Ion-enhanced etching of Si(100) with molecular chlorine: Neutral and ionic product yields as a function of ion kinetic energy and molecular chlorine flux

Article abstract of DOI:10.1021/jp993278q

Time-of-flight mass spectrometry (TOFMS) is used to measure neutral and ionic silicon etch products evolved during argon ion-enhanced etching of room temperature Si(100) with molecular chlorine. The yields of these neutral and ionic etch products are examined as a function of ion energy, ion flux, and molecular chlorine flux. For the neutral products, an Ar⁺ ion energy range of 275-975 eV is used, while the ionic product measurements are continued down to 60 eY The atomic Si, SiCl, and SiCl₂ neutral etch products are measured without complications due to fragmentation by using 118-nm laser single-photon TOFMS. Atomic Si and SiCl are the major observed etch products. The ionic Si⁺ and SiCl⁺ etch products are also measured using TOFMS; however, the SiQ₂⁺ species is not observed. The similarities between neutral and ionic Si and SiCl etch products as a function of various parameters suggest a model based on direct collisional desorption. For the observed neutral SiCl₂ product, the absence of SiCl₂⁺ suggests a different mechanism than that for Si and SiCl. For SiCl₂, formation models based on thermal heating or reaction and desorption of neutral species at chemically active surface sites, which are ruled out for Si and SiCl, should be considered.

Full text of DOI:10.1021/jp993278q

J. Phys. Chem. B 2000, 104, 3261-3266
3261
Ion-Enhanced Etching of Si(100) with Molecular Chlorine: Neutral and Ionic Product
Yields as a Function of Ion Kinetic Energy and Molecular Chlorine Flux†
N. Materer,‡ Rory S. Goodman, and Stephen R. Leone*,§
JILA, UniVersity of Colorado and National Institute of Standards and Technology, and
Department of Chemistry and Biochemistry, UniVersity of Colorado, Boulder, Colorado 80309-0440
ReceiVed: September 14, 1999; In Final Form: December 16, 1999
Time-of-flight mass spectrometry (TOFMS) is used to measure neutral and ionic silicon etch products evolved
during argon ion-enhanced etching of room temperature Si(100) with molecular chlorine. The yields of these
neutral and ionic etch products are examined as a function of ion energy, ion flux, and molecular chlorine
flux. For the neutral products, an Ar+ ion energy range of 275-975 eV is used, while the ionic product
measurements are continued down to 60 eV. The atomic Si, SiCl, and SiCl2 neutral etch products are measured
without complications due to fragmentation by using 118-nm laser single-photon TOFMS. Atomic Si and
SiCl are the major observed etch products. The ionic Si+ and SiCl+ etch products are also measured using
TOFMS; however, the SiCl2+ species is not observed. The similarities between neutral and ionic Si and SiCl
etch products as a function of various parameters suggest a model based on direct collisional desorption. For
the observed neutral SiCl2 product, the absence of SiCl2+ suggests a different mechanism than that for Si and
SiCl. For SiCl2, formation models based on thermal heating or reaction and desorption of neutral species at
chemically active surface sites, which are ruled out for Si and SiCl, should be considered.
Introduction
molecular chlorine flux, additional insight into the product
formation mechanisms is obtained.
Etching of silicon is a critical manufacturing process in the
production of microelectronic devices.1-3 Etching is generally
carried out in halogen-containing plasma environments. In these
environments, energetic ion bombardment and reactive gases
act in unison to form volatile silicon halide species that desorb
from the silicon surface.1-3 The synergistic action of reactive
gases and energetic ions is often termed ion-enhanced etching
or reactive ion etching. To produce higher performance devices,
more stringent requirements are being placed on the etching
process. Objectives such as reduced surface damage and smaller
etch features will require additional refinements in current
process technologies.
To develop a microscopic understanding of the ion-enhanced
etching of silicon, many studies of silicon etching have been
performed.4-11 Recently, studies have been carried out with ion
beam energies less than 200 eV.12,13 From previous investiga-
tions, it is generally known that in ion-enhanced etching
molecular chlorine chemically affects the substrate by forming
new chlorine-silicon bonds and disrupting silicon-silicon
bonds.9 This disruption reduces the total number of chemical
bonds that must be broken to collisionally remove silicon-
containing species by Ar+ ion bombardment and increases the
total sputtering yield.9 During ion-enhanced etching, both neutral
and ionic etch products are formed. The ionic product must be
formed during the energetic impact. In contrast, neutral products
can be formed during impact or after impact of the Ar+ ion
due to collisional heating of the surface and at chemically
reactive defect sites created by the impact. By comparing neutral
to ionic species as a function of bombardment energy and
Here we employ time-of-flight mass spectrometry (TOFMS)
to study ion-enhanced etching of Si(100) with molecular
chlorine. We detect the ionic etch products directly and use
single-photon ionization time-of-flight mass spectrometry (SPI-
TOFMS) to detect the neutral etch products. The neutral etch
products are ionized without dissociation of the neutral parent
molecules by using a gentle 118-nm single-photon ionization
process.14 This detection scheme avoids the mass spectral
cracking problem normally associated with electron impact
ionization. The ability to identify the neutral reaction products
allows a direct comparison of the experimental data to molecular
dynamics (MD) simulations. In principle at 118 nm, all
chlorinated silicon etch products can be detected, with the
exception of SiCl4. The lower ionization potential of neutral
silicon atoms (Si) as compared to carbon monoxide (CO) and
nitrogen (N2) permits the Si etch product to be detected without
interferences by CO and N2, which have the same mass. SPI-
TOFMS has been used to detect Si, SiCl, and SiCl2 cleanly
during thermal etching of Si(100)14 and during ion-enhanced
etching15 with molecular chlorine.
Previous molecular beam studies have examined etch products
under ion-enhanced etching conditions.16-21 These results are
summarized in Table 1. Although the etch conditions, chlorine/
ion flux ratio, ion identity, and ion energy are similar in the
studies mentioned above, there are many discrepancies in these
results. Several experimental limitations may contribute to these
conflicting results. One problem is related to the use of electron
impact ionization for the mass spectrometer detection. In many
studies, complex cracking patterns for the etch products are
observed, and the actual product distributions have to be
inferred. Another difficulty stems from the reactivity of radical
etch products. Reactions of these reactive species on the vacuum
chamber walls with halogens and other adsorbed species can
† Part of the special issue “Gabor Somorjai Festschrift”.
‡ Present address, Oklahoma State University, Department of Chemistry.
§ Quantum Physics Division, National Institute of Standards and
Technology.
10.1021/jp993278q CCC: $19.00 © 2000 American Chemical Society
Published on Web 02/23/2000

3262 J. Phys. Chem. B, Vol. 104, No. 14, 2000
Materer et al.
TABLE 1: Summary of Previous Experimental Results of
Ion-Enhanced Etching of Silicon
with molecular chlorine. The silicon sample is positioned
perpendicular to the flight tube and within the extraction region
of a Wiley-McLaren TOFMS.26 For the neutral products,
single-photon ionization at 118 nm (ninth harmonic of a pulsed
Nd:YAG laser) is used to gently ionize the neutral species
without fragmentation.14 The nanosecond pulses of 118-nm light
travel parallel to the substrate and through the center of the
extraction region of the TOFMS. In the case of positive ionic
products, no additional ionization is required. The Ar+ ion beam
is directed normal to the sample.
study
ion energy (keV)
productsa
reference
Ar+, Cl2
Ar+, Cl2
Ar+, Cl2
Ar+, Cl2
Ar+, Cl2
Ar+, Cl2
Ar+, Xe+, Cl2
0.275-1
Si, SiCl, (SiCl2)
SiCl2, SiCl4, (SiCl)
SiCl4
15
16
18
21
35
19
29
1
0.3-3
0.125-0.8
SiCl2, SiCl4
1
3
Si, SiCl, SiCl2
Si, SiCl, SiCl2
SiCl, SiCl2 (SiCl3, SiCl4)
0.05-1
a Parentheses indicate minor products.
The experiments are conducted in an ultrahigh vacuum
chamber with a typical base pressure of 5 × 10-7 Pa. The
chamber pressure rises to 10-5 Pa when argon gas is introduced
into the ion source but before the introduction of chlorine gas.
A chlorine gas background is introduced into the main chamber
through a precision leak valve. Chlorine flux to the sample,
calculated from the pressure measured with a nude ionization
gauge without correction, is varied from 0 to 1.2 of a Cl2
monolayer per second (ML s-1). The Ar+ ion beam current
ranges between 0.2 and 2 µA at the sample surface depending
on the beam energy and source conditions. The full-width at
half-maximum beam current, measured with a Faraday cup,
corresponds to a beam width of ∼0.45 cm. The resulting Ar+
beam flux is 1.2 × 10-2 ML s-1 (µA of current)-1. The Ar+
ion beam is modulated by applying a 1-kHz unsymmetrical
square wave with a 200-V amplitude to one of the deflectors.
Measurements of the current collected by the 1-cm2 sample
during the application of the deflector voltage establishes that
the Ar+ ion beam is deflected off the sample in approximately
400 ns. These ion pulses of 550 µs duration and 1 kHz repetition
rate are used for the experiments.
Figure 1. Schematic diagram of the apparatus employed to study ion-
enhanced etching. The Wiley-McLaren TOFMS is depicted with
separate extraction and acceleration regions. The silicon sample position
is indicated relative to the mutually orthogonal, normal incident ion
beam, TOFMS flight tube, and 118-nm ionization laser.
The Ar+ ion-enhanced neutral and ionic etch products are
mass detected using the TOFMS. The positively charged species
are detected by a channel electron multiplier after traveling
through a 1-m long flight tube. The positively charged etch
products are directly injected into this tube by applying an
electrical pulse (between +300 and +500 V) to the spectrom-
eter’s repeller plate. This pulse is turned on between 3 and 8
µs after the Ar+ ion beam is deflected off the sample. The effects
of the repeller plate voltage and the delay between the deflection
of the ion beam and the repeller pulse on the measurement were
examined, and with the exception of changing the absolute
intensities, these parameters had little effect on the ion product
species observed versus chlorine flux and ion energy. To obtain
a mass spectrum of the ionic products, approximately 500 pulses
at a 1-kHz rate are averaged on a digital oscilloscope.
obscure the true identities of these products. In previous studies,
modulated ion beams were employed to minimize this effect;
however, the modulation rates can sometimes be too low to
exclude species generated by wall reactions. In contrast, SPI-
TOFMS can be used to detect directly the neutral Si, SiCl, and
SiCl2 etch products without ambiguity during ion-enhanced
etching of silicon with a molecular chlorine background.15
Molecular dynamics simulations have also recently elucidated
the microscopic mechanisms of ion-enhanced etching of silicon
with ion energies below 200 eV.22-25 Feil et al. examined the
formation of surface roughness from chemical sputtering of a
chlorinated silicon surface.23 They observed that chlorine
passivation of the surface and the resulting high barriers for
surface diffusion resulted in significant surface roughness.
Molecular dynamics simulations by Barone et al. using F+ and
Cl+ ions found a tendency toward less halogenated products
with increasing collision energy.22 Kubota et al. simulated Ar+-
induced etching of clean and chlorinated silicon and reported
SiCl as the most abundant product for silicon substrates with
an initial chlorine coverage.24 However, differing amounts of
Si and SiCl2 are also found depending on surface disorder and
chlorine coverage. In agreement with the study of Kubota et
al.,24 a MD simulation of ion-enhanced etching of silicon with
chlorine ions by Hanson et al.25 indicated that Si, SiCl, and SiCl2
are the main etch products. At the highest energy examined
(200 eV), atomic silicon was reported as the most abundant
product.25
Detection of neutral products is more difficult. The ionic
products interfere with the detection of the single-photon-ionized
neutral products. To eliminate the charged products from the
extraction region prior to laser ionization, an additional short
voltage pulse (+550 V) is applied first to the extraction grid to
eject the charged etch products from the extraction region. A
voltage (+400 V) is applied to the repeller plate just before
laser ionization. The resulting electrostatic field extracts the
products ionized by the 118-nm radiation into the flight tube.
The timing and experimental details of this process have been
described.15 Typically, 1000 laser pulses at 10 Hz are required
to obtain a spectrum. The larger number of pulses required for
the neutral species with respect to the ionic species is a direct
result of the approximately 103 times greater intensity of the
raw signals for the ionic species. Figure 2 shows a typical
spectrum of the neutral products. Because the ionization energy
of ground-state Ar is greater than the 118-nm radiation, the
observation of Ar+ produced by the 118-nm light in Figure 2
Experimental Section
Figure 1 shows the experimental apparatus15 used to deter-
mine the yields of both neutral and ionic etch products during
Ar+ ion-enhanced etching of a single-crystal Si(100) surface

Ion-Enhanced Etching of Si(100)
J. Phys. Chem. B, Vol. 104, No. 14, 2000 3263
Figure 2. Typical TOFMS spectrum under ion-enhanced etching
conditions. The four peaks are assigned to Si, Ar, SiCl, and SiCl2. The
presence of Ar indicates the possible production of metastable argon
during the sputtering event. The SiCl2 peak is only observed at room
temperature and Ar+ ion energies above 700 eV.
Figure 3. Si+ and SiCl+ ionic etch product yields plotted versus Ar+
ion current measured on the sample. A molecular chlorine background
pressure was maintained at 4 × 10-4 Pa, and an Ar+ energy of 975 eV
was utilized. Over the range of Ar+ ion current, the Cl2 to Ar+ flux
ratio varied from 35 to 140.
implies the production of metastable, or long-lived, electronically
excited neutral Ar atoms during the sputtering event. The weak
SiCl2 peak is observed only at room temperature and fluctuates
considerably from run to run.
The silicon sample is initially cleaned by sputtering with a
2-keV Ar+ ion beam followed by high-temperature annealing.
The cleanliness is verified by Auger electron spectroscopy. The
samples are lightly P-doped (1.2 Ω cm) Si(100) wafers that
have been coated on the backside with approximately 300 nm
of molybdenum. The edges of the silicon sample are sandwiched
between two 25-µm-thick tantalum tabs pressed on by tungsten
clamps. The tantalum tabs are then mechanically attached to a
copper sample holder. This arrangement is similar to a method
previously described.27
Results and Discussion
The neutral and ionic etch products are first examined as a
function of the ion flux under high chlorine/ion flux ratio
conditions. This is followed by measurements of the product
distribution and product ratios as a function of the kinetic energy
of the Ar+ ion. Finally, the variations of the neutral and ionic
etch products with molecular chlorine flux are presented and
discussed. All experiments are performed at room temperature
where thermal etching is negligible.
Ion Flux Dependence of the Etch Products. Figure 3 shows
a linear dependence of the Si+ and SiCl+ ionic etch product
yields versus beam current measured on the sample. A molecular
chlorine background pressure was maintained at 4 × 10-4 Pa.
For this measurement, an Ar+ beam energy of 975 eV was
utilized to provide the maximum variation in current. Over the
Ar+ ion current range utilized in Figure 3, the chlorine/ion flux
ratio varied from 35 to 140. A similar linear dependence of the
neutral etch products as a function of the Ar+ ion flux, with
chlorine/ion flux ratios g10, has been observed previously for
the neutral products.15 However, under conditions of chlorine/
ion flux ratios <10, the rate of production of neutral etch
products has been shown to depend on the ratio of chlorine to
energetic ions.12,16 For the neutral product formation under these
high flux ratio conditions (chlorine/ion flux ratios g10),
Goodman et al. proposed that the region affected by an ion
impact is rechlorinated before a subsequent Ar+ impact within
the same region.15 This proposed explanation implies that under
Figure 4. Si neutral and ionic ion-enhanced etch product yields are
shown as a function of the kinetic energy of the Ar+ ion. The yields of
the products are normalized so that the value at the highest energy is
1. The Cl2 flux was maintained at 0.5 ML s-1
.
these conditions the room temperature silicon surface maintains
a saturation coverage of chlorine during the ion bombardment.
This interpretation is further supported by the observation
here that the yields of the ionic etch products exhibit a linear
dependence on the Ar+ beam current. Because the ions must
be produced by direct collision processes,28 they directly reflect
the surface conditions. Although yields of both neutral15 and
ionic etch products are linear functions of the Ar+ ion flux under
high molecular chlorine partial pressures, this result alone does
not imply that these products are produced via the same
mechanisms. As described below, the linear behavior of the etch
products with respect to the ion current will also break down
as the chlorine/ion flux ratio is decreased below 10.
Neutral and Ionic Products as a Function of Ion Energy.
The yields of neutral and ionic Si and SiCl, plotted as a function
of Ar+ kinetic energy with constant chlorine background
pressure, are presented in Figures 4 and 5. The neutral species
are recorded at ion energies between 275 and 975 V. A detailed
consideration of the relative sensitivities shows that SiCl is by
far the dominant product compared to the other observed etch
products, Si and SiCl2.15 For the ionic species, the more sensitive

3264 J. Phys. Chem. B, Vol. 104, No. 14, 2000
Materer et al.
the range of kinetic energies reported here, it is reasonable to
assume that the probability of secondary ion formation is
approximately independent of energy. At much higher energies,
additional ionization pathways are possible. For example, the
threshold bombardment energy to generate ions by inner-shell
excitations is approximately 4 keV.28 However, over the range
of 60-1000 eV, no additional ionization channels are expected.
Thus, it is unlikely that the secondary ion formation probability
would rapidly change under the conditions studied. Given this
assumption, the similarity between the ionic and neutral etch
products suggests that thermal and defect-mediated pathways
for the neutrals are minimal, and the results support a direct
collision formation mechanism for the Si and SiCl products.
Neutral SiCl2 products can be observed with Ar+ bombard-
ment energies g700 eV (see Figure 2); it is possible that even
higher kinetic energies are required to desorb the surface species
responsible for SiCl2 formation. However, temperature pro-
grammed desorption31 and steady-state thermal etching14 results
suggest that the SiCl2 species is more weakly bound than SiCl.
Yet, SiCl is observed at lower Ar+ kinetic energies than SiCl2.
In addition, if the production of neutral SiCl2 occurs through a
Figure 5. SiCl neutral and ion etch product yields are shown as a
function of the kinetic energy of the Ar+ ion. The yields of the products
are normalized so that the value at the highest energy is 1. The Cl2
flux was maintained at 0.5 ML s-1
.
+
collision mechanism, an ionic SiCl2 product should also be
detection threshold allows measurements down to 60 eV. To
minimize the effect of the changing chlorine/ion flux ratio due
to the variation in ion beam current with beam energy, a
chlorine/ion flux ratio g10 was maintained during these TOFMS
measurements. The product intensities were normalized to the
ion current at each energy. In addition, each plot is indepen-
dently normalized to a value of 1 at the highest energy, although
it must be noted that the absolute ion yields are expected to be
much less than the neutral yields.28
No correction is made for possible changes in the velocity
distributions of the sputtered particles with changes in the Ar+
ion energy. The average kinetic energy of the sputtered Si and
SiCl etch products is expected to be reduced at lower Ar+ kinetic
energies. This lower velocity would increase the ion density in
the extraction region and enhance the measured signal. Molec-
ular beam measurements show a change in the velocity
distribution of the sputtered particle energy from a Maxwell-
Boltzmann-like distribution at low energies to a collision-
cascade at high energies.29 Therefore, some caution should be
exercised with regard to the quantitative yields shown in Figures
4 and 5 as a function of energy.
The results of Figures 4 and 5 can be qualitatively understood.
As the ion energy is increased, more energy is deposited onto
the surface and a greater number of neutral and ionic desorption
events from the chlorinated surface occur. At energies below
275 eV, the neutral product yields fall below the detection limit.
However, the higher detection sensitivity of the ionic products
allows measurements down to 60 eV. Total yield measurements
show that the energy-dependent decrease in sputter yield varies
with E1/2 throughout the energy range.30 There have been very
few measurements of the sputter yield below 100 eV;12,13
however, these measurements show the same functional depen-
dence. Assuming that neutral SiCl is the major product15 and
given both the estimated error bars and the relatively smooth
variation of the yield over the 275-1000 eV energy range for
the neutral species, the yields shown in Figure 5 for SiCl are
consistent with previous measurements.
+
observed. However, the ionic SiCl2 etch product is not
observed, and any possible SiCl2+ product is below detectable
levels. From our calibrations, any possible SiCl2+ ion signal is
less than 1/1000 that of Si+ and less than 1/100 that of SiCl+.
In contrast, the neutral SiCl2 signal is approximately 1/10 of
the Si and SiCl etch products. The virtual lack of any appreciable
+
flux of the SiCl2 ion implies that the neutral SiCl2 product is
possibly formed by a different mechanism than that responsible
for the Si and SiCl etch products. Subsequent thermal reactions
at defect sites are one possibility. In this case, the SiCl2+ product
would not necessarily be generated in conjunction with the
neutral SiCl2. At higher Ar+ bombardment energies, a larger
number of such sites are formed.
The possible importance of thermal reactions at defect sites
for the formation of SiCl2 has important ramifications for MD
studies, which typically follow the reaction only out to several
picoseconds after the collision. One MD study of atomic layer
etching of silicon estimated the time scale for the postcollision
generation of these products during simultaneous exposure to
neutral species and ion beams to be on the order of hundreds
of milliseconds.32 Some progress has been made to better
simulate these processes during ion-enhanced etching. The MD
simulation by Barone et al. utilized a first-order thermal
desorption rate expression to approximate desorption events at
longer times than the simulation time of 1.2 ps.22 In addition, a
recent study by Hanson et al. examined the effect of background
atomic chlorine on the mechanisms of product formation.25
Neutral and Ionic Si to SiCl Product Ratios. Further
information can be gained from the changes in the SiCl/Si and
SiCl+/Si+ intensity ratios as a function of Ar+ ion energy. The
ratios calculated from the intensities shown in Figures 4 and 5,
uncorrected for particle velocities, are plotted versus the ion
beam energy in Figure 6. Because the intensities of the neutral
and ionic products are normalized to 1 at the highest energy,
the neutral and ionic ratios are equal to 1 at the highest energy.
No additional normalization is applied to the data. Both neutral
and ionic ratios of SiCl/Si increase with decreasing primary ion
beam energy. These ratios will depend on the coverage of the
various singly and multiply chlorinated silicon surface species
and the differences between their surface binding energies. The
results show that either the saturation chlorine coverage increases
with decreasing beam energy or that, as the energy provided
The rate of ion formation during sputtering need not neces-
sarily follow the neutral product formation. Secondary ions are
produced directly during the Ar+ collisions.28 In contrast, the
neutral products can also be formed in postcollision processes
by thermal reactions at hot spots generated by the impact or at
chemically reactive defects created during bombardment. Over

Ion-Enhanced Etching of Si(100)
J. Phys. Chem. B, Vol. 104, No. 14, 2000 3265
Figure 6. SiCl/Si and SiCl+/Si+ product yield ratios are plotted versus
the kinetic energy of the Ar+ ion. The Cl2 flux was maintained at 0.5
ML s-1
.
by the Ar+ ions is decreased, there is a preference for the
desorption of more weakly bound SiCl species. An argument
can be considered based on the changing penetration depth of
the Ar+ ion. However, MD studies of Si(100) sputtered with 1
keV Ar+ ions have shown that while the openness of the Si-
(100) surface allows the first four layers to be clearly visible to
the incident Ar+ ion, 98% of the Si products detected come
from only the first three layers.33 Thus, a mechanism based on
changing penetration depths is unlikely.
Figure 7. Cl2 background pressure dependence of (A) Si and Si+
product yields and (B) SiCl and SiCl+ product yields are shown. These
data are collected with an Ar+ energy of 975 eV and flux of 0.01 ML
s-1
.
neutral Si and SiCl ion-enhanced etch products based on the
neutral chlorine/ion flux product dependence.15 This model takes
into account various differences between the Si and the SiCl
etch products and is similar to ones proposed by Backer et al.34
and Levinson et al.13 The observed similarity between the neutral
and the ionic species as a function of molecular chlorine flux
suggests that the neutral and the ionic etch products are created
by similar mechanisms.
For the neutral SiCl product (see Figure 7), the yield initially
increases rapidly from zero to almost its maximum value with
increasing molecular chlorine flux. After this rapid rise, the yield
increases much more slowly with increasing pressure. This
behavior is explained by a simple kinetic model based on two
reactions. One reaction is the adsorption of molecular chlorine
on free silicon surface sites to yield a chlorinated precursor.
The other involves the collisional removal of this precursor to
form a chlorinated etch product and to generate additional free
silicon surface sites.15 The collisional removal of a surface
precursor is further supported by the similarities of the neutral
and ionic SiCl etch products.
The variation with Cl2 pressure shown in Figure 7 for neutral
Si can also be qualitatively understood. A detailed analysis over
a wider range of pressures and ion energies has been presented
for the neutral Si species.15 Goodman et al. found that the
generation of the atomic Si species was more complex than that
of the SiCl product.15 The ejected Si product can arise from
both physical sputtering of unreacted areas of the Si surface
and chemical sputtering from a chlorinated silicon precursor.
In addition, Si produced from chemical sputtering has a much
higher detection probability than that from physical sputtering,
possibly due to the different velocity distributions imparted to
the Si product.15 Initially, the Si yield is composed of physically
sputtered Si. With the introduction of Cl2, there is a rapid
increase in yield due to the formation of chemically sputtered
Si, which may indicate it is detected more efficiently than the
physically sputtered Si. The yield increases to a maximum value.
Following this maximum, a slow decrease in the neutral Si yield
occurs. By careful fitting of the atomic Si yields versus
molecular chlorine over an extensive range of molecular chlorine
flux and at several different Ar+ ion energies, the decrease has
been ascribed to the increasingly chlorinated silicon surface.15
This decrease results in a reduction in the atomic Si product in
favor of the more chlorinated etch products.15
At energies less than approximately 400 eV, the amount of
chlorinated product with respect to the atomic etch product
seems to increase much more rapidly. This effect is clearly seen
in the ionic product ratio because these measurements are
possible at much lower energies. The possible changes in the
relative velocity distribution between the Si and SiCl discussed
above may be responsible for some of this more rapid increase
in the product ratios. However, the large increase also supports
a rapid increase in amount of chlorinated species on the surface.
Evidence for the latter explanation comes from the MD
simulation by Hanson et al. over an energy range from 50 to
200 eV,25 which also indicates an increasing flux of desorbed
chlorinated species with decreasing energy. These observations
suggest that there is a clear difference in the sputtering
mechanisms that occur at lower ion beam energies. Additional
experimental studies of the etch products are clearly required
to connect the large volume of work in the keV range (see Table
1) to the lower energy studies. Only recently have there been
several total yield studies of Ar+ ion-enhanced etching with
molecular chlorine and ion beam energies less than 200 eV.12,13
Molecular Chlorine Flux Dependence. Varying the chlorine/
Ar+ ion flux from 0 to 10 allows for the examination of the
sputtering processes as the surface transforms from a clean
silicon surface to a chlorinated silicon surface at saturation
coverages. Under conditions of constant ion flux, the chlorine
coverage will depend only on the chlorine flux to the surface.
Figure 7A shows the measured Cl2 pressure dependence for the
Si and Si+ etch products with an Ar+ ion energy of 975 eV.
Figure 7B shows similar results for the SiCl and SiCl+ etch
products also at an Ar+ energy of 975 eV. Because absolute
yields are unknown, the neutral and ionic products are scaled
to provide the best relative fit between the variations of both
products with the Cl2 pressure. In addition, the variation of the
ionic product signals with chlorine flux observed at 200 eV is
found to be very similar to the variations at 975 eV.
Reaction kinetics of ion-enhanced etching were previously
examined by measurement of the product yields as a function
of chlorine pressure with constant ion flux.15 Goodman et al.
proposed a simple kinetic model for the formation of both

3266 J. Phys. Chem. B, Vol. 104, No. 14, 2000
Materer et al.
Conclusions
Office, Physics Division. Additional equipment and support were
provided by the National Science Foundation and the National
Institute of Standards and Technology.
Neutral and ionic silicon etch products that are evolved during
Ar+ ion-enhanced etching of room temperature Si(100) with
molecular chlorine are measured using TOFMS. The yields of
neutral and ionic etch products are examined as a function of
ion energy, ion flux, and molecular chlorine flux. The neutral
product yields are measured over an Ar+ ion energy range of
275-975 eV, while the ionic product measurements start at 975
and continue down to 60 eV. For the neutral etch products, the
Si and SiCl species dominate the TOFMS spectra with a minor
contribution of SiCl2 at the higher Ar+ ion kinetic energies. For
the positively charged ionic etch products, only Si+ and SiCl+
are observed. In addition, the yield of each product increases
with increasing ion energy. Although neutral SiCl2 products can
be observed at Ar+ bombardment energies above 700 eV, the
References and Notes
(1) Suzuki, K.; Itabashi, N. Pure Appl. Chem. 1996, 68, 1011.
(2) Flamm, D. L. Pure Appl. Chem. 1990, 62, 1709.
(3) Donnelly, V. M.; Flamm, D. L. Solid State Technol. 1981, 24, 161.
(4) Hamers, R. J.; Wang, Y. J. Chem. ReV. 1996, 96, 1261.
(5) Engel, T. Jpn. J. Appl. Phys., Part 1 1996, 35, 2403.
(6) Leone, S. R. Jpn. J. Appl. Phys., Part 1 1995, 34, 2073.
(7) Waltenburg, H. N.; Yates, J. T., Jr. Chem. ReV. 1995, 95, 1589.
(8) Rhodin, T. Prog. Surf. Sci. 1995, 50, 131.
(9) Winters, H. F.; Coburn, J. W. Surf. Sci. Rep. 1992, 14, 161.
(10) Yu, M. L.; DeLouise, L. A. Surf. Sci. Rep. 1994, 19, 285.
(11) Zalm, P. C. Vacuum 1986, 36, 787.
(12) Chang, J. P.; Arnold, J. C.; Zau, G. C. H.; Shin, H.-S.; Sawin, H.
H. J. Vac. Sci. Technol., A 1997, 15, 1853.
(13) Levinson, J. A.; Shaqfeh, E. S. G.; Balooch, M.; Hamza, A. V. J.
Vac. Sci. Technol., A 1997, 15, 1902.
(14) Materer, N.; Goodman, R. S.; Leone, S. R. J. Vac. Sci. Technol.,
+
ionic SiCl2 product is not observed. It is possible that the
neutral SiCl2 product is formed by subsequent thermal reactions.
This result has important implications for MD studies. In
addition, an apparently rapid increase in the SiCl/Si and the
SiCl+/Si+ product ratios at approximately 400 eV occurs. A
propensity for increasingly chlorinated etch products is possible
as the Ar+ ion kinetic energy decreases.
A 1997, 15, 2134.
(15) Goodman, R. S.; Materer, N.; Leone, S. R. J. Vac. Sci. Technol.,
A 1999, 17, 3340.
(16) Coburn, J. W. J. Vac. Sci. Technol., B 1994, 12, 1384.
(17) Kolfschoten, A. W.; Haring, R. A.; Haring, A.; de Vries, A. E. J.
Appl. Phys. 1984, 55, 3813.
The Cl2 flux, energy, and current dependencies of the neutral
and the ionic products show clear similarities to those predicted
by a previously proposed model that describes the formation
of neutral Si and SiCl etch products based on product formation
by direct collisions.15 Under a molecular chlorine background
and Cl2/Ar+ flux ratios <200, a direct collisional mechanism
is currently the best explanation for the experimental data
presented here and in ref 15. The observation that the ion
products follow the neutral products supports this assumption,
since the ionic products must be directly produced. Additional
theoretical modeling will provide more insight. One clear
challenge for the MD simulations is the influence of the
molecular chorine background on the surface processes. In
addition, these simulations should be extended to higher energies
(>200 eV) to allow better comparison with experiments.
Similarly, the challenge for experimental studies is the deter-
mination of the etch products at lower Ar+ ion energy (<200
eV) and higher Cl2/Ar+ flux ratios.
(18) McNevin, S. C.; Becker, G. E. J. Vac. Sci. Technol., B 1985, 3,
485.
(19) van Zwol, J.; van Laar, J.; Kolfschoten, A. W.; Dieleman, J. J.
Vac. Sci. Technol., B 1987, 5, 1410.
(20) Barker, R. A.; Mayer, T. M.; Pearson, W. C. J. Vac. Sci. Technol.,
B 1983, 1, 37.
(21) Rossen, R. A.; Sawin, H. H. J. Vac. Sci. Technol., A 1987, 5, 1595.
(22) Barone, M. E.; Graves, D. B. J. Appl. Phys. 1995, 78, 6604.
(23) Feil, H.; Dieleman, J.; Garrison, B. J. J. Appl. Phys. 1993, 74, 1303.
(24) Kubota, N. A.; Economou, D. J.; Plimpton, S. J. J. Appl. Phys.
1998, 83, 4055.
(25) Hanson, D. E.; Voter, A. F.; Kress, J. D. J. Appl. Phys. 1997, 82,
3552.
(26) Wiley, W. C.; McLaren, I. H. ReV. Sci. Instrum. 1955, 26, 1150.
(27) Koehler, B. G.; Mak, C. H.; Arthur, D. A.; Coon, P. A.; George,
S. M. J. Chem. Phys. 1988, 89, 1709.
(28) Yu, M. L. In Sputtering by Particle Bombardment III; Behrish, R.,
Wittmaack, K., Eds.; Springer-Verlag: Berlin, Heidelberg, 1991; p 91.
(29) Oostra, D. J.; Haring, A.; van Ingen, R. P.; de Vries, A. E.; van
Veen, G. N. A. J. Appl. Phys. 1988, 64, 315.
(30) Steinbruchel, C. Appl. Phys. Lett. 1989, 55, 1960.
(31) Szabo, A.; Farrall, P. D.; Engel, T. Surf. Sci. 1994, 312, 284.
(32) Athavale, S. D.; Economou, D. J. J. Vac. Sci. Technol., A 1995,
13, 966.
Acknowledgment. N.M. would like to acknowledge experi-
ence during his graduate studies gained under the guidance of
G. A. Somorjai. The authors would like to acknowledge the
expert help of the JILA Instrument Shop. They gratefully
acknowledge support of this work by the U.S. Army Research
(33) Smith, R.; Harrison, D. E., Jr.; Garrision, B. J. Phys. ReV. B 1989,
40, 93.
(34) Backer, R. A.; Mayer, T. M.; Pearson, W. C. J. Vac. Sci. Technol.,
B 1983, 1, 37.
(35) Sanders, F. H. M.; Kolfschoten, A. W.; Dieleman, J.; Haring, R.
A.; Haring, A.; de Vries, A. E. J. Vac. Sci. Technol., A 1984, 2, 487.