, D.; Zaera, F. J. Phys. Chem. 2001, 105, 1003-1011.
(25) (a) Uvdal, P.; Karlsson, P.-A.; Nyberg, C.; Andersson, A.; Rich-
ardson, N. V. Surf. Sci. 1988, 202, 167-182. (b) Hollins, P. Surf. Sci. Rep.
1992, 16, 51-94.
at 2091 cm-1 assigned to CO by comparing it with the spectrum
of CO-saturated (θCO ) 0.57) Cu(100).25 The integrated CO
intensity, I(CO) = ∫(∆R/R) dν, is 0.20 ( 0.01 cm-1 compared
to 0.30 ( 0.01 cm-1 for saturation CO on clean Cu (not shown).
In the Discussion section, we estimate the coverage of the 1:1
η3-allyl(c) and Cl(c) overlayer using the CO intensities from these
experiments.
Figure 2 shows RAIRS, each from a separate experiment, of
a saturation 1:1 η3-allyl(c) and Cl(c) prepared as described aboves

2992 J. Am. Chem. Soc., Vol. 123, No. 13, 2001
Celio et al.
Figure 1. RAIRS 1:1 mixture of η3-allyl(c) and Cl(c) (A), followed by
dosing 2.0 L CO (B) and 2 L CO on Cl-covered Cu(100) after
desorption of η3-allyl(c) groups (C). The inverted peaks in spectra A
and B near the 2100 cm-1 region are due to background CO adsorption
on clean Cu(100). All spectra were taken at 110 K, and the ordinate in
the region between 800 and 1100 cm-1 was multiplied by 5.
Figure 3. Bar graphs comparing vibrational bands calculated for a Cr
metal complex (A) and observed for η3-allyl on Cu(100) (B). The
proposed bonding structures for each case are also displayed.
+ Cl(c) structure. The relatively narrow 886 cm-1 band is
characteristic of a homogeneous environment and suggests an
intimately mixed and ordered overlayer of η3-C3H5(c) + Cl(c).
In particular, between 85 and 160 K, this band red-shifts by no
more than 1 cm-1, the fwhm increases slowly from 7.8 to 11.8
cm-1, and there is no detectable line shape change. The
integrated intensity, (9.5 ( 0.3) × 10-3 cm-1, also shows little
variation. The other bands are too weak for us to assess the
effect of substrate temperature, but there is no evidence for
significant changes. Discussion of the weak band at ∼947 cm-1
is postponed to the Discussion section.
The band assignments in Figure 2 are based on a density
functional theory study of an (η3-allyl)CrBr2 complex41 that
assigns the 887, 966, and 986 cm-1 transitions to Fω(CH2)sym
,
Fω(CH), and ν(CCC)sym modes with A’ symmetry (Cs local
symmetry27). Figure 3 compares the positions and integrated
intensities of the two cases (normalized to the strong band
between 868 and 886 cm-1). The Cu(100) data were fit using
Lorentzian (886 cm-1 peak) and Gaussian (966 and 986 cm-1
peaks) functions.28 Agreement is excellent, lending support to
the η3-allyl structural assignment on Cu(100).
The (η3-allyl)Cr(Br)2 complex (geometry indicated in Figure
3A) contains a delocalized C-C-C framework (C-C-C bond
angle is 122° and C-C bond lengths are ∼1.4 Å) π-bonded
(dashed lines) to the Cr center.41 The plane of the allyl moiety
is canted at an angle of 5-10° with respect to the coordination
polyhedron around the Cr atom. This calculated structure and
bonding configuration is in excellent agreement with well-
established π-bonded allyl complexes, e.g., bis(η3-allyl) nickel.42
The orientation of the η3-allyl on Cu(100) comes from a near-
edge X-ray absorption fine structure (NEXAFS) measurement
that places the C-C-C plane nearly parallel to the metal
Figure 2. Reflection absorption infrared spectra (RAIRS) of a
saturation 1:1 mixture of η3-C3H5(c) and Cl(c) on Cu(100) at temperatures
between 85 and 160 K. The adsorbate layer was prepared by dosing
1.0 L C3H5Cl at 110 K, annealing at 200 K to desorb undissociated
C3H5Cl, and cooling to the indicated temperatures. Spectra were
acquired by adding 1500 scans at 4.0 cm-1 resolution.
dose at 110 K, anneal to 200 K, and cool to indicated T for
RAIRS. While taken at 10 K intervals, only four spectra are
shown. The bands of the 85 K spectrum in Figure 2 support a
previously proposed,9 single-isomer, homogeneously bound η3-
C3H5(c) structure consistent with that found for metal complexes.
The same bands appear upon dissociating allyl bromide on
Cu(100).9 There is little temperature dependence, from which
we conclude that a 200 K anneal followed by cooling to
temperatures between 85 and 160 K leads to the same η3-C3H5(c)
(26) Sheppard, N.; De La Cruz, C. AdV. Catal. 1996, 41, 1-122.
(27) In Cs symmetry, the 18 vibrational modes of an allyl group are
distributed as 10A′ and 8A′′, where A′ represents the totally symmetric
modes and A′′ represents the non-totally symmetric modes. Based on the
surface selection rule, all of the A′ modes are infrared active and the A′′
modes are infrared inactive.
(28) The peaks were fit with a commercial software package: Peak Fit;
Jandel: Chicago, IL, 1991.
(29) Durig, J. R.; Jalilian, M. R. J. Phys. Chem. 1980, 84, 3543-3547.

Synthesis of 1,5-Hexadiene on Cu(100)
J. Am. Chem. Soc., Vol. 123, No. 13, 2001 2993
Figure 5. (Top) TPD spectra of a 1.4 L dose of C3H5Br on (A) ∼0.2
ML Br-covered and (B) clean Cu(100). (Bottom) (C) 1.0 L 1,5-
hexadiene dosed on Br-covered Cu(100) and (D) 2.0 L 1,5-hexadiene
dosed on clean Cu(100). The adsorption temperature for each spectrum
Figure 4. RAIR spectra of allyl bromide (0.6 L) dosed at different
temperatures on Cu(100) saturated with a 1:1 mixture of η3-C3H5(c)
and Cl(c). Reaction of C3H5Br(p) and η3 -C3H5 produces 1,5-hexadiene,
C4H10(p). Bands assigned to various species are marked.
was 110 K, and the heating rate was 2.5 K s-1
.
surface.30 Based on these comparisons and previous work,9 an
analogous bonding and orientation of C3H5(c) on Cu(100) is
proposed (bottom of panel B). In passing, we note that the
intensities of the remaining seven A′ modes lie below the
detection limit of the MCT detector (peak-to-peak noise
∼0.004%).
at 900 cm-1. We ascribe the shoulder to η3-allyl perturbed (red-
shifted) by Br(c).23 Above 160 K, RAIRS indicated no changes
in the η3-allyl intensity or line shape as a result of dosing 0.6
L of allyl bromide.
Further insight is gained (Figure 5) by comparing TPD after
doses at 110 K of C3H5Br (upper panel) and C6H10 (lower panel)
on clean and Br-covered (Br prepared as described in the
Experimental Section) Cu (100). In each case, the dose exceeds
that needed to saturate the first layer. The 121 and 67 amu
signals monitor C3H5Br and C6H10 desorption, respectively. For
a Br-covered surface, C3H5Br appears in TPD after dosing
C3H5Br (spectrum A). The 133 and 146 K peaks are assigned
to the second and first layers of C3H5Br adsorbed on Br-
passivated Cu. Importantly, there is no C6H10 desorption (not
shown).
On the other hand, when C3H5Br is dosed on clean Cu(100),
C6H10 desorbs with peaks at 131 and 148 K (spectrum B). These
peaks align with those from directly dosed C6H10, spectrum C,
indicating that a desorption-limited, rather than reaction-limited,
process is detected in spectrum B. These TPD results are
consistent with RAIRS; C6H10 forms during C3H5Br doses at
110 K on clean Cu(100).9 This highlights an important distinc-
tion in the behavior of C3H5Cl and C3H5Br. Neither RAIRS
nor TPD detected 1,5-hexadiene formation from allyl chloride
on Cu(100) under any conditions. These reactivity differences
are attributable to the C-Cl bond being ∼50 kJ mol-1 stronger
than C-Br.
With the above assessment of the structure taken by the
saturated 1:1 η3-C3H5(c) and Cl(c) layer, we turn to the central
topic of this paper: reaction of C3H5Br with this layer. RAIRS
spectra after 0.6 L doses of C3H5Br, Figure 4, at the indicated
temperature reflect the temperature-dependent loss of η3-C3H5(c)
and gain of 1,5-hexadiene, C6H10(p). Each spectrum involves a
separate surface preparation and was taken at the indicated
temperature by scanning RAIRS for 13.5 min immediately after
dosing allyl bromide. The reference for these spectra is RAIRS
of a Cl(c) layer prepared as described in the Experimental
Section.
At 77 K, bands assigned to η3-C3H5(c) (886 cm-1) and
C3H5Br(p) (1213 cm-1) are strong. As the dosing temperature
rises, the 1213 cm-1 intensity decays to negligible values at
110 K, whereas the 886 cm-1 intensity decays between 77 and
110 K but then recovers and broadens. The decays are
accompanied the emergence of new bands at 915 and 994 cm-1
,
assigned unambiguously to 1,5-hexadiene (C6H10) by compari-
son with 1,5-hexadiene directly dosed on ∼0.2 ML bromine-
covered Cu(100) at 110 K.9 The evidence is clear: at 110 K
some, but not all, of the incident flux of C3H5Br reacts with
η3-C3H5(c) to form C6H10 that remains adsorbed.
In striking contrast to the 110 K results, those at 120 K show
much less 1,5-hexadiene and more η3-allyl. Above 130 K, very
little 1,5-hexadiene is detected, while the 886 cm-1 band of
η3-allyl remains intense but becomes asymmetric with a shoulder
Comparing TPD of directly dosed C6H10, spectra C and D, it
is clear that ∼0.2 ML Br is sufficient to replace the 190 K
desorption with a peak at 148 K that has, within 4%, the same
peak area. The 190 K peak is attributed to C6H10 in contact
with Cu and the 148 K peak to contact with Br(c). In neither
case is there evidence for dissociation of C6H10; i.e., the surface
is clean after TPD and no other hydrocarbons or H2 was detected
(30) Gurevich, A. B.; Teplyakov, A. V.; Yang, M. X.; Bent, B. E.;
Holbrook, M. T.; Bare, S. R. Langmuir 1998, 14, 1419-1427.

2994 J. Am. Chem. Soc., Vol. 123, No. 13, 2001
Celio et al.
in TPD. Finally, the 130 ( 3 K peaks are ascribed to C6H10
adsorbed on itself.
allyl halide bands are present: 933 (dCH2 wag), 988 (dCH2
twist), 1194 cm-1 (dCH2 rock), and 1638 cm-1 (CdC stretch;
not shown). These bands are lower by no more than 3-6 cm-1
than those observed for solid C3H5Br.29 Further, there is no
evidence for dissociation to C3H5(c) and Br(c), a process readily
occurring on clean Cu(100) at 77 K.9 Not surprisingly, in view
of the absence of CO adsorption, we conclude that C3H5Br is
physisorbed onto the η3-C3H5(c)-Cl(c) adlayer, with negligible
interaction with the underlying Cu.
The area of the 148 K peak in spectrum B is 0.65 of that in
spectrum C, while the total C6H10 desorbing in spectrum B is
0.81 of spectrum C. The results in spectrum B are consistent
with the RAIRS data; i.e., reaction to form C6H10 occurs readily
at 110 K. The total amount formed is equal, within 20%, to the
saturation first layer coverage of C6H10 on clean Cu(100),
spectrum D.
As the dosing temperature is increased, the fall and rise of
vibrational intensity associated with η3-C3H5, and the reverse
for intensity associated with C6H10, indicate that the latter’s
formation is controlled by two temperature-dependent processes.
Below 100 K, unreacted allyl bromide accumulates, indicating
that the overall C6H10 formation rate is limited by reaction
between C3H5Br(p) and η3-C3H5(c). At 90 K, for example, only
30% of the initial η3-C3H5(c) reacts in 100 min (not shown).
Between 110 and 160 K, an increasing amount of η3-C3H5(c)
remains after the C3H5Br dose, while the amount of accumulated
C6H10 drops. Since the TPD onset for first-layer C6H10 is 130
K and since the RAIRS C6H10 intensity is 30% lower at 120 K
than at 110 K, we conclude that the reaction becomes controlled
by the C3H5Br(p) residence time; i.e., the desorption rate of
weakly held allyl bromide becomes competitive with its reaction
IV. Discussion
Coverage of η3-Allyl, Cl, and Br. On the basis of the CO
adsorption results of Figure 1, we estimate the coverage of η3-
allyl as follows. Saturation 1:1 C3H5(c) and Cl(c) coverage
precludes adsorption of CO, whereas the clean surface adsorbs
0.57 CO per surface Cu (0.57 ML), all into atop sites.25
Removing C3H5(c) while leaving Cl(c) restores CO uptake
capacity to 0.66 of its saturation value (0.38 ML), from which
we immediately conclude that C3H5(c) blocks more CO than Cl(c).
Assuming Cl(c) occupies a 4-fold site, as it does when Cl2 is
dosed,31 and blocks the four surrounding atop sites, we estimate
that 0.25 ML of Cl(c) will inhibit all CO uptake. Assuming a
linear relation, 0.38 ML of CO implies a Cl coverage of 0.08
ML and, by stoichiometry, an η3-C3H5 coverage of 0.08 ML.
Multiplying by 1.53 × 1015 atoms cm-2, the surface atomic
density of Cu(100), the estimated absolute coverage of C3H5Cl
required is 1.2 × 1014 cm-2. This estimate is reasonable
compared to the coverage (3.7 × 1014 cm-2) required to pack
C3H5Cl into a layer equivalent to the density (0.94 g cm-3) of
liquid C3H5Cl.
rate with η3-C3H5(c)
.
Both RAIRS and TPD support the formation of physisorbed
1,5-hexadiene, i.e., C6H10(p). The peaks in the 110 K RAIR
spectrum are within 5 cm-1 of those found for C6H10 in the gas
phase32 and adsorbed on halogen-covered Cu(100).9 Larger shifts
occur for C6H10 on clean Cu(100); for example, the CH2
wagging mode is at 900 cm-1, a shift of 15 cm-1 compared to
the gas phase (not shown). TPD (Figure 5C,D) shows that the
C6H10 desorption energy is higher on clean Cu(100), 190 K peak,
than on Br-covered Cu(100), 148 K peak.
Extending this CO titration method, the coverage of Cl (from
repeated doses of allyl chloride and annealing cycles) saturates
at 0.20 ML, while Br (from allyl bromide) saturates at 0.22
ML. The latter is consistent with the C6H10 TPD data of Figure
5.
As part of the process that forms C6H10(p), we propose that
the Br released occupies sites previously occupied by η3-C3H5(c)
.
Structure and Reactions of η3-Allyl. The RAIRS results,
Figure 2, show that after dosing, annealing to 200 K, and
recooling to a selected temperature, allyl chloride has either
desorbed or dissociated. We reproduced earlier TPD results30
showing that for higher temperatures, 470 K, 80% of the initial
allyl desorbs; the remaining adsorbed material partially dehy-
drogenates to hydrogenate neighboring species to form and
desorb propylene and an unidentified C3H4 species. The Cl(c)
desorbs as CuClx at ∼950 K.
Such chemical displacement processes are facile on copper
surfaces.33 Based on the peak temperature and a preexponential
factor of 1013 s-1, the desorption activation energy for C6H10
from clean Cu is 49 kJ mol-1, much weaker than the Cu-Br
bond energy, which exceeds 260 kJ mol-1 based on TPD,
showing that CuBr, not Br, desorbs from Br-covered Cu with
a peak at 975 K; i.e., 1,5-hexadiene cannot compete with Br
for Cu sites.
Formation of C6H10 has also has been observed on Ag(111)
and Ag(110). On Ag(110), formation is interpreted in terms of
a Langmuir-Hinshelwood process linking two η3-C3H5(c)
groups34 that, in agreement with calculations,35 is limited by
surface diffusion of η3-C3H5(c). On Ag(111), C3H5Cl and C3H5Br
dissociate between 77 and 200 K to form η3-C3H5(c) that, through
two paths, forms C6H10.23 Above 200 K, diffusion-limited
coupling dominates but, unlike Ag(110), some C6H10 (only from
C3H5Br) forms and desorbs between 140 and 150 K.23 The
Ag(110) features mimic those reported here, but the yield on
Cu(110) is a factor of at least 2 higher for reasons yet to be
established.
As noted above, vibrational modes are assigned to η3-C3H5
and agree with those from dissociation of allyl bromide on clean
Cu(100)9 and with previous studies of allyl halides on coinage
and other transition metal surfaces, as well as organometallic
complexes.26 In addition, there is good agreement between the
relative intensities of the observed modes of allyl on Cu(100)
and those of a Cr complex (Figure 3).41
Comparison of spectra taken at different temperatures indi-
cates that the 947 cm-1 band belongs to the Fω(CH2)sym mode
of η3-allyl groups strongly perturbed by coadsorbed Cl.9 This
mode in other π-bonded hydrocarbons, such as ethylene and
propylene, adsorbed on Cl-modified Ag(111) is sensitive to the
presence of halogen atoms, as evidenced by a >30 cm-1 blue-
shift compared to adsorption on clean Ag(111).17
Allyl Bromide Dosed on η3-C3H5(c) + Cl(c) at 77 K. When
C3H5Br is dosed onto η3-C3H5(c) + Cl(c) at 77 K, molecular
adsorption is clearly evidenced (Figure 4). The band at 1213
cm-1 is the -CH2 wag mode of the bromomethyl group.29 Other
Precursor Model for Kinetics at T > 110 K. In view of
the competition between reaction and desorption of C3H5Br,
we apply a simple precursor model to the RAIRS data, Figure
(32) NIST Chemistry Web Book; http://webbbok.nist.gov/chemistry/.
(33) Kash, P. W.; Yang, M. X.; Teplyakov, A. V.; Flynn, G. W.; Bent,
B. E. J. Phys. Chem. B 1997, 101, 7908-7918.
(34) Carter, N. R.; Anton, A. B.; Apai, G.J. Am. Chem. Soc., 1992, 114,
4410-4411.
(31) Nakakura, C. Y.; Zheng, G.; Altman, E. I. Surf. Sci. 1998, 401,
173-184.
(35) Shustorovich, E. Surf. Sci. 1992, 279, 355-366.

Synthesis of 1,5-Hexadiene on Cu(100)
J. Am. Chem. Soc., Vol. 123, No. 13, 2001 2995
Figure 7. On a Cu(100) surface, the reaction probability, Sr, as a
function of temperature, for the C-C bond-forming reaction that
synthesizes 1,5-hexadiene from physisorbed C3H5Br reacting with
chemisorbed η3-C3H5(c). The dashed line is the calculated reaction
probability. The inset is an Arrhenius plot based on eq 4 for reaction
temperatures between 110 and 160 K, and a linear fit (dashed line) is
also shown.
Figure 6. Schematic one-dimensional potential energy profile for the
reaction, illustrated in the inset, of C3H5Br(g), denoted R(g), with η3-
C3H5(c) to form C6H10(g), denoted P(g). Adsorption first forms R(p) that
is physisorbed over the 1:1 mixture of η3-C3H5(c) and Cl(c). In
competition with desorption, thermal activation of R(p) forms a transition
state, denoted R†, from which product P(p), physisorbed 1,5-hexadiene
adsorbed on Cl(c) + Br(c), is synthesized.
Since RAIRS shows no C3H5Br accumulation above 110 K, it
either reacts with η3-C3H5(c) or desorbs into the gas phase during
the 150 s dose. Under these conditions, the instantaneous
coverage of physisorbed C3H5Br, θp, is very small and a steady-
state approximation is valid, i.e., (dθp/dt) ) 0.0. Under these
conditions, the reaction probability, Sr, is given by the ratio
between the rate of reaction (Rr) and the impinging flux (F).
Straightforward algebraic manipulation gives
6. Along the reaction coordinate, the adsorbed reactant complex,
R(p), defined as C3H5Br(p) physisorbed on η3-C3H5(c) mixed with
Cl(c), passes through a transition state barrier at d ) 0 to form
the adsorbed product complex, P(p), defined as C6H10(p) phys-
isorbed on Cl(c) mixed with Br(c). The inset schematically
illustrates the structural transformation.
Integrated intensities, I, of the 886 cm-1 wagging mode of
η3-C3H5(c) are used to describe the extent of reaction involved
in the fixed doses of C3H5Br at the selected substrate temper-
atures (Figure 4). An average reaction probability (Sr) character-
izing the C3H5Br dose is defined as
kr
R
Sr ) R
)
(3)
(
)
kr + kd
1 + kd/kr
The reaction probability is given by the product of the
trapping probability R and, once trapped, the fraction of C3H5Br
that reacts. Because the stoichiometry is 1:1, Pr is also the
reaction probability of η3-C3H5(c). Taking the Polanyi-Wigner
form for both kr and kd, the temperature dependence of Pr is
given by
(I° - I)
Sr )
(2)
I°
where I° and I are intensities measured before and after dosing
C3H5Br (Figure 4). In calibration experiments based on TPD
of C3H5, we demonstrated that the RAIRS intensity increases
linearly with the η3-C3H5(c) coverage. The expected uncertainty
in Sr is determined by the signal-to-noise ratio (SNR) of the
RAIR band. The average SNR of the CH2 wagging mode of
η3-C3H5 is ∼35 (Figure 2).
R
Sr )
(4)
k(d0)
d - Er)/RT
1 +
e-(E
(0)
( )
kr
For data taken between 85 and 160 K, Sr rises from 0.04 to
0.62 between 85 and 110 K and then falls to 0.03 at 160 K
(Figure 7). At 85 K the reaction rate is neglible; only 4% of the
η3-allyl reacted in 16 min. At 110 K, no allyl bromide
accumulates and, based on Sr ) 0.62, 62% of the initial η3-
allyl coverage (0.08 ML) reacted to form 1,5-hexadiene.
The decay of Sr above 110 K is taken as a signature of a
precursor-mediated surface process analyzable using the fol-
lowing phenomenological rate equations:
Following previous work,6 we assume the trapping probability
is unity and temperature independent. This is not unreasonable
since the average incident translational energy of C3H5Br(g) is
less than one-third of the physisorption well depth for C3H5Br(p).
Using eq 4, a plot of ln[(R/Sr) - 1] as a function of T-1 is
nicely linear (inset of Figure 7), and from the slope, Ed - Er )
12 kJ mol-1; from the intercept, kd(0)/kr(0) ) 9.4 × 104. These
values were used to construct the dashed curve through the data
(and the extrapolation) in the body of Figure 5. Using Redhead
analysis36 with kd(0) ) 1013 s-1, and data for TPD of C3H5Br(p)
on ∼0.2 ML Br-covered Cu(100) (Figure 5), the calculated
desorption activation energy is 38 kJ mol-1. This is a suitable
estimate for Ed in the case where desorption occurs from η3-
C3H5(c) mixed with Cl(c), i.e., kd ) 1013 exp{-38 kJ mol-1/RT}
s-1. From this, kr ) 1.1 × 108 exp{-26 kJ mol-1/RT} s-1. Table
1 summarizes the rate parameters for desorption and reaction
C3H5Br(g) - RF f C3H5Br(p)
C3H5Br(p) - kd f C3H5Br(g)
C3H5Br(p) + η3-C3H5(c) - kr f C6H10(p) + Br(c)
where R is the trapping probability of C3H5Br into a physisorbed
state and F is the incident flux of C3H5Br. The rate coefficients,
kd and kr, describe C3H5Br desorption and reaction, respectively.
(36) Redhead, P. A. Vacuum 1962, 12, 203-211.

2996 J. Am. Chem. Soc., Vol. 123, No. 13, 2001
Table 1. Experimentally Determined Rate Parameters
Celio et al.
Scheme 1
Ed
Er
a
adsorbate
(kJ mol-1
)
k(d0) (s-1
)
(kJ mol-1
)
k(r0) (s-1
1 × 108
)
allyl bromide
1,5-hexadiene
η3-allyl
38
39
107
250
260
1013
1013
1013
1013
1013
26
241.4c
nad
na
na
6 × 1014
c
na
na
na
b
Cl(c)
Br(c)b
a Assumed values. b Reference 39. c Central C-C bond scission in
gas phase; ref 37. d na, not applicable.
including parameters for the formation of allyl radicals from
the thermal decomposition of 1,5-hexadiene37 and the desorption
energies of chemisorbed Br and Cl atoms from Cu(100).38
As anticipated for precursor-assisted reactions,39 the experi-
mental reaction probability (Figure 7) decreases as the surface
temperature increases since, with respect to the bottom of the
physisorption potential energy well, the barrier to desorption
(Ed ) 38 kJ mol-1) is about the same as the barrier to reaction
with η3-C3H5(c) (Er ) 26 kJ mol-1). In a one-dimensional
representation (Figure 6), the activated complex, R† ) C3H5Br
‚‚‚η3-C3H5(c), lies 26 kJ mol-1 above R(p).
The reaction probabilities in Figure 7 indicate that the reaction
does not occur via a direct mechanism; i.e., C3H5Br does not
react on the inbound trajectory. Rather, energy is transferred to
a trapped adsorbate state, and a reaction, competing with
desorption, occurs from this state. The fit to the extrinsic
precursor model is certainly adequate.
not allow determination of whether electron transfer, bond
breaking, and bond formation are concerted or stepwise, we
propose that the surface reaction reported here involves an
electron transfer from η3-C3H5(c) to the C-Br bond of C3H5Br(p)
that results in the transient formation of C3H5Br-. The latter
reacts, with significant probability, with the surface η3-C3H5(c)
to form 1,5-hexadiene, C6H10(p), and Br(c). The physisorbed state
of 1,5-hexadiene is the result of chemical displacement by the
more aggressive bromine atom, as discussed above.
IV. Conclusion
The reaction between chemisorbed η3-allyl, C3H5(c), and
transiently physisorbed allyl bromide, C3H5Br(p), to form ad-
sorbed 1,5-hexadiene, C6H10(p), has been characterized by
RAIRS and TPD. Between 110 and 160 K, the kinetics are
described in terms of an extrinsic precursor trapping-mediated
model. As expected for such a model, the reaction probability
decreases rapidly with increasing substrate temperature. The
difference in the activation energies for desorption and reaction
of C3H5Br(p) (Ed - Er) is 12 kJ mol-1. We expect this to be
generalizable to other metal surfaces under conditions where
η3-C3H5(c) forms and the activation energy for reaction with
electrophiles, e.g., organic halides, is low enough to compete
with alternative reaction paths, e.g., desorption and decomposi-
Proposed Electron-Transfer Mechanism. The reaction
described here has analogues in organometallic chemistry:
transition metal-catalyzed carbon-carbon bond-forming reac-
tions between organometallic nucleophiles and organic halide
electrophiles are common (Scheme 1).40 These proceed by
electron transfer from the nucleophile to the carbon-halogen
bond and subsequent formation of a radical anion intermediate.
For example, organic halides (RX) react with (η3-allyl)-nickel
halides in a so-called nucleophilic radical chain reaction, SRN1,
mechanism that involves the short-lived (∼1 µs) radical anion,
RX-.41 While the time scale of the RAIRS measurements does
(37) Roth, W. R.; Bauer, F.; Beitat, A.; Ebbrecht, T.; Wu¨stefeld, M.
Chem. Ber. 1991, 124, 1453-1460.
(38) Nakakura, C. Y.; Phanse, V. M.; Altman, E. I. Surf. Sci. 1997, 370,
L149-L157.
tion of η3-C3H5(c)
.
(39) Rettner, C. T.; Ashfold, M. N. R. Dynamics of Gas-Surface
Interactions; Royal Society of Chemistry: London, 1991.
(40) Hegedus, L. S.; Thompson, D. H. P. J. Am. Chem. Soc. 1985, 107,
5663-5669.
(41) Swang, O.; Blom, R. J. Organomet Chem. 1998, 561, 29-35.
(42) Goddard, R.; Kru¨ger, C.; Mark, F.; Stansfield; R.; Zhang X.
Organometallics 1985, 4, 285-290.
Acknowledgment. This work was supported in part by the
Robert A. Welch Foundation and by the U.S. Department of
Energy, Office of Basic Energy Sciences. We also thank
Professor Buddie Mullins and C. Reeves for helpful discussions.
JA0011414