Chemistry of methoxonium on (2 × 1)Pt(110)

Article abstract of DOI:10.1021/jp011184m

The chemistry of methoxonium was studied on (2 × 1)Pt(110). Methoxonium formed when hydrogen and methanol were coadsorbed on the surface. Three reactions occurred on the surface, i.e., methanol dehydrogenation, methanol dehydration, and dimethyl ether formation. Methoxonium was important intermediates for the last two reactions. Methanol dehydration reaction involved methoxonium decomposition as a rate-determining step. Dimethyl ether formation was improved by increasing the predosed hydrogen amount. Methoxonium on surfaces was a true cation reacting almost identically to a cation in solution.

Full text of DOI:10.1021/jp011184m

J. Phys. Chem. B 2001, 105, 8583-8590
8583
Chemistry of Methoxonium on (2 × 1)Pt(110)
Chang Lu, Fred S. Thomas, and Richard I. Masel*
Department of Chemical Engineering, UniVersity of Illinois, Urbana, Illinois 61801
ReceiVed: March 29, 2001; In Final Form: June 27, 2001
In a previous paper, (Chen, N.; Blowers, P.; Masel, R. I. Surf. Sci. 1998, 418, 329-341.) we found that
+
methoxonium ions (CH
we characterize the chemistry of methoxonium on (2 × 1)Pt(110) using mainly TPD. We observe three
reaction pathways: a simple dehydrogenation to CO and hydrogen; an S 1 pathway to form water, methane,
and adsorbed CH groups; and an S 2 pathway yielding dimethyl ether. The S 1 pathway shows a strong
secondary isotope effect implying that the transition state is ionic. The S 2 pathway is seen at high hydrogen
coverages and shows kinetics typical for an S 2 process. This work demonstrates for the first time that cationic
species on surfaces react via the same S 1 and S 2 pathways seen during reactions of cations in solution.
3 2
OH ) can form when methanol and hydrogen coadsorb on Pt(110). In this paper,
N
x
N
N
N
N
N
N
Introduction
In recent years, a number of investigations have demonstrated
We will demonstrate that methoxonium shows similar chemistry
to its behaviors in solution. Methoxonium can react via an SN1
or an SN2 pathway. A secondary isotope effect is observed
during an SN1 step, suggesting that methoxonium on the surface
is fully charged. We will also examine what role hydrogen plays
during the reactions. These results are the first clear evidence
that cationic pathways are important during reactions on
transition-metal surfaces.
1
-6
that cationic species can exist on transition-metal surfaces.
1
3
5
Wagner et al., Kizhakevariam et al., and Chen et al. suggested
that a species with an HREELS spectrum characteristic of
hydronium ions forms during water and hydrogen coadsorption
4
on Pt(111), Pt(100), and Pt(110). Chen et al. and Thomas et
al. provided evidence that a new species formed after adsorbing
6
methanol on hydrogen pre-covered (2 × 1)Pt(110) and then
annealing the crystal to 190 K. The species was assigned to
Experimental Section
+
methoxonium ions (CH3OH2 ) on the basis of a comparison of
The experiments were done using the apparatus and proce-
dures described previously. The UHV system in which all the
TPD work was done was of standard design with a working
measured HREELS spectrum to those calculated for methoxo-
8
7
nium and several other possible radical species.
Still, up to this point, very little is known about the properties
of cationic species on transition-metal surfaces. Prior to our
work, no investigation showed how much charge the surface
cations carry. Are they only slightly charged or fully charged
as most cations in liquid phase are? Also, it is not clear if the
cationic species on transition-metal surfaces are only spectators
or intermediates of some important reactions. If they are
intermediates, how do they participate in the reactions? How
do ionic pathways on transition-metal surfaces compare with
similar reactions in liquid phase? Methoxonium reacts via two
different reaction pathways in liquids, an SN1 pathway:
-10
base pressure of 1 × 10 Torr. The system was equipped with
a PHI 4-161 sputter gun, a PHI 10-155 cylindrical mirror
analyzer for Auger electron spectroscopy (AES), a UTI-100C
quadrupole mass spectrometer, and an RVL6-120 reverse low-
energy electron diffraction (LEED) system. A Pt(110) single-
crystal sample was cut from a Metron single-crystal rod. The
sample was then polished with diamond paste and mounted in
a UHV system. The sample was then oxidized, sputtered, and
annealed until no impurities could be detected by AES. A sharp
(
2 × 1) LEED pattern was seen at this stage. We also checked
12-14
the surface structure with H2 TPD because previous work
+
+
shows that the H2/TPD spectrum is very sensitive to the surface
order. The clean (2 × 1)Pt(110) surface yielded thermal
desorption spectra (Figure 8) for hydrogen in excellent agree-
(
CH OH ) f (CH ) + H O
(1)
3
2
3
2
and an SN2 pathway:
CH OH + (CH OH ) f (CH OHCH ) + H O (2)
1
2-14
ment with the TPD spectrum expected
for a well-ordered
+
+
(2 × 1)Pt(110) surface.
3
3
2
3
3
2
Methanol (CH3OH, 99.9%, Aldrich) and methyl-d3 alcohol
CD3OH, 99 atom % D, Aldrich) were freeze-thaw degassed
(
The chemistry of cations on transition-metal surfaces would be
conceptually important and very helpful for explaining surface
reactivity of organic molecules.
and then used without further purification. During a TPD run,
the sample was first cleaned, and its cleanliness was verified
by AES and sharp (2 × 1) LEED pattern. The surface was then
dosed with a measured amount of hydrogen and methanol (or
deuterated methanol CD3OH) through a capillary array. Hy-
drogen was adsorbed on the surface first unless otherwise
described. Next, the sample was heated with a fixed rate of 15
K/s under computer control, and a TPD spectrum was recorded.
6
In a previous paper, we provided evidence that methoxonium
can react via an SN1 pathway on Pt(110)(2 × 1). In this study,
we will expand the findings. In particular, we will examine the
reactivity of methoxonium on (2 × 1)Pt(110) using mainly TPD.
*
To whom correspondence should be addressed.
1
0.1021/jp011184m CCC: $20.00 © 2001 American Chemical Society
Published on Web 08/21/2001

8584 J. Phys. Chem. B, Vol. 105, No. 36, 2001
Lu et al.
simultaneously with increasing methanol coverage suggesting
that water, methane, and adsorbed carbon are formed via a
similar reaction pathway. We have also discovered an effect of
ordering. A well-annealed surface produces less water and
methane than a surface that was sputtered and then annealed
for only a few minutes.
The data in Figure 1 shows that methane and water form
during hydrogen and methanol coadsorption on (2 × 1)Pt(110).
However, we do not know whether the methane and water are
formed via an SN1 pathway or some other process.
2
2,23
Fortunately, there is a standard test
that we can do to tell
whether the reaction goes via an SN1 pathway or some other
process.
The test is based on looking for a secondary kinetic isotope
effect. Consider the two overall reactions:
CH OH + H f CH + H O
(7)
(8)
3
2
4
2
CD OH + H f CD H + H O
3
2
3
2
If reactions 7 and 8 go via reactions 3, 4, 5, and 6 with reaction
rate-determining, the hybridization of the carbon atom will
3
Figure 1. A composite TPD spectra taken by exposing a clean (2 ×
)Pt(110) to 5 L of hydrogen, adding 1 L of methanol at 110 K. Heating
rate is 15 K/s. The scale factors in the figure are the raw scale factors,
uncorrected for differences in mass spectrometer sensitivity.
3
2
1
change from sp to sp during the rate-determining step. The
change in hybridization shifts the wagging modes of the
hydrogen in a way that lowers the barrier to reaction. The
lowering of the barrier is smaller with deuterated species.
Consequently, if reactions 7 and 8 occur via steps 3, 4, 5, and
6 with reaction 3 rate-determining, one would expect reaction
8 to be slower than reaction 7. In contrast, if reaction 3 is not
rate-determining, or if the overall reaction goes via an adsorbed
Results: Evidence for an SN1 Pathway
First, we will try to look for an SN1 pathway. Recall that if
an adsorbed methoxonium undergoes an SN1 reaction on
platinum, one would expect to produce water and methyl groups.
3
CH3 radical or some other species where the sp hybridization
+
+
(
[CH OH ] ) f ([CH ] ) + H O
(3)
(4)
of the CH3 is preserved, one would not expect to observe a
secondary kinetic isotope effect.
3
2 ad
3 ad
2
+
-
23
Experimentally, Maskill finds that reactions which go via
(
[CH ] ) + e f [CH ]
3 ad 3 ad
an SN1 pathway through a fully charged transition state show a
secondary kinetic isotope effect of a factor of 1.2 to 1.3 at 300
K and a factor of 1.4 to 1.8 at 150 K. In contrast, reactions
going via an SN2 pathway, or those involving neutral species,
usually show a rate that changes by at most a factor of 1.05.2
Consequently, one can use the presence of a secondary kinetic
isotope effect as strong evidence for an SN1 reaction.
The methyl groups can then be either hydrogenated or dehy-
drogenated.9
,10
2-25
[
CH ] + H f CH
(5)
(6)
3
ad
ad
4
[
CH ] f C + 3H
3 ad ad
ad
Deuterated methanol (CD3OH) was used in coadsorption with
hydrogen (H2) in replacement of normal methanol (CH3OH).
Figure 2 shows all the products. The 35 amu spectrum contains
a sharp peak at 150 K, a shoulder extending up to 240 K, and
a second peak at 320 K. The 18 amu spectrum shows a peak at
150 K corresponding to a cracking fraction of CD3OH and a
smaller peak at 180 K associated with water desorption. Analysis
on an expanded scale shows that a tiny peak is present at about
220 K corresponding to a cracking fragment of CD3H. The 19
amu spectrum shows a peak at 180 K of HOD desorption and
a feature at 230 K corresponding to CD3H desorption. H2 as
well as a small amount of HD and D2 were found coming out
of the surface. There is no deuterated dimethyl ether detected
at the particular coverages and sensitivities shown. Figure 3
shows a closeup comparison of yield in CD3H and CH4, which
are from the deuterated methanol reaction and the normal
methanol reaction, respectively. The CH4 (16 amu) peak is about
1.8 times larger than the CD3H (19 amu) peak. Another
comparison is made between water yields from the two reactions
in Figure 4. The difference spectrum shows the water peak left
after the deuterium methanol cracking is deducted. There is also
some amount of HOD from deuterated methanol reacting with
hydrogen via some exchange mechanism. After taking all these
factors into consideration, the ratio between water yield from
Therefore, if an SN1 reaction occurs, one should observe
methane and water desorption from the surface.
In previous work, we found that no methane and water
production is observed during methanol decomposition on a
1
5-17
clean (2 × 1)Pt(110) sample.
However, methane and water
production is observed during methanol and hydrogen coad-
6
sorption on (2 × 1)Pt(110). Here, we have reproduced the
findings. Figure 1 shows a composite TPD spectrum taken after
coadsorption of 5 L hydrogen and 1 L methanol on (2 × 1)-
Pt(110). There are six species coming out of the surface during
the thermal desorption: methanol, hydrogen, CO, water,
methane, and dimethyl ether. Methanol partially desorbs mo-
lecularly at 150 K. Hydrogen has a broad desorption peak with
the range from 220 to 300 K. CO desorbs at 520 K. A water
peak shows up at 174 K. Three major peaks can be seen in the
1
6 amu spectrum: A methanol desorption peak and a water
desorption peak are merging with each other around 170 K,
methane comes out at 230 K, and CO desorption takes place
around 520 K. There also is a tiny dimethyl ether peak at 175
K. AES also reveals that carbon contamination is left on the
surface after each run.
In other work, we have found that at modest hydrogen
coverages, the water, methane, and adsorbed carbon grow

Chemistry of Methoxonium on (2 × 1)PT(110)
J. Phys. Chem. B, Vol. 105, No. 36, 2001 8585
2
Figure 4. A comparison of H O yield (18 amu) during the coadsorption
of CH (a) during the coadsorption of deuterium methanol
3
OH and H
2
(
CD OH) and H (b). (b) is a difference spectrum taken by starting
3
2
Figure 2. A composite TPD spectra taken by exposing a clean (2 ×
3 2
with the 18 amu spectrum for coadsorption of CD OH and H (c), and
then subtracting away the expected cracking fraction of CD OH.
1
)Pt(110) to 5 L of hydrogen, adding 1 L of deuterium methanol (CD
3
-
3
OH) at 110 K. Heating rate is 15 K/s. The scale factors in the figure
are the raw scale factors, uncorrected for differences in mass spec-
trometer sensitivity.
Figure 5. The effect of methanol exposure on the yield of methane
(9) and water (2). The methane and water peaks were determined by
TPD. The exposures were estimated from uncalibrated ion gauge
readings.
are shown in Figure 5. The data shows a linear production rate
of methane and water as a function of the methanol coverage,
as expected for an SN1 pathway.
Results: SN2 Pathway
Next, we will look for evidence for an SN2 pathway during
methoxonium decomposition on (2 × 1)Pt(110). Recall that if
an adsorbed methoxonium undergoes an SN2 reaction on
platinum, one would expect to produce water and dimethyl ether.
Figure 3. A comparison of the methane peak (a) at 16 amu produced
during the coadsorption of CH
amu produced during the coadsorption of deuterium methanol (CD
OH) and H
3 2 3
OH and H to the CD H peak (b) at 19
3
-
2
.
+
+
CH OH + ([CH OH ] ) f ([CH OHCH ] ) + H O
3
ad
3
2 ad
3
3 ad
2
these two reactions is around 1.8 to 2.0, which is close to the
ratio between methane and CD3H yielded.
(9)
The implication of Figures 3 and 4 is that reaction 7 shows
a secondary isotope effect, suggesting that the reaction occurs
via an SN1 pathway. This idea will be amplified in the
discussion.
We have also measured the rate of methane and water
production as a function of the methanol exposure. The data
-
+
e + ([CH OHCH ] ) f ([CH OCH ] ) + H
(10)
3
3 ad
3
3 ad
ad
Further, the rate should increase monotonically with increasing
hydrogen coverage and increase nonlinearly with increasing
methanol coverage. We observe dimethyl ether formation in

8586 J. Phys. Chem. B, Vol. 105, No. 36, 2001
Lu et al.
Figure 6. The effect of predosed hydrogen on the yield of dimethyl
ether. In the figure, the X axis is the amount of hydrogen dosed onto
the sample. The Y axis is the dimethyl ether yield (9). The amount of
1 2
â hydrogen determined from separate H TPD experiments is also
shown (2). During the experiments, the sample was dosed with various
amounts of hydrogen, 1 L of methanol was added, and then the sample
was heated at 15 K/s.
Figure 8. A TPD spectra taken by exposing hydrogen (0.2 L ∼ 60 L)
to a clean (2 × 1)Pt(110) at 110 K. Heating rate is 15 K/s.
CO as a coverage reference. We find that the dimethyl ether
yield varies nonlinearly with the methanol coverage. Little
dimethyl ether formation is seen at methanol coverages below
1
4
2
2
× 10 molecules/cm , but the yield grows rapidly at higher
14 2
coverages up to coverages 5 × 10 molecules/cm .
Figure 7. The effect of methanol coverage on the yield of dimethyl
ether. In this figure, the yield was determined by integrating the area
under the dimethyl ether TPD peak. The methanol coverage was
determined by AES using a saturated layer of CO as a coverage
standard.
The results in Figures 6 and 7 show that the dimethyl ether
product expected for the SN2 pathway is observed during
methanol and hydrogen coadsorption on (2 × 1)Pt(110), and
the reaction shows the nonlinear behavior expected for an SN2
pathway (see the discussion).
Figure 1, suggesting the possibility of an SN2 pathway. Next,
we wish to take the critical data to see if the SN2 pathway
actually occurs.
To illustrate the role that hydrogen plays during the reactions,
the hydrogen desorption spectra in Figure 8 were also obtained.
Figure 8 is in very good agreement with previous literature.1
Three different adstates for hydrogen show up in the spectra.
â2 appears with a constant peak temperature around 307 K and
saturates around 1 L. â1 begins to show up when exposure is
higher than 1 L, and its peak temperature varies in the range
from 270 to 230 K when exposure goes up to 60 L. We also
have an R peak at exposure beyond 20 L. In Figure 9, spectrum
(a) is taken after 2 L of hydrogen is adsorbed on the clean Pt-
(110) surface alone. After the reaction between hydrogen and
methanol (b) takes place, another spectrum (c) was taken for 2
L hydrogen desorption from the reaction-contaminated surface.
It is interesting that the â2 hydrogen peak disappears in spectrum
2-14
Figure 6 shows how the yield of dimethyl ether varies with
the amount of hydrogen dosed onto the sample prior to the run.
The amount of â1 hydrogen seen in H2 exposure is also included
for reference. In the figure, the H2 coverage was determined
by integrating the area under the H2 TPD peak, using a saturated
layer of H2 as a coverage reference.11 The dimethyl ether yield
was determined by integrating the area under the dimethyl ether
TPD peak. We report the yield of dimethyl ether in arbitrary
units, since the absolute yield of dimethyl ether also varied with
how carefully the sample was annealed. Qualitatively, the
maximum dimethyl ether in Figure 6 corresponds to between 1
1
3
13
2
×
10 and 4 × 10 molecules/cm of dimethyl ether.
(c). In contrast, the â1 peak does not change significantly.
Notice that the dimethyl ether peak seems to track the â1
hydrogen peak, suggesting that â1 hydrogen is involved in the
process to produce dimethyl ether.
Discussion
We have also examined the variation in the water and
methane peaks as the hydrogen exposure changes. The size of
the water and methane peaks change when we vary the hydrogen
The results above show four overall reactions happening on
the surface.
14
2
exposure from 0 to 5 × 10 molecules/cm , but there are only
CH OH f CO + 2H
(11)
(12)
(13)
(14)
3
2
small increases in water and methane yield when we vary
1
4
2
CH OH + H f CH + H O
3 2 4 2
hydrogen coverage from 5 × 10 molecules/cm to saturation.
Figure 7 shows how the rate of dimethyl ether formation
varies with the methanol coverage. In the figure, we determined
the methanol coverages using AES, using a saturated layer of
CH OH f C + H + H O
3 ad 2 2
2
CH OH f CH OCH + H O
3 3 3 2

Chemistry of Methoxonium on (2 × 1)PT(110)
J. Phys. Chem. B, Vol. 105, No. 36, 2001 8587
1
4
2
up to coverages of 5 × 10 molecules/cm . This suggests that
the methanol is somehow activated by the presence of hydrogen.
We can speculate how hydrogen catalyzes the reaction. Acidic
hydrogens are known to catalyze reactions 12 and 13 in solution.
The protons react with methanol to form a methoxonium ion
+
(
CH3OH2) .
+
+
H + CH OH f (CH OH )
(15)
3
3
2
The methoxonium then undergoes an SN1 reaction
+
+
(
CH OH ) f (CH ) + H O
(16)
3 2 3 2
We observe an EELS spectrum characteristic of methoxonium.4
Therefore, it is possible that water and methane form via a
reaction pathway similar to that in reactions 15 and 16, that is,
+
-
CH OH + H(ad) f ([CH OH ] ) + (e )
(ad)
(17)
(18)
3
(ad)
3
2
(ad)
+
+
(
[CH OH ] ) f [CH3(ad)] + H O
3 2 (ad) 2
Figure 9. (a) TPD spectrum of 2 L hydrogen on a clean (2 × 1)Pt-
(
2
110). (Tads ) 110 K, â ) 15 K/s). (b) TPD spectrum taken by exposing
L hydrogen and then adding 1 L methanol to a clean (2 × 1)Pt(110)
+
-
[
CH3(ad)] + (e ) f CH3(ad)
(19)
(ad)
at 110 K. Heating rate is 15 K/s. (c) TPD spectrum of 2 L hydrogen
on a reaction-contaminated (2 × 1)Pt(110) surface (after a run of (b)).
(Tads ) 110 K, â ) 15 K/s).
CH3(ad) + H(ad) f CH4
(20)
(21)
Reaction 11 is the primary pathway observed previously.8,15-20
It is observed even when there is no hydrogen on the surface.
CH3(ad) f C(ad) + 3H(ad)
8
Reaction 12 is the primary reaction pathway on (1 × 1)Pt(110).
Here, we observe the reaction pathway on (2 × 1)Pt(110), but
only when there is significant hydrogen on the surface. Even
then, the pathway is small. At the best conditions, we observe
only about 10% as much water and methane production on (2
The data in Figures 1-4 provide strong support for this
mechanism. We observe a factor of 1.9 reduction in the net
water production and a factor of 1.8 reduction in the net methane
production when we switch from CH3OH to CD3OH. In the
latter case, our main products are H2O and CD3H. We do not
detect CD4, and we detect only a small amount of HOD.
Therefore, we are changing the rate by deuterating the methyl
group. Such a change is characteristic of a secondary kinetic
isotope effect.
×
1)Pt(110) as on (1 × 1)Pt(110). Reactions 13 and 14 are
smaller reaction pathways. We postulate them because we
observe residual carbon in AES and dimethyl ether formation
in TPD. Generally, reaction 11 is the fastest reaction in this
system. Reactions 12 and 13 occur at about an order of
magnitude slower than Reaction 11. Reaction 14 only occurs
at high hydrogen coverages, and even then, it is more than 2
orders of magnitude slower than reaction 11.
Next, we wish to consider the implications of a secondary
kinetic isotope effect of a factor of 1.8. Secondary kinetic isotope
effects are subtle. When we switch from CH3OH to CD3OH,
we are not changing the molecule significantly. According to
Reaction 11 is a common pathway for methanol decomposi-
tion on Pd or Pt single-crystal surfaces. Wang and Masel found
that the methanol first decomposes to methoxy (CH3O) and
hydrogen atoms, and then the methoxy sequentially loses its
hydrogen to form CO.1
2
3
24
Maskill and Collins and Bowman, SN1 reactions with fully
charged transition states with the charge on the carbon show
large secondary kinetic isotope effects, but other reactions which
preserve the hybridization of the CH3 do not show large
secondary isotope effects. At room temperature, Banks et al.25
measure a factor of 1.2-1.3 change in rate for a series of SN1
reactions where a fully ionized carbocation forms. In contrast,
SN2 reactions and reactions involving an uncharged species show
at most a factor of 1.05 change in rate because of a secondary
5-17
Reactions 12 and 13 are less common. These reactions were
reported previously when methanol is absorbed on (1 × 1)-
8
Pt(110). There also have been some reports of reactions 12
and 13 occurring on Pd(111),1
9,20
but other investigators did
18
not reproduce the findings. Reactions 12 and 13 involve C-O
bond scission and further dehydrogenation to step produce
carbon contamination (Cad).
2
2,24-25
kinetic isotope effect.
Banks et al. do not present any
measurements below 274 K, so one does not know how large
of an effect will occur. However, if we extrapolate their data
Reaction 14 has not been reported on any transition-metal
surfaces, but it is commonly observed on strong Lewis acids
2
3
using the model in Maskill, we find that one would expect a
factor of 1.4-1.8 change in rate if the dehydration process goes
via an SN1 reaction with a fully ionized carbocation intermediate.
We observe a factor of 1.8 change in rate. Therefore, it is clear
that the rate-determining step in methanol dehydration on
Pt(110) is an SN1 process with a cationic transition state starting
with a cationic intermediate. Further, even if there is an error
in our measurements and the change in rate were only a factor
of 1.4, we would still conclude that the rate-determining step
2
1
such as alumina. Here, we observe a tiny amount of dimethyl
ether formation, which suggests the possibility that (2 × 1)-
Pt(110) has some Lewis acid character.
Next, we want to consider the mechanism of reactions 12
and 13. Reactions 12 and 13 are not seen during methanol
decomposition on a clean (2 × 1)Pt(110) surface. They are
observed on a hydrogen-covered surface, and the methane and
water yield increases monotonically with the hydrogen coverage

8588 J. Phys. Chem. B, Vol. 105, No. 36, 2001
Lu et al.
in methane formation on Pt(110) is an SN1 process with a
cationic transition state.
an E2 mechanism via a methoxonium intermediate. One can
postulate a similar mechanism here
The data in Figure 5 are also consistent with an SN1 reaction
pathway. Notice that the rate of methane and water production
is linear with exposure, as expected for an SN1 reaction. In
solution, SN1 reactions show linear dependences at low con-
centrations and nonlinear dependences at high concentrations.
Evidently, all of our data are taken in the linear regime.
+
-
CH OH + H(ad) f ([CH OH ] ) + (e ) (23)
(ad)
3
(ad)
3
2
(ad)
+
(
[CH OH ] ) + CH OH
f
3
2
(ad)
3
(ad)
+
([CH
OHCH
] )(ad) + H
O (24)
3
3
2
Next, it is useful to note that the secondary isotope effect
occurs upon deuterating the carbon. One would observe a
secondary isotope effect only if the transition state had a positive
charge on the carbon. Consequently, the transition state must
be a true carbocation and not simply a cation with charge
elsewhere on the molecule.
+
-
([CH OHCH ] ) + (e ) f CH OCH + H
(ad)
(25)
3
3
(ad)
(ad)
3
3
Our data can be explained if dimethyl ether forms when the
reaction occurs via reactions 23, 24, and 25. Figure 6 implies
that the reaction is catalyzed by hydrogen. The methoxonium
cation intermediate ([CH3OH2] )(ad) is observed in spectroscopy
and the dimethyl ether product is seen in Figure 2. Further,
Figure 7 shows that the dimethyl ether peak varies nonlinearly
with the methanol coverage, as expected for an SN2 or an E2
pathway.
We have not been able to find a radical mechanism that would
explain the data. Recall that our previous data shows that
increases in the hydrogen coverage decrease the amount of
methanol that decomposes, which decreases the availability of
+
4
Next, we want to consider the likely intermediates during
the reaction. Note that the rate-determining step in the dehydra-
tion process must produce methyl cations or closely related
species. We observe a factor of 1.8 change in rate of water
formation upon deuteration of the methyl groups. That would
+
only occur if CH3(ad) was forming during the rate-determining
step of the dehydration process.2 Similarly, we suggest that
water is probably produced in the rate-determining step of the
dehydration process. We observe water desorption at 180 K. In
contrast, when we coadsorb water and methanol, the water
desorbs at about 170 K. As a result, the formation of water in
the data in Figure 2 must be largely reaction-limited. Therefore,
it seems likely that water is being produced in the rate-
determining step of the methanol dehydration process.
2,24
8
chemisorbed radicals such as methoxy (CH3O), formaldehyde
-11
(
H2CO), or formyl (HCO). The concentration of CH39
decreases as well because the hydrogen reacts quickly with
methyls. Consequently, increases in the hydrogen coverage
should decrease the concentrations of chemisorbed radicals,
which should suppress the reaction pathways involving these
species. Yet the data here shows that the yield of dimethyl ether
increases with increasing hydrogen coverage. Therefore, we
conclude that the formation of dimethyl ether does not proceed
via an adsorbed methoxy (CH3O), formaldehyde (H2CO), formyl
One can imagine a series of possible SN1 reactions forming
water, methyl cations, and possibly other species. All of the
+
+
reactions must be of the form (CH3OH2 )R f CH3(ad) + H2O
+
+
R to yield CH3(ad) and H2O. Only water, methanol, and a
trace of dimethyl ether desorb below 200K. Consequently, there
are not very many possible reactions.
(
HCO), or methyl (CH3) intermediate.
In contrast, the concentration of methoxonium cations
The only feasible possibilities that we have been able to find
are
increases with increasing hydrogen exposure as expected from
the mechanism in reactions 23-25. A mechanism where
methoxonium reacts via an SN2 pathway to form dimethyl ether
is consistent with the data in Figure 6. A related E2 reaction
would also be consistent with the data.
+
[
CH OH ]+ f CH
+ H O
(18)
3
2(ad)
3(ad)
2
+
[
(CH OH) (CH OH ) ]+ f CH
+ H O + nCH OH
However, methoxonium could also conceivably react via an
SN1 pathway to yield dimethyl ether, that is,
3
n
3
2 (ad)
3(ad)
2
3
(22)
CH OH ]+ f CH
+
+ H O
(18)
(26)
(25)
[
3
2(ad)
3(ad)
2
Both reactions start with methoxonium-like intermediates.
Therefore, the TPD data provide strong evidence that the rate-
determining step during methanol dehydration on Pt(110) is an
SN1 process starting with a methoxonium intermediate or some
closely related species. We postulate that the mechanism goes
through steps 17-20. We have spectroscopic evidence for
reaction 17 and kinetic evidence for reaction 18. We do not
have any direct evidence for reaction 19, but methyl groups
are such strong electron acceptors that thermodynamically it
does not seem likely that they would stay charged on Pt(110).
Reactions 20 and 21 have been observed previously by White
and co-workers.9 Therefore, we suggest that methane forma-
tion occurs via an SN1 pathway on Pt(110).
+
+
(
[CH ] ) + CH OH f ([CH OHCH ] )
3
(ad)
3
(ad)
3
3
(ad)
+
-
(
[CH OHCH ] ) + (e ) f CH OCH + H
3
3
(ad)
(ad)
3
3
(ad)
The evidence, however, shows that this is not the pathway for
dimethyl ether formation. First, the kinetics are wrong. In
solution, for example, SN1 reactions show first-order dependence
on the reactant concentration. Figure 7 shows that the depen-
dence is clearly second or higher order on the methanol
,10
+
concentration. Second, a pathway via a ([CH3] )(ad) intermediate
seems unlikely since ([CH3] )(ad) has such a high electron
+
Next, we wish to change topics and consider dimethyl ether
formation (reaction 14). Figure 6 shows that yield increases
dramatically as we add hydrogen to the surface prior to dosing
methanol. Therefore, we conclude that hydrogen somehow
catalyzes dimethyl ether formation on (2 × 1)Pt(110). We can
speculate how the process occurs. In solution, acidic hydrogens
are known to catalyze dimethyl ether formation via an SN2 or
affinity that it should be quickly neutralized on the surface to
form ([CH ])(ad). Coadsorption of CH3(ad) and CH OH does not
3
3
8
produce significant dimethyl ether on (1 × 1)Pt(110). Further,
9,10
White has found that methyl groups are quickly hydrogenated
on Pt(111) so that if dimethyl ether were formed by reactions
18, 25, and 26, one would expect hydrogen additions to decrease
rather than increase the rate of reaction. The data in Figures 6

Chemistry of Methoxonium on (2 × 1)PT(110)
J. Phys. Chem. B, Vol. 105, No. 36, 2001 8589
and 7 is not consistent with an SN1 pathway or any pathway
involving adsorbed radicals.
ready source of electrons and hydrogen atoms. Also, the reaction
happens at much lower temperatures. Overall, methoxonium on
Pt(110) behaves very similar to methoxonium in solution.
Consequently, it appears that we can think of the species as a
true ion on a metal surface.
The data is consistent with an SN2 pathway, reactions 23-
5. The rate increases as the hydrogen concentration increases,
2
as expected if methoxonium is a key intermediate. The dimethyl
ether yield varies nonlinearly with the methanol coverage, as
expected for an SN2 pathway reaction. We fit the data in Figure
Summary
7
to a power law and found that the rate varies approximately
The chemistry of methoxonium has been studied on (2 ×
as the third order in the methanol coverage, while SN2 reactions
usually show second-order behavior in dilute solutions. How-
ever, we do not have a dilute solution on our surface, and third-
order behavior is often seen in SN2 reactions where the reactant
helps stabilize an ionic transition state. Our data is consistent
with the SN2 reactions 23-25 and not consistent with the SN1
pathway reactions 18, 25, and 26 or any reactions of any of the
chemisorbed radical species which are observed during methanol
adsorption and decomposition on platinum. Therefore, we
conclude that dimethyl ether forms via an SN2 pathway on Pt
1
)Pt(110). Methoxonium forms when hydrogen and methanol
are coadsorbed on the surface, which is proved by characteristic
4
vibrational spectrum obtained using EELS. Three reactions
happen on the surface: methanol dehydrogenation, methanol
dehydration, and dimethyl ether formation. Methoxonium is
important intermediates for the last two reactions.
According to the results:
(1) Methanol dehydration reaction involves methoxonium
decomposition as a rate-determining step. The step is SN1 and
the intermediate involved is fully charged. This has been
confirmed by the observation of secondary kinetic isotope effect
of a factor of 1.8.
(110).
The one confusing feature is that the methoxonium intermedi-
ate and the dimethyl ether product do not grow simultaneously
with increasing hydrogen coverage. We observe methoxonium
spectroscopically at modest hydrogen coverages. However, that
methoxonium seems to initially decompose via reactions 18-
(
2) Dimethyl ether formation is improved by increasing the
predosed hydrogen amount. Such behavior would be expected
if the reaction occurred via a pathway involving methoxonium
cations, but not one involving any of the radical species is seen
on the surface. The process follows the kinetics expected for
an SN2 process, strongly suggesting that the reaction is SN2.
In a larger way, our results show that the methoxonium on
surfaces is a true cation that reacts almost identically to a cation
in solution. This is the first time that S 1 and S 2 pathways
21 to yield methane, water, and adsorbed carbon. Dimethyl ether
formation is only seen at high hydrogen exposures.
Figure 8 shows a thermal desorption spectra of hydrogen on
clean (2 × 1)Pt(110). Three peaks show up when hydrogen
exposure is as high as 20 L: â1, â2, and R adstates. Figure 6
shows that the dimethyl ether peak grows into the spectrum
simultaneously with the â1 hydrogen peak. These results suggest
that the â1 hydrogen is intimately involved with the process
which forms dimethyl ether, although the details are not clear.
One possibility is that â1 and â2 hydrogen are bound to two
different sites on the surface, and reaction 14 (dimethyl ether
formation) takes place specifically on the â1 sites. There has
been speculation in the literature13 that â1 and â2 hydrogen are
bound to two different sites. Another possibility is that the â1
hydrogen is suppressing the decomposition of methanol, so the
species stay on the surface to higher temperatures. Cong and
Masel26 found that when they saturated a clean Pt(331) surface
with hydrogen, the methanol decomposed at higher temperatures
than on a clean surface. Reactions 23 and 24 require molecular
methanol, so when the methanol decomposition is suppressed,
the rate of reaction 24 should increase.
N
N
have been identified during reactions of small molecules on
metal surfaces.
Acknowledgment. This material is based on work supported
by the Department of Energy under Award No. DEGF-02-
99ER14993. Any opinions, findings, conclusions, or recom-
mendations expressed in this publication are those of the authors
and do not necessarily reflect the views of the Department of
Energy.
References and Notes
(1) Wagner, Frederick, T.; Moylan, Thomas, E. Surf. Sci. 1988, 206,
187-202.
(2) Sass, J. K.; Lackey, D.; Schott, J.; Straehler, B. Surf. Sci. 1991,
2
47, 239-247.
(
3) Kizhakevariam, N.; Stuve, E. M. Surf. Sci. 1992, 275, 223-236.
(4) Chen, N.; Blowers, P.; Masel, R. I. Surf. Sci. 1998, 418, 329-341.
Still, in either case, the reaction looks SN2. We observe the
methoxonium intermediate expected for an SN2 pathway, we
can rule out radical pathways, and we find that the kinetics of
the process are consistent with an SN2 pathway and not, for
example, an SN1 pathway. Therefore, all of the data points
strongly to a true SN2 reaction, although we cannot prove
reactions 23-25 are the correct SN2 pathway because we have
(5) Chen, N.; Blowers, P.; Masel, R. I. Surf. Sci. 1999, 419, 150-157.
6) Thomas, F. S.; Lu, C.; Lee, I. C.; Chen, N. S.; Masel, R. I. Catal.
001, 72, 167-175.
7) Blowers, P.; Masel, R. I. J. Phys. Chem. A 2000, 104 (1), 34-44.
(8) Wang, Jianhua; Masel, R. I. J. Am. Chem. Soc. 1991, 113, 5850-
856.
9) Costello, S. A.; Roop, B.; Liu, Z. M.; White, J. M. J. Phys. Chem.
988, 92, 1019-1020.
10) Zhou, Y.; Feng, W. M.; Henderson, M. A.; Roop, B.; White, J. M.
(
2
(
5
(
1
(
+
not yet observed the ([CH3OHCH3] )(ad) intermediate spectro-
J. Am. Chem. Soc. 1988, 110, 4447-4448.
(
(
11) Van Spaendonk, Vincent. Ph.D. Thesis, University of Illinois, 1994.
12) Anger, G.; Berger, H. F.; Luger, M.; Feistritzer, S.; Winkler, A.;
scopically.
Comparison to the methoxonium in solution. In a larger
way, all of our data suggests that methoxonium on Pt(110) reacts
very similarly to methoxonium in solution. We have conclusive
evidence for an SN1 pathway and convincing evidence for an
SN2 pathway. Both pathways strongly resemble the behavior
of methoxonium in solution. We also find that just as in solution,
the SN2 reaction has a much lower rate than the SN1 reaction
because the SN2 reaction needs a special geometry to occur.
There are some differences. Methane formation is more rapid
on the surface than in solution because the surface provides a
Rendulic, K. D. Surf. Sci. 1989, 219, L583-L589.
(13) Kristen, E.; Parschau, G.; Stocker, W.; Rieder, K. H. Surf. Sci. Lett.
1990, 231, L183-L188.
(
(
(
14) Shern, C. S. Surf. Sci. 1992, 264, 171-176.
15) Wang, Jianhua; Masel, R. I. J. Catal. 1990, 126, 519-531.
16) Wang, Jianhua; Masel, R. I. Surf. Sci. 1991, 243, 199-209.
(17) Wang, Jianhua; Masel, R. I. J. Vac. Sci. Technol., A 1991, 9 (3),
1
3
879-1884.
(
18) Guo, X.; Hanley, L.; Yates, J. T. Jr. J. Am. Chem. Soc. 1989, 111,
155-3157.
19) Levis, R. J.; Jiang, Z.; Winograd, N. J. Am. Chem. Soc. 1988, 110,
4431-4432.
(

8590 J. Phys. Chem. B, Vol. 105, No. 36, 2001
Lu et al.
(
20) Levis, R. J.; Jiang, Z.; Winograd, N. J. Am. Chem. Soc. 1989, 111,
605-4612.
21) Ozkan, U. S.; Kueller, R. F.; Moctezuma, E. Ind. Eng. Chem. Res.
990, 29, 1136-1142.
22) Streitwieser, A.; Jagow, R. H.; Fahey, R. C.; Suzuki, S. J. Am.
Chem. Soc. 1958, 80, 2326.
23) Maskill, H. J. Am. Chem. Soc. 1976, 98, 8482.
(24) Collins, C. J.; Bowman, N. S. Isotope effects in chemical reactions;
ACS Monograph 167; Van Nostrand Reinhold: New York, 1970.
4
(
(25) Banks, R. M.; Maskill, H.; Natarajan, R.; Wilson, A. A. J. Chem.
1
Soc., Perkin Trans. 1980, 2, 427.
(
(26) Cong, Y.; van Spaendonk, V.; Masel, R. I. Surf. Sci. 1997, 385,
246-258.
(