Collision-induced activation of the β-hydride elimination reaction of isobutyl iodide dissociatively chemisorbed on Al(111)

Article abstract of DOI:10.1063/1.476294

The collision-induced activation of the endothermic surface reaction of isobutyl iodide chemisorbed on an Al(111) surface is demonstrated using inert-gas, hyperthermal atomic beams. The collision-induced reaction (CIR) is highly selective towards promoting the β-hydride elimination pathway of the chemisorbed isobutyl fragments. The cross section for the collision-induced reaction was measured over a wide range of energies (14-92 kcal/mol) at normal incidence for Ar, Kr, and Xe atom beams. The CIR cross section exhibits scaling as a function of the normal kinetic energy of the incident atoms. The threshold energy for the β-hydride elimination reaction calculated from the experimental results using a classical energy transfer model is ～1.1 eV (～25 kcal/mol). This value is in excellent agreement with that obtained from an analysis of the thermally activated kinetics of the reaction. The measured cross section shows a complex dependence on both the incident energy of the colliding atom and the thermal energy provided by the surface where the two energy modes are interchangeable. The dynamics are explained on the basis of an impulsive, bimolecular collision event where the β-hydride elimination proceeds via a possible tunneling mechanism. The threshold energy calculated in this manner is an upper limit given that it is derived from an analysis which ignores excitations of the internal modes of the chemisorbed alkyl groups.

Full text of DOI:10.1063/1.476294

Collision-induced activation of the β-hydride elimination reaction of isobutyl iodide
dissociatively chemisorbed on Al(111)
Shrikant P. Lohokare, Elizabeth L. Crane, Lawrence H. Dubois, and Ralph G. Nuzzo
Citation: The Journal of Chemical Physics 108, 8640 (1998); doi: 10.1063/1.476294
View online: http://dx.doi.org/10.1063/1.476294
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/108/20?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Hot precursor reactions during the collisions of gas-phase oxygen atoms with deuterium chemisorbed on Pt(100)
J. Chem. Phys. 126, 134704 (2007); 10.1063/1.2713111
Quantum study of Eley-Rideal reaction and collision induced desorption of hydrogen atoms on a graphite
surface. I. H-chemisorbed case
J. Chem. Phys. 124, 124702 (2006); 10.1063/1.2177654
Surface induced dissociations of protonated ethanol monomer, dimer and trimer ions: Trimer break-down graph
from the collision energy dependence of projectile fragmentation
J. Chem. Phys. 118, 7090 (2003); 10.1063/1.1556851
The chemisorption and dissociation of ethylene on Pt{111} from first principles
J. Chem. Phys. 110, 4699 (1999); 10.1063/1.478356
Collision induced desorption and dissociation of O 2 chemisorbed on Ag(001)
J. Chem. Phys. 109, 2490 (1998); 10.1063/1.476820
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

JOURNAL OF CHEMICAL PHYSICS
VOLUME 108, NUMBER 20
22 MAY 1998
Collision-induced activation of the ␤-hydride elimination reaction
of isobutyl iodide dissociatively chemisorbed on Al„111…
Shrikant P. Lohokare and Elizabeth L. Crane
School of Chemical Sciences, Department of Materials Science and Engineering and the Frederick Seitz
Materials Research Laboratory, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801
Lawrence H. Dubois
Defense Advanced Research Projects Agency, Arlington Virginia 22203, and MIT Lincoln Laboratories,
Lexington, Massachusetts 02173
Ralph G. Nuzzo
School of Chemical Sciences, Department of Materials Science and Engineering and the Frederick Seitz
Materials Research Laboratory, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801
͑Received 25 September 1997; accepted 19 February 1998͒
The collision-induced activation of the endothermic surface reaction of isobutyl iodide chemisorbed
on an Al͑111͒ surface is demonstrated using inert-gas, hyperthermal atomic beams. The
collision-induced reaction ͑CIR͒ is highly selective towards promoting the ␤-hydride elimination
pathway of the chemisorbed isobutyl fragments. The cross section for the collision-induced reaction
was measured over a wide range of energies ͑14–92 kcal/mol͒ at normal incidence for Ar, Kr, and
Xe atom beams. The CIR cross section exhibits scaling as a function of the normal kinetic energy
of the incident atoms. The threshold energy for the ␤-hydride elimination reaction calculated from
the experimental results using a classical energy transfer model is ϳ1.1 eV (ϳ25 kcal/mol). This
value is in excellent agreement with that obtained from an analysis of the thermally activated
kinetics of the reaction. The measured cross section shows a complex dependence on both the
incident energy of the colliding atom and the thermal energy provided by the surface where the two
energy modes are interchangeable. The dynamics are explained on the basis of an impulsive,
bimolecular collision event where the ␤-hydride elimination proceeds via a possible tunneling
mechanism. The threshold energy calculated in this manner is an upper limit given that it is derived
from an analysis which ignores excitations of the internal modes of the chemisorbed alkyl groups.
© 1998 American Institute of Physics. ͓S0021-9606͑98͒02420-9͔
environment.8 For example, kinetic-energy enhanced activa-
tion of neutral species finds important applications in the
technologies used for the deposition10–12 and etching of
silicon.13 Translational activation of molecular precursors is
not a completely general strategy for materials growth, how-
ever, since in many cases thermal decomposition may result
from the use of high nozzle temperatures in the beam source.
Complementing translational activation is the closely re-
lated process of collision-induced activation, in which en-
ergy transfer to the adsorbate is effected by an energetic
collision involving a second gas-phase species of high trans-
lational energy. An early discussion of the dynamics of
collision-induced activation of surface bound adsorbates was
presented by Zeiri, Low, and Goddard.14 These results, based
on classical stochastic trajectory theory, have been expanded
in more recent studies by Zeiri et al.,15,16 and Rabitz
et al.17,18 From an experimental stand point, Ceyer et al.
demonstrated both the collision-induced desorption
INTRODUCTION
The effects of surface structure and temperature on the
reactivity of adsorbates under equilibrium conditions have
been studied extensively.1–3 These studies have yielded a
good understanding of thermally activated surface processes
important in diverse areas of technology, especially as re-
gards materials growth and catalysis.1–7 The dynamical ef-
fects of energetic atom or molecule collisions with surfaces
have received much less attention ͑in part due to the complex
experimental techniques needed to study such phenomena͒,
and as a result are less well understood. Molecular beams
producing high velocity atoms or molecules in the hyperther-
mal energy range can be coupled with sensitive, ultrahigh
vacuum spectroscopies in order to investigate the effect of
collision-induced processes at the microscopic level.8 Super-
sonic beams have two useful qualities: They can be gener-
ated with a narrow velocity distribution and a wide range of
translational energies.9 The utility of the beam as an experi-
mental tool stems from the fact that the kinetic energy of the
collision can be utilized to overcome the activation barrier of
a surface reaction. The conditions achieved can approximate
the high energy collisions which occur at high pressure, tem-
perature, or in plasmas and allow activated and/or low prob-
ability processes to be simulated inside an ultrahigh vacuum
19–23
21–25
͑CIDE͒
and collision induced dissociation ͑CID͒
of
methane on a Ni͑111͒. They proposed a model for the kinetic
energy transfer from the incident inert gas atom to the phy-
sisorbed CH4 molecule involving an impulsive, bimolecular
collision which induces the activated dissociative adsorption
of the CH4. These investigations have inspired more recent
0021-9606/98/108(20)/8640/11/$15.00
8640
© 1998 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

J. Chem. Phys., Vol. 108, No. 20, 22 May 1998
Lohokare et al.
8641
studies on other small, symmetric adsorbates, including
ammonia26,27 and ethylene28 on Pt͑111͒ and water on
Ru͑001͒.29 Levis et al., made a direct measurement of the
threshold energy for collision-induced desorption of NH3
from the Pt͑111͒ surface by a beam of neutral, hyperthermal
Ar atoms and found that the total energy scaling of the de-
sorption cross section holds for various angles of
incidence.26 In contrast, Asscher et al., observed efficient ki-
netic energy dissipation by the hydrogen bonded network of
a dense monolayer of water on the Ru͑001͒ surface in an
investigation of H2O desorption effected by collisions with
energetic Kr atoms.29 Despite this considerable progress, the
feasibility of effecting collision-induced surface reactions in
large, multiatomic adsorbates ͑both physisorbed and chemi-
sorbed͒ remains a largely unexplored area of research. One
expects intuitively that the probability of bond dissociation
by means of an energetic collision will be reduced given the
increased possibilities for internal energy dissipation and the
exceedingly rapid rates of both energy transfer and quench-
ing. However, it must be pointed out that the cross sections
for certain types of selective bond scission can be substan-
tial. Levis and Velic,28 for example, have demonstrated that
the collision-induced desorption of ethylene from the Pt͑111͒
surface is selective in desorbing the ␲-bonded form of C2H4
over that of the ␴-bonded species. The ability to activate
reactions in complex adsorbates would be of great utility in
exploring activated chemisorption phenomena such as those
involved in the technologically important areas of catalysis3,8
and materials growth and processing.4,5,8,30–32 The present
study explores the feasibility of using collision-induced pro-
cesses to understand the dynamics and energetics of reac-
tions involved in the chemical vapor deposition ͑CVD͒ of
aluminum.
FIG. 1. A schematic sketch of the molecular beam source, differentially
pumped stages and the scattering chamber.
The thermal chemistries of the surface bound isobutyl
and iodine moieties were originally studied by Bent et al.37
In the study described here, neutral, inert-gas ͑Xe, Kr, and
Ar͒ beams with energies in the range of 0.5–4 eV were scat-
tered off of a monolayer of isobutyl groups chemisorbed on
Al͑111͒. The experimental results suggest that the CID pro-
cess predominantly promotes the ␤-hydride elimination reac-
tion. The value of the activation barrier for the process is
estimated to be ϳ1.1 eV (ϳ25 kcal/mol), in good agree-
ment with the thermal barrier measured by temperature pro-
grammed reaction spectroscopy ͑TPRS͒.
EXPERIMENTAL APPARATUS AND TECHNIQUES
The apparatus used in these investigations, illustrated in
Fig. 1, combines a supersonic molecular beam with several
UHV surface analytical tools. Specific details of the latter
techniques have been described elsewhere.38 The features
pertinent to the CID studies are described below.
Aluminum metallization in very large scale integrated
͑VLSI͒ devices is of significant technological importance.5,33
One of the fabrication processes of current commercial inter-
est is the chemical vapor deposition and growth of aluminum
thin films by the surface-mediated thermal decomposition of
triisobutylaluminum ͑TIBA͒.5,6,33–35 The dissociation and re-
As shown in Fig. 1, the chamber has three main vacuum
stages—two stages are employed in the beam generation and
the third serves as the scattering and analytical chamber. The
supersonic molecular beam source is housed in the first stage
of the differentially pumped two stage beam chamber. The
beam is formed by the continuous adiabatic expansion of the
gas at high stagnation pressures from a 25 ␮ molybdenum
aperture ͑Ladd Optics͒ mounted by furnace brazing in a
stainless steel seat which in turn is welded to steel tubing
conveying the gas. The aperture seat and the connected tub-
ing is enclosed inside a cylindrical copper block with
grooves holding tungsten wire insulated with ceramic. The
wire was used to ͑resistively͒ heat the beam source from
room temperature to ϳ850 K to adjust the incident beam
energy. The entire source assembly was mounted on a
X–Y –Z translational manipulator. Temperatures were mea-
sured using a chromel-alumel thermocouple spot welded to
the nozzle edge. The beam source chamber was pumped by a
10 in. diffusion pump. A 400 ␮ diameter nickel skimmer
͑model 1, Beam Dynamics͒ mounted on the inside wall of
the first stage facing the source was used to form the beam.
The skimmer-nozzle distance was р1 cm and was adjusted
to optimize the beam intensity and velocity distribution by
minimizing skimmer interference effects.39,40 A collimator
centered in a double sided flange separating the differentially
pumped stages from the scattering chamber was used to fur-
action of TIBA on a non-native substrate ͑e.g., Si or
33,36
SiO2͒
is activated and thus an understanding of the
nucleation chemistry and nucleation morphology is not easy
to study in UHV. This deficit is significant given that the
growth-rate hysteresis and morphology are strongly deter-
mined by these early events and thus of central importance in
determining the lifetimes ͑which are sensitive to microstruc-
ture͒ of devices so constructed.5,32 A goal of the research
reported here is to develop a better understanding of the
mechanisms and energetics involved in the reaction-limited
nucleation of Al–CVD ͑and other͒ thin films. In this report,
we examine the feasibility of using hyperthermal, inert gas
atom beams to activate large surface-bound organometallic
fragments, such as those that might be found in Al–CVD
processes involving TIBA.6 The beam-induced reactions of a
covalently bound isobutyl group are used as a model to ex-
plore this possibility. The present work is advantaged by the
fact that useful models of isobutyl surface intermediates can
be prepared easily by the dissociative chemisorption of
isobutyl iodide on an aluminum surface.37
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

8642
J. Chem. Phys., Vol. 108, No. 20, 22 May 1998
Lohokare et al.
ther collimate the expanding beam to a diameter of ϳ1.5 cm
at the sample. This ensures that the crystal is flooded with a
uniform gas flow. A manual shutter attached to a rotary mo-
tion feed-through was used to shunt the beam flow through
the skimmer in order to measure and minimize the back-
ground gas load. A similar shutter was also housed at the
entry point inside the scattering chamber in order to measure
the effusive load contributed by the beam source.
The scattering chamber is pumped by a 10 in. diffusion
pump ͑5000 L/s͒ with a liquid nitrogen trap. Additional
pumping is provided by a titanium sublimation pump and a
540 L/s turbomolecular pump. The base pressure of the UHV
chamber is typically р2ϫ10Ϫ10 Torr. The chamber is
equipped with two quadrupole mass spectrometers ͑VG SXP
300 Quadrupoles͒, one in line with the beam axis and the
other at 45° with reference to the beam axis, a cylindrical
mirror analyzer ͑CMA͒ with a coaxial electron gun ͑Physical
Electronics͒ for Auger electron spectroscopy ͑AES͒, an ion
gun ͑Physical Electronics͒ for sputtering the crystal, a
Bayard-Alpert type ionization gauge ͑Varian͒, and a preci-
sion leak valve ͑Duniway͒. Both mass spectrometers are
shrouded and differentially pumped by 30 L/s ion pumps
͑Varian͒ and liquid nitrogen cold shields. The coaxial mass
spectrometer was used for the time of flight ͑TOF͒ velocity
distribution analysis of the supersonic molecular beam and
for temperature programmed reaction spectroscopy. Specific
details regarding sample preparation and TPRS acquisition
are provided elsewhere.41
FIG. 2. TPR spectra of m/eϭ41 as a function of exposures of isobutyl
iodide to an Al͑111͒ surface at 300 K.
concentrations. They were further tested for purity by mass
spectroscopy and also by monitoring the exposed crystal
with Auger electron spectroscopy for adsorbed contaminants
deposited by the beam. Both these control tests showed the
presence of little or no contaminants ͑such as H2O, CO2, and
O2͒.
The kinetic energy distribution of the beam was deter-
mined by a standard time of flight method.39 The time of
flight was measured between a rotating chopper in the sec-
ond stage, at a distance of ϳ15 cm from the nozzle, and the
line-of-sight mass spectrometer in the scattering chamber.
The distance between the chopper and the mass spectrometer
is ϳ98 cm. The chopper is an aluminum disk with two di-
agonally opposite, precision machined slits ͑1 cm long, 1
mm wide͒ mounted on a two way 800 Hz ac synchronous
motor ͑Globe͒ which is controlled by a power supply with a
variable frequency control. The chopper modulates the beam
at 175 Hz with a gating function of 5.85 ␮s FWHM. The
output pulses of the mass spectrometer signal were collected
in a pulse-counting mode using a multichannel scalar
͑EG&G Ortec MCS PCB͒ with 256 channels and temporal
resolution of 1 ␮s. The time base of the scalar is triggered by
the signal from a light emitting diode/photodetector which
determines the rotational position of the chopper slit and
hence the span of the beam pulse in real time. The signal is
averaged over typically 10 000 chopper counts to obtain the
time of flight spectrum. The kinetic energy spread ͑⌬E/E at
FWHM͒ was typically ϳ40%. To estimate the absolute
beam flux, the mass spectrometer was calibrated for its rela-
tive sensitivity to Ar, Kr, and Xe. The flux weighted distri-
bution was calculated using the Jacobian for conversion from
time to velocity space.39,42 The signals for the most common
isotopes for Ar ͑40͒, Kr ͑84͒, and Xe ͑131͒ were calibrated
against a fixed background pressure.43 The absolute flux was
calculated to be of the order of ϳ(2Ϯ1)
ϫ1013 atoms/cm2/s. All gas mixtures ͑1% in 99.99995%
pure He͒ were obtained from Matheson and were of certified
RESULTS
A study of the thermally activated reactions of isobutyl
iodide on aluminum has been described.37 The thermal dis-
sociation of isobutyl iodide occurs efficiently on aluminum
surfaces.37 Physisorbed isobutyl iodide desorbs at ϳ180 K
but the chemisorbed alkyl and iodine fragments are stable on
the surface to much higher temperatures (Ͼ450 K). The
thermal decomposition of the bound isobutyl group is facile
at temperatures above 450 K. The reaction-limited evolution
of the ␤-hydride elimination reaction product, isobutene,
provides a convenient measure of the reaction kinetics. The
depletion of the surface-bound isobutyl groups also results
͑to a lesser degree͒ from complex, associative metal etching
reactions which yield both hydrido- and haloaluminum and
isobutylaluminum complexes ͓AlI, diisobutylaluminum hy-
dride ͑DIBAH͒, disobutylaluminum iodide ͑DIBAI͔͒.41
A
common fragment useful for the mass spectroscopic detec-
tion of these desorption products ͑excluding AlI͒ is the ion
C3Hϩ5 (m/eϭ41). The desorption yield detected at this mass
can be used to estimate the cross sections of the collision-
induced reactions described in this report. Figure 2 shows the
TPRS data for m/eϭ41 as a function of the increasing isobu-
tyl iodide exposure to the Al͑111͒ surface at 300 K. This
data is included for the convenience of the reader. We note
for later reference that the peak temperature shifts towards
lower values with increasing coverages. The TPRS wave
contains contributions from several products. The leading
edge intensity is weighted by the associative desorption of
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

J. Chem. Phys., Vol. 108, No. 20, 22 May 1998
Lohokare et al.
8643
FIG. 3. TPR spectra of m/eϭ41 taken before and after an exposure of ϳ1ϫ1016 collisions/cm2 of a ͑a͒ 1.4 eV, ͑b͒ 2.7 eV, and ͑c͒ 4.0 eV xenon beam to
an Al͑111͒ surface at 300 K and saturated with isobutyl iodide.
diisobutyl iodide ͑DIBAI, Tmax at ϳ490 K͒.41 The latter por-
beam energy obtained was 2.7 eV while that for Ar was 1.4
eV. The thermally activated barrier for the ␤-hydride elimi-
nation reaction is estimated by TPRS to be in the range of
ϳ1 eV.37,41 On varying the incident beam energy for Ar
from 1 to 1.4 eV, only negligible changes were seen in the
TPRS data for m/eϭ41 ͑data not shown͒. This suggests an
insignificant cross section for CID by the Ar atoms in this
energy range. The net energy transfer is expected to depend
on the incident mass and consequently so should the reaction
cross section. Small, but measurable, reaction probabilities
were observed for both 1.4 eV Kr and Xe beams. The rep-
resentative TPR spectra shown in Fig. 3 were collected for
Xe beams incident at three distinct energies. Figures 3͑a͒–
3͑c͒ show the effect of the normal energy component of the
incident atom on the TPR data for a constant exposure of
ϳ1ϫ1016 collisions/cm2. Isobutyl iodide physisorbs with an
appreciable strength on the walls of the chamber and thus a
low-level flux from the background is difficult to eliminate.
The integrated beam fluxes used were restricted in part mini-
mize errors due to adsorption from the background. Three
major effects are evident in the data shown. First, there is a
marked shift of the desorption peak towards higher tempera-
tures with increasing beam energy. The measured shifts var-
ied from ϳ10 K for the 1.4 eV beam to ϳ50 K for the 4 eV
beam. Second, integration of the peak intensity suggests the
yield of the desorbing products is reduced by ϳ2%, 11%,
and 18%, respectively, with increasing beam energy ͑at a
constant integrated flux͒. Third, the collision process pro-
duces large changes in the TPRS line shapes. The magnitude
of this change scales with the beam energy and, as we show
below, the integrated flux as well.
tion of the peak is more heavily weighted by the desorption
of isobutene ͑which is believed to be the predominant ther-
mal reaction product͒ and diisobutyl aluminum hydride
͑DIBAH, Tmax at ϳ515 K͒. The product partitioning is sen-
sitive to coverage, but weakly varies at coverages within
ϳ20% of saturation. For this reason, a saturation coverage
of isobutyl iodide was used in all the collision-induced reac-
tion experiments reported here.
The collision-induced reaction experiment
The collision-induced reaction experiments were per-
formed at surface temperatures above 200 K to minimize
effects due to the adsorption from the background gases dur-
ing an experimental run. The exposures were made at 300 K
and the sample subsequently cooled. The velocity distribu-
tion of each gas was recorded prior to the collision experi-
ments. After reaching the desired exposure to the beam, TPR
spectra were recorded at m/eϭ41. Control experiments were
performed for each beam energy and exposure used by
blocking the beam path with the shutter and exposing either
the clean or adsorbate covered surface to the beam flux back-
ground. The TPR spectra obtained were identical to those
taken for the same adsorbate coverage but without the beam
exposure. This showed that no reaction or contamination re-
sults from indirect effusive gas load of the beam. Since the
seeded beam mixture contained predominantly helium, we
also examined the ͑remote͒ possibility of a reaction induced
by helium atom collisions. In this control experiment, the
adsorbate covered surface was exposed to a pure helium
beam at various nozzle temperatures. No effect on the adsor-
bates was observed ͑the energy contained in the helium beam
is Ͻ0.4 eV͒.
The response of the TPRS data to varying exposures to
the high energy Xe beam at 4 eV is demonstrated by the data
shown in Fig. 4. The number of Xe collisions was varied
over 1 order of magnitude in the range of 1ϫ1015 to 1
ϫ1016 atoms/cm2. Both the maximum desorption tempera-
ture and the integrated desorption intensity vary progres-
sively over the range of integrated fluxes used. These effects
Observation and results of collision-induced reaction
Collision experiments were carried out in the energy
range of 1–4 eV for the Xe beam. For Kr, the maximum
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

8644
J. Chem. Phys., Vol. 108, No. 20, 22 May 1998
Lohokare et al.
energetic Xe collisions on the relative desorption yield for
each of these products in order to more fully understand the
product partitioning seen in this reaction. Appropriate ioniza-
tion fragments were chosen for monitoring the reactions
yielding the most important products: m/eϭ41 ͑isobutene,
isobutyl compounds͒; m/eϭ2 ͑dihydrogen͒; m/eϭ141
͑DIBAH, DIBAI͒; m/eϭ154 ͑AlI͒; m/eϭ168 ͑DIBAI͒. The
␤-hydride elimination reaction occurs concurrently with the
associative desorption of diisobutyl aluminum hydride and
dihydrogen39 and serves as the required source of surface-
bound hydrogen. Selective activation of the ␤-hydride elimi-
nation reaction at 300 K by the collisional energy transfer is
expected to influence the kinetics of ͑and hence TPRS data
for͒ these reactions. Figures 5͑a͒–5͑c͒ show TPR spectra for
each of the noted masses taken before and after a 4 eV xenon
beam exposure ͑corresponding to an integrated flux of ϳ1
ϫ1016 collisions/cm2͒ at 300 K. From Fig. 5͑a͒, it can be
seen clearly that there is a large ͑almost 70%͒ reduction in
the integrated peak intensity of m/eϭ141 desorption profile.
The two peak profile of the initial state ͑due to a closely
competitive desorption of diisobutyl aluminum iodide at
ϳ490 K and diisobutyl aluminum hydride at ϳ515 K͒ is
lost; it is replaced by a broad feature centered near ϳ515 K.
Figure 5͑b͒ shows a similar and confirming decrease in the
TPR spectra measured at m/eϭ168, an ion specific to
DIBAI. A reduction in the integrated intensity ͑again of
about 70%͒ is seen for the main peak. A simultaneous emer-
gence of another, albeit smaller, peak at ϳ425 K is also evi-
denced. It is noteworthy that this feature is generated only at
the upper end of the incident beam energies studied. The data
suggest that xenon bombardment predominantly depletes
states yielding DIBAI, although the promoted state at 425 K
suggests the effects are kinetically very heterogeneous. Fig-
FIG. 4. TPR spectra of m/eϭ41 as a function of exposures of a 4 eV xenon
beam to an Al͑111͒ surface at 300 K and saturated ͑8 L͒ with isobutyl
iodide. The values next to each spectrum indicate the total flux
(collisions/cm2) exposures of the beam.
also are convolved with substantial changes in the line
shapes of the TPR spectra. We defer further comments on
these points until later.
As noted earlier, the thermal decomposition of isobutyl
iodide on Al͑111͒ produces, in addition to isobutene and
dihydrogen, a variety of aluminum etching products, includ-
ing AlI, DIBAI, and DIBAH, whose yields depend strongly
on adsorbate coverage. We examined of the influence of the
FIG. 5. TPR spectra of ͑a͒ m/eϭ141, ͑b͒ m/eϭ154, and ͑c͒ m/eϭ168 taken before ͑top spectra͒ and after ͑bottom spectra͒ an exposure of ϳ1
ϫ1016 collisions/cm2 flux of a 4 eV xenon beam to an Al͑111͒ surface at 300 K and saturated with isobutyl iodide.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

J. Chem. Phys., Vol. 108, No. 20, 22 May 1998
Lohokare et al.
8645
FIG. 6. TPR spectra of m/eϭ2 taken before ͑top͒ and after ͑bottom͒ an
exposure of ϳ1ϫ1016 collisions/cm2 flux of 4 eV xenon beam to an
Al͑111͒ surface at 300 K and saturated with isobutyl iodide.
FIG. 7. TPR spectra of m/eϭ41 for an Al͑111͒ surface ͑at 300 K͒ saturated
with isobutyl iodide and exposed to
a
constant flux of
1
ϫ1016 collisions/cm2 for krypton at 2.7 eV.
ure 5͑c͒ shows data recorded for m/eϭ154, the AlIϩ frag-
ment. The data suggest, again, that low energy states yield-
ing this fragment ͑ones due to DIBAI͒ are depleted. It is
most interesting to note that the higher energy states ͑pre-
sumably representing the desorption of AlI as a stable di-
atomic molecule͒ are enhanced. The broad tailing seen on
the high temperature feature is not understood but may be
due to desorption of AlI from heterogeneous surface sites.
The complex line shapes make integration difficult and, thus,
no attempt has been made to quantify the changes in inten-
sity seen here. The remaining product which results from the
␤-hydride elimination process is dihydrogen. Figure 6 shows
desorption spectra of m/eϭ2 taken before ͑upper curve͒ and
after ͑lower curve͒ collision-induced activation with a 1
ϫ1016 atoms/cm2 exposure of the 4 eV xenon atom beam.
The ‘‘simple’’ pattern of the spectrum shown in the upper
trace has been analyzed elsewhere.39 The complexity of the
desorption profile seen in the post collision spectrum ͑lower
trace͒ is both intriguing and presently not understood. The
possibility of the hydrogen coming from the adsorption of
water from the background gases was ruled out by both mass
spectroscopy ͑no increase in water peak intensity is observed
in the residual gas analyzer during beam exposure͒ and AES
͑which shows no signs of any surface oxygen or Al–O bond-
ing͒. One possibility we have considered but have not been
able to firmly establish is that of multiple desorption states of
hydridoaluminum halides (AlHxI3Ϫx). We defer further
comment on this point to later.
ference viz. krypton at this energy. This conclusion was fur-
ther supported by monitoring the behavior of the m/eϭ154
desorption profile in an identical set of experiments ͑data not
shown͒. The intensity and line shape changes seen in the 490
and 625 K peaks were similar.
An adsorbate covered surface maintained at a constant
surface temperature during a collision experiment is under
thermal equilibrium. Thermal energy participation is incon-
sequential for a process where the colliding species hits the
surface from a distance in a rapid action. For example, in the
collision-induced dissociation of methane on Ni͑111͒,25 the
methane is propelled towards the surface with the energy
acquired from the collision with the incident inert gas atom.
In this billiard-ball-like mechanism, the surface energy plays
no role ͑to a first approximation͒ in the dissociation process.
We carried out collision-induced reaction experiments at two
different surface temperatures ͑200 and 300 K͒ to test this
hypothesis for isobutyl groups on Al͑111͒. These data are
shown in Fig. 8. For both experiments, a 4 eV Xe beam was
used. At the lower surface temperature, the measured reduc-
tion in integrated intensity for a fixed integrated flux of Xe
was always slightly less. Perhaps more interesting, the mea-
sured shift in peak maximum was ϳ50 K at the higher sur-
face temperature case but approaches ϳ80 K in the 200 K
experiment. Differences in peak broadening were also noted.
It thus appears that a complex coupling of the incident and
surface thermal energies is observed in this process.
The influence of the mass of the incident atoms in the
collision dynamics was investigated using a substitution of
Kr ͑Fig. 7͒ for Xe ͓Fig. 3͑b͔͒ in the seeded gas at fixed
kinetic energy ͑2.7 eV͒ and gas dose (ϳ1
ϫ1016 collisions/cm2). Comparison of these two figures
shows that the more massive xenon has only a marginal dif-
DISCUSSION
The absolute cross section for collision-induced
reaction of isobutyl iodide on Al„111…
Every reaction has a finite probability of occurrence for
a given set of controlling parameters determined by the re-
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

8646
J. Chem. Phys., Vol. 108, No. 20, 22 May 1998
Lohokare et al.
FIG. 8. Plot of reduction in integrated desorption intensity of m/eϭ41 as a
function total flux of 4 eV xenon beam incident on an Al͑111͒ surface
saturated with isobutyl iodide at 300 K ͑top curve͒ and 200 K ͑bottom
curve͒, respectively.
FIG. 9. Plot of percentage reduction in the integrated desorption intensity of
m/eϭ41 as a function of total flux exposure of 1.4, 2.7, and 4.0 eV xenon
beam to an Al͑111͒ surface at 300 K and saturated with isobutyl iodide.
where ͚
is the absolute cross section for the collision
action cross section.44 The most-likely collision-induced re-
action ͑CIR͒ in the present case ͑see below͒ is the unimo-
lecular ␤-hydride elimination ͑i.e., dissociative conversion͒
of the isobutyl species to isobutene. The energy transfer and
conversion therefore are given as
CIR
induced reaction, ⍜IB is the coverage of isobutyl groups at
time tϭ0 and tϭt, and FB is the absolute flux of the inert
gas. The results are presented in Fig. 10.
Typically, the collision energy dependence of the reac-
tion cross section, ͚R(E), can be approximated by an equa-
tion of the form45
B E ͒ϩiso-C H9ads→C4H8ads/gasϩB E ͒ϩH
,
ads
͑1͒
͑
͑
i
4
f
where Ei and Ef are the initial and the final energies of the
colliding atom, B. We present a more detailed discussion of
the dynamics and the constraints which preclude the thermal
activation of more complex pathways via collisions in the
hyperthermal energy range below. As noted above, Ar colli-
sions of 1.4 eV were essentially incapable of activating any
reaction for the integrated fluxes used. Both Kr and Xe col-
lisions showed larger apparent cross sections, however. Fig-
ure 9 shows a more quantitative depiction of the energy scal-
ing seen. The plot shows the experimentally measured
reduction ͑in percent͒ in the integrated m/eϭ41 TPRS inten-
sity as a function of the integrated flux ͑i.e., the collision
density͒ of the xenon beam at three different energy values
(Tsurfϭ300 K). We discuss this energy scaling in detail be-
low.
Collision-induced activation and the energy transfer
dynamics
Using the data in Fig. 9 obtained at an integrated flux of
ϳ7ϫ1015 collisions/cm2 at each of the energies, the varia-
tion of the cross section was calculated as a function of the
incident atom kinetic energy using the expression25,26
FIG. 10. Plot of ␤-hydride elimination reaction cross section (Å2) as a
function of the normal kinetic energy of the xenon beam incident on an
Al͑111͒ surface. The sample was dosed using a 8 L exposure of isobutyl
iodide at 300 K. A constant flux of ϳ0.7ϫ1016 collisions/cm2 was used to
collect each data point.
ln ⍜tϭ0/⍜tϭt
͒
IB
͑
IB
⌺
E ͒ϭ
i
,
͑2͒
͑
CIR
FBt
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

J. Chem. Phys., Vol. 108, No. 20, 22 May 1998
Lohokare et al.
8647
n
num effective mass is used to account for the recoil effect.
The energy that would be retained in the recoiling adsorbate
species is given by
EϪE ͒
͑
0
⌺
E͒ϭA
,
͑3͒
͑
R
E
where E is the collision energy and A and n are fitting pa-
rameters. The translational energy threshold E0 for a chemi-
cal reaction can be defined as the minimum kinetic energy of
the colliding species required for the reaction. In the present
case, E0 corresponds to the minimum translational energy
required of a colliding inert gas atom to overcome the acti-
vation barrier. Equation ͑3͒ is derived from a statistical treat-
ment of simple gas-phase collision-induced dissociation
events of the type AϩBC→AϩBϩC.46–50 Fitting the data
in Fig. 10 to this model, we estimate the threshold energy for
the CID of the isobutyl groups, ͑presumably via a ␤-hydride
elimination reaction͒ on the Al͑111͒ surface to be ϳ1.1 eV.
The fitting parameter, n, has a value of 2 (Ϯ0.2). Values of
n between 1–2.5 have been reported for a variety of gas
phase, collision-induced dissociation systems based on both
experimental27,51–53 and theoretical studies.54 The calculated
threshold agrees well with that obtained from TPRS mea-
surements of the decomposition of isobutyl iodide on single-
crystalline aluminum surfaces,37 where values of ϳ1 eV on
Al͑111͒ and ϳ1.2 eV on Al͑100͒ were obtained. The close
correspondence of the latter values to the threshold energy
calculated here provides strong support for the inference that
the CID of isobutyl moieties involves activation of only the
unimolecular decomposition pathway ͑i.e., ␤-hydride elimi-
nation͒.
The structures of the adsorbed isobutyl and iodine moi-
eties ͑viz. coverage and, for the former, orientation͒ are not
precisely known. Even so, it can be assumed that two types
of collision events take place between the surface bound spe-
cies and the colliding atoms. The collisions of the inert gas
atoms occurring with either the isobutyl group or the iodine
atom are bimolecular in nature. Collisions may also occur
with the substrate atoms ͑a process whose formal molecular-
ity depends on the kinetic energy of the incoming inert gas
atom and the nature of the substrate͒. We consider first bi-
molecular collisions between the surface adsorbates and the
inert gas atom. To a first approximation, the ͑direct͒ collision
events can be modeled as being impulsive, where the energy
transfer occurs in a hard sphere fashion without providing
time for energy transfer to the internal modes. The corruga-
tion depth of the attractive potential is not significant in com-
parison with the translational energy of the incoming atoms.
The energy transfer from the inert gas atom to the adsorbate
thus can be estimated as55
EBf ϭEiB 1Ϫ4m m / m ϩm
͒,
͑5͒
2
͑
͓
͔
B
eff
B
eff
where meff is the effective surface mass and is an approxi-
mation of the collective mass participating in the collision
event.56,57 This analysis predicts that the energy lost by the
isobutyl group to the surface effective mass is ϳ100% while
that for the iodine atom is ϳ87%. It therefore does not re-
quire an explicit calculation to show that the recoil energy is
negligible in both the cases and is insufficient to overcome
the adsorbate-surface bond strength.6
For both near-threshold desorption and sputtering, the
collision events occur at either the topmost or second layer
of the substrate and the minimum energy loss process corre-
sponds to a mechanism in which the particle with the small-
est mass moves within it.58 The estimated threshold value for
the incident energy which can bring about such a desorption
of an isobutyl radical or iodine atom ͑where the bond ener-
gies are ϳ63 kcal/mol35 and ϳ80 kcal/mol,59 respectively͒
is much larger than the experimentally determined value and,
thus, there appears little likelihood of effecting collision-
induced desorption of either these species at these beam en-
ergies.
The third excitation process noted above involves the
direct interaction of the colliding atom with the aluminum
substrate atoms. A theoretical treatment of energy transfer
between a hyperthermal inert gas atom and copper surface60
yields a value of ϳ80% in the range of energies appropriate
to this study. Also, it was observed experimentally that the
energy lost by a xenon atom with an incident energy of
у4 eV, to a Pt or Au surface was ϳ60%. Thus, for the
scattering of Xe from Al, the energy transfer should be quite
efficient. Assuming ͑based on these precedents͒ a broad
range for the energy transfer to the surface of 60%–80%
implies a value between 2.4–3.2 eV for the highest energy ͑4
eV͒ xenon beam. As a result, a ‘‘hot spot’’ could be formed
and this, in turn, could result in an indirect energy transfer to
the adsorbate. Zeiri et al.15,16 noted for CIDE studies of N2
on a W surface that the importance of such a mechanism was
insignificant. They proposed that the only indirect mecha-
nism contributing to the desorption process was of a collision
cascade type and only accounted for ϳ10% of the total de-
sorption yield.61 We have shown above that, for a collision
cascade mechanism to occur and cause either the desorption
of aluminum atoms alone or of aluminum isobutyl species,
the required threshold energy would be much larger than the
values under consideration here. These arguments thus sug-
gest that CID in this instance must predominantly involve a
direct process.
EiBϭEiA 4m m / m ϩm ͒,
͑4͒
2
͑
͓
͔
A
B
A
B
where A and B correspond to the inert gas atom and the
adsorbate species, respectively. We estimate the maximum
energy transfer to be ϳ85% for the isobutyl group and
ϳ100% for the iodine atom, respectively.
Having acquired this energy, the adsorbed moieties
translate toward the aluminum substrate until they encounter
the repulsive wall of the potential energy surface. It is there-
fore necessary to estimate how much energy the adsorbates
lose to the surface. To determine this value, a secondary
interaction is defined using a hard cube model.55 The alumi-
The most intriguing fact noted in this work is the appar-
ent coupling of the surface temperature to the translational
energy transfer, as was suggested by the similar cross sec-
tions found for a 4 eV beam at a surface temperature of 200
K and a 2.7 eV beam at a surface temperature of 300 K. This
implies that the lower kBT factor is at least partially compen-
sated by the translational energy of the colliding beam. A
more quantitative analysis of the present data is complicated
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

8648
J. Chem. Phys., Vol. 108, No. 20, 22 May 1998
Lohokare et al.
both by the multiple reaction pathways which operate and
the complex line shapes and peak shifts which accompany
the beam exposure. The possibility of temperature and colli-
sion dependent modifications in structure of the adsorbed
alkyl species has to be taken into consideration in a quanti-
tative analysis since it would be a vital factor in determining
the reaction kinetics and CID cross sections. These issues not
withstanding, it appears likely to us that the barrier to CID
originates in the deformation and repulsive energies required
to move the dissociating species close to the substrate ͑and is
discussed below͒.21
Dubois et al., have investigated the thermal stability and
organization of alkyl groups on an aluminum surface by
means of reflection-absorption infrared spectroscopy
͑RAIRS͒.62 The surface bound isobutyl groups were derived
both from the dissociation of isobutyl iodide as well as from
the adsorption of triisobutyl aluminum ͑TIBA͒ on the sur-
face. They report that the alkyl species maintained a tem-
perature insensitive organization up to surface temperatures
approaching the point of decomposition (ϳ450 K). The
TPRS data measured here ͑viz. line shape and Tmax͒ show a
significant sensitivity to the beam and thus suggest that
changes in the orientation and organization of the chemi-
sorbed isobutyl groups ͑and iodine atoms͒ likely result from
the hyperthermal beam impact. These effects cannot be due
to a simple annealing since thermal cycles below the decom-
position threshold do not produce comparable changes in the
TPR spectra.63 The data therefore firmly establish that effects
͑presumably structural͒ other than those seen during thermal
activation result from the CID process. These effects pro-
gressively complicate the thermal decomposition kinetics
measured for the remaining adsorbed species by TPRS. We
note, though, that the qualitative aspects of the product par-
titioning appear unchanged ͑at least for the conditions used
here͒. It is interesting that, whatever the precise structural
features which operate here, thermal annealing does not
‘‘undo’’ the lineshape and peak temperature effects seen in
the post-exposure TPRS data. We believe this rate-structure
sensitivity suggests an important role for nonequilibrium ad-
sorbate overlayer structures in this system.
FIG. 11. Auger electron spectra of Al͑111͒ surface; saturated with isobutyl
iodide at 300 K and saturated with isobutyl iodide followed by exposure to
a 1ϫ1016 collisions/cm2 flux of 4 eV xenon.
centered at ϳ625 K ͑corresponding to AlIx͒ after the beam
exposure. If the beam were inducing an iodine consuming
reaction, the reduced iodine concentration on the surface
would also have affected the formation of this product ad-
versely. A further, albeit qualitative confirmation of this in-
ference came from monitoring the surface by AES. Figure 11
shows AES data taken for an Al͑111͒ surface under a variety
of conditions ͑a low beam current was used to minimize
electron-stimulated desorption͒. The upper panel is for a sur-
face at 300 K saturated with isobutyl iodide. The second is a
spectrum taken after exposing a similarly prepared surface to
a 4 eV xenon beam with an integrated flux of ϳ1
ϫ1016 collisions/cm2. The spectra, scaled to a common alu-
minum peak intensity to enable comparison of the relative
intensities of C and I, show a clear reduction in the intensity
of the C͑KLL͒ peak at 272 eV and almost no change in
intensity for the I͑MNN͒ peak at 511 eV after the beam
exposure. The decrease seen in the carbon coverage is ap-
proximately the same as that suggested by the TPRS results.
The reduced concentration of the surface isobutyl groups as
suggested by both AES and TPRS is expected to affect the
associative product forming reactions in a quantitatively pro-
portional manner. Post beam exposure desorption intensities
͑Fig. 5͒ do indeed show that the yields of aluminum contain-
ing ions are greatly reduced ͑m/eϭ141 and 168͒, with
ϳ80% reductions being seen for the maximum exposures
used. Comparing this with the coverage dependent yields
Mechanism of collision-induced reaction and
selectivity
We discussed the three pathways of energy transfer in
the collision induced activation of isobutyl and iodine spe-
cies in the previous section. The possibilities of a sputtering
mechanism and radical desorption were ruled out on the ba-
sis of kinematic arguments. It is seen in the plot of cross
section vs. incident beam energy ͑Fig. 9͒, that the threshold
energy required for collision-induced activation is similar to
the activation barrier for a ␤-hydride elimination reaction.
However, the thermal chemistry of this system has other re-
actions with similar free energies of activation that also con-
sume both aluminum and surface bound species at rates com-
petitive with this particular process. Even so, iodine
consuming reactions do not appear to be important. For ex-
ample, the data presented in Fig. 3͑c͒ shows that, although
there is a change in line shape, there is little or no change
seen in the integrated intensity of the broad set of peaks
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

J. Chem. Phys., Vol. 108, No. 20, 22 May 1998
Lohokare et al.
8649
seen in the TPRS experiments indicates that it matches well
with the results obtained for an initial 2 L exposure of isobu-
tyl iodide.64
6 L. H. Dubois, B. E. Bent, and R. G. Nuzzo, Surface Reactions, edited by
R. J. Madix, ͑Springer, Berlin, 1994͒, Vol. 34, p. 135.
7 M. K. Weldon and C. M. Friend, Chem. Rev. 96, 1391 ͑1996͒.
8 S. T. Ceyer, D. J. Gladstone, M. McGonigal, and M. T. Schulberg, Physi-
cal Methods of Chemistry, 2nd ed., edited by B. W. Rossiter, and R. C.
Baetzold ͑Wiley, New York 1993͒.
The final issue we address is the mechanism of the
collision-induced ␤-hydride elimination reaction. We noted
earlier that the energy transfer is mass dependent and, for a
large mass atom such as xenon, the impact should essentially
crush the isobutyl groups. A similar mechanism was pro-
posed by Ceyer et al.,25 in which a dissociating methane
molecule is trapped between the Kr ͑the highest mass used in
that study͒ and a nickel surface. The deformation of the tet-
rahedral structure of the methane molecule caused by the
impact brings the carbon closer to the metal surface and al-
lows a hydrogen from the deformed C–H bond to tunnel to
the metal surface. We believe a similar mechanism may be
operating in the present system as well. Such mechanisms
appear to be general, at least on the basis of the reports
available in the literature which describe the CID of
methane65–67 and ethane65,68,69 on various metal surfaces. It
is interesting to note that, for the latter, the preferred disso-
ciation is that of the C–H and not the C–C bond. Madix and
McMaster68 have shown that the experimental results are in
excellent agreement with those predicted from the Eckart
tunneling model70 where the motion along the one-
dimensional reaction coordinate for the lowest energy tun-
neling pathway are well represented by the Eckart potential.
The activation of the ␤-hydride elimination reaction with a
threshold energy as small as ϳ1.1 eV, which coincides with
the thermal activation barrier for the reaction, suggests that
the reaction proceeds via a non-RRKM type mechanism.44 If
the reaction followed an equilibrium energy dissipation path-
way, where the energy was distributed uniformly to all the
internal modes, then the total energy required to activate the
C–H bond would be much larger than is feasible via energy
transfer from a hyperthermal neutral gas molecular beam.
Since the interaction time of the colliding atom with the
adsorbate is of the order of the vibrational lifetime, the col-
lision can be viewed as being totally impulsive in nature. It is
possible that the isobutyl fragment is oriented on the surface
with its methyls pointing away from the surface ͑the RAIRS
data are consistent with this picture͒.63 A collision-based dis-
tortion would bring the activatable beta hydrogen and the
surface into close proximity. Further development of this ar-
gument requires a more detailed knowledge of the structural
aspects of the surface bound isobutyl groups than is currently
available.
9 C. R. Arumainayagam and R. J. Madix, Prog. Surf. Sci. 38, 1 ͑1992͒.
10 L. Q. Xia and J. R. Engstrom, J. Chem. Phys. 101, 5329 ͑1994͒.
11 L. Q. Xia, M. E. Jones, J. N. Maity, and J. R. Engstrom, J. Chem. Phys.
103, 1691 ͑1995͒.
12 K. A. Pacheco, B. A. Ferguson, C. Li, S. John, S. Banerjee, and C. B.
Mullins, Appl. Phys. Lett. 67, 2951 ͑1995͒.
13 S. R. Leone, Jpn. J. Appl. Phys., Part 1 34, 2073 ͑1995͒.
14 Y. Zeiri, J. J. Low, and W. A. Goddard III, J. Chem. Phys. 84, 2408
͑1986͒.
15 Y. Zeiri, Surf. Sci. 231, 404 ͑1990͒.
16 Y. Zeiri and R. R. Lucchese, J. Chem. Phys. 94, 4055 ͑1991͒.
17 E. Vilallonga and H. Rabitz, J. Chem. Phys. 97, 1562 ͑1992͒.
18 K. Yang, H. Cheng, E. Vilallonga, and H. Rabitz, Surf. Sci. 326, 177
͑1995͒.
19 J. D. Beckerle, A. D. Johnson, and S. T. Ceyer, Phys. Rev. Lett. 62, 685
͑1989͒.
20 J. D. Beckerle, A. D. Johnson, and S. T. Ceyer, J. Chem. Phys. 93, 4047
͑1990͒.
21 S. T. Ceyer, Science 249, 133 ͑1990͒.
22 S. T. Ceyer, Langmuir 6, 82 ͑1990͒.
23 J. D. Beckerle, A. D. Johnson, Q. Y. Yang, and S. T. Ceyer, Solvay
Conference on Surface Science ͑Springer-Verlag, Berlin, 1988͒.
24 J. D. Beckerle, Q. Y. Yang, A. D. Johnson, and S. T. Ceyer, J. Chem.
Phys. 86, 7236 ͑1987͒.
25 J. D. Beckerle, A. D. Johnson, Q. Y. Yang, and S. T. Ceyer, J. Chem.
Phys. 91, 5756 ͑1989͒.
26 G. Szulczewski and R. J. Levis, J. Chem. Phys. 101, 11070 ͑1994͒.
27 G. Szulczewski and R. J. Levis, J. Chem. Phys. 103, 10238 ͑1995͒.
28 D. Velic and R. J. Levis, J. Chem. Phys. 104, 9629 ͑1996͒.
29 L. Romm, T. Livneh, and M. Asscher, J. Chem. Soc. Faraday Trans. 91,
3655 ͑1995͒.
30 J. R. Engstrom, D. A. Hansen, M. J. Furjanic, and L. Q. Xia, J. Chem.
Phys. 99, 4051 ͑1993͒.
31 M. E. Gross, L. R. Harriott, and R. L. Opila, Jr., J. Appl. Phys. 68, 4820
͑1990͒.
32 R. Jairath, A. Jain, M. J. Hampden-Smith, and T. T. Kodas, CVD of Metal,
edited by M. J. Hampden-Smith and T. T. Kodas ͑VCH, Weinheim,
1995͒, p. 1.
33 B. E. Bent, R. G. Nuzzo, and L. H. Dubois, Mater. Res. Soc. Symp. Proc.
101, 177 ͑1988͒.
34 B. E. Bent, R. G. Nuzzo, and L. H. Dubois, Mater. Res. Soc. Symp. Proc.
131, 327 ͑1989͒.
35 B. E. Bent, R. G. Nuzzo, and L. H. Dubois, J. Am. Chem. Soc. 111, 1634
͑1989͒.
36 B. E. Bent, C. T. Kao, B. R. Zegarski, L. H. Dubois, and R. G. Nuzzo, J.
Am. Chem. Soc. 113, 9112 ͑1991͒.
37 B. E. Bent, R. G. Nuzzo, B. R. Zegarski, and L. H. Dubois, J. Am. Chem.
Soc. 113, 1137 ͑1991͒.
38 B. C. Wiegand, S. P. Lohokare, and R. G. Nuzzo, J. Phys. Chem. 97,
11553 ͑1993͒.
39 G. Scoles, D. Bass, U. Buck, and D. Laine, Atomic and Molecular Beam
Methods ͑Oxford University Press, New York, 1988͒, Vol. 1.
40 J. B. Anderson, Molecular Beams and Low Density Gas Dynamics, edited
by P. P. Wegener ͑Marcel Dekker, New York, 1974͒.
41 S. P. Lohokare, E. L. Crane, L. H. Dubois, and R. G. Nuzzo, Langmuir 14,
1328 ͑1998͒.
ACKNOWLEDGMENT
We are grateful for the funding support for this project
by the National Science Foundation ͑Grant Nos. CHE-
9300995 and CHE-9626871͒.
42 J. B. Anderson and J. B. Fenn, Phys. Fluids 8, 780 ͑1965͒.
43 J. A. O’Hanlon, User’s Guide to Vacuum Technology ͑Wiley, New York,
1980͒.
44 R. D. Levine and R. B. Bernstein, Molecular Reaction Dynamics and
Chemical Reactivity ͑Oxford University Press, New York, 1987͒.
45 A. G. Urena, J. Phys. Chem. 96, 8212 ͑1992͒.
1 G. A. Somorjai, Chemistry in Two Dimensions ͑Cornell University Press,
Ithaca, 1981͒.
46 R. D. Levine, Chem. Phys. Lett. 11, 109 ͑1971͒.
47 R. D. Levine and R. B. Bernstein, J. Chem. Phys. 56, 2281 ͑1972͒.
48 C. Rebick and R. D. Levine, J. Chem. Phys. 58, 3942 ͑1973͒.
49 E. E. Park, A. Wagner, and S. Wexler, J. Chem. Phys. 58, 5502 ͑1973͒.
50 A. F. Wagner and E. K. Parks, J. Chem. Phys. 65, 4343 ͑1976͒.
51 E. K. Parks, N. J. Hansen, and S. Wexler, J. Chem. Phys. 58, 5489 ͑1973͒.
2 D. A. King, CRC Crit. Rev. Solid State Mater. Sci. 7, 167 ͑1978͒.
3 B. E. Bent, Chem. Rev. 96, 1361 ͑1996͒.
4 J. G. Ekerdt, Y. M. Sun, A. Szabo, G. J. Szulczewski, and J. M. White,
Chem. Rev. 96, 1499 ͑1996͒.
5 J. R. Creighton and J. E. Parmeter, Crit. Rev. Solid State Mater. Sci. 18,
175 ͑1993͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32

8650
J. Chem. Phys., Vol. 108, No. 20, 22 May 1998
Lohokare et al.
52 S. K. Loh, D. A. Hales, L. Lian, and P. B. Armentrout, J. Chem. Phys. 90,
5466 ͑1989͒.
(ϭ2kBT). The parameters ␣g and ␣s are derived on the basis of energy
and momentum conservation ␣ ϭ(1Ϫ 4␮/(␮ϩ1)2 cos2 ␪), ␣ ϭ␮(2
͓
͔
g
s
53 P. d. S. Claire, G. H. Peslherbe, and W. L. Hase, J. Phys. Chem. 99, 8147
͑1995͒.
Ϫ␮)/(␮ϩ1)2, where mϭm/M, with m being the effective mass of the
surface atoms influenced by the collision and M is the mass of the collid-
ing atom. The value of kBT for Tϭ273 K is only 2.35ϫ10Ϫ2 eV ͑Ref.
57͒. It can be easily seen as a result that the contribution from the surface
temperature to the final energy of the scattering atom is not significant as
compared to the incident translational energy of that atom. Hence, for all
practical purposes the surface temperature factor can be neglected safely
as we have already done in Eq. ͑3͒. H. Gades and H. M. Urbassek, Appl.
Phys. A 61, 39 ͑1995͒.
54 W. J. Chesnavich and M. T. Bowers, J. Phys. Chem. 83, 900 ͑1979͒.
55 E. K. Grimmelmann, J. C. Tully, and M. J. Cardillo, J. Chem. Phys. 72,
1039 ͑1980͒.
56 Effective masses were determined experimentally through detailed scat-
tering of inert gas atoms from platinum and gold surfaces. H. F. Winters,
H. Coufal, C. T. Rettner, and D. S. Bethune, Phys. Rev. B 41, 6240 ͑1990͒.
57 An approximate value of the effective mass can be obtained from the
phonon dispersion and velocity of sound in the metal, however. The ex-
pression for the velocity of sound in metals is given by the Bohm-Staver
relation: c2ϭ(1/3)Z(m/M)v2F, where c is the velocity of sound, Z is the
61 Y. Zeiri, A. Redondo, and W. A. Goddard III, Surf. Sci. 131, 221 ͑1983͒.
62 B. R. Zegarski and L. H. Dubois, Surf. Sci. Lett. 262, 129 ͑1992͒.
63 Iodine might act to block sites in what would be considered a ‘‘cage
effect’’ on the remaining species. The consequence this would have on
surface diffusion and reaction has not been established. C. J. Jenks, A.
Paul, L. A. Smoliar, and B. E. Bent, J. Phys. Chem. 98, 572 ͑1994͒.
64 This is reflected in the reduction obtained for m/eϭ41 TPRS intensities as
well. It is observed that the intensity reduction for this mass is ϳ18% in
the case of CID which is in close agreement with the difference between
the integrated TPRS intensities corresponding to 8 and 2 L exposures
(ϳ20%) obtained from the exposure dependent thermal experiments.
65 R. W. Verhoef, D. Kelly, C. B. Mullins, and W. H. Weinberg, Surf. Sci.
311, 196 ͑1994͒.
mass number, m is the mass of the electron, M is the mass of the atom and
is the Fermi velocity. Since the electron-ion mass ratio is typically of
vF
the order 10Ϫ4 or 10Ϫ5, this predicts a sound velocity about 1% of the
Fermi velocity, or of order 104 m/s. Therefore, in the typical interaction
time period of ϳ30–50 fs, the phonons interact with a region 3–5 Å
wide. The lattice constant for aluminum is 4.02 Å. Hence, at the high
energies under consideration, the effective mass of the metal is only of the
order of 1–2 atomic masses. By way of reference, an effective mass of
one Al atom was estimated from the collision-induced dissociation dy-
namics of Al clusters. N. W. Ashcroft and N. D. Mermin, Solid State
Physics ͑Cornell University Press, Ithaca, 1976͒.
58 Y. Yamamura, T. Takiguchi, and H. Kimura, Nucl. Instrum. Methods
Phys. Res. A 78, 337 ͑1993͒.
66 G. R. Schoofs, C. R. Arumainayagam, M. C. McMaster, and R. J. Madix,
Surf. Sci. 215, 1 ͑1989͒.
59 M. W. M. Hisham and S. W. Benson, J. Am. Chem. Soc. 91, 3631 ͑1987͒.
60 The role of the surface temperature in the energy exchange can be esti-
mated on the basis of a hard-cube model analysis along the lines of one
performed for argon scattering from a tungsten surface by Grimmelmann
67 M. B. Lee, Q. Y. Yang, and S. T. Ceyer, J. Chem. Phys. 87, 2724 ͑1987͒.
68 M. C. McMaster and R. J. Madix, Surf. Sci. 294, 420 ͑1993͒.
69 D. J. Oakes, H. E. Newell, F. J. M. Rutten, M. R. S. McCoustra, and M.
A. Chesters, J. Vac. Sci. Technol. A 14, 1439 ͑1996͒.
et al. ͑Ref. 55͒. The final energy of the scattering atom is given by EЈ
͗
͘
g
70 R. P. Bell, The Tunnel Effect in Chemistry ͑Chapman and Hall, London,
1978͒.
ϭ␣g
E ϩ␣ E , where EЈ is the final average energy of the scat-
͘ ͗ ͘ ͗ ͘
g s s
g
͗
tered atom, Eg is the incident energy, Es is the surface energy
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.138.73.68 On: Mon, 22 Dec 2014 11:33:32