Gas-surface chemical reactions at high collision energies?
ABSTRACT Most gas-surface chemical reactions occur via reaction of adsorbed species to form a thermal-energy ( approximately kT) product; however, some instances exist where an energetic projectile directly reacts with an adsorbate in a single-collision event to form a hyperthermal product (with a kinetic energy of a few eV). Here we show for the first time that 30-300 eV F(+) bombardment of fluorinated Ag and Si surfaces produces "ultrafast" F(2)(-) products with exit energies of up to 90 eV via a multistep direct-reaction mechanism. Experiments conclusively show that the projectile F atom ends up in the fast molecular product despite the fact that the impact energy is far greater than typical bond energies.
Gas-Surface Chemical Reactions at High Collision Energies?
Michael J. Gordon,†Xiangdong Qin,‡Alex Kutana,‡and Konstantinos P. Giapis*,‡
Department of Chemical Engineering, UniVersity of CaliforniasSanta Barbara, Santa Barbara,
California 93106-5080, and DiVision of Chemistry and Chemical Engineering,
California Institute of Technology, Pasadena, California 91125
Received October 3, 2008; E-mail: email@example.com
Abstract: Most gas-surface chemical reactions occur via reaction of adsorbed species to form a thermal-
energy (∼kT) product; however, some instances exist where an energetic projectile directly reacts with an
adsorbate in a single-collision event to form a hyperthermal product (with a kinetic energy of a few eV).
Here we show for the first time that 30-300 eV F+bombardment of fluorinated Ag and Si surfaces produces
“ultrafast” F2-products with exit energies of up to 90 eV via a multistep direct-reaction mechanism.
Experiments conclusively show that the projectile F atom ends up in the fast molecular product despite the
fact that the impact energy is far greater than typical bond energies.
Gas-surface reactions typically proceed via the Langmuir-
Hinshelwood mechanism, where atoms or molecules adsorb onto
a surface, diffuse around, and subsequently react to form a
product molecule that leaves the surface (desorbs) at thermal
energies (∼kT).1However, another class of reactions exists
where an adsorbate directly reacts with an impinging gas
molecule (or energetic projectile) through a collision-based
mechanism to form a “fast” productsthe so-called Eley-Rideal
(ER) mechanism.2-6ER products are distinct because they are
frequently excited (rotation-vibration) and tend to leave the
surface at hyperthermal energies (up to a few eV) in specific
directions that correlate with the incoming projectile geometry.7
Our current understanding of ER reactions, from the perspective
of detailed collision dynamics and energy transfer, is largely
qualitative. Most observations suggest that a translationally hot
product can be generated in two limiting situations: (i) a light
atom strikes a heavy adsorbate, becoming trapped in the
attractive well of the heavy-atom recoil as it leaves the surface
or (ii) a heavy projectile scatters off the surface and abstracts
(“scoops up”) a lighter atom on the outgoing trajectory path.
Therefore, ER products would seem to form in single-collision
events only when (i) the partners are mismatched in mass and
(ii) the post-collision kinetic energy of each fragment that makes
up the product molecule must be less than typical bond energies
(otherwise the product molecule would not remain intact as it
leaves the surface). As such, ER reactions should not occur at
high impact energies. Here, we present experimental evidence
and theoretical support for a new class of ER reactions at high
impact energies (up to 300 eV) that involve equal-mass partners
and multiple collision events and form “ultrafast” products (up
to 90 eV) that surprisingly remain intact after leaving the surface.
Specifically, we have used F+projectiles (30-300 eV), which
undergo ER abstraction of F atoms from fluorinated Ag and Si
surfaces, to form F2-products with substantial kinetic energies
(up to ∼90 eV). Energy analysis of F2-leaving the surface
reveals two distinct exit channels: one low-energy component
attributed to sputtering and another higher-energy component
that directly correlates with the F+incident energy. The fast-
F2--exit component only occurs for F+bombardment, indicating
that the product must contain the projectile F atom. Molecular
dynamics also predicts fast-F2 formation at the obserVed
experimental energy through a direct-collision sequence that
results in two F species in close proximity leaving the surface
with large, yet comparable, exit velocities.
Experimental evidence for ER mechanisms can take several
forms:2(i) nonthermal translational energies of products, (ii)
product energies that depend on the projectile energy (i.e.,
momentum memory), (iii) rotationally and/or vibrationally
excited products, and (iv) product distributions that depend on
impact geometry or peak at off-normal angles. Documented
examples of ER mechanisms usually involve projectiles at “low”
collision energies accessible by molecular-beam methods.8
Notable examples include D- and Cl-atom abstraction by H
atoms from D/Cu(111),9Cl/Au(111),10and D/Si(100)11as well
as proton “pickup” from Pt(111) by N(C2H4)3N.12Other studies
with low-energy ion beams (<50 eV) have demonstrated direct
abstraction of protons from pyridine-covered Ag(111) (pyridine
†University of CaliforniasSanta Barbara.
‡California Institute of Technology.
(1) Weinberg, W. H. Dynamics of Gas-Surface Interactions; Rettner,
C. T., Ashford, M. N. R., Eds.; Royal Society of Chemistry: London,
(2) Maazouz, M.; Barstis, T. L. O.; Maazouz, P. L.; Jacobs, D. C. Phys.
ReV. Lett. 2000, 84, 1331.
(3) Kratzer, P. J. Chem. Phys. 1997, 106, 6752.
(4) Dinger, A.; Lutterloh, C.; Kuppers, J. Chem. Phys. Lett. 1999, 311,
(5) Kim, J. Y.; Lee, J. Phys. ReV. Lett. 1999, 82, 1325.
(6) Yi, S. I.; Weinberg, W. H. Surf. Sci. 1998, 415, 274.
(7) Rettner, C. T.; Auerbach, D. J. Science 1994, 263, 365.
(8) Rettner, C. T.; Auerbach, D. J.; Tully, J. C.; Kleyn, A. W. J. Phys.
Chem. 1996, 100, 13021.
(9) Rettner, C. T.; Auerbach, D. J.; Lee, J. J. Chem. Phys. 1996, 105,
(10) Rettner, C. T. J. Chem. Phys. 1994, 101, 1529.
(11) Buntin, S. A. J. Chem. Phys. 1998, 108, 1601.
(12) Kuipers, E. W.; Vardi, A.; Danon, A.; Amirav, A. Phys. ReV. Lett.
1991, 66, 116.
Published on Web 01/20/2009
10.1021/ja807672n CCC: $40.75 2009 American Chemical Society
J. AM. CHEM. SOC. 2009, 131, 1927–1930 9 1927
projectiles);13,14H, F, CH3, and C2H3 groups from self-
assembled monolayers by pyrazine and pyrene;15O from
Ag(111) by NO+,2and CO and CO2from Pt(111) by Cs+.16
Energetic O+(10-60 eV) has also been shown to remove O
atoms from oxidized Si to form fast O2-(∼8 eV).17In all of
these cases, product energies are comparable to or a few times
greater than the bond energy; this situation is understandable
because the relative kinetic energy of the fragments that make
up the final molecule cannot be greater than typical bond
energies. In fact, a single collision event, when it occurs at high
energy, can never result in a scattered projectile and recoil with
small relative kinetic energies (i.e., relative to the center-of-
mass of the product molecule). To date, no ER mechanisms
that result in highly energetic products (>10 eV) have been
observed. The reason may be found in the collision cascade:
when the collision energy is raised, gas-surface reactions
transition into hard-sphere scattering processes in which the
projectile cascade sputters atoms and clusters from the surface.18
At such energies, chemical “reactions” per se become secondary
compared to momentum-driven sputtering processes; direct
reactions (even if they occur through multiple collisions) are
simply washed out. However, we show here that direct reactions
do occur at high impact energies, yielding highly energetic
(10-100 eV) products; furthermore, these experiments suggest
that ultrafast products should be formed in many other collision
Scattering experiments were conducted in a custom-built, low-
energy ion scattering system that has been described elsewhere.19
Briefly, the system utilizes an inductively coupled plasma (ICP)-
source, high-voltage beamline (20 keV) with a 60° magnetic mass
filter and deceleration optics to deliver high fluxes of isotopically
pure species at low energies (20-1500 ( 2 eV fwhm) onto a
grounded target under UHV conditions. The ion-beam energy and
spatial profile were characterized with a 180° electrostatic sector
and 2D wire sensor at the target position. For F+and22Ne+beams,
the ICP was run at 2 mTorr/600 W with pure Ne or CF4/O2mixes.
Scattered species were analyzed with a triply differentially pumped
detector system with an electrostatic sector and quadrupole mass
filter to measure the energy distribution and charge state of each
species leaving the target. Experiments were conducted at a
laboratory angle θL) 90° in specular reflection with beam currents
in the 1-10 µA range over 2 mm. Product spectra were recorded
under steady-state scattering conditions at room temperature; all
of the samples were sputter-cleaned with 5 keV Ar+, and the target
surface could be flooded with XeF2gas through a doser pipe situated
∼1 cm from the surface.
Results and Discussion
Typical spectra for F-and F2-leaving an Ag surface under
F+bombardment at various projectile energies (E0) 30-175
eV) are shown in Figure 1a,b, respectively. Intensities were
normalized by beam current and scattering cross section (F-
data only) for an F + Ag collision at θL ) 90° using the
Thomas-Fermi-Molie `re (TFM) potential.20,21Several trends
in the data are worth noting: (i) F2-is produced, (ii) two distinct
peaks can be seen for both F-and F2-, (iii) the low-energy
peak (∼10 eV) for each ion appears to be invariant with
projectile energy (reminiscent of cascade sputtering18), and (iv)
the high-energy F-and F2-peaks shift upward as the impact
energy is increased. The charge-exchange picture of F scattering
in this case is efficient Auger or resonant neutralization of the
projectile on the incoming path and hard collision (which may
lead to F*, F+, or F-) followed by electron capture (or
detachment) on the outgoing path. Copious F-is not unexpected
because the anion level of F shifts22well below the Fermi level
of the target as a result of image-charge effects when Rmin< 4
Å. The intensity of negative ions leaving the target surface
(relative to that of F+) was quite large (>2-3 orders of
magnitude) for all impact energies. In general, the F-yield (total
negative ions) goes through a maximum. This behavior is due
to perpendicular velocity effects23at low Eexit(i.e., more anions
survive because of shorter contact time on the outgoing path)
and cross-section/subplantation24effects at higher E0(i.e., the
projectile preferentially scatters forward and/or penetrates deeper
into the target, so the 90° flux is lower).
(13) Wu, Q.; Hanley, L. J. Phys. Chem. 1993, 97, 8021.
(14) Wu, Q.; Hanley, L. J. Phys. Chem. 1993, 97, 2677.
(15) Morris, M. R.; Riederer, D. E.; Winger, B. E.; Cooks, R. G.; Ast, T.;
Chidsey, C. E. D. Int. J. Mass Spectrom. Ion Processes 1992, 122,
(16) Kim, J. H.; Lahaye, R. J. W. E.; Kang, H. Surf. Sci. 2007, 601, 434.
(17) Quinteros, C. L.; Tzvetkov, T.; Jacobs, D. C. J. Chem. Phys. 2000,
(18) Sigmund, P. Phys. ReV. 1969, 184, 383. Thompson, M. W. Phil. Mag.
1968, 18, 377.
(19) Gordon, M. J.; Giapis, K. P. ReV. Sci. Instrum. 2005, 76, 083302.
(20) Gordon, M. J.; Mace, J.; Giapis, K. P. Phys. ReV. A 2005, 72, 012904.
(21) Mace, J.; Gordon, M. J.; Giapis, K. P. Phys. ReV. Lett. 2006, 97,
(22) Ustaze, S.; Guillemot, L.; Esaulov, V. A.; Nordlander, P.; Langreth,
D. C. Surf. Sci. 1998, 415, L1027.
(23) Maazouz, M.; Guillemot, L.; Esaulov, V. A. Phys. ReV. B 1997, 56,
(24) Rabalais, J. W. Principles and Applications of Ion Scattering
Spectrometry: Surface Chemical and Structural Analysis; John Wiley
and Sons: Hoboken, NJ, 2003.
Figure 1. Energy distributions of (a) F-and (b) F2-leaving the surface
when Ag is bombarded with F+projectiles at various energies (shown as
E0) for θL) 90°.
1928 J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009
Gordon et al.
Although yield data are informative, energy distributions of
species leaving the target are more useful in determining the
actual scattering mechanism. For instance, both F-and F2-
leaving the surface have well-defined, high-energy peaks that
appear to scale with the incoming projectile energy. These peak
locations are summarized in Figure 2, along with the elastic
prediction20,21for a single-scatter F + Ag or F + Si collision
giving a 90° exit. As shown, the fast-F-exit is in excellent
agreement with elastic binary collision theory; similar results
for F+exits off the same surfaces have been shown previously.21
However, the most notable feature in the data is the highly
nonthermal F2-, which is suggestive of a direct (ER) mechanism.
This peak could originate from (i) “sputtering” of F2off the
surface by the projectile cascade or (ii) ER-like abstraction (i.e.,
the projectile F atom ends up in the fast-F2-exit via direct or
scattering-mediated abstraction of a surface-bound F). Process
(i) can be eliminated on several grounds. The high exit energy
(.surface binding energy), the shift in the distribution maximum
with E0, and the lack of a high-energy tail are all contrary to
standard sputtering theories and experiment.18Thus, it seems
that direct reaction [process (ii)] could be responsible for the
fast F2-. To test this hypothesis further, the target was flooded
with XeF2gas from a doser pipe during F+or22Ne+bombard-
ment. In these experiments, the scattering chamber pressure was
increased from <10-9to 7 × 10-7Torr by leaking in XeF2to
saturate the target surface with fluorine. We specifically chose
22Ne+projectiles instead of20Ne+to remove all possibilities of
20Ne+contamination in the F+beam or “false counting” of mass
20 when the quadrupole in the scattered-product detector was
set to mass 19.
Figure 3 summarizes typical ion-energy distribution functions
for the XeF2dosing experiments for a 60 eV projectile. It can
be clearly seen that fast F2-is absent for the22Ne+projectile
case. The data also show that (i) F+and22Ne+generate similar
amounts of low-energy sputtered F-when the surface is
fluorinated, (ii) more low-energy F-is generated when the
surface is flooded with thermal F (i.e., XeF2dosing) than for
F+projectiles alone, and (iii) small amounts of F2-at low energy
(3-5 eV) are sputtered from the surface by22Ne+. All of these
trends make sense with respect to increased sputtering due to
higher surface fluorination, sputtering that is projectile-insensi-
tive (only momentum is at play), and production of fast F2-
via an ER mechanism. This experiment eliminates the possibility
of scattering-mediated abstraction,17where the incoming pro-
jectile creates an F recoil that directly “scoops up” another F
on the outgoing path. Such a mechanism should be operable
for both F+and22Ne+projectiles; however, this is not seen. By
the process of elimination, it appears that fast F2-is formed
through an ER-type direct reaction where the projectile F atom
ends up in the F2-exit.
Additional support for a direct mechanism comes from
molecular dynamics (MD) simulations.25Figure 4a shows a
trajectory “history” that generates a fast, vibrationally excited
F2exit when F hits a fluorinated Ag surface at 50 eV. Fast F2
is made via five steps: (i) projectile P (black) hits F atom A,
after which P forward-scatters and A recoils (red); (ii) P collides
hard with an Ag atom and scatters at ∼95-100°; (iii) P nudges
the moving F atom A again; (iv) P deflects off another F, leaving
the surface; and (v) the moving F atom A nudges another surface
F atom B and leaves with a velocity vector similar to that of P.
The two exiting F atoms combine and leave the surface in a
vibrationally hot state (the trajectory oscillation should be noted:
further timesteps clearly show the F2vibration). This collision
sequence is clearly direct (Eley-Rideal), retains a considerable
portion of the projectile momentum (i.e., P hits its final partner
two times), and the resulting F2molecule contains the projectile
F atom. The role of the surface in this case is to act as a
kinematic mediator that enables the fast-F2channel by preparing
the trajectory and energy of the projectile and its final partner.
For the case shown (and similar direct sequences), the F2final
energy falls in the 24-28 eV range for a 50 eV projectile.
When all of the MD trajectories resulting in F2are binned
according to the final F2energy (Figure 4b), a large peak near
26 eV is seen, exactly matching that measured in our experi-
ments. The fast-F2peak width difference between the experi-
mental and MD results is easily explained by the finite-energy
width of the F+projectile beam in the experiment (50 ( 2 eV
(25) MD was carried out using a superposition of modified Morse-Molie `re
potentials for F + F and F + Ag. At each E0, 107primaries were
launched at the F-on-Ag target [one F and two Ag(111) layers] with
45° incidence versus the surface normal, random impact parameter,
and random azimuth. The cascade was followed until all species had
kinetic energies of <10 eV. An “intact” F2was defined as Etot
mass system and Utot
F2from density-functional theory (DFT) calculations were -99.7208
and -199.4968 hartree, respectively; the optimized bond length for
F2was found to be 2.6549 bohr. See the Supporting Information for
F2< 0, where Ktot
F2is the kinetic energy of F2in the center-of-
F2is the potential energy. Total energies for F and
Figure 2. Exit energies of fast F-and F2-off Ag and Si along with binary
elastic predictions for F + Ag and F + Si at θL) 90°.
Figure 3. F-and F2-product energy distribution functions for F+and
22Ne+projectiles when an Ag surface is flooded with thermal XeF2.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009
Gas-Surface Chemical Reactions at High Collision Energies?
fwhm). Differences at low F2exit energies are related to electron
detachment from F2-at low exit velocities, i.e., a slow F2-
leaving the surface easily loses its electron back to the surface
and is not seen in the experiment (i.e., perpendicular velocity
effects). As added evidence, the fast F2 peak from MD
calculations was tallied for different impact energies; Figure
4c demonstrates that this “direct” collision sequence from MD
agrees extremely well with experiment. We should mention that
a scattering-mediated abstraction reaction (i.e., projectile F hits
a surface F, which then “scoops up” another F on the exit) was
found via MD; however, this collision sequence was eliminated
given that the negative control experiment with22Ne+and XeF2
on Ag (see Figure 3) did not give fast F2-.
Finally, it is worth commenting on the intensity of scattered
F2-as a function of F+impact energy. The decreasing F2-yield
at high E0results from increased vibrational excitation of ER-
created F2-, leading to autodetachment17(in which case the F2
neutral would not be seen), or complete dissociation of
vibrationally hot F20/-. The MD results for fast exits consistently
show rotation and vibration in the F2product. For instance,
Figure 5 gives a summary of the results for four vibrationally
excited fast F2molecules leaving the surface for 50 eV F impact
as a result of direct reactions similar to that shown in Figure
4a. The insets show the F-F bond length of the final F2product
as a function of the collision time; short times (<50-100 fs),
where the bond length is physically impossible (>2.1-2.2 Å),
are associated with the projectile approach to the surface and
“pre-molecule-formation” collisions (see Figure 4a). These
classical trajectory calculations predict a wide range of vibra-
tional excitation in the F2 product, all the way up to the
dissociation limit. In addition, the anharmonicity in the potential
energy curve near the dissociation threshold becomes apparent
in the temporal variation of the F-F bond length (Figure 5,
upper inset). Although the potential energy curve changes when
the F2becomes an anion, it is highly probable that the fast F2-
product is vibrationally excited.
In summary, we have shown that Eley-Rideal abstraction
involving F+projectiles and fluorinated Ag and Si surfaces can
occur at high impact energies (30-300 eV), resulting in
hyperthermal F2-ions leaving the surface at energies of tens of
electron volts. Energy analysis of the F2-products as well as
bombardment of externally fluorinated surfaces with22Ne+rather
than F+shows that F+projectiles are required for fast-F2-
formation. MD results also show that a direct-reaction sequence
involving multiple collisions can form an F2product with a high
translational energy that correlates exactly with experiment.
Acknowledgment. This study was based on work supported
by the National Science Foundation under Grant CTS-0613981.
Applied Materials, Inc. donated equipment essential for the
Supporting Information Available: Details of MD and DFT
calculations. This material is available free of charge via the
Internet at http://pubs.acs.org.
Figure 4. Fast F2formation. (a) MD trajectory history for a 50 eV F
projectile (P; black line) reacting directly with a fluorinated Ag surface to
generate F2at a high exit energy (27 eV). See the text for the collision
sequence. (b) Experimental F2-exit energy (solid) and MD (histogram)
results for F2leaving the surface for 50 eV incident F (see ref 25). (c)
Fast-F2-peak location for various F+impact energies on fluorinated Ag:
(solid) experimental and (open) MD results. The solid line is to guide the
Figure 5. Vibrational excitation of fast F2leaving a fluorinated Ag surface
for 50 eV F+impact, as determined from MD. The potential energy curve
and associated vibrational levels (classical) of four different exits are shown.
The insets show the F2bond length vs time for two specific fast-F2exits.
1930 J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009
Gordon et al.