Thermal and Low-Energy Electron-Driven Chemistry of Biacetyl on Ag(111)

Article abstract of DOI:10.1021/jp9605820

The thermal and electron-induced chemistry of biacetyl (CH3COCOCH3) on Ag(111) has been studied using temperature-programmed desorption (TPD), X-ray photoelectron spectroscopy, and ultraviolet photoelectron spectroscopy.No thermal decomposition of biacetyl occurs, confirming that Ag(111) is inert with respect to the breaking of C-C, C=O, and C-H bonds.There are five molecular biacetyl desorption peaks in TPD-180, 178, 174, 188, and 215 K.The peak at 180 K is attributed to monolayer adsorption, and its saturation peak area is used to scale other TPD biacetyl peaks.The peak at 188 K is assigned to multilayers and that at 215 K to desorption from defect sites.The peaks at 174 and 188 K are discussed in terms of coverage dependent reorientation and bilayers.Nonthermal excitation pathways by which the surface chemistry of biacetyl may be directed were explored by irradiating 1 ML of biacetyl with 50 eV electrons.During irradiation, CO, CH3, ketene (H2C=C=O), and C2H6 desorb.After irradiation, five new post-irradiation TPD peaks appear.These are identified as H2 at 210 K, CH4 at 235 and 315 K, H2C=C=O at 240 K, and reaction-limited CH3COCOCH3 at 440 K.XPS shows C_(a) and O_(a) remain on the surface after heating to 700 K.

Full text of DOI:10.1021/jp9605820

15890
J. Phys. Chem. 1996, 100, 15890-15899
Thermal and Low-Energy Electron-Driven Chemistry of Biacetyl on Ag(111)
E. D. Pylant, M. J. Hubbard, and J. M. White*
Department of Chemistry and Biochemistry, Center for Materials Chemistry, UniVersity of Texas,
Austin, Texas 78712
ReceiVed: February 27, 1996; In Final Form: July 2, 1996X
The thermal and electron-induced chemistry of biacetyl (CH3COCOCH3) on Ag(111) has been studied using
temperature-programmed desorption (TPD), X-ray photoelectron spectroscopy, and ultraviolet photoelectron
spectroscopy. No thermal decomposition of biacetyl occurs, confirming that Ag(111) is inert with respect to
the breaking of C-C, CdO, and C-H bonds. There are five molecular biacetyl desorption peaks in TPDs180,
178, 174, 188, and 215 K. The peak at 180 K is attributed to monolayer adsorption, and its saturation peak
area is used to scale other TPD biacetyl peaks. The peak at 188 K is assigned to multilayers and that at 215
K to desorption from defect sites. The peaks at 174 and 188 K are discussed in terms of coverage dependent
reorientation and bilayers. Nonthermal excitation pathways by which the surface chemistry of biacetyl may
be directed were explored by irradiating 1 ML of biacetyl with 50 eV electrons. During irradiation, CO,
CH3, ketene (H2CdCdO), and C2H6 desorb. After irradiation, five new post-irradiation TPD peaks appear.
These are identified as H2 at 210 K, CH4 at 235 and 315 K, H2CdCdO at 240 K, and reaction-limited
CH3COCOCH3 at 440 K. XPS shows C(a) and O(a) remain on the surface after heating to 700 K.
1. Introduction
desorption of oxygenates typically occur with little or no
dissociation on Ag(110) and Ag(111).3-8 In such cases,
nonthermal excitation and subsequent reaction chemistry become
obvious through the appearance of reaction products. For
instance, acetone, CH3COCH3, on Ag(111) shows no thermal
activation and, below 5 eV, no photon activation, but when
irradiated with 50 eV electrons, adsorbed acetyl, CH3C(a)O,9
and methyl, C(a)H3, fragments form.5 Similarly, electron
activation of methyl formate, CH3OCHO, on Ag(111) leads to
the formation of C(a)H3 and formate, O(a)CHO.4 In this context,
biacetyl is of interest because it offers the possibility of
selectively forming adsorbed acetyl groups through nonthermal
excitation.
Studied over many years in the gas phase and in solution,
biacetyl exhibits rich, wavelength dependent photochemistry,
breaking and making bonds, and photophysics, intersystem
crossing from singlet to triplet states, and light emission from
singlet and triplet states.10-19 To our knowledge, there have
been no investigations of adsorbed biacetyl at the monolayer
level on any substrate. There are two studies involving adsorbed
multilayers. In the first, quenching of light emission from
photoexcited adsorbed biacetyl was studied as a function of its
distance from Ag(111), using an NH3 spacer layer and 440 nm
photons.20 In the second study, multilayers of biacetyl on quartz
and on silver films were irradiated with 266 nm photons; CH3
and CH3CO radicals were produced.21
This paper describes the results of two kinds of
activationsthermal and electron. Thermal activation is chemi-
cally simple; low-temperature adsorption and subsequent thermal
desorption occurs without dissociation. This is evidenced by
ultraviolet photoelectron spectroscopy (UPS) and X-ray photo-
electron spectroscopy (XPS), work function change (∆Φ), and
temperature-programmed desorption (TPD). These results
provide valuable benchmarks for electron-induced chemistry,
reported here, and photon-induced chemistry to be reported
elsewhere.22 Setting aside possible coverage variations, we
focus on a single coverage, motivated by the observation that
photon irradiation of this coverage causes decomposition and
prompt ejection, but no retention at the surface, of fragments.
As demonstrated here, there is retention and ejection of products
Surface chemical reactions, initiated by thermal and nonther-
mal means, are of great scientific and technological interest.
Thermal activation of surface reactions is by far the most widely
studied and, in applications, the most widely used. Excited
electronic states generally do not play a significant role in
thermal excitation; as the system is heated, other processes, e.g.,
bond rearrangement through vibrational activation of the ground
electronic state, generally intervene before significant electronic
excitation takes place.
Exploration of electronically altered states is interesting and
challenging and motivates the work described here. Photon and
electron activation are initiated quite differently than thermally
activated processes. For example, when visible and ultraviolet
photon excitation of a molecule occurs, the light couples to the
electronic coordinates of the molecular system and in the
Franck-Condon description leaves the nuclear coordinates
initially unaltered. Subsequently energy typically flows into
these coordinates and drives bond rearrangements. Similar
considerations apply to activation using electrons. By system-
atically varying the energy of the electrons or photons, the newly
formed excited state configuration can be changed, and in
principle, specific chemistry, not governed by thermal distribu-
tions, can be realized. Simply put, specific electronic states
can be populated which are not significantly populated by
heating. Beyond intrinsically interesting issues of excitation
and energy flow, we have used electron excitation to prepare,
selectively, spectroscopically significant concentrations of ad-
sorbed species derived from molecular adsorbates, e.g., vinyl
and phenyl.1
As one component of extending to oxygenates previous work
in our laboratory that involves nonthermal excitation of hydro-
carbons and organic halides,2 we have undertaken a series of
experiments involving biacetyl, CH3COCOCH3, adsorbed on
single-crystal silver, Ag(111). Other oxygenates under study
include methanol,3 methyl formate,4 and acetone.5 Unless O(a)
is present, the low-temperature, ca. 100 K, adsorption and
X Abstract published in AdVance ACS Abstracts, September 15, 1996.
S0022-3654(96)00582-5 CCC: $12.00 © 1996 American Chemical Society

Chemistry of Biacetyl on Ag(111)
J. Phys. Chem., Vol. 100, No. 39, 1996 15891
under 50 eV electron excitation. Previously, we have identified
impact ionization as the activation process leading to these
products.23
2. Experimental Section
All experiments were carried out in an ultrahigh-vacuum
chamber described elsewhere.24 Briefly, the chamber is equipped
with a quadrupole mass spectrometer (UTI 100C) for temper-
ature-programmed desorption and residual gas analysis (RGA)
and a double-pass cylindrical mirror analyzer (PHI 15-255GAR)
for Auger electron spectroscopy (AES) and photoelectron
measurements. Excitation sources include a VG Instruments
X-ray source for X-ray photoelectron spectroscopy and a
differentially pumped helium discharge He I and He II source
for ultraviolet photoelectron spectroscopy and work function
changes. The chamber is pumped with an ion pump, an
auxiliary titanium sublimation pump, and a 170 L/s turbo-
molecular pump.
The typical base pressure during experiments was 1 × 10-10
Torr. Substrate temperatures were measured using a chromel-
alumel thermocouple welded to a tantalum loop inserted into a
hole drilled in the edge of the crystal. By attachment to a liquid
nitrogen reservoir, the sample could be cooled to 110 K and
heated resistively to 1000 K. The crystal was cleaned by cycles
of Ar+ ion sputtering and annealing to 675 K until no impurities
were detected by AES.
Since electrons from the mass spectrometer filament influ-
enced the results and since our instrumentation did not allow
biasing the sample during TPD, during thermal desorption the
sample was rotated about 90° away from line-of-sight with
respect to the mass spectrometer. XPS made use of Mg KR
radiation at 1253.6 eV and an analyzer pass energy of 50 or
100 eV. XPS scan windows were 15 eV wide, and data
collection time was typically 15 min. Core level binding
energies were referenced to the Ag (3d5/2) photoelectron peak
centered at 367.9 eV (peak width at half-maximum, ∆E1/2, is
1.7 eV). Binding energy uncertainties are ≈0.3 eV. XPS curves
were fit using a least-squares algorithm.25
The purity of biacetyl (Aldrich Chemical Corporation, 97%
pure) was verified by mass spectrometry after several freeze-
pump-thaw cycles. It was dosed through a directed micro-
capillary array placed 0.5 cm from the sample. To dose, the
biacetyl partial pressure was first increased, indicated ∆P ) 4
× 10-10 Torr at the system ion gauge, with the sample turned
away from the doser. After the pressure stabilized, the dose
was initiated by rotating the sample in front of the doser for a
measured time. To terminate the dose, the sample was turned
away from the doser and the leak valve promptly closed. The
exposures, as monitored by TPD, although very reproducible,
are not known in langmuirs because there is a molecular flux
enhancement of approximately 100 (with respect to background
biacetyl partial pressure rise) when the surface is positioned in
front of the doser. Dosing times are used as a reliable, but
relative, measure of exposure. Biacetyl coverages are defined,
as described in section 3, in terms of the saturation of a
characteristic TPD peak.
For electron activation, we used the mass spectrometer
filament. As measured by the CMA, the energy distribution
peaked at 50 ( 1 eV, and the current from the sample to ground
was ≈16 µA (1.0 × 1014 electrons s-1). Since the sample area
is 1 cm2, the electron flux becomes 1.0 × 1014 electrons cm-2
s-1. Lower energy electrons were generated by adjusting the
filament voltage of the mass spectrometer. As the energy was
lowered, the current tended downward; constant fluences were
maintained by increasing the irradiation time. During electron
Figure 1. Temperature-programmed desorption of biacetyl, CH3-
COCOCH3, as a function of coverage, after dosing at 110 K on Ag(111).
The 43 amu signal, the most intense cracking fragment of biacetyl,
was followed. The curve marked 1 ML is, as described in the text,
defined as 1 ML and based on XPS amounts to 4.3 × 1014 molecules
cm-2. The inset replots the 7.5 ML curve and, overlain with it, TPD
for 20 ML.
irradiation, the sample was placed in line-of-sight with the mass
spectrometer. The flux uniformity over the crystal was inves-
tigated by measuring the sample-to-ground current as a function
of the angular position of the sample away from its line-of-
sight position. The angle was varied using the rotary motion
capability of the manipulator which moves the sample on a 2
in. radius circle with respect to the manipulator’s axis of rotation.
The electron flux varied by no more than 10% over an angular
range three times larger than that subtended by the sample with
respect to the axis of rotation. Reaction rates are discussed in
terms of cross sections (section 3). The latter are upper limits,
since primary electrons reflected from the sample or secondary
electrons generated within the substrate are not properly counted.
Using 70 eV electrons, the fragmentation of gas-phase
biacetyl in the ionizer of our mass spectrometer is dominated
by three species: (1) 86 amu (CH3CO)2+, the parent ion, (2)
43 amu (CH3CO+), the result of breaking the central C-C bond,
and (3) 15 amu (CH3+), from breaking a terminal C-C bond.
At the detector, these give the following fragmentation pat-
terns86 amu:43 amu:15 amu::22:100:42, in reasonable agree-
ment with 15:100:35 reported for 70 eV electrons.26 Typically,
parent TPD spectra are characterized by following the 43 amu
signal, since it is the most intense.
3. Thermal Chemistry
3.1. Biacetyl Adsorption and Desorption. During both
adsorption at 110 K and subsequent TPD, there is no evidence
for thermal decomposition of biacetyl; desorption mass spectra
are fully accounted as parent biacetyl, and the substrate is clean,
as judged by XPS, after heating to 300 K. This confirms
conditions for which Ag(111) is inactive with respect to
dissociation of biacetyl. For coverages ranging from submono-
layer to multilayer, TPD spectra, Figure 1, were taken following
the most intense ion fragment at 43 amu, CH3CO+. The spectra
have been corrected systematically for background and system
pumping speed.27 The coverages, marked on each curve, are
based on defining the low-coverage saturated 180 K peak as 1

15892 J. Phys. Chem., Vol. 100, No. 39, 1996
Pylant et al.
ML. The reasonableness of this assignment is established below
on the basis of XPS results. Above 1 ML, the peaks are not
readily resolvable into monolayer and multilayer contributions.
As the biacetyl coverage increases, bottom to top in Figure 1,
four desorption peaks are identifiables(1) one peak up to 1
ML, 179-180 K, (2) a split pair, 178 and 185 K, at and above
1.5 ML, (3) a shoulder, 174 K, at 2 ML, and (4) an unsaturable
peak setting in at about 183 K (>4.5 ML) and moving to 188
K at 20 ML. Focusing on the 7.5 ML case, the shoulder is at
position (c), the multilayer at position (d), and the split pair at
positions (a) and (b). Below (Figure 8) for a somewhat higher
pumping speed, we also resolve a small peak at 215 K which
is attributed to adsorption at defect sites, on the basis of
adsorption after sputtering with no annealing (not shown) which
gave a much more intense peak at 215 K. With the ML
definition in mind and the thick multilayer established, we are
left with two further assignments. At 1.5 ML there are two
peaks; the lower temperature peak is assigned, as outlined below,
to desorption from local regions where the number density is
equivalent to that of the monolayer case, while the higher
temperature peak is assigned to desorption from local regions
where the number density is higher. Whether these two peaks
involve reorientations of first-layer-adsorbed CH3COCOCH3 or
bilayers of CH3COCOCH3 or some composite involving both
cannot be fully resolved.
Figure 2. Biacetyl TPD peak area (mass 43) as a function of dose
time. Biacetyl was dosed at 110 K after raising the background
CH3COCOCH3 pressure to 4 × 10-10 Torr.
basis of the Chan-Aris-Weinberg method,29 the calculated
The inset shows the 20 ML data and a replot of the 7.5 ML
case. It is of interest that the shoulder at 174 K disappears as
the coverage increases suggesting restructuring of the whole
adsorbate layer. The two curves are superimposable, except in
the shoulder region. This is taken to indicate zero-order
desorption kinetics away from the shoulder region. Assuming
zero-order kinetics, analysis of the 20 ML curve from 177 to
185 K gives an activation energy of 75.6 ( 5 kJ mol-1. There
is one other multilayer system for comparison; biacetyl on
Si(100) exhibits, for 5 ML, a peak at 190 K28 which is in
acceptable agreement with Figure 1.
activation energy for desorption of 1 ML was 47 ( 5 kJ mol-1
.
To ascertain how the uptake varied with dosing time, we
calculated the desorption peak areas in ML units and correlated
them with the dosing time, Figure 2. The linear correlation
over the full range of doses and TPD profiles shown in Figure
1 indicates a constant sticking coefficient, probably unity.
The multilayer peak is at a slightly higher temperature than
the monolayer peak, perhaps indicating that biacetyl interacts
more strongly with itself than with the silver surface. However,
since the peak temperatures are separated by no more than 5
K, interaction energy differences are, at most, small. In fact,
the small peak temperature difference may be due to changes
in the preexponential factor. Several studies have observed
similar behavior for molecular desorption peaks in such systems
as SO2/metals, C6H6/metals, and SF6 on Ni(111).30-34
In the monolayer regime, we suggest biacetyl adsorbs with
both oxygen atoms toward the Ag(111) in an eclipsed confor-
mation, thereby maximizing the oxygen lone-pair interactions
with unfilled metal orbitals and minimizing the repulsion of
the methyl groups with the surface. Gas-phase biacetyl exists
in two conformers, eclipsed and trans, with a small barrier
(20 kJ mol-1) separating them.35 At 300 K, the trans form is
favored. The eclipsed geometry is intuitively favored for
adsorption on Ag since it allows occupied lone pairs on both
oxygen atoms to interact with unfilled orbitals on Ag.
3.2. X-ray Photoelectron Spectroscopy. XPS results are
consistent with nondissociative adsorption and desorption of
biacetyl. Only one O(1s) chemical state is observeds532.7 eV
with a fwhm of 2.0 eV (Figure 3). This peak is typical of
molecular carbonyl groups (Table 1). In passing from 1 to 24
ML, this peak width remains at 2.0 eV while the peak position
shifts 0.4 eV higher, 533.1 eV, a typical shift for reduced core
hole screening in thick multilayers. Similarly, the C(1s) spectra,
Figure 4, shift higher by 0.3 eV. As expected, there are two
distinct C(1s) peaks at all coverages, 285.2 and 288.1 eV, typical
of molecular alkyl and carbonyl groups, respectively. These
have equal intensities and widths, 2.0 eV, at all coverages,
consistent with molecular adsorption. Using sensitivity factors
for C(1s) and O(1s), the C-to-O ratio is 2.0 ( 0.1 over the range
With these thick multilayer results in hand, we turn to the
low-coverage cases. When the difference between the 1 and
1.5 ML spectra is taken, the low-temperature peak at 180 K is
preserved and the additional 0.5 ML coverage appears at higher
temperatures. Compared to desorption of an initial coverage
of 0.3 ML, the lowest curve in Figure 1, the desorption of the
last 0.3 ML from an initial coverage of 1.5 ML occurs at
measurably higher temperatures, ca. 6 K (179 vs 185 K). This
strongly suggests that the 0.3 ML coverages have different
structures in the two cases and, in turn, that passing from 1 to
1.5 ML at 110 K leads to restructuring of the entire adsorbate
layer.
As for the inset data of Figure 1, these changes are subtle,
have very little effect on the desorption activation energy, and
are not analyzable further with TPD alone. As indicated below,
XPS and UPS, while helpful, do not address these subtle
changes. In all likelihood, changes of molecular orientation are
involved. Characterizing them using coverage dependent
intensity variations of infrared vibrational bands is planned.
The lowest coverages, 0.3 and 1 ML, are consistent with first-
order desorption with a slight amount of attractive adsorbate-
adsorbate interaction. The latter is evidenced in a very detailed
analysis of the peak position, which increases ∼2 K between
0.3 and 1 ML, and the peak shape which broadens on the high-
temperature side. In passing from 0.3 to 1 ML, a model based
on first-order kinetics with a coverage dependent activation
energy gives no more than 11 kJ mol-1 ML-1 for the attractive
interaction parameter. Thus, it is reasonable to use a first-order
desorption model for analysis of the 1 ML TPD peak. On the

Chemistry of Biacetyl on Ag(111)
J. Phys. Chem., Vol. 100, No. 39, 1996 15893
TABLE 1: Binding Energy Assignments for 1 ML
Biacetyl/Ag(111)
speciesa
C(1s)
0(1s)
532.7
ref
-CdO(1 ML biacetyl)
-CH3(1 ML biacetyl)
CH3COCH3
288.1
285.2
287.9
this work
this work
36
PhCOCOPhb
532.7
37
CH3COCH3
284.7
36
a Atoms of interest denoted with boldface type. b Ph ) phenyl.
Figure 5. He(II) UPS of 1 and 21.5 ML of biacetyl on Ag(111). The
Ag substrate contribution has been subtracted out of the 1 ML, but not
the 21.5 ML, spectrum. Molecular orbital assignments are: (a) 10ag;
(b) 9bu; (c) 2bg, 2au, 9ag; (d) 8bu, 8ag, 1bg; (e) 7bu, 1au, 7ag; (f) 6bu; (g)
6ag; (h) 5bu, 5ag. Binding energies are given in Table 2.
TABLE 2: UPS (He II) Binding Energy Assignments for 1
ML Biacetyl/Ag(111)
1 ML
biacetyl/
Ag (eV)
multilayer
biacetyl/
Ag (eV)
gas-phase
refs4.4 eVa
(eV)
assignments
10ag (nO+
9bu (nO-
Figure 3. O(1s) XPS for three coverages of biacetyl on Ag(111) at
110 K.
)
4.5
6.5
8.1
4.3
6.4
8.0
5.2
7.1
8.2-9.6
)
2bg (πCO-
)
)
2au (πCO+
9ag
8bu,8ag,1bg
7bu,1au,7ag
6bu
6ag
5bu,5ag
9.5
10.6
11.5
15.1
18.4
9.4
11.0
11.9
15.2
19.7
10.1-10.8
11.6-12.1
13.0
16.2
19.6
a Gas-phase spectra adjusted to the Fermi level of the 1 ML/Ag(111)
spectrum. Add 4.4 eV to obtain their gas-phase values referenced to
the gas-phase vacuum level.
saturation coverage of iodine results in a (x3 × x3)-R30°
packing arrangement on Ag(111) corresponding to a I/Ag ratio
of 1/3 or a coverage of 4.6 × 1014 atoms cm-2 38
.
This calibration
was used, in conjunction with known atomic sensitivity factors
and the measured C(1s) and O(1s) peak areas of 1 ML CH3-
COCOCH3, to calculate that 4.3 × 1014 molecules cm-2
comprise monolayer biacetyl. This is a very reasonable result;
if one uses biacetyl’s liquid-phase density and approximates
biacetyl molecules as cubes, a monolayer would contain 3.6 ×
1014 molecules cm-2
.
3.3. Ultraviolet Photoelectron Spectroscopy. To probe the
valence orbitals of adsorbed CH3COCOCH3, ultraviolet pho-
toelectron He(II) spectra were taken for 1 and 21.5 ML (Figure
5). The Ag contribution to the 1 ML, but not the 21.5 ML,
spectrum was subtracted, and Fermi level referencing was
utilized. Adsorbate-derived signals are clearly evident in both
cases, and on the basis of gas-phase biacetyl UPS data,39 orbital
assignments (Table 2) are indicated by lower case letters and
vertical lines. The gas-phase orbital energies, measured with
respect to the vacuum level, have been shifted by 4.4 eV to
reference them to the Fermi level of the adsorbate system.
Consistent with the TPD and XPS results, there is nothing in
Figure 4. C(1s) XPS for three coverages of biacetyl on Ag(111) at
110 K; the O(1s) spectra in Figure 3 were taken at the same time.
of coverages shown in Figures 3 and 4. Importantly, when the
adsorbate-substrate system was heated, all the C(1s) and O(1s)
intensity disappeared by 300 K (not shown).
XPS peak areas, corresponding to the 1 ML assignment of
Figure 1, were used to calculate the absolute coverage of biacetyl
in molecules per square centimeter. XPS coverages are based
on atomic iodine. CF3I, which dissociates with ejection of CF3,
was dosed at 600 K to saturate the surface with iodine. A

15894 J. Phys. Chem., Vol. 100, No. 39, 1996
Pylant et al.
either of these spectra that is not readily attributable to adsorbed
CH3COCOCH3.
In passing from the multilayer to the monolayer, there is
evidence of modest differential shifting in peaks (a), (e), and
(h). Such differential shifts are generally attributed to coverage
dependent variations of the coupling of the particular orbitals
to the substrate, in either or both the initial and final states.
That the shifts in the highest lying valence orbitals are, at most,
small is taken as further evidence of very weak interactions
between CH3COCOCH3 and Ag(111). Referring to Table 2,
orbital (a) at 4.5 eV for 1 ML is derived from the 10ag(nO+
)
orbital of CH3COCOCH3, commonly referred to as a lone-pair
orbital strongly localized on a carbonyl oxygen. The 0.3 eV
shift of orbital (a) to lower binding energy (BE) between 1 and
21.5 ML is opposite the expectations for final state relaxation
variations and is, thus, attributed to a small initial state variation
consistent with a low-coverage adsorption model in which the
two oxygens of CH3COCOCH3 are both weakly linked to
Ag(111), i.e., CH3CO(a)CO(a)CH3. The shifts in orbitals (e) and
(h) to higher BE with coverage is consistent with final state
relaxation effects.
3.4. Work Function Change. One monolayer of biacetyl
lowers the work function of Ag(111) by 0.3 eV. The lowered
work function is consistent with a shift of electron density from
CH3COCOCH3 toward Ag(111) as anticipated for the adsorption
model proposed above.
Figure 6. Overview of the proposed chemical reaction paths beginning,
at the top, with 50 eV electron irradiation of 1 ML biacetyl molecularly
adsorbed on Ag(111) at 110 K. Products ejected into the gas phase
are indicated to the upper right, and products retained as adsorbates at
110 K are indicated beneath, in the second Ag(111) depiction.
Subsequent TPD results are depicted further down the figure. The
temperature range is arbitrarily divided into two segments, below and
above 300 K. There is no detectable desorption above 500 K, but C
and O are retained, at least to 700 K. Biac and ac denote biacetyl and
acetyl, respectively.
4. Electron-Induced Chemistry
4.1. Overview of Electron-Driven Processes. While no
surface reactions of CH3COCOCH3 on Ag(111) occur thermally,
electron irradiation leads to a number of products, some of which
are ejected during irradiation and others which are retained and
contribute to products that desorb during TPD. The results are
not describable by a single pathway, and because of this
complexity, we introduce this section with an overview of our
proposed reaction path model. As noted already in the
Introduction, the focus is on 1 ML coverage because, unlike
photon-driven chemistry of 1 ML, products of electron irradia-
tion are retained. Examination of coverage variations, below
and above 1 ML, is postponed for future work.
In schematic form, Figure 6 accounts for the observed
products and outlines the proposed paths leading to them.
During electron irradiation, relatively large amounts of CH3 and
CO are ejected along with smaller amounts of ketene
(CH2dCdO) and C2H6. These products demand that electron
activation leads to both C-C and C-H bond breaking. Since
the cross section for loss of biacetyl is negligible for energies
below its ionization potential (9.55 eV),40 the dominant activa-
tion mechanism is the impact ionization of biacetyl.23 In
subsequent TPD, CH4, H2, CH2CO, and CH3COCOCH3 are
observed. To account for these products and their peak
temperatures, our model includes the retention of CH3(a), H(a),
CH3C(a)dO, Cx(a)Hy, and CH2C-O(a)CdOCH3. While there is
evidence for rehybridization of carbonyls to form species with
the O bound to Ag and a C-O single bond, there is no evidence
for electron-induced complete fragmentation of the carbonyl to
give adsorbed atomic O. Between 500 and 700 K, atomic forms
of C and O remain (<8% of the initial amounts) according to
XPS; this implies some C-O bond breaking during TPD.
4.2. Ejection during Electron Irradiation. Having sum-
marized our model, we turn our focus to those species which
desorb during irradiation. In a very interesting set of experi-
ments, denoted isoscans, we adsorbed 1 ML of biacetyl, held
the temperature at 110 K, and monitored desorbing products
during electron irradiation. After dosing, the sample was biased
-90 V to repel electrons and then rotated in front of the electron
Figure 7. Desorption during isothermal 50 eV electron irradiation of
1 ML biacetyl at 110 K. Desorption products are CO, CH3, CH2dCdO,
and C2H6. The sharp rises and falls in the data are due to switching
the electron irradiation on and off by unbiasing and biasing the sample
-90 volts, respectively.
source. After the background pressure in the chamber was
allowed to stabilize, the isoscan was started (Figure 7). After
60 s, the bias was removed and the sample grounded, allowing
50 eV electrons to strike the sample. After another 60 s, the

Chemistry of Biacetyl on Ag(111)
J. Phys. Chem., Vol. 100, No. 39, 1996 15895
bias was again applied to stop the electron-stimulated desorption.
This cycle was repeated for 10 min. The isoscans have
contributions at 12, 13, 14, 15, 16, 28, 30, and 42 amu, but
importantly, there is no contribution at either 86 or 43 amu,
i.e., no electron-stimulated parent desorption. A blank experi-
ment repeated the 10 min measurement with the -90 V bias
constantly applied. The resulting background signals were
subtracted, and Figure 7 shows the results for 15, 28, 42, and
30 amu. The prompt on-off character and the fact that the
second (third) exposure begins with the intensity measured at
the end of the first (second) exposure indicate that no detectable
reaction occurs when the electron beam is off. We conclude
that thermal effects do not make a significant contribution; the
thermocouple temperature increased by no more than 2 K. This
data is satisfactorily described by first-order decay; the resulting
cross sections are presented below.
We attribute the observed masses to four desorption prod-
ucts: CH3, CO, ketene (CH2dCdO), and C2H6. The CH3 and
1
CO signals dominate, and the CH3 signal is /4 that of CO.
Ketene, identified on the basis of a literature standard fragmen-
tation pattern, is ≈50 times smaller than the CO signal. The
30 amu signal, barely discernible above the noise, is ≈100 times
smaller than the CO signal and is attributed to C2H6. Other
ion signals, expected for C2H6 fragmentation (26-29 amu), are
not discernible above background. After subtracting the
contribution of ketene to mass 14, the remaining intensity for
masses 12-15 is attributed mainly to CH3 desorption. But since
the cracking pattern for CH3 is not known for our system, we
cannot rule out a minor contribution from CH2. Ketene (mass
42) desorption gives rise to some of mass 14 and all of mass
42. CO desorption accounts for all of mass 28 and most, if not
all, of mass 16, when the cracking pattern of CO is taken into
account. Consequently, although a small amount of methane
desorption cannot be ruled out during electron irradiation, it is,
at most, a very minor product. That methane does not make a
significant contribution is confirmed by the fact that the mass
15 intensity exceeded that for mass 16 by a factor of 4. Methane
gives a cracking ratio for 15/16 amu of ≈0.85.
Figure 8. Post-irradiation TPDs of 1 ML biacetyl (43 amu signal) for
increasing, top to bottom, exposures to 50 eV electrons.
isoscan was started with the bias applied, and (4) after 60 s, the
bias was removed for the rest of the scan, 600 s. Assuming
the electron stimulated desorption reactions are first order, the
following equation can be used to obtain the cross section, σ
ln{∆Px(g)(t)/∆Px(g)(0)} ) -σFe
(1)
where ∆Px(g)(0) is the intensity attributed to the initial pressure
of species x and ∆Px(g)(t) is the intensity attributed to the pressure
of species x at time t.42 Fe is the fluence of electrons irradiating
the sample. The calculated cross sections for CH3 and CO
desorption were both (3.4 ( 0.3) × 10-17 cm2, that for ketene
was (4.1 ( 0.6) × 10-17 cm2, and that for ethane was (3.0 (
0.6) × 10-17 cm2. The uncertainties for ketene and ethane are
larger due to the small amount of product desorbing (see Figure
7).
4.3. Loss of Parent TPD Signal. Another view is obtained
by examining, for a series of experiments, how the parent TPD
peak area decays with electron exposure. From this decay, a
cross section was calculated for the total loss of biacetyl by all
active channels, i.e., desorption of species plus retention of
decomposition products. Post-irradiation TPDs for the desorption
of biacetyl (mass 43) are shown in Figure 8 for numerous
electron fluences. Clearly, the 180 K peak decays rapidly. At
large fluences, a tiny biacetyl peak appears at 440 K (not
shown). It is not due to unreacted biacetyl molecules and,
therefore, is not considered in the following discussion. The
cross section for total loss of biacetyl was calculated from
CH3 and CO ejection require electron-induced C-C bond
breaking in adsorbed CH3COCOCH3. There are two possible
pathways. First, the central C-C bond of biacetyl could be
broken through irradiation, and the resulting acetyl group might
decompose into CH3 and CO. One might then, contrary to
observation, expect to see acetyl fragments, 43 amu, desorb
during electron irradiation. The second possibility, favored
because 43 amu fragments are retained, is electron-induced
cleavage of one of the terminal C-C bonds to promptly eject
CH3. CO ejection would then arise from prompt decomposition,
just before or just after neutralization of the resulting adsorbed
CH3CO-CO+ fragment. This decomposition step forms CO,
which will not adsorb to Ag(111) at 110 K,41 and adsorbed
acetyl, CH3C(a)O.
The only other ejected products, small amounts of C2H6 and
CH2CO, are attributed to reaction of two CH3 fragments and to
electron-induced C-H bond breaking followed by cleavage of
the central C-C bond in CH2CO(a)COCH3. As depicted in
Figure 6 and discussed below, there is evidence that both C(a)H3
and CH2CO(a)COCH3 play important roles in the subsequent
TPD.
ln{I(t)/I(0)} ) -nσ(E)t ) -(iet/eA)σ(E) ) -Feσ(E) (2)
where I(t)/I(0) is the ratio of the biacetyl TPD peak areas before
and after irradiation time t, ie is the current density (A/cm2), A
is the sample surface area, e is the electron charge, and Fe is
the electron fluence (electrons/cm2).43
As shown in Figure 9, the resulting slope is not constant with
respect to electron fluence; the cross section drops as the fluence
increases. An adequate fit is realized with two lines, i.e., two
cross sectionss(8.0 ( 0.2) × 10-17 cm2 and (4.0 ( 0.2) ×
10-17 cm2. The decline is attributed to the accumulation of
products which begin to inhibit the electron-driven decomposi-
tion of the remaining biacetyl for fluences greater than 1 ×
From the time dependence of the decay of ion signals like
those of Figure 7, cross sections were calculated. The data (not
shown) was generated using the following procedure: (1) a
monolayer of biacetyl was adsorbed at 110 K, (2) the sample
was biased -90 V and rotated in front of the mass spectrometer,
(3) after allowing the background to stabilize, a multiplexed

15896 J. Phys. Chem., Vol. 100, No. 39, 1996
Pylant et al.
Figure 10. Post-irradiation experiment on 1 ML of biacetyl adsorbed
at 110 K and irradiated with 6.4 × 1016 electrons cm-2
. Proposed
reactions leading to desorption of each peak are listed to the side of
the plot. Biac ) biacetyl, and H(ac) ) a hydrogen from the dehydro-
genation of an acetyl group.
Figure 9. Semilogarithmic plot of the fractional decrease in 1 ML
biacetyl (mass 43) TPD peak area as a function of electron fluence.
The indicated cross sections for removal of biacetyl were calculated
from the slopes.
1016 electrons cm-2. Interestingly, the smaller cross section is
comparable to those, as outlined above, that describe the
electron-initiated ejection of products. However, the total initial
cross section for destruction of biacetyl is twice as large. This
implies the reactions leading to adsorbed fragments are inhibited
as the surface becomes cluttered with decomposition products,
but the ejection reactions continue on with the same cross
section.
4.4. Thermal Reactions of Species Retained after Electron
Irradiation. This section deals with desorption after 1 ML of
biacetyl was irradiated with 50 eV electrons. Consistent with
our model, there is evidence for electron-induced C-C, C-H,
and possibly even minor amounts of CdO bond breaking.
Several major desorption channels appear in the post-irradiation
TPD spectra. These products are identified as: H2 between
210 and 230 K, ketene (H2CdCdO) at 240 K, methane at 235
and 315 K, and reaction-limited biacetyl desorption at 440 K,
Figure 10. Figure 11 summarizes the desorption areas as a
function of fluence.
In the biacetyl desorption plots, (Figure 8), interesting features
include: (1) the parent desorption temperature decreases from
180 to 174 K with increasing electron fluence, (2) the defect-
site peak temperature (215 K) increases to 230 K with increasing
fluence, and (3) indicative of defects inhibiting biacetyl’s
reactions with electrons, the defect-site peak decays ap-
proximately four times slower than the decay of the 1 ML peak.
The mass 42 desorption state at 240 K was identified as
ketene. No higher mass ions (>42 amu) were observed at 240
K, and all detected ion intensities are in accord with the cracking
fragments of gas-phase ketene. Further support for assignment
to ketene is found by subtracting from the mass 14 peak the
expected contribution from methane. The remaining area for
mass 14 is in a ratio with mass 42 that is consistent with ketene
desorption at 240 K.
In the region where mass 42 peaks, masses 15 and 16 also
peak, with maxima at 235 K, Figure 10. When the area for
desorption of biacetyl from defect sites is taken into account,
the 15 and 16 amu peak areas are in the correct ratio for methane
desorption. The peak at 315 K is also attributed to methane
Figure 11. Accumulation of post-irradiation TPD products as a
function of increasing electron fluence. These plots are each normalized
to unity at the maximum. The mass 16 (total) curve combines the
peak areas for both the 235 and 315 K desorption states.
desorption on the basis of fragmentation ratios. No masses
higher than 16 were observed in the 315 K region of the TPD
spectra. Since methane does not adsorb on silver at 110 K, it
must form through a rate-determining surface reaction for both
peaks in Figure 10. Another interesting point is the peak at
315 K is larger than the 235 K peak. Since both peaks are
attributed to the same desorption species, this indicates that the
concentration of species leading to the 315 K desorption channel
is higher than for the 235 K channel.
At 440 K, only peaks for masses 15, 43, and 86 were
observable in the TPD spectra. The cracking ratios for each of
these peaks follows what is expected for biacetyl. However,
since adsorbed biacetyl desorbs below 300 K, the desorption
state at 440 K is attributed to the reaction-limited desorption of
biacetyl; i.e., surface fragments assemble to form biacetyl.
The H2 desorption spectra at different electron fluences
(Figure 12) is intriguing. At low electron fluences the peak
maximum is at 210 K. With increasing fluence, 5.3 × 1016
electrons cm-2, the peak shifts higher in temperature to 230 K.
Finally, at a fluence of 6.8 × 1016 electrons cm-2 the peak
maximum is once again at 210 K. This indicates the chemistry
of H2 desorption is complex and possibly involves several
desorption channels, as does our model (Figure 6).

Chemistry of Biacetyl on Ag(111)
J. Phys. Chem., Vol. 100, No. 39, 1996 15897
Figure 14. O(1s) post-irradiation XPS anneal series of 1 ML biacetyl
Figure 12. Post-irradiation TPD of dihydrogen (mass 2) for several
electron fluences.
on Ag(111) at 110K with 6.4 × 1016 electrons cm-2
. The “dose”
spectrum was taken after dosing one monolayer. The “irrad.” spectrum
was taken after irradiation. For each subsequent spectrum, the sample
was flashed to the indicated temperature on the plot and then taken
after the sample cooled back to 110 K.
TABLE 3: Binding Energy Assignments for 1 ML
Biacetyl/Ag(111)+Electrons
C(1s) 0(1s) C(1s) 0(1s)
exptl exptl
(eV) (eV) (eV) (eV)
ref
ref
speciesa
CdO (biacetyl)
CdO (EID products)b
ref
288.1 532.7
286.9 531.8 286.4 532.0 47
this work
dC-O-Ag (EID products) 286.9 530.1 287.2 530.5 48, 49
-CH3
285.2
285.0
284.7
47
-CH2
50
533.4 51
52
C-O-C
C(graphitic)
carbide
533.6
284.3
283.2
284.3
280.8
50
-282.8
a Atoms of interest denoted with boldface type. b EID ) electron-
induced decomposition.
to produce ketene and surface hydrogen. Indications for this
part of our model include: (1) evidence for the production of
ketene by the dehydrogenation of acetyl groups is also found
in the 50 eV electron irradiation of acetone/Ag(111);5 (2) since
all surface hydrogen produced by irradiation should have
desorbed by 210 K,44 the peak area above 210 K and extending
to 260 K in the H2 desorption spectra is attributed to recombina-
tion of hydrogen atoms made by the loss of hydrogen from
CxHyOz groups, e.g., acetyl; (3) because no surface hydrogen
produced by irradiation should remain on the surface above 210
K, the desorption peak for methane (mass 16) at 235 K is
attributed to CH3(a) species combining with H atoms formed
by dehydrogenation of acetyl fragments to desorb as methane;
(4) since 1 ML ketene desorbs at 120 K,45 the ketene (mass 42)
desorption at 240 K must be due to a reaction-limited step; (5)
indicative of multiple reaction products having the same rate-
limiting step, i.e., the dehydrogenation of acetyl groups, methane
(mass 16), ketene (mass 42), and hydrogen (mass 2) desorption
tails end at 260 K.
Figure 13. C(1s) post-irradiation XPS anneal series of 1 ML biacetyl
on Ag(111) at 110 K with 6.4 × 1016 electrons cm-2
. The “dose”
spectrum was taken after dosing one monolayer. The “irrad.” spectrum
was taken after irradiation. For each subsequent spectrum, the sample
was flashed to the indicated temperature on the plot and then taken
after the sample cooled back to 110 K.
As shown in Figure 11, all the TPD peak areas, normalized
to their maxima, rise smoothly with increasing electron fluence.
Assuming first-order kinetics, the cross section for accumulation
of these products is (2 ( 1) × 10-17 cm2.
Figure 10 shows proposed reactions leading to TPD products.
These reactions involve the species produced by impact ioniza-
tion, Figure 6; CH3, H, CH3C(a)dO, and CH2C-O(a)CdOCH3
are the major products. Surface hydrogen made during irradia-
tion was expected, as observed, to recombine and desorb below
210 K.44
The desorption peak for methane (mass 16) at 235 K is
attributed to two pathways, one of which is CH3(a) species
combining with surface H formed during irradiation. Other
proposed reactions include the dehydrogenation of acetyl groups
Above 260 K, no CH3(a) groups remain on the surface. If
they did, recombination of methyl groups would lead to ethane
desorption at 273 K.46 As the sample is heated further, some
of the remaining hydrocarbon species decompose to produce

15898 J. Phys. Chem., Vol. 100, No. 39, 1996
Pylant et al.
TABLE 4: XPS Peak Area Ratios of C and O Left on the Surface by Chemical State (Normalized to Initial Amount of C and
O on the Surface)
original C remaining
original O remaining
peak positions
283.2
284.3
285.2-285.0
286.9
288.1
530.1
531.8
532.7
533.6
dose at 110 K
irradiation at 110 K
200 K
275 K
350 K
0
0
0
0
0
0
0.05
0.51
0.39
0.32
0.33
0.27
0
0
0.49
0.14
0.08
0
0
0
0
0
1
0
0.13
0.13
0.08
0.06
0
0.13
0.16
0.15
0.14
0
0.17
0.17
0.26
0.25
0.04
0.33
0.35
0.27
0.24
0
0.27
0.12
0
0
0
0.05
0.06
0.02
0.06
0.04
700 K
TABLE 5: XPS Peak Area Ratios of C and O Lost during
Irradiation and Subsequent Anneals
TABLE 6: XPS Peak Area Ratios of C and O Left on the
Surface
C/O ratio of desorbing species
C/O ratio of
original C original O
remaining species remaining remaining
dose-irradiation
irradiation-200 K
200 K-275 K
275 K-350 K
350 K-700 K
2
1.9
1.8
dose at 110 K
irradiation at 110 K
200 K
275 K
350 K
1.9
1.8
1.8
1.8
1.6
1.1
1.0
1.0
0.81
0.69
0.56
0.48
0.05
0.82
0.71
0.56
0.55
0.08
only C desorption
1.7
700 K
more methane at 315 K. Finally, at 440 K, CH2C-O(a)CdOCH3
scavenges a hydrogen atom from other decomposition species
remaining on the surface to desorb as biacetyl. After this occurs,
heating to 700 K leaves, as shown by XPS, small amounts of
atomic carbon and oxygen (<8% of the initial amounts).
Future studies are planned to examine the effects of the initial
coverage of biacetyl on the post-irradiation TPD product
distribution; in particular, the relative amounts of methane and
ketene desorption at 240 K may provide insight into whether
there is a situation in which the C-C bond between the carbonyl
groups is preferentially cleaved. Since signal-to-noise is an issue
when there are many decomposition pathways and very little
starting material, submonolayer studies are postponed here.
desorbs, and CH2dC-O(a)CdOCH3, which leads to the high-
temperature biacetyl desorption.
Upon annealing to 350 K, O(1s) does not change but the C(1s)
intensity drops, consistent with CH4 desorption. At 350 K,
remaining O(1s) and C(1s) signals are consistent with some
CH2dC-O(a)CdOCH3. After heating to 700 K, only small
broad C(1s) and O(1s) signals remain consistent with desorption
of reaction-limited biacetyl desorption at 430 K and retention
of some atomic C and O. We propose that the CH2dC-
O(a)CdOCH3 species in our model acquires a hydrogen atom
from one of the hydrocarbon-containing species left behind on
the surface and desorbs as biacetyl.
4.5. Post-irradiation XPS. XPS supports the argument that
the post-irradiation desorption products are ketene, methane, and
reaction-limited biacetyl. After irradiation of 1 ML of biacetyl
with 6.4 × 1016 e- cm-2, an XPS anneal series, Figure 13 and
14, was undertaken. The sample was heated to a selected
temperature and cooled to 110 K, and XPS was acquired.
Except for the 700 K spectra acquired at 100 eV pass energy
because of the low intensity, all spectra were acquired using
50 eV pass energy and were fit with features having a fwhm of
2.0 ( 0.1 eV using a curve-fitting routine with an 85% Gaussian/
Lorentzian mixture. The 700 K spectra were fit with a fwhm
of 3.8 eV in the oxygen region and 3.2 eV in the carbon region.
Table 6 lists the C/O ratios at the surface for various
conditions. Columns 3 and 4 of Table 6 give the fraction of
the initial amounts of C and O that remain after various
treatments. According to the interpretation given in the previous
paragraph, the biacetyl recombination channel contributes more
than a third of the total retained reaction products. This is
consistent with TPD.
Flashing the sample to 700 K leaves a small amount of carbon
(5% of the initial amount) and a comparable amount of oxygen
on the surface. Since atomic oxygen by itself would recombine
and desorb between 490 and 550 K,53,54 it is interesting that a
small amount of oxygen remains to 700 K. No O2 desorption
is observed in our TPDs as we annealed from 350 to 700 K.
There appear to be two states in the O(1s) spectrum at 700 K,
both involving oxygen bound to carbon. One is assigned to
C-O-C bonding, and the other to C-O-Ag species. The
carbon remaining is, on the basis of the measured BE, graphitic
in nature.52
To make sure the remaining carbon and oxygen signals in
the XP spectra were not due to background adsorption during
the anneal series and/or X-ray-induced decomposition of biacetyl
over extended periods of time, the sample was sputtered clean
and annealed. One monolayer of biacetyl was then dosed and
held for the same time used in the electron irradiation experi-
ments, but with the electrons off. The sample was then flashed
immediately to 700 K. No C or O was detected; i.e., the spectra
were not distinguishable from the clean spectra 13 and 14.
4.6. UPS and Work Function Changes. Following the
same annealing procedure, UPS shows changes in intensity and
energy distribution consistent with the above model. Because
there are a number of adsorbed species contributing, these
valence band spectra are not analyzable in terms of species
present. After heating to 700 K, the UPS spectrum is dominated
The spectra, immediately after irradiation, are adequately fit
with four carbon (283.2, 285.1, 286.9, and 288.1 eV) and four
oxygen chemical states (530.1, 531.8, 532.7, and 533.6 eV).
The chemical assignments are summarized in Table 3. Surface
carbide (283.2 eV) indicates some multiple-bond breaking
initiated by electrons. Upon heating to 200 K, loss of biacetyl
occurs in TPD, and in agreement, the C(1s) intensities at 285.2
and 288.1 eV and O(1s) at 532.7 eV decline. Quantitation of
the peak areas is given in Table 4. Moreover, the intensity lost
is consistent with removal of two carbons for each oxygen, as
expected for biacetyl. Table 5 includes the average C/O ratios
for the material removed in successive anneals. In the second
anneal, 275 K, the 288.1 eV C(1s) and 532.7 eV O(1s) peaks
become negligible, consistent with complete removal of mo-
lecular biacetyl. Accompanying this loss, the 531.8 eV O(1s)
intensity, assigned to CdO in an acetyl group, decays, while
the 530.1 eV signal, assigned to O singly bound to both C and
Ag, increases. At the same time, there is as slight downward
shift (0.2 eV) of the 285.2 eV C(1s) peak. To account for these
XPS changes and the desorption of ketene, we propose a surface
reaction in which acetyl groups react to form ketene, which

Chemistry of Biacetyl on Ag(111)
J. Phys. Chem., Vol. 100, No. 39, 1996 15899
(15) Noyes, W. A., Jr.; Mulac, W. A.; Matheson, M. S. J. Chem. Phys.
1962, 36, 880.
by Ag-derived features, but the influence of impurities, i.e., C
and O, is evident in the non-negligible alteration of the Ag
d-band region. After annealing to 700 K, the work function
was, however, indistinguishable from that of clean Ag(111).
(16) Padnos, N.; Noyes, W. A., Jr. J. Phys. Chem. 1964, 68, 464.
(17) Sidebottom, H. W.; Badock, C. C.; Calvert, J. G.; Rabe, B. R.;
Damon, E. K. J. Am. Chem. Soc. 1972, 94, 13.
(18) Kaya, K.; Harshbarger, W. R.; Robin, M. B. J. Chem. Phys. 1974,
60, 4231.
5. Conclusions
(19) Imamura, T.; Onitsuka, O.; Murai, H.; Obi, K. J. Phys. Chem. 1985,
89, 4921.
One monolayer biacetyl adsorbs at 110 K and desorbs
molecularly on Ag(111) at 180 K. Fifty-electronvolt electron
irradiation of biacetyl leads to the breaking of C-C and C-H
bonds with very little, if any, electron-induced formation of
atomic oxygen. Consequently, our model includes CH3, H,
CH3CdO, CH2dCdO, CO, and CH2C-O(a)CdOCH3 as the
major reaction products (shown more fully in Figure 6).
Ejection of CO, CH3, ketene (CH2dCdO), and ethane into the
gas phase occurs during irradiation. CO and CH3 are the major
desorption species. No parent desorption is observed during
irradiation. We propose that impact ionization leads to the
cleavage of the H3C-CO bond and prompt desorption of CH3,
leaving an excited -COCOCH3 species which decomposes to
form acetyl and CO fragments, the latter desorbing. After
irradiation, there is evidence for the following adsorbed spe-
cies: CH3, H, CH3CdO, and CH2C-O(a)CdOCH3. We
propose two reaction pathways for the methane desorption peak
at 235 K: (1) the CH3 group combines with H which was
formed on the surface during irradiation by the breaking of a
C-H bond in biacetyl, and (2) the methyl fragment recombines
with a hydrogen atom from the dehydrogenation of an acetyl
group and, consequently, results in desorption of ketene and
CH4. Ketene desorption at 240 K supports this claim. Methane
desorption at 315 K is attributed to decomposition of CxHyOz,
a process that leaves C(a) and O(a); the O(1s) XPS intensity does
not change when this CH4 desorbs. Finally, some partially
dehydrogenated biacetyl fragments rehydrogenate by scavenging
H from the remaining hydrocarbon species and desorb at 440
K as biacetyl. This leaves only minor quantities, e8% of
original amounts, of adsorbed C and O up to 700 K.
(20) Alvisatos, A. P.; Waldeck, D. H.; Harris, C. B. J. Chem. Phys.
1985, 82, 541.
(21) Qin, Q.; Zhang, Z.; Yang, X.; Zheng, Q. Proc. SPIE-Int. Soc. Opt.
Eng. 1995, 2547, 389.
(22) Pylant, E. D.; Junker, K. H.; Szulczewski, G.; Hubbard, M. J.;
White, J. M. To be submitted for publication.
(23) Pylant, E. D.; Hubbard, M. J.; White, J. M. J. Vac. Sci. Technol. A
1996, 14, 1684.
(24) Koel, B. E.; Peebles, D. E.; White, J. M. Surf. Sci. 1983, 125, 709.
(25) XPS peak-fitting program: ISTFIT, by C. Truong.
(26) McLafferty, F. W.; Stauffer, D. B. In The Wiley/NBS Registry of
Mass Spectral Data; Wiley: New York, 1989.
(27) The following procedure was used: (1) the 43 amu signal at 450
K was taken as background level and subtracted uniformly, (2) the resulting
signal was plotted in semilog form, i.e., ln(43 amu signal) versus time (not
temperature), and the slope between 220 and 320 K (70-100 s) was used
as the pumping speed characteristic, (3) from the slope determined in step
2, a pumping speed correction was computed and subtracted from the 43
amu signal for temperatures higher than the peak, and finally, (4) a constant
residual was subtracted to force to zero the average signal between 110
and 120 K.
(28) Armstrong, J. L.; Pylant, E. D.; White, J. M. To be submitted for
publication.
(29) Chan, C.-M.; Aris, R.; Weinberg, W. H. Appl. Surf. Sci. 1978, 1,
360.
(30) Sun, Z.-J.; Mackay, R. S.; White, J. M. Surf. Sci. 1993, 296, 36.
(31) Wassmuth, H.-W.; Ahner, J.; Ho¨fer, M.; Stolz, H. Prog. Surf. Sci.
1993, 42, 257.
(32) Ahner, J.; Wassmuth, H.-W. Surf. Sci. 1993, 287, 125.
(33) Jakob, P.; Menzel, D. Surf. Sci. 1989, 220, 70.
(34) Klekamp, A.; Umbach, E. Surf. Sci. 1991, 249, 75.
(35) Tyrrell, J. J. Am. Chem. Soc. 1979, 101, 3766.
Further investigations are underway to examine biacetyl
adsorption geometries by FTIR, the variation, with initial
biacetyl coverage, of the electron-induced products, and the
photon-induced chemistry of adsorbed biacetyl.
(36) Gelius, V.; Heden, P. F.; Hedman, J.; Lindberg, B. J.; Manne, R.;
Nordberg, R.; Nordling, C.; Siegbahn, K. Phys. Scr. 1970, 2, 70.
(37) Wagner, C. D.; Zatko, D. A.; Raymond, R. H. Anal. Chem. 1980,
52, 1445.
(38) Gerenser, L. J.; Baetzold, R. C. Surf. Sci. 1980, 89, 259.
(39) Von Niessen, W.; Bieri, G.; Asbrink, L. J. Electron Spectrosc. Relat.
Phenom. 1980, 21, 175.
Acknowledgment. This work is supported in part by NSF
Grant CHE9319640 and by the Robert A. Welch Foundation.
E.D.P. would like to thank M. F. Arendt and A. L. Schwaner
for their helpful discussions.
(40) Walzl, K. N.; Xavier, I. M., Jr.; Kuppermann, A. J. Chem. Phys.
1987, 86, 6701.
(41) Roop, B.; Blass, P. M.; Zhou, X.-L.; White, J. M. J. Chem. Phys.
1989, 90, 608.
References and Notes
(42) Zhou, X.-L.; Coon, S.; White, J. M. J. Chem. Phys. 1990, 92, 1504.
(43) Madey, T. E.; Yates, J. T., Jr. J. Vac. Sci. Technol. 1971, 8, 525.
(44) Zhou, X.-L.; White, J. M.; Koel, B. E. Surf. Sci. 1989, 218, 201.
(1) Zhou, X.-L.; Schwaner, A. L.; White, J. M. J. Am. Chem. Soc.
1993, 115, 4309.
(2) See, for example: (a) Zhou, X.-L.; White, J. M. J. Phys. Chem.
1992, 96, 7703. (b) Zhou, X.-L.; Zhu, X.-Y.; White, J. M. Surf. Sci. Rep.
1991, 13, 73. (c) Liu, Z.-M.; Zhou, Z.-L.; White, J. M. Chem. Phys. Lett.
1992, 198, 615. (d) Zhou, X.-L.; White, J. M. Surf. Sci. 1991, 241, 244.
(3) Schwaner, A. L.; White, J. M. To be submitted for publication.
(4) Schwaner, A. L.; Fieberg, J. E.; White, J. M. To be submitted for
publication.
(45) Buchanan, D. A. Doctoral Dissertation, University of Texas at
Austin, 1991.
(46) White, J. M. Langmuir 1994, 10, 3946.
(47) Tolman, C. A.; Riggs, W. M.; Linn, W. J.; King, C. M.; Wendt, R.
C. Inorg. Chem. 1973, 12, 2772.
(48) Barteau, M. A.; Madix, R. J. Surf. Sci. 1982, 120, 262.
(5) Converse, S.; Szulczewski, G.; Szabo, A.; Mahoney, S.; White, J.
M. To be submitted for publication.
(49) Bukhtiyarov, V. I.; Boronin, A. I.; Savchenko, V. I. Surf. Sci. 1990,
232, L205.
(6) Wachs, I. E.; Madix, R. J. Surf. Sci. 1978, 76, 531.
(7) Ayre, C. R.; Madix, R. J. Surf. Sci. 1994, 303, 297.
(8) Sault, A. G.; Madix, R. J. Surf. Sci. 1986, 176, 415.
(9) We use the subscript (a) to indicate which atom(s) bind(s) to the
substrate.
(10) Calvert, J. G.; Pitts, J. N., Jr. In Photochemistry; Wiley and Sons:
New York, 1966.
(11) Noyes, W. A., Jr.; Porter, G. B.; Jolley, J. E. Chem. ReV. 1956, 56,
49.
(50) Muilenberg, G. E. In Handbook of X-ray Photoelectron Spectros-
copy; Perkin-Elmer: Eden Prairie, MN, 1979.
(51) Clark, D. T.; Thomas, H. R. J. Polym. Sci., Polym. Chem. Ed. 1978,
16, 791.
(52) Johansson, G.; Hedman, J.; Berndtsson, A.; Klasson, M.; Nillsson,
R. J. Electron Spectrosc. Relat. Phenom. 1973, 2, 295.
(53) Campbell, C. T. Surface Sci. 1985, 157, 43.
(12) Sidman, J. W.; McClure, D. S. J. Am. Chem. Soc. 1955, 77, 6461.
(13) Sheats, G. F.; Noyes, W. A., Jr. J. Am. Chem. Soc. 1955, 77, 1421.
(14) Bell, W. E.; Blacet, F. E. J. Am. Chem. Soc. 1954, 76, 5332.
(54) Grant, R. B.; Lambert, R. M. Surf. Sci. 1984, 146, 256.
JP9605820