Cs+ reactive scattering from a Si(111) surface adsorbed with water

Article abstract of DOI:10.1063/1.474572

Reactive scattering of hyperthermal Cs ion is examined from a Si(111) surface adsorbed with water. Collision of Cs⁺ beams with the Si surface at the energy of 10-100 eV produces Cs⁺-bound cluster ions as scattering products, including CsOH⁺, CsOH₂⁺, CsSi⁺, CsSiH_n⁺ (n = 1,2), and CsSiO⁺. The yields for these clusters are examined as a function of Cs⁺ beam energy and water exposure. Kinetic energy distributions for the clusters are measured. The reactive scattering process is explained in terms of collision-induced desorption of adsorbate, followed by ion-molecule association between the scattered Cs⁺ and the desorbed molecule. The probability that Cs⁺ undergoes reactive scattering is 5 × 10^-4-2×10^-3 for 50 eV collision energy. The corresponding probability for the Cs⁺-molecule association reaction is in the order of 5 × 10^-3-2×10^-2 or slightly lower. It is proposed that CsOH⁺ and CsOH₂⁺ are formed from OH and H₂O adsorbates, respectively, via direct collisional desorption. CsSiO⁺ formation can be related to desorption of surface oxide species or, at high energy, to collisional dissociation of adsorbates. Several aspects of using Cs⁺ reactive scattering for surface adsorbate detection are discussed.

Full text of DOI:10.1063/1.474572

Cs + reactive scattering from a Si(111) surface adsorbed with water
M. C. Yang, C. H. Hwang, and H. Kang
Citation: The Journal of Chemical Physics 107, 2611 (1997); doi: 10.1063/1.474572
View online: http://dx.doi.org/10.1063/1.474572
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/107/7?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
H/D isotopic exchange between water molecules at ice surfaces
J. Chem. Phys. 121, 2765 (2004); 10.1063/1.1770548
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 formation and ejection of endohedral Cs @ C 60 + by low energy collisions (35–220 eV) of Cs + ions with
surface adsorbed C 60 molecules
J. Chem. Phys. 117, 3484 (2002); 10.1063/1.1491898
Neutralization and negative ion conversion of low-energy proton scattered from Ar, Kr, and Xe condensed on the
Pt(111) surface
J. Chem. Phys. 115, 1522 (2001); 10.1063/1.1379738
Ionic-to-metallic layer transition in Cs adsorption on Si (111)-(7×7): Charge-state selective detection of the
adsorbates by Cs + reactive ion scattering
J. Chem. Phys. 112, 8660 (2000); 10.1063/1.481467
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:
136.165.238.131 On: Fri, 19 Dec 2014 11:30:22

Cs؉ reactive scattering from a Si(111) surface adsorbed with water
M. C. Yang, C. H. Hwang, and H. Kanga)
Department of Chemistry, Pohang University of Science and Technology, Pohang, Gyeongbuk 790-784,
South Korea
͑Received 4 April 1997; accepted 7 May 1997͒
Reactive scattering of hyperthermal Cs ion is examined from a Si͑111͒ surface adsorbed with water.
Collision of Csϩ beams with the Si surface at the energy of 10–100 eV produces Csϩ-bound cluster
ions as scattering products, including CsOHϩ, CsOH2ϩ, CsSiϩ, CsSiHϩn (nϭ1,2), and CsSiOϩ. The
yields for these clusters are examined as a function of Csϩ beam energy and water exposure. Kinetic
energy distributions for the clusters are measured. The reactive scattering process is explained in
terms of collision-induced desorption of adsorbate, followed by ion–molecule association between
the scattered Csϩ and the desorbed molecule. The probability that Csϩ undergoes reactive scattering
is 5ϫ10Ϫ4 –2ϫ10Ϫ3 for 50 eV collision energy. The corresponding probability for the
Csϩ–molecule association reaction is in the order of 5ϫ10Ϫ3 –2ϫ10Ϫ2 or slightly lower. It is
proposed that CsOHϩ and CsOH2ϩ are formed from OH and H2O adsorbates, respectively, via direct
collisional desorption. CsSiOϩ formation can be related to desorption of surface oxide species or, at
high energy, to collisional dissociation of adsorbates. Several aspects of using Csϩ reactive
scattering for surface adsorbate detection are discussed. © 1997 American Institute of Physics.
͓S0021-9606͑97͒02131-4͔
detailed experimental results and analyses for Csϩ–surface
I. INTRODUCTION
reactive scattering. Various Csϩ-bound cluster ions emanat-
During the past years many endeavors have been made
ing from a water-adsorbed Si͑111͒ surface are identified, and
their yields are measured as a function of beam collision
energy and water exposure. The kinetic energies of the clus-
ter ions are measured. These results are analyzed in order to
understand the dynamics of reactive scattering.
in order to understand gas–surface collisional phenomena.
The collisional nature varies depending on the particle en-
ergy from trapping and desorption in the thermal regime to
binary collision ͑BC͒ in the keV energy range.1 Between
these two extremes, there exists a transitional region, often
called hyperthermal energy ͑1–100 eV͒. Gas–surface inter-
actions in this energy region offers an interesting subject to
explore, not only because such investigation can bridge the
gap between thermal and BC collisional behaviors, but be-
cause chemical transformation energy falls into this region.
Recent investigations of hyperthermal gas–surface interac-
tions to pursue such aims have been well responded. Hyper-
thermal collision, induces dissociation of molecular
projectiles,2–10 desorption from surface,11–19 chemical reac-
tion with functionalized surface,3,4 and surface modification
and film deposition in the atomic scale.20–23
The water-adsorbed Si surface is an important system to
study because it is related to the initial stage of wet Si oxi-
dation, a process useful in semiconductor industry. Despite
numerous studies in the past years, subtle questions still re-
main regarding the adsorption state of water on Si. An ex-
cellent review on the interaction of water with Si surfaces
has been done by Thiel and Madey.26 We will only briefly
mention several representative studies on this subject. Ultra-
violet photoelectron spectroscopic ͑UPS͒ studies have sug-
gested that H2O adsorbs molecularly on a Si surface,27,28
together with some hints for partial dissociation.29 On the
other hand, high resolution electron energy loss spectroscopy
͑HREELS͒ indicates mostly dissociated OH,30,31 probably
due to instrumental difficulty of detecting molecular H2O.32
Laser induced thermal desorption ͑LITD͒ ͑Refs. 33,34͒ has
detected H2O and SiOH moieties desorbing from a Si͑111͒-
7ϫ7 surface, supporting partial dissociation of H2O. Scan-
ning tunneling microscopic ͑STM͒ study35 suggests molecu-
lar adsorption at low water coverages ͑␪Ͻ0.05 ML͒ while
dissociative adsorption at high coverages ͑␪Ͼ0.1 ML͒. Re-
cent X-ray photoelectron spectroscopic ͑XPS͒ study using
synchrotron radiation36 shows that H2O not only dissociates
into H and OH on the ͑111͒-7ϫ7 surface, but oxidation of Si
to oxides can also occur at high water exposures. To sum-
marize the results to the present, at least a certain fraction of
water molecules are believed to exist in dissociated forms on
a room temperature Si͑111͒-7ϫ7 surface. For high water ex-
Recently, we reported that Csϩ ions of hyperthermal en-
ergy undergo reactive scattering from a solid surface,24,25
producing Csϩ-bound clusters ͑CsXϩ, where X is an adsor-
bate or a surface atom͒. We proposed that the reactive scat-
tering process can conceptually be divided into two
steps,24,25
Csϩ g͒ϩX–surface→Csϩ g͒ϩX g͒ϩsurface ͑reaction 1)
͑
͑
͑
Csϩ g͒ϩX g͒→CsXϩ g͒.
reaction 2͒
͑
͑
͑
͑
In reaction 1, collision of hyperthermal Csϩ causes desorp-
tion of X from the surface. Then, the desorbed X is attracted
to the scattered Csϩ via electrostatic attraction forces, and
forms a CsXϩ complex ͑reaction 2͒. In this paper, we present
a͒
Author to whom correspondence should be addressed. FAX: ϩ82-562-
279-3399. Electronic mail: surfion@vision.postech.ac.kr
J. Chem. Phys. 107 (7), 15 August 1997
0021-9606/97/107(7)/2611/8/$10.00
© 1997 American Institute of Physics
2611
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:
136.165.238.131 On: Fri, 19 Dec 2014 11:30:22

2612
Yang, Hwang, and Kang: Csϩ scattering from Si(111)
posure, a small amount of surface oxides may also be
formed.
II. EXPERIMENT
The experiment was performed using the ion–surface
scattering chamber equipped with a low-energy ͑1–300 eV͒
ion beamline. This apparatus has previously been described
in detail.7,37 Csϩ ions were produced from CsCl powder
heated inside the Colutron ion source. The ion beam was
mass-selected by a Wien filter and electrostatically deflected
to eliminate possible neutral impurities before it reached the
sample surface. A Faraday cup was used to spatially probe
the profile of the beam and to measure ion current density.
The current density of the Csϩ beam was 1–10 nA cmϪ2
,
and its energy spread was normally less than 2 eV. The
sample, an intrinsic phosphorus-doped Si͑111͒ wafer with its
area of 1 cmϫ1 cm, was placed inside an electric field-free
region of the scattering chamber. The native oxide layer was
removed from the sample surface by cycles of Arϩ sputtering
at 2 keV and heating at 1100 °C using electron bombardment
from the backside in ultrahigh vacuum ͑UHV͒. The sample
prepared in this fashion has a ͑111͒-7ϫ7 reconstructed
surface.22 Once a clean surface was prepared in UHV, heat-
ing to 1100 °C was sufficient to remove the impurities ad-
sorbed during the experiment. Surface cleanness was
checked by Auger electron spectroscopy ͑AES͒ and Csϩ re-
active scattering. The cleaned surface was then exposed to
water vapor at a pressure of 1ϫ10Ϫ6 Torr at room tempera-
ture.
The ions scattered from the surface were detected using
a quadrupole mass spectrometer ͑QMS͒ operated in a posi-
tive ion sampling mode. Kinetic energy of the ions was mea-
sured by placing a retarding field analyzer ͑RFA͒ of a 4-grid
type in front of the mass spectrometer nose. The angle be-
tween an incident ion beam and a detector was fixed at 90°,
and the incidence beam direction was 45° to the surface. The
total Csϩ beam dose was kept less than 1ϫ1014 ions cmϪ2 in
most experiments. The base pressure of the UHV scattering
chamber was 3ϫ10Ϫ10 Torr. Through differential pumping
of the beamline, the chamber pressure was maintained in the
10Ϫ10 Torr range during Csϩ beam exposure.
FIG. 1. Mass spectra of the positive ions emitted upon Csϩ impact of a
Si͑111͒ surface adsorbed with water at saturation coverage. The intensity of
the scattered Csϩ peak is reduced to a factor 1ϫ10Ϫ3. The angles of an
incident Csϩ beam and a detector are 45° to the surface normal. ͑a͒ 15 eV
collision on H2O-adsorbed surface. ͑b͒ 15 eV collision on D2O-adsorbed
surface. ͑c͒ 50 eV collision on H2O-adsorbed surface.
identified as CsOHϩ ͑150͒, CsOHϩ2 ͑151͒, and CsSiOϩ ͑177
amu͒ ͓Fig. 1͑a͔͒. The intensities of these products are
10Ϫ3 –10Ϫ4 times smaller than the elastically scattered
Csϩ. In order to confirm the mass identification, a control
experiment has been performed by exposing a surface with
D2O. The CsOHϩ and CsOH2ϩ peaks become shifted to 151
and 153 amu, representing CsODϩ and CsOD2ϩ, respectively
͓Fig. 1͑b͔͒. The CsSiOϩ peak remains in the same position
͑177 amu͒. A small peak at 152 amu is attributed to
CsOHDϩ coming from adsorbed HDO impurity. It is be-
cause a small HDO partial pressure was measured during
D2O exposure, possibly produced from isotope scrambling
reactions on stainless steel chamber walls. The isotopic sub-
stitution does not change the intensity ratio
III. RESULTS
A. Mass identification of Cs؉-bound clusters
Collision of hyperthermal Csϩ ions onto a water-
adsorbed Si͑111͒ surface gives rise to various CsXϩ cluster
ions. Figure 1 shows the cluster species produced from the
surface exposed to water vapor up to saturation ͑1000 L,
1 Lϭ1ϫ10Ϫ6 Torr s) at room temperature. The mass spectra
are displayed for the region of 130–200 amu. In the lower
mass region, no peaks appear with measurable intensity ex-
cept Hϩ and alkali metal ions produced from surface impu-
rities. The spectra of Fig. 1 are characterized by an intense
Csϩ peak at 133 amu and the higher mass peaks representing
CsXϩ cluster ions. Upon 15 eV Csϩ collision onto the
H2O-adsorbed surface, three cluster ions appear that are
I(CsODϩ)/I(CsOD2ϩ) from I(CsOHϩ)/I(CsOHϩ2 ). CsDϩ
ions ͑135 amu͒ are not emitted or have only negligible in-
J. Chem. Phys., Vol. 107, No. 7, 15 August 1997
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:
136.165.238.131 On: Fri, 19 Dec 2014 11:30:22

Yang, Hwang, and Kang: Csϩ scattering from Si(111)
2613
FIG. 3. The reactive scattering yield, Yrs͑X͒, for XϭOH ͑ᮀ͒, H2O ͑᭺͒, Si
͑᭹͒, and SiO ͑᭡͒ as a function of Csϩ beam energy for the range of 10–100
eV. Yrs͑X͒ is normalized to the Csϩ flux that scatters into a detector solid
angle. The curves for Yrs͑OH͒ and Yrs͑H2O͒ represent the collisional desorp-
tion function ͑see the text͒.
FIG. 2. Ion emission intensities for Csϩ ͑᭡͒, CsOHϩ ͑᭝͒, CsOH2ϩ ͑᭿͒,
CsSiϩ ͑᭹͒, and CsSiOϩ ͑᭺͒ as a function of Csϩ beam energy for the range
of 10–100 eV. The intensities are normalized to the Csϩ current reading at
the sample.
A more useful way of presenting the ion intensity data is
to show the reactive scattering yield or reaction probability,
Yrs͑X͒, which is defined as the ratio I(CsXϩ)/I(Csϩ).
Yrs͑X͒ is normalized to the Csϩ flux that scatters toward a
detector angle, and has two advantages over I(CsXϩ). First,
uncertainty of measuring the incident Csϩ flux is canceled
out in Yrs͑X͒. Due to the very low efficiency for Csϩ neu-
tralization on Si, it is difficult to deduce the number of inci-
dent Csϩ ions from sample current. Second, the unknown
angular distribution for scattered Csϩ is not problematic, be-
cause only the scattering fluxes moving in the same direction
͑the detector angle͒ are included in this expression. These
two features allow easier analysis for the scattering data in
terms of the reactive scattering mechanism. Since the reac-
tive scattering process can conceptually be divided into two
steps, collision-induced desorption of X and Csϩ–X associa-
tion reaction ͑reactions 1 and 2͒, we can write
tensity. Although CsHϩ signal would be buried in the tail of
scattered Csϩ peak, the CsDϩ position is well separable from
the Csϩ tail.
An increase of Csϩ collision energy to 50 eV strongly
increases the yield for CsSiOϩ relative to CsOHϩ and
CsOHϩ2 ͓Fig. 1͑c͔͒. At the same time, the peaks correspond-
ing to CsSiϩ, CsSiHϩ, and CsSiHϩ2 start to appear at 161,
162, and 163 amu, respectively. The intensity ratio
I(CsOHϩ)/I(CsOHϩ2 ) remains unchanged upon this energy
increase.
B. Energy-dependency of Cs؉ reactive scattering
The dependency of CsXϩ production yield on Csϩ col-
lision energy is closely related to the dynamics of reactive
scattering. The intensities for Csϩ, CsOHϩ, CsOH2ϩ,
CsSiϩ, and CsSiOϩ are measured as a function of collision
energy for 10–100 eV from the water-saturated surface.
I(CsXϩ), the quantity presented in Fig. 2, represents the
scattered ion intensity normalized to the incident Csϩ current
measured at the sample. I(Csϩ) initially increases with beam
energy, then reaches a maximum at about 30 eV. At higher
energies I(Csϩ) decreases again. The energy-dependencies
for I(CsOHϩ) and I(CsOHϩ2 ) approximately follow the
track of the Csϩ curve. In contrast, I(CsSiϩ) and
I(CsSiOϩ) are small at low collision energies, but continu-
ously increase with increasing energy.
I CsXϩ͒
͑
Yrs͑X͒ϭ
ϭYd X͒Y
͑X͒.
͑1͒
͑
assoc
I Csϩ͒
͑
Here, Yd(X) is the probability for collisional desorption of
X, I(X)/I(Csϩ), and Yassoc͑X͒ is the probability for
Csϩ–X association after X desorption, I(CsXϩ)/I(X).
The plot of Yrs͑X͒ as a function of beam collision energy
is shown in Fig. 3. Yrs͑OH͒ and Yrs͑H2O͒ increase sharply
below 15 eV, then flatten above 20 eV. Yrs͑Si͒ and
Yrs͑SiO͒ exhibit quite different behaviors from Yrs͑OH͒ and
Yrs͑H2O͒. Yrs͑Si͒ is virtually zero below 20 eV, and in-
J. Chem. Phys., Vol. 107, No. 7, 15 August 1997
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:
136.165.238.131 On: Fri, 19 Dec 2014 11:30:22

2614
Yang, Hwang, and Kang: Csϩ scattering from Si(111)
creases slowly and continuously with increasing energy.
Yrs͑SiO͒ exhibits an onset at about 10 eV, then a strong
increase.
The Yrs͑OH͒ and Yrs͑H2O͒ data are attempted for a fit
with a functional form for collisional desorption cross sec-
tion. Note that Yrs͑X͒ gives a measure for Yd(X) ͓Eq. ͑1͔͒,
under approximation that Yassoc͑X͒ is constant. Yd(X) can be
replaced with desorption cross section ␴d(X) for a fixed sur-
face coverage ␪, which is the case in the present experiment.
We use a collisional desorption model, ␴d(E)ϭa(E
ϪEth)b/E,38 where E is the collision energy, Eth is the
threshold energy for desorption, and a and b are fitting pa-
rameters. This model has been used to determine Eth for
adsorbate desorption induced by Xe collision.19 From the
fitting we obtain Ethϭ8.0Ϯ1.4 eV for desorption of OH and
7.1Ϯ1.8 eV for H2O. The optimized fitting constants are a
ϭ1.2Ϯ0.5 and bϭ0.7Ϯ0.1 for OH, and aϭ0.4Ϯ0.2 and
bϭ1.0Ϯ0.1 for H2O. Whether or not an experimental
threshold energy for desorption can be converted into an ab-
solute surface binding energy is probably left to further
investigation.39 At least, this analysis indicates that both
H2O and OH have comparable desorption energies. The
curves for Yrs͑Si͒ and Yrs͑SiO͒ cannot be fitted into the
␴d(E) form for the entire energy range.
C. Dependency on adsorbate coverage
The variation in Yrs͑X͒ with water exposure is presented
in Fig. 4 for OH, H2O, and SiO. The amount of water expo-
sure was varied from 0 to 1000 L at room temperature. The
water vapor pressure was read from an ionization gauge
without further calibration. The measurements were per-
formed at 30 eV collision energy, at which the ion intensities
are substantial for CsSiOϩ as well as for CsOHϩ and
CsOHϩ2 . The Yrs͑X͒ curves for these species qualitatively
follow the Langmuir adsorption behavior, with an initial stiff
increase and a saturation at higher exposures. The curves are
fitted by the Langmuir adsorption equation,
FIG. 4. Yrs͑X͒ for XϭOH ͑a͒, H2O ͑b͒, and SiO ͑c͒ as a function of H2O
exposure in Langmuir. Csϩ beam energy is 30 eV. The solid lines represent
Langmuir adsorption curves.
tensities at the present energy and coverage to accumulate
the energy distributions. A 45°/45° specular scattering geom-
etry is employed. The scattered Csϩ has an energy distribu-
tion peaking at about 10 eV. This value indicates that 90% of
the Csϩ beam energy is transferred to the Si surface during
100 eV collision, in agreement with our previous
measurement.40 CsSiϩ, CsSiOϩ, and Csϩ2 are ejected with
relatively lower kinetic energies, peaking at 3–5 eV. Mono-
mer Siϩ ions are also ejected with substantial intensity at this
relatively high collision energy. The energy distribution for
Siϩ exhibits a maximum at about 6 eV, which is slightly
higher than the cluster species.
␪ϭ1ϪeϪcL
,
͑2͒
where ␪ is adsorbate coverage, c fitting constant, and L the
amount of water exposure in Langmuir. Yrs͑OH͒ and
Yrs͑H2O͒ fit to the Langmuir equation with cϭ0.01 LϪ1
.
The curve for Yrs͑SiO͒ rises much more steeply and reaches
an earlier saturation ͑Ͻ100 L͒. The Yrs͑SiO͒ fit gives c
ϭ0.2 LϪ1. The early saturation for SiO implies that forma-
tion of surface silicon oxides, which are the likely precursors
for SiO desorption ͑Sec. IV C͒, follows different kinetics
from OH and H2O adsorption.
Although the maximum kinetic energies for the CsXϩ
clusters are much lower than that for Csϩ, there are small but
sizable overlaps between the energy distributions for Csϩ
and the clusters. The cluster energy distributions are similar
to those of sputtered particles, for instance, the sputtered Ni
atoms in 100 eV Xe collision on Ni͑100͒ ͑Ref. 41͒ and the
sputtered Rh atoms in keV Arϩ bombardment on Rh͑111͒.42
The kinetic energies for Siϩ appears overall shifted to the
high-energy side compared to the clusters or the sputtered
atoms. The high-energy shift of ionic Siϩ distribution is
probably due to preferential neutralization of the sputtered,
D. Kinetic energy of scattered ions
Kinetic energies of the scattered ions are measured using
a RFA method. A kinetic energy distribution is obtained
from raw RFA data through differentiation with respect to
energy. Figure 5 shows kinetic energy distributions for
Csϩ, Siϩ, CsSiϩ, CsSiOϩ, and Csϩ2 emitted during 100 eV
Csϩ collision onto a Si͑111͒ surface with partial H2O adsorp-
tion. CsOHϩ and CsOHϩ2 have relatively too low signal in-
J. Chem. Phys., Vol. 107, No. 7, 15 August 1997
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:
136.165.238.131 On: Fri, 19 Dec 2014 11:30:22

Yang, Hwang, and Kang: Csϩ scattering from Si(111)
2615
TABLE I. Desorption yields upon impact of heavy projectiles on light-atom
surfaces at 50–100 eV reported in the literature.
Projectile-surface
Impact
energy
Krϩ–Ge(s)a
Xe–Ni͑100͒
c
Hgϩ–Ni(s)a Xeϩ–N/W(s)b
50 eV
100 eV
0.02
0.10
0.01
0.05
0.11
0.24
0.0͑0.0͒
0.25͑0.0͒
aReference 12. Ion incidence direction is random.
bReference 13. The desorption yield for adsorbed N atom. Ion incidence
direction is random.
cReference 14. Beam incidence angle is 45°. The normal incidence values
are in parentheses.
Relevant literature data for the desorption yield, Yd , are
listed in Table I for impact energies of 50 and 100 eV. Yd
values for 50 eV ion impact lie between 0.01–0.02 for sur-
face atom desorption from metal ͑Ni͒ and semiconductor
͑Ge͒ surfaces. For desorption of chemisorbed N atom from
W by Xeϩ collision, it is 0.11. These values are obtained for
ion incidence at random directions. The Yd values are ex-
pected to be somewhat different for 45° beam incidence, but
probably have the same order. The desorbed species in the
present work are Si atoms or chemisorbed species on Si. For
such cases, we may assume that Yd values are similar to or
greater than those of Table I, which gives the order Yd(Si)
ϳ0.01 and Yd͑adsorbate͒ϳ0.1 as lower limits for 50 eV col-
lision.
From the ion intensities of Fig. 1͑c͒, we obtain the prob-
ability of reactive scattering, Yrs͑X͒, to be 5ϫ10Ϫ4 ͑XϭOH
and H2O͒, 2ϫ10Ϫ4 ͑Si͒, and 2ϫ10Ϫ3 ͑SiO͒ for 50 eV
Csϩ–surface collision. Since Yrs͑X͒ϭYd(X)Yassoc͑X͒ ͓Eq.
͑1͔͒, one can deduce from these numbers the probability of
Csϩ–X association, Yassoc͑X͒. Yassoc͑X͒ is estimated for the
above species to be in the range 5ϫ10Ϫ3 –2ϫ10Ϫ2, if we
use Yd(Si)ϳ0.01 and Yd(adsorbate)ϳ0.1.
FIG. 5. Kinetic energy distributions for scattered Csϩ, sputtered Siϩ, and
the cluster species ͑CsSiϩ, CsSiOϩ, and Csϩ2 ͒ produced from a Si surface
partially covered with H2O. The incident Csϩ energy is 100 eV in a 45°/45°
specular scattering geometry. The area under the curves represents relative
intensity for the ions.
low energy Siϩ at the surface.43 Such neutralization effect
has been found40,44 to deplete the low-energy part of the
energy distribution for sputtered ions.
Kinetic energy measurement requires extensive signal
accumulation be done in order to obtain reasonable statistics
for an energy distribution. In such an experiment, a pro-
longed Csϩ beam exposure ͑Ͼseveral min͒ deposits a sub-
stantial amount of Cs on the surface. The deposited Cs is
indicated by the Csϩ2 emission, formed via reactive scattering
from surface Cs. In the presence Cs deposits, an alternative
reaction could occur between the sputtered Csϩ and the de-
sorbed X, also forming the CsXϩ products. A control experi-
ment using 100 eV Arϩ bombardment on the same surface,
covered with both Cs and water, gives extremely small
yields for CsXϩ emission. Therefore, we conclude that most
of the CsXϩ clusters are produced by reacting with incident
Csϩ beams even in the presence of surface Cs.
B. Ion–molecule association reaction
Since Csϩ is a chemically inert species, the binary asso-
ciation reaction between the scattered Csϩ and the desorbed
X is to be driven by electrostatic ion–molecule interactions.
The binding energy for a Csϩ–OH2 pair is 0.61 eV,45 which
stands for a typical ion–dipole attraction energy. Csϩ–X as-
sociation will occur when Cs and X have similar velocities
and moving directions such that the energy of their relative
motions can be smaller than the ion–molecule binding en-
ergy. Figure 5 shows that a major portion of scattered Csϩ
ions has kinetic energy higher than that of CsXϩ clusters.
Thus, only a small, low energy portion of the Csϩ ions,
roughly 10% when estimated from the overlapped area be-
tween the energy distributions for Csϩ and CsXϩ, appears to
meet the energy requirement for cluster formation.
IV. DISCUSSION
In the following sections, we will first address two key
steps of Csϩ reactive scattering, collision-induced desorption
and ion–molecule association reaction. Then, the experimen-
tal observations are analyzed in view of this scattering
mechanism. Finally, prospects of doing surface analysis with
Csϩ reactive scattering are discussed.
A. The probability for desorption induced by
hyperthermal collision
Nonetheless, Csϩ is the most suitable projectile for reac-
tive scattering among the alkali metal ions. It is because
Csϩ scatters from a surface with substantially lower kinetic
energy than other ions due to its heavy mass.40 We observed
in a control experiment with Kϩ that the yield for KXϩ pro-
duction was much smaller.
Desorption induced by low energy ion impact of a sur-
face has been investigated by several workers.11–13,18,39 Par-
ticularly useful to the present study are the reports on desorp-
tion yield upon heavy projectile-light surface collision.
J. Chem. Phys., Vol. 107, No. 7, 15 August 1997
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:
136.165.238.131 On: Fri, 19 Dec 2014 11:30:22

2616
Yang, Hwang, and Kang: Csϩ scattering from Si(111)
The association reaction will occur with highest prob-
pact causes the secondary particles to be ejected preferen-
tially as a result of complex collision cascades rather than via
direct collisions. However, the situation can change for hy-
perthermal collision. It has been found13,41 that as the impact
energy is lowered ͑Ͻ100 eV͒, the directly knocked-off par-
ticles have a more dominating portion in the sputtered flux.
It is worthwhile to consider the possibility for the inter-
conversion between H2O and OH during Csϩ–surface colli-
sion. A heavy projectile like Csϩ distributes its incidence
energy to many surface atoms during the collision,40 and thus
the amount of energy transferred to individual atoms should
be very small. For this reason, it will be difficult to break
HO–H bond D0(HO–H͒ϭ5.2 eV by low energy Csϩ col-
lision, for instance, at Ͻ20 eV. A small relative collision
energy between Csϩ and H atoms ͑0.8% of the Cs beam
energy͒ coming from their extreme mass difference is an-
other drawback for collisional dissociation. More impor-
tantly, OH desorption requires dissociation of a HO–H bond,
and then desorption of the OH fragment, because H2O is
chemisorbed on Si through O–Si bonding.26 Such simulta-
neous fragmentation and ejection requires even higher en-
ergy. In Fig. 3, the ratio Yrs͑OH͒/Yrs͑H2O͒ stays fairly un-
changed over a wide energy range. If HO–H bond
dissociation occurred to substantial degrees, then it would be
reflected by a change in the Yrs͑OH͒/Yrs͑H2O͒ ratio. All
these considerations indicate that both OH and H2O are
present on the surface, which desorb by the collision retain-
ing their original state.
Yrs͑Si͒ and Yrs͑SiO͒ exhibit different energy-
dependencies from Yrs͑OH͒ and Yrs͑H2O͒ ͑Fig. 3͒, suggest-
ing that desorption of Si and SiO involves different mecha-
nism. The curves for Yrs͑Si͒ and Yrs͑SiO͒ resemble that for
secondary particles ejected via so called substrate sputtering
mechanism,13,41 which is initially low at small impact energy
but increases continuously with energy. The substrate sput-
tering occurs such that the impact energy is initially trans-
ferred to substrate atoms, and then eventually to the ejecting
species.13 Subsurface atoms are ejected exclusively via this
mechanism.13,41 For desorption of Si atoms, as they are lo-
cated underneath the adsorbate layer on the present surface,
the substrate sputtering mechanism provides a reasonable ex-
planation.
Desorption of SiO can be related to the presence of sili-
con oxides on the surface. Recent XPS study36 has suggested
that insertion of O into Si–Si bond occurs upon H2O expo-
sure on Si͑111͒-7ϫ7. For saturation exposure at 300 K, the
ratio of Siϩ/Si2ϩ is 1/0.18. Here, Siϩ is assigned to the sur-
face Si atoms on which only OH is bonded, and Si2ϩ to the
OH-decorated adatoms with additional O atom insertion into
their Si–Si backbonds.36 According to this proposed struc-
ture, the oxide units ͑Si–O–Si͒ are located in the subsurface
layer. To be consistent with this structure and the Yrs͑SiO͒
behavior, we suggest that SiO is generated from the Si–O–Si
units via substrate sputtering mechanism. SiO species are
known to readily desorb upon heating a water-exposed Si
surface,34 indicating that SiO is a volatile species.
ability while the scattered Csϩ and the desorbed X are in
close locations near the surface. If Csϩ and X are sufficiently
close to each other at the initial stage of attraction, then the
nascent Csϩ–X pair will remain stable along its outgoing
trajectory. Otherwise, the CsXϩ needs to be energetically
stabilized by interacting with the surface and by losing its
excess energy. Molecular dynamics ͑MD͒ calculations of Xe
scattering from a Si surface40,41 show that 10 eV Xe atoms
stay within 3 Å distance from the surface for a period of 2
ϫ10Ϫ13 s. This implies that the Csϩ–X association and the
energy quenching is completed most likely within 1 ps. Pre-
viously, formation of Cun clusters has been studied using a
MD method during 600 eV Arϩ bombardment of a Cu
surface.46 The Cun clusters are generated only from the sput-
tered atoms in this case, but this analogous study has also
shown that cluster formation occurs mostly within 4 Å dis-
tance from the surface and within 2ϫ10Ϫ13 s.
͓
͔
The probability for Csϩ–X association is estimated to be
5ϫ10Ϫ3 –2ϫ10Ϫ2 in Sec. IV A. An order-of-magnitude
calculation may be useful to check the validity of this
Yassoc value. The Langevin cross section for reaction between
an ion and a nonpolar molecule, ␴ϭ␲e(2␣/Erel)1/2, has an
order of 10 Å2 for relative motion energy (Erel) of 1 eV and
molecular polarizability ͑␣͒ of 1 Å3. We assume that about
10% of scattered Csϩ has kinetic energy small enough to
form a stable complex by Csϩ–X interaction, a value esti-
mated from the energy distribution curves of Fig. 5. In this
case, in order to give the ion-molecule reaction probability of
5ϫ10Ϫ3 –2ϫ10Ϫ2 for scattered Csϩ, molecular gas density
of 0.1–0.01 molecule ÅϪ3 is required. 10–100 Å3 is a rea-
sonable volume of space in which a desorbing molecule
spreads out during the time scale of reactive scattering, and
is consistent with the cluster formation in the near-surface
region suggested by the MD calculation.46
C. Dynamics of collisional desorption
The energy-dependencies of Yrs͑OH͒ and Yrs͑H2O͒ fit
well to the functional form for collisional desorption cross
section,38 ␴d(E)ϭa(E-Eth)b/E. This function is derived
from a line-of-centers collision model assuming direct, hard-
sphere collisions between a projectile and surface atoms.
Thus, its application is limited most likely to desorption of
top-layer species. OH and H2O adsorb on top of a Si
surface.26 The energy-dependencies of Yrs͑OH͒ and
Yrs͑H2O͒, a stiff increase at low energy with a plateau at
high energy, resemble that of N atom ejection from a W
surface during Heϩ bombardment reported by Winters and
Sigmund.13 Based on trajectory calculation, they analyzed
that the surface N atoms are ejected mostly via direct knock-
off collisions with Heϩ. The agreement of Yrs behavior with
the line-of-centers collision model and the N atom ejection
from W suggests that OH and H2O desorb by direct knock-
off collisions.
The knock-off ejection of adsorbates by Csϩ collision
might appear unreasonable in view of the sputtering mecha-
nism proposed for keV ion bombardment;47 heavy ion im-
At high collision energy ͑Ͼ30 eV͒, Yrs͑SiO͒ becomes
larger than Yrs͑OH͒ or Yrs͑H2O͒. Since the surface popula-
J. Chem. Phys., Vol. 107, No. 7, 15 August 1997
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:
136.165.238.131 On: Fri, 19 Dec 2014 11:30:22

Yang, Hwang, and Kang: Csϩ scattering from Si(111)
2617
tion of the oxides should be considerably smaller than OH
and H2O according to the XPS study,36 the observation of
large Yrs͑SiO͒ is difficult to reconcile. One possible explana-
tion is that the efficiency for SiO ejection drastically in-
creases at high collision energy. Note that desorption via
substrate sputtering much prevails at high energy over direct
knock-off ejection of adsorbate.13,41 Another possibility is
collisional conversion of surface OH or H2O into SiO. SiO
formation is thermally a facile reaction on this surface.34
Thus, if collision cascades are generated in the surface region
by high energy impact, then it is plausible to induce SiO
desorption.
tion still plays an important role, as indicated by desorption
of H2 and SiO.
Secondary ion mass spectrometry ͑SIMS͒ from a Si sur-
face reacted with O2 shows a number of ions including
Siϩ, Siϩ2 , SiOϩ, Si2Oϩ, Si2O2ϩ, Si3Oϩ, and etc.48 Despite
numerous superior features of SIMS as a surface analysis
technique, identification of molecular adsorbate structure
from such a complex spectrum is presently an almost impos-
sible task. One will need to fully account for the probability
of secondary particle ionization and the mechanisms of mo-
lecular ion ejection and fragmentation occurring during keV
ion bombardment.
When compared to the above-mentioned TPD, LITD,
and SIMS results, hyperthermal Csϩ reactive scattering
promises has several unique features for adsorbate detection.
First, by detecting Csϩ-bound cluster ions, the problem of
secondary particle ionization in conventional SIMS can
be overcome. The probability of Csϩ–X association
(5ϫ10Ϫ3 –2ϫ10Ϫ2) is higher than the efficiencies of mol-
ecule ionization normally achievable by keV ion sputtering
or by electron impact ionization. The formation of CsXϩ
clusters has previously been reported (Csϩ-SIMS),49–51 and
has been utilized for surface compositional analysis. How-
ever, the destructive nature of keV sputtering in Csϩ-SIMS
hampers molecular adsorbate identification. Second, the re-
active scattering occurs in subpicosecond time scale, which
is faster than the desorption time in LITD ͑уns͒ by at least
three orders. This means that the scattered Csϩ ions pick up
only promptly desorbed molecules. In certain cases, Csϩ col-
lision may initiate surface chemical reactions that occur in
slower time scales. The delayed reaction products, however,
will not be able to form CsXϩ, and thus not be detected. This
scenario explains why thermally unstable species like OH
can be monitored by Csϩ reactive scattering, while it is not
the case in LITD and TPD. Third, hyperthermal collision
induces molecular fragmentation to a lesser degree than keV
sputtering. Hence, molecular identity of an adsorbate is bet-
ter preserved during the collisional desorption. For the sys-
tems where CO, H2O, and C6H6 are chemisorbed on a
Ni͑100͒ surface, all these adsorbates are detected in their
molecular forms by Csϩ reactive scattering.52 We expect that
there may be exceptions to exclusive molecular desorption.
The strong emission of CsSiOϩ observed at high collision
energy possibly indicates adsorbate fragmentation. Competi-
tion between molecular desorption and fragmentation will be
controllable with Csϩ beam energy depending on chemical
property of a system.
D. Identification of surface adsorbate
In the previous sections, formation of CsOHϩ,
CsOHϩ2 , CsSiϩ, and CsSiOϩ has been explained in the
framework of collisional desorption of surface adsorbates or
atoms followed by association reaction with Csϩ. CsOHϩ
and CsOHϩ2 are formed by direct desorption of chemisorbed
OH and H2O, respectively. CsOHϩ and CsOH2ϩ show com-
parable ion intensities. Provided that Yd and Yassoc are ap-
proximately equal for OH and H2O, the ratio
I(CsOHϩ)/I(CsOHϩ2 ) stands for the relative surface popula-
tion of the two species. CsSiϩ production represents sputter-
ing of Si atoms. CsSiOϩ produced at low impact energy
͑Ͻ20 eV͒ is related to SiO desorption from the Si–O–Si
units present on a water-saturated Si surface. However, the
strong increase of CsSiOϩ emission with collision energy
suggests that other reaction channels possibly intervene dur-
ing high energy collision.
Atomic hydrogens are believed29 to exist on the water-
exposed Si͑111͒ surface. CsSiHϩn (nϭ1,2) peaks of Fig. 1͑c͒
indicate the presence of SiHn species on the surface. Con-
trarily, CsDϩ peak is not observed in Fig. 1͑b͒. The absence
of CsDϩ peak indicates the difficulty of dissociating the
Si–D bond by Csϩ collision, and likewise the Si–H bond.
Because of an extremely small efficiency for energy-transfer
from Csϩ to H, the collision may cause SiHn desorption
rather than dissociation of the Si–H bond. It is also probable
that the small molecular polarizability of H atom reduces the
Csϩ–H association yield.
It is interesting to compare the Csϩ reactive scattering
with other desorption or sputtering techniques. Temperature
programmed desorption ͑TPD͒ from the same surface34 gives
rise to H2O, H2, and SiO at desorption temperatures of 160,
800, and 950 K, respectively. The existing adsorbates such
as OH and H are not thermally desorbed. Apparently, a slow
heating rate (Ͻ10 K sϪ1) in TPD has caused thermal reac-
tions to occur, desorbing only thermally available products
like H2 and SiO. Laser induced thermal desorption ͑LITD͒
͑Refs. 33,34͒ produces H2 ͑100%͒, SiO ͑80%͒, SiOH ͑44%͒,
V. SUMMARY AND CONCLUSIONS
͑1͒ Hyperthermal collision of Csϩ ions onto a Si͑111͒ sur-
face reacted with water gives rise to emission of
CsOHϩ, CsOHϩ2 , CsSiϩ, CsSiHϩn (nϭ1,2), and
CsSiOϩ ions. These cluster species are formed as a result
of collisional desorption of surface species, followed by
ion–molecule association between the low-energy scat-
tered Csϩ and the desorbed molecule.
H2O ͑22%͒, Si ͑16%͒, and Si͑OH͒ ͑14%͒, the numbers in
2
parentheses representing the relative branching ratios. The
desorbed products like H2O, SiOH, and Si͑OH͒n(nϭ1,2) re-
flect the presence of H2O and OH on the surface. Due to the
rapid heating rate (Ͼ1010 K sϪ1), thermal reactions are no-
ticeably suppressed in LITD. However, thermal decomposi-
J. Chem. Phys., Vol. 107, No. 7, 15 August 1997
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:
136.165.238.131 On: Fri, 19 Dec 2014 11:30:22

2618
Yang, Hwang, and Kang: Csϩ scattering from Si(111)
16 U. Diebold and P. Varga, Surf. Sci. 241, L6 ͑1991͒.
17 J. D. Beckerle, A. D. Johnson, and S. T. Ceyer, J. Chem. Phys. 93, 4047
͑1990͒.
͑2͒ CsOHϩ and CsOHϩ2 are proposed to be formed from
desorption of OH and H2O adsorbates, respectively,
which occurs via direct knock-off collision with Csϩ. It
follows that the OH and H2O species coexist on the sur-
face with comparable surface concentrations. CsSiϩ and
CsSiOϩ emission suggests desorption of Si and SiO, re-
spectively, via substrate sputtering mechanism.
18 C. Kim, J. R. Han, and H. Kang, Surf. Sci. 320, L76 ͑1994͒.
19 D. Velic and R. J. Levis, J. Chem. Phys. 104, 9629 ͑1996͒.
20 J. L. Robertson, S. C. Moss, Y. Lifshitz, S. R. Kasi, J. W. Rabalais, G. D.
Lempert, and E. Rapoport, Science 243, 1047 ͑1989͒.
21 H. J. Steffen, D. Marton, and J. W. Rabalais, Phys. Rev. Lett. 68, 1726
͑1992͒.
͑3͒ The probability of reactive scattering is 5ϫ10Ϫ4 –2
ϫ10Ϫ3 for 50 eV Csϩ collision on the present surface.
The corresponding probability of Csϩ–X association re-
action is in the order 5ϫ10Ϫ3 –2ϫ10Ϫ2 or slightly
lower.
22 B. C. Kim, H. Kang, C. Y. Kim, and J. W. Chung, Surf. Sci. 301, 295
͑1994͒.
23 J. R. Hahn, H. Kang, S. Song, and I. C. Jeon, Phys. Rev. B 53, R1725
͑1996͒.
24 M. C. Yang, H. W. Lee, and H. Kang, J. Chem. Phys. 103, 5149 ͑1995͒.
25 M. C. Yang, C. H. Hwang, J. K. Ku, and H. Kang, Surf. Sci. 366, L719
͑1996͒.
͑4͒ Unique capabilities of Csϩ reactive scattering as a sur-
face analysis method have been demonstrated. They are
͑i͒ detection of desorbed neutral species, ͑ii͒ fast
probing-time compared to thermal reaction and LITD,
and ͑iii͒ possibility for molecular adsorbate detection.
26 P. A. Thiel and T. E. Madey, Surf. Sci. Rep. 7, 211 ͑1987͒.
27 K. Fujiwara, Surf. Sci. 108, 124 ͑1981͒.
28 D. Schmeisser, F. J. Himpsel, and G. Hollinger, Phys. Rev. B 27, 7813
͑1983͒.
29 S. Ciraci and H. Wagner, Phys. Rev. B 27, 5180 ͑1983͒.
30 H. Ibach, H. Wagner, and D. Bruchman, Solid State Commun. 42, 457
͑1982͒.
ACKNOWLEDGMENTS
31 M. Nishijima, K. Edamoto, Y. Kubota, S. Tanaka, and M. Onchi, J. Chem.
Phys. 84, 6458 ͑1986͒.
The authors thank K. D. Kim and K. Y. Kim for their
assistance. This project was financially supported in parts by
KOSEF ͑CMS͒, ETRI, and School of Environmental Engi-
neering.
32 D. Schmeisser and J. E. Demuth, Phys. Rev. B 33, 4233 ͑1986͒.
33 C. H. Mak, B. G. Koehler, and S. M. George, Surf. Sci. 208, L42 ͑1989͒.
34 B. G. Koehler, C. H. Mak, and S. M. George, Surf. Sci. 221, 565 ͑1989͒.
35 M. Chander, Y. Z. Li, J. C. Patrin, and J. H. Weaver, Phys. Rev. B 48,
2493 ͑1993͒.
36 C. Ponccy, F. Rochet, G. Dufour, H. Roulet, F. Sirotti, and G. Panaccione,
Surf. Sci. 338, 143 ͑1995͒.
1 F. O. Goodman and H. Y. Wachman, in Dynamics of Gas–Surface Scat-
tering ͑Academic, New York, 1976͒.
37 W. Y. Choi, T. H. Kang, and H. Kang, Bull. Kor. Chem. Soc. 11, 290
͑1990͒.
2 S. R. Kasi, H. Kang, C. S. Sass, and J. W. Rabalais, Surf. Sci. Rep. 10, 1
͑1989͒.
38 A. G. Urena, J. Phys. Chem. 96, 8212 ͑1992͒.
39 W. Eckstein, C. Garcia-Rosales, and J. Roth, Nucl. Instrum. Methods
Phys. Res. B 83, 955 ͑1993͒.
3 M. R. Morris, Jr., D. E. Riederer, B. E. Winger, R. G. Cooks, T. Ast, and
C. E. D. Chidsey, Int. J. Mass Spectrom. Ion Processes 122, 181 ͑1992͒.
4 R. G. Cooks, T. Ast, T. Pradeep, and V. Wysocki, Acc. Chem. Res. 27,
316 ͑1994͒.
40 M. C. Yang, C. Kim, H. W. Lee, and H. Kang, Surf. Sci. 357, 595 ͑1996͒.
41 C. Kim, H. Kang, and S. C. Park, Nucl. Instrum. Methods Phys. Res. B
95, 171 ͑1995͒.
5 H. Akazawa and Y. Murata, J. Chem. Phys. 88, 3317 ͑1988͒.
6 S. Schubert, U. Imke, W. Heiland, K. J. Snowdon, P. H. F. Reijnen, and
A. W. Kleyn, Surf. Sci. 205, L793 ͑1988͒.
42 B. J. Garrison, C. T. Reimann, N. Winograd, and D. E. Harrison, Jr., Phys.
Rev. B 36, 3516 ͑1987͒.
7 K. H. Park, B. C. Kim, and H. Kang, J. Chem. Phys. 97, 2742 ͑1992͒.
8 U. van Slooten, O. M. N. D. Teodore, A. W. Kleyn, J. Los, D. Taillet-
Billy, and J. P. Gauyacq, J. Chem. Phys. 179, 227 ͑1994͒.
9 J. S. Martin, J. N. Greeley, J. R. Morris, B. T. Feranchak, and D. C.
Jacobs, J. Chem. Phys. 100, 6791 ͑1994͒.
43 R. F. Garrett, R. J. MacDonald, and D. J. O’Connor, Surf. Sci. 138, 432
͑1984͒.
44 H. W. Lee and H. Kang, Bull. Kor. Chem. Soc. 16, 101 ͑1995͒.
45 P. Kebarle, Annu. Rev. Phys. Chem. 28, 445 ͑1977͒.
46 B. J. Garrison, N. Winograd, and D. E. Harrison, Jr., J. Chem. Phys. 69,
1440 ͑1978͒.
10 J. A. Burroughs, S. B. Wainhaus, and L. Hanley, J. Chem. Phys. 103,
6706 ͑1995͒.
47 P. Sigmund, Phys. Rev. 184, 383 ͑1969͒.
11 G. K. Wehner, Phys. Rev. 108, 35 ͑1957͒; ibid. 112, 1120 ͑1958͒.
12 N. Laegreid and G. K. Wehner, J. Appl. Phys. 32, 365 ͑1961͒; R. V. Stuart
and G. K. Wehner, ibid. 33, 2345 ͑1962͒.
48 A. Benninghoven, W. Sichtermann, and S. Storp, Thin Solid Films 28, 59
͑1975͒.
49 H. A. Storms, K. F. Brown, and J. D. Stein, Anal. Chem. 49, 2023 ͑1977͒.
50 Y. Gao, J. Appl. Phys. 64, 3760 ͑1988͒.
13 H. F. Winters and P. Sigmund, J. Appl. Phys. 45, 4760 ͑1974͒.
14 J. W. Rabalais, T. R. Schuler, and O. Grizzi, Nucl. Instrum. Methods
Phys. Res. B 28, 185 ͑1987͒.
51 K. Wittmaack, Nucl. Instrum. Methods Phys. Res. B 85, 374 ͑1994͒.
52 H. Kang, M. C. Yang, K. D. Kim, and K. Y. Kim, Int. J. Mass Spectrom.
Ion Processes ͑submitted͒.
15 J. Murakami and I. Kusunoki, Nucl. Instrum. Methods Phys. Res. B 33,
560 ͑1988͒.
J. Chem. Phys., Vol. 107, No. 7, 15 August 1997
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:
136.165.238.131 On: Fri, 19 Dec 2014 11:30:22