Full text of DOI:10.1557/JMR.2002.0148

Etching-enhanced ablation and the formation of a
microstructure in silicon by laser irradiation in an
SF6 atmosphere
S. Jesse, A.J. Pedraza,a) and J.D. Fowlkes
Department of Materials Science and Engineering, The University of Tennessee,
Knoxville, Tennessee 37996-2200
J.D. Budai
Solid State Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6056
(Received 25 April 2001; accepted 5 February 2002)
Sequential pulsed-laser irradiation of silicon in SF6 atmospheres induced the formation
of an ensemble of microholes and microcones. Profilometry measurements and direct
imaging with an intensifying charge-coupled device camera were used to study the
evolution of this microstructure and the laser-generated plume. Both the partial
pressure of SF6 and the total pressure of an SF6-inert gas mixture strongly influenced
the maximum height that the microcones attained over the initial surface. The cones
first grew continuously with the number of pulses, reached a maximum, and then
began to recede as the number of laser pulses increased further. The growth of the
cones was closely connected with the evolution of the laser-generated plume.
explain the formation of these structures.14 In the case of
I. INTRODUCTION
the formation of cones under sulfur hexafluoride (SF6)
atmospheres in silicon this explanation is not sufficient
because the cones protrude above the initial surface.11,12
Moreover, we have studied laser-induced silicon micro-
structuring and have observed that it strongly depends on
the irradiation atmosphere.2–5,12 By contrast, under the
same irradiation conditions of wavelength, pulse fluence,
and accumulated fluence, no columns are formed if the
irradiation is performed under 1 atm of high-purity nitro-
gen or argon.2,5
We have proposed a mechanism to explain this relief
in SF6 as well as the morphological details of the micro-
structure shown in Fig. 1. This mechanism involves the
ablation of silicon from regions surrounding the emerg-
ing features and the enhanced redeposition of silicon on
top of them.5,11,12 It was proposed that the deep grooves
and pits between cones result from a laser-induced abla-
tion phenomenon and that this receding part of the sur-
face surrounding each of the cones is the source of an
intense flux of silicon-rich vapor. Growth was explained
to occur through a synergism: the pulsed-laser melting of
the cone tips and deposition of silicon from the vapor
produced by pulsed-laser ablation/etching. The reason
for the tips of the cones being strongly preferred sites for
deposition is that they are melted; thence a very high
axial growth rate could ensue. We have inferred the
growth process just outlined from experimental studies
reported in a series of papers and pointed out that it
is based on the theory of film deposition.15 Similar
Mechanical spallation, expulsion of small droplets
from a transiently laser-melted layer, desorption of atoms
and ions, and evaporation have been identified as some
of the processes responsible for laser-induced ablation.1
When a reactive atmosphere is present, ablation can be
enhanced if volatile compounds within the target mate-
rial are produced and/or if the reaction yields a surface
layer that is easier to remove by laser irradiation than the
target material itself. In this paper this process is identi-
fied as etching-enhanced laser ablation. Generally, part
of the material removed from the target can be collected
on a substrate and this constitutes the basic step for the
deposition of thin films by pulsed laser ablation. In this
process, laser irradiation can produce very drastic
changes in the topography of the target. Commonly,
these changes strongly decrease the deposition rate.
When silicon is irradiated in air, or more generally in
O2-rich atmospheres, with 1000 laser pulses at a
power of 130 MW/cm2, a dense array of high aspect
ratio columns forms on the surface. These columns are
20 to 70 ␮m high and 2–3 ␮m in diameter.2–10 An array
of conical microstructures is produced when laser irra-
diation is performed under an SF6 atmosphere.11–13
Formation of conelike structures due to pulsed-laser
irradiation is well documented for a wide range of ma-
terials.14 In many instances preferential etching could
a)e-mail: apedraza@utk.edu
1002
J. Mater. Res., Vol. 17, No. 5, May 2002
© 2002 Materials Research Society
http://journals.cambridge.org
Downloaded: 16 Mar 2015
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
pressure. It was concluded that the profuse ablation of
etching products under SF6 was responsible for cone
growth. A variety of other reported experiments and very
simple calculations convinced us that the ablation prod-
ucts, which include small clusters and particles, are re-
sponsible for cone growth.5,11,12 The microstructuring
initial stages, however, can be traced to surface pertur-
bations of the laser-induced melt that are generated in the
liquid, but cone growth begins only once microholes ex-
ist. A very detailed study of these initial stages and their
aftermath together with a calculation that predicts
their spacing is presented in another paper.19
In this paper we report on the topographical evolution
as a function of the number of laser pulses at various
pressures. In situ studies help to establish a relationship
FIG. 1. Micrograph of a typical (100) Si microstructure obtained after
1500 laser pulses at a fluence of 1.5 J/cm2.
between the evolution of the fluorescent plume and the
growth and dissolution of cones, which are accompanied
concepts have been used to explain the mechanism of
growth of silicon whiskers by the vapor–liquid–solid
method (VLS).16,17 In the present case we proposed that
the pulsed-laser irradiation has two, almost simultaneous,
effects: it provides the flux of silicon-containing mol-
ecules and it melts the tips of the cones. At variance with
VLS, the laser process does not require the presence of a
catalyst.5
During laser irradiation, there are liquid, solid, and
vapor phases involved, and they interact in a very com-
plex manner. Sanchez et al. proposed that a hydrody-
namic instability was responsible for the growth of
columns in Si irradiated in air.8–10 We examined the
possibility of this process using our experimental results
for oxygen-rich atmospheres and concluded that it was
unlikely that straight columns would grow by this
mechanism.2
This paper is part of a system of studies that focused
on the effects of SF6-rich atmospheres on the production
of a surface microstructure in silicon. In this case, it was
not very difficult to separate the liquid effects from the
ablation products effects by simply switching atmo-
spheres from SF6 to He, as previously reported.12 Using
scanning electron microscopy (SEM), we have followed
the evolution of a set of identified cones in well devel-
oped microhole/microcone structures grown under SF6
and observed that switching to inert Ar gas caused the
cone growth to stop. When irradiation was subsequently
resumed in the SF6 atmosphere, the growth of micro-
cones resumed. If the motion of liquid would be respon-
sible for growth of the cones, growth should not be
stopped as we change atmospheres, contrary to what was
observed. It is well recorded in the literature that SF6
etches silicon during irradiation.18 Indeed, we always
observed the formation of microholes preceding the
formation of microcones during irradiation of SF6. No
effects of this sort are observed when the irradiation is
done in Ar under similar conditions of energy and gas
by the deepening of microholes. Our focus here is to
study the interrelationship between the formation of
microholes, the growth of microcones and the evolution
of the plume with the aim of firmly establishing the
overall growth mechanisms. Studies of the plume evolu-
tion during laser irradiation at high background gas
pressure are presented for the first time. At difference
with other studies of the plume, under these high pres-
sures the plume is confined to microscopic distances
from the target and a long focal microscope is required to
image it.
II. EXPERIMENTAL TECHNIQUES
Silicon substrates were irradiated using a Lambda
Physik LPX-305i (Göttingen, Germany) excimer laser
operating at 248-nm wavelength with a pulse duration of
25 ns. All laser treatments were carried out at nearly
normal laser incidence. The laser beam emerging from
the laser cavity was directed to an optic bench using a
MgF2-coated fused silica mirror.
The optic bench contained an aperture, a fused-silica
lens array, and the irradiation chamber. The aperture,
1 cm × 2.5 cm, partly removed the low-energy tails of
the trapezoidal laser beam. The laser fluence was varied
by adjusting the position of the fused-silica lens array
with respect to the sample holder in the irradiation
chamber.
Test grade p-type, boron doped silicon wafers of
(100) and (111) orientations were used in the experi-
ments. The specified thickness range of the wafers was
475–525 ␮m, and the resistivity ranges were 1.0–6.2 and
1–100 ⍀ cm for the (100) and (111) wafers, respectively.
The native oxide present on each polished substrate was
not removed prior to laser irradiation. The difference in
crystalline orientation did not translate into any signifi-
cant difference in the growth of cones.
J. Mater. Res., Vol. 17, No. 5, May 2002
Downloaded: 16 Mar 2015
1003
http://journals.cambridge.org
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
The laser treatments were carried out in two different
background gases: inert He and reactive SF6, both
99.95% pure. The irradiation chamber was pumped to a
base pressure of 10−7 torr and then filled with the gas of
choice. In situ imaging of the ablation process was per-
formed using a Questar QM100 (Montpelier, MD) long
focal distance microscope attached to a Princeton Instru-
ments PI-MAX (Tucson, AZ) intensifying charge-
coupled device (ICCD) camera. The focal distance of the
microscope was set to approximately 24 cm. All lenses
within the microscope are fused silica and transparent to
the 248-nm laser radiation. This arrangement is required
because most of the light coming from the plume is in the
UV range. The charge-coupled device is comprised of a
512 × 512 imaging array, and the minimum acquisition
time (pulse width) for the camera is 2 ns. The maximum
resolution achieved using this setup was approximately
1 ␮m/pixel.
III. PROFILOMETRY STUDIES
All of the profilometry measurements were done on
laser-microstructured (100) silicon substrates. The verti-
cal resolution of the profilometer is better than 10 nm,
but the lateral resolution is much poorer because the tip
diameter is 5 ␮m. Figure 2 shows the profiles of the
grooves on the surface of two silicon specimens after
irradiation with 200 pulses at a fluence of 2.9 J/cm2 in
two different atmospheres, He and SF6. It can be seen
FIG. 2. Profiles of a Si substrate irradiated with 200 consecutive
that no columns or cones have yet evolved. The grooves
produced under He are very shallow indicating that
very little ablation took place [Fig. 2(a)]. By contrast,
0.25-␮m-deep trenches were produced under an atmos-
phere of SF6 [Fig. 2(b)], showing that more material was
removed during the irradiation in this atmosphere. As
previously analyzed, the etching is most probably made
possible by the decomposition of SF6 during laser irra-
diation.11,12 The fluorine produced by the decomposition
is absorbed in the near-surface region of the specimen
where it reacts with Si producing a volatile compound.
At temperatures higher than 800 K the etching product
with the highest concentration is SiF2.18 Thus, the laser
irradiation causes decomposition of SF6, heats the sur-
face of the silicon, and promotes desorption of the vola-
tile products generated in the near surface region of the
target. This set of phenomena provides the signature of
an etching-enhanced ablation process: the irradiation-
induced formation of a compound that is easily ablated
by irradiation as well.
pulses at a laser fluence of 2.9 J/cm2, in 1 bar of (a) He and (b) SF6.
cone in the central region of the laser spot grew to a
height of 10 ␮m [Fig. 3(b)]. The average height of the
cones increased steadily with the number of pulses reach-
ing a maximum of approximately 40 ␮m after 1000
pulses [Fig. 3(c)]. Thence, there was a steady decrease in
height as the number of laser pulses was further increased
[Figs. 3(d)–3(h)]. After 2250 pulses the average height
decreased to 10 ␮m [Fig. 3(h)].
Similar results were obtained by varying the pressure
of SF6 between 0.125 and 1 bar. The maximum height
was reached after 1000 to 1500 pulses and a significant
decrease in cone height was observed after 2250 pulses.
As the background pressures of SF6 were increased be-
tween 0.125 and 0.5 bar, irradiation produced an increase
in the cones’ height. On the other hand, when the pres-
sure was increased between 0.5 and 1 bar the cones’
height decreased. Figures 4(a)–4(h) show height profiles
as a function of pressure after 2000 pulses at a laser
fluence of 2.5 J/cm2. The serrated pattern is a clear indi-
cation of the presence of microcones. The microcones
grew in every instance around microholes as shown in
the SEM micrograph of Fig. 5.
Profilometry measurements were used to follow the
growth of the cones (Fig. 3). These measurements were
performed on specimens irradiated at 2.5 J/cm2 under a
total background pressure of 0.5 bar of SF6. No cone
growth above the initial surface was detected with
500 pulses [Fig. 3(a)], but with 750 pulses the average
1004
J. Mater. Res., Vol. 17, No. 5, May 2002
Downloaded: 16 Mar 2015
http://journals.cambridge.org
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
FIG. 3. Profiles of a Si substrate irradiated at a laser fluence of 2.5 J/cm2 under 0.5 bar of SF6 at an increasing number of pulses: (a) 500; (b) 750;
(c) 1000; (d) 1250; (e) 1500; (f) 1750; (g) 2000; (h) 2250.
To better understand the effect of ambient atmosphere
on redeposition, irradiations were performed using dif-
ferent proportions of SF6 and Ar while maintaining the
total background gas pressure at a constant value of 1 bar.
The cones’ height increased and then decreased with an
increase in the partial pressure of SF6 [Figs. 6(a)–6(d)].
The maximum height was attained at a partial pressure
of 0.5 bar of SF6 [Fig. 6(c)]. This was similar to the
dependence detected when a background gas of SF6
alone was used. However, the maximum height of the
cones in this case was greater than when the specimens
were irradiated under a gas mixture with a partial pres-
sure of 0.5 bar of SF6.
IV. EVOLUTION OF THE LASER GENERATED
PLUME AND ITS RELATION WITH THE
SURFACE MICROSTRUCTURE
The evolution of a cone structure on a (111) silicon
target over multiple pulses of laser irradiation was fol-
lowed using the ICCD-microscope setup. The irradia-
tions were performed at a fluence of 3.1 J/cm2 and at
various background pressures of SF6 ranging from
0.1 to 0.9 bar. Figure 7 illustrates the relative orientation
of target, microscope image plane, and incident beam.
All the images were taken nearly edge-on; that is, the
irradiated surface was at an angle of approximately 5°
J. Mater. Res., Vol. 17, No. 5, May 2002
Downloaded: 16 Mar 2015
1005
http://journals.cambridge.org
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
with the normal to the image plane of the microscope. In
this way the line of cones at the edge of the substrate
could be clearly observed without significant interfer-
ence from the cones formed beyond the target edge. The
surface was fully irradiated up to the edge as the center
of the laser spot was positioned at the edge of the target.
The edge of the sample was undercut to eliminate reflec-
tion of the incident beam into the camera setup and also
FIG. 4. Profiles of different Si substrates irradiated at a laser fluence of 2.5 J/cm2 with 2000 sequential pulses, under SF6 pressures of the
following: (a) 1.0; (b) 0.875; (c) 0.75; (d) 0.625; (e) 0.5; (f) 0.375; (g) 0.25; (h) 0.125 bar.
FIG. 5. Cross-section SEM of a Si substrate irradiated at a laser fluence of 2.5 J/cm2 after 2000 pulses under 0.5 bar of SF6.
1006
J. Mater. Res., Vol. 17, No. 5, May 2002
http://journals.cambridge.org
Downloaded: 16 Mar 2015
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
FIG. 6. Profiles of different Si substrates irradiated at a laser fluence of 2.5 J/cm2 with 2000 pulses at a total pressure of 1 bar of SF6
and Ar mixtures. The partial pressure of SF6 is the following: (a) 1.0; (b) 0.75; (c) 0.50; (d) 0.25 bar.
500 laser pulses. This result is congruent with pro-
filometry measurements that reveal a relatively smooth
surface and an ablation rate of only 1.2 nm/pulse up to
400 pulses.
The ICCD camera first detects microcones and mi-
croholes, together with the formation of a plume at
550 pulses. The development of cones and the evolu-
tion of the plume were followed at a constant delay
time of 55 ns and a pulse width of 4.5 ns for a total of
22,000 pulses. After 55 ns of delay time the laser light,
which has a FWHM of 25 ns, has essentially ceased, but
the substrate continues to be immersed in a fluorescent
plume. Figures 8(a)–8(h) show selected images taken re-
spectively after 520, 550, 600, 700, 800, 900, 1300,
1700, and 22,000 pulses. The imaged plume extends
across the entire width of the exposed area. The leading
edge of the plume is clearly delineated in these images.
This light is due only to the plume components that de-
cay from excited electronic states emitting visible and
UV photons. Nonemitting atoms, ions, molecules, or
clusters are not detected.20 The tip of the emerging cones
in Fig. 8(b)–8(d) can be seen as small bright spots el-
evated above the surface of the substrate. In these first
images the edge of the fluorescent plume extends up to
40 to 60 ␮m beyond the tips of the cones, and it is higher
in the region where cones show fastest growth.
FIG. 7. Schematic drawing illustrating the relative orientation of tar-
get, microscope image plane, and incident beam.
make possible the detection of microholes formed near
the edge during the ablation process. In the schematic
drawing they are represented by small dots on the un-
dercut surface.
The nominal time that the ICCD gate is open is defined
as the pulse width and is equal to the time taken to
acquire one image. The time elapsed between the initial
arrival of the laser beam to the target and the acquisition
of the image is defined as the delay time. One method-
ology of image acquisition was to maintain a constant
delay time for each image. This acquisition mode is re-
ferred as “continuous” mode, and it is used primarily to
characterize changes associated with the number of laser
pulses. Using this acquisition mode, no plume was de-
tected until the sample was irradiated with approximately
In the early stages of growth we observed that the
plume height is not the same throughout the irradiated
area. It is higher in the regions where the cones are grow-
ing faster. When the cones reach a certain height the
plume tends to become higher in another area, and cones
J. Mater. Res., Vol. 17, No. 5, May 2002
Downloaded: 16 Mar 2015
1007
http://journals.cambridge.org
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
note that the ablation rate is not constant throughout the
growth of the microhole. In the initial stages of formation
of the hole, the ablation rate is comparatively small. As
the depth increases, the number of internal reflections
increases focusing more light to the bottom of the mi-
crohole. This in turn increases the ablation rate. As the
hole deepens even further, light is dispersed more evenly
on the walls of the microhole, maximum intensity drops,
and as a consequence ablation slows. This can also be
clearly realized by comparing the sequence of ICCD im-
ages shown in Fig. 8. In Fig. 8(f) two distinctive holes
emerge in the center of the image that were not seen
400 laser pulses earlier, Fig. 8(e). For these, the average
growth rate through the 400 laser pulses is approximately
200 nm/pulse. This result agrees very well with the ab-
lation rate previously measured at the bottom of micro-
holes using cross-sectional SEM images.11 However, if
the images of emerging holes in Fig. 8(h) are compared
with those in Fig. 8(e), the hole mean deepening rate is
now 20 nm/pulse. At this stage the cone structure is well
under the initial surface.
Images of the target during irradiation can also be
acquired by sequentially increasing the delay time from
one exposure to the next. This mode is called sequential,
and it is often used to characterize events during a single
laser pulse. To study the evolution of the plume 20 to
40 images were acquired in sequential mode. Since the
target is being modified with every pulse, it must
be assumed that all the irradiation pulses produce iden-
tical results. This is a good approximation provided im-
ages are acquired over the span of few pulses. Because
the modification of the target is appreciable only at in-
tervals of hundreds of pulses, we can safely assume that
very small changes take place at intervals of a few tens of
pulses thus qualifying the results as representative of the
changes within a single pulse. The plume evolution was
studied by recording images in sequential mode between
800 and 900 pulses. Each image was captured 1 ns later
than the preceding one. Figure 9 shows a set of images
with delay times increasing from 45 to 90 ns showing the
expansion of the plume under 0.5 bar of SF6 and a laser
fluence of 3 J/cm2. The distance from the cone tips to the
plume edge was measured using 30 images, and it was
plotted as a function of the delay time (Fig. 10). We
found that the drag model for describing the propagation
of a laser-generated plasma fits the data well. This model
predicts that the plume front should advance with time
according to20,21
FIG. 8. Evolution of microcones and microholes in Si irradiated at
3.1 J/cm2 in SF6 at a pressure of 0.5 bar. ICCD images were taken in
continuous mode with a delay time of 55 ns. Parts (a) through (h) show
the microstructure and the plume at 520, 550, 600, 700, 800, 900,
1300, 1700, and 22,000 sequential pulses, respectively.
start to develop further in this area. After 900 pulses
[Fig. 8(e)] the cone profiles are well resolved by the long
focal distance microscope. Together with the growing
microcones, deepening microholes are observed in
Figs. 8(e)–8(i) as bright spots below the initial surface,
and when holes perforate the lateral surface, the bright
spots at the bottom of the holes reveal a strong increase
in fluorescence [Figs. 8(f)–8(i)]. The maximum ablation
rate in the holes measured by the ICCD images was
1.4 ␮m/pulse, a value much larger than the 1.2 nm/pulse
measured on flat surfaces. This measurement was per-
formed by following the progression of a hole through
the substrate during 20 sequential pulses. The advance
of the hole is followed from the point in its wall that first
emerges until the opposite side in its wall emerges, and
thus, not further advance is detected. It is important to
x ס xf(1 − e−␤t
)
.
(1)
Fitting this equation to our data yielded the values
xf ס 46 ␮m and ␤ ס 47.5 ␮s−1.
In Fig. 11, the plume front propagation distance was
plotted as a function of background pressure for various
delay times. For a given delay time, the plume reaches a
1008
J. Mater. Res., Vol. 17, No. 5, May 2002
Downloaded: 16 Mar 2015
http://journals.cambridge.org
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
V. X-RAY ANALYSIS
local maximum at 0.5 bar. The initial velocity as a func-
tion of pressure is plotted in Fig. 12. Although the maxi-
mum velocity was reached at 0.5 bar, remarkably the
initial plume velocity increased again as the pressure was
decreased below 0.2 bar. The profilometry measurements
indicating a maximum cone height at either a partial or
total pressure of 0.5 mbar of SF6 correlate with the fact
that both the plume front distance and the velocity also
are at a maximum when the pressure is 0.5 bar.
In strong contrast with the columns that grow when
silicon is pulsed laser irradiated in air, cones grown under
an SF6 atmosphere are single crystals with the same
orientation as the target. What is even more remarkable
is that no significant line broadening has been detected in
the x-ray analysis. Rocking curve analyses show that the
original target had a line broadening of 0.0142° while
FIG. 9. Evolution of a laser-generated plume during ablation of Si at 3.1 J/cm2 at a pressure of 0.5 bar of SF6. Images were captured in sequential
mode. Parts a through i correspond to delay times of 40, 50, 55, 60, 65, 70, 75, 80, and 95 ns.
FIG. 11. Position of plume front edge relative to target surface as a
FIG. 10. Propagation distance of plume front edge from target surface
as a function of time (0.5 bar SF6, 3.1 J/cm2, 800th–900th shot).
function of background gas pressure for various delay times
(3.1 J/cm2).
J. Mater. Res., Vol. 17, No. 5, May 2002
Downloaded: 16 Mar 2015
1009
http://journals.cambridge.org
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
rate in the microholes produced during multiple irradia-
tion in SF6 is of the order of 1.4 ␮m/pulse, 3 orders of
magnitude higher than for a flat surface.
The profilometry results show that both the pressure
and the nature of the gases present affect the microstruc-
ture produced during multiple pulsed-laser irradiation.
Profilometry measurements also show that the cones’
height attains a maximum when irradiation is conducted
at a background pressure of 0.5 bar of SF6. These results
correlate with the in situ measurements of the maximum
height of the plasma relative to the cone tips and the
average velocity of the plasma as a function of SF6 pres-
sure. Both the plume front distance from the cone tips
and the velocity are at a maximum when the SF6 pressure
is 0.5 bar. Although the maximum velocity was reached
at 0.5 bar, the plume velocity increased again as the pres-
sure was decrease below 0.2 bar (Fig. 12). These results
suggest that SF6 fulfills two roles: first, it acts as an
etcher of silicon, and second, it exerts the background
pressure that tends to restrict the expansion of the plume.
An increase in the ambient background pressure pro-
duces an increase in etching rate thus increasing the pres-
sure produced by the laser-generated plume. This
increase in pressure is reflected in an increase in the ve-
locity of the laser-generated plasma. However, as the
background pressure is increased, the role of SF6 as an
etcher is increasingly counteracted by its role as a back-
ground gas that tends to slow the plume front. When a
gas mixture of SF6 and Ar at a constant total pressure of
1 bar is used, the maximum height of the cones is reached
when the partial pressure of SF6 is also 0.5 bar. However,
the maximum height reached in this case is significantly
lower than for the case where only 0.5 bar of SF6 is
present. This behavior is to be expected because the ad-
dition of Ar further contributes to contain the plasma
without altering the etching power of the background gas.
The dynamics of growth and dissolution as a function
of the number of laser pulses very clearly show a corre-
lation between cone and plume evolutions. The cones
grow continually until approximately 1200 pulses and at
the same time the microholes continually deepen. Up to
this point, the material ablated from within the micro-
holes and redeposited at the tips of the cones is larger
than the amount of silicon that is ablated from the cones’
tips. Like any other laser-exposed area on the substrate
surface, the cones were continually etched and ablated,
but the plasma that surrounded them supplied the tip and
sides with enough silicon to yield a net cone growth. The
laser-melted silicon both at the cone’s side and tip can
efficiently collect both the silicon-rich vapor and small
clusters that form at these relatively high ambient pres-
sures. According to deposition theory a negligible super-
saturation is required to initiate the deposition on the
substrate.15 The accommodation coefficient for mol-
ecules and atoms in a liquid is very close to one.15 For
FIG. 12. Initial velocity of plume front as a function of SF6 pressure,
calculated using the drag model fit (3.1 J/cm2).
the ablated target has a line broadening of 0.0179°. Both the
lengthening and thickening of the cones are related to
their ability to collect silicon from the silicon-rich vapor
as well as small clusters that can form under the high-
pressure background atmosphere.11 Some of the volatile
products that are being generated during the ablation
process must be the constituents of the cones. Thus, we
have proposed that the SiF2 formed at the bottom of the
holes and removed by the irradiation could react later at
the tip and sides of the cones according to18
2SiF2 → SiF4 + Si
.
This is a well-known reaction used in describing
chemical vapor deposition of Si.22 It can be expected that
this decomposition will take place in the gas under the
high supersaturation conditions prevalent during irradia-
tion. The top and sides of the cones melted during irra-
diation are completely resolidified after 100–200 ns
following irradiation. The solidification of silicon after
nanosecond irradiation takes place by the advance of a
solid/liquid interface moving from the nonmelted sub-
strate.23,24 If Si atoms and small clusters are incorporated
while the sides and tips of the cones remain melted,
resolidification should generate a single crystal with the
same orientation as the substrate.
VI. SUMMARY AND DISCUSSION
The amount of silicon evaporated by laser irradiation
in a He atmosphere at a fluence of 3 J/cm2 is fairly small.
The ablation rate of 1.2 nm/pulse in SF6 on relatively flat
surfaces is significantly larger than the laser-induced
evaporation during irradiation in the noble gas atmos-
phere. We have named this process of ablation boosting
etching-enhanced laser ablation. The maximum ablation
1010
J. Mater. Res., Vol. 17, No. 5, May 2002
Downloaded: 16 Mar 2015
http://journals.cambridge.org
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
this reason the deposition takes place preferentially at the
cone tips which remain melted for a longer time than any
other part of the cones.
the shape of the microhole, the absorption and reflections
of the beam and the change in reflectivity with the
angle of incidence. We assume that the incident light is
not polarized. The shape of a typical microhole having
a depth of approximately 140 ␮m was determined
from cross-sectional SEM images, and it is outlined in
Fig. 13(a) together with the path of some of the rays
reflected inside the cavity. Because of the high reflectiv-
ity of a silicon surface to UV light and the steep slopes of
the cavity walls that form an angle of approximately 5°
with the incident beam, there is very strong concentration
of laser energy at the bottom of the holes. Figure 13(b)
shows the variation of intensity of the incident light
along the surface of the hole. At the bottom of the mi-
crohole, this focusing effect produces a fourfold increase
in the light intensity [Fig. 13(b)]. For a nominal laser
fluence of 3 J/cm2, this implies that the intensity at the
bottom of the microholes shown in Figs. 8(d)–8(h) would
be over 12 J/cm2. Bright spots are seen at the bottom of
the microholes in these figures. An intense absorption
of laser light and a high rate of decomposition of SF6
produced this hot plasma. The enhanced production of
fluorine and possibly further ionization should result in
a strong increase of the etching efficiency. Even in a
vacuum background pressure, this very high laser inten-
sity should produce a high ablation rate, as it is shown in
the following.
As the holes deepen, the maximum plume height de-
creases as well. After approximately 1700 pulses, the
fluorescent plume does not reach beyond the tips of
the cones. If the fluorescent plume is an indication of the
amount of ablated material present, less silicon-rich mol-
ecules or clusters are reaching the tips of the cones after
1700 shots. The balance between material ablated and
material redeposited at the cone’s tip becomes tilted to-
ward the former and the cones begin to recede. At a
sufficient number of pulses, the plume is concentrated
at the bottom of the over 300-␮m-deep holes, the
silicon-rich plume is completely trapped, and the cones
recede below the initial surface. The cones that once
protruded 30 ␮m above the original surface after the first
1700 pulses are fully ablated after 22,000 pulses
[Fig. 8(h)]. The bright spots at the bottom of the
holes [Figs. 8(g)–8(i)] due to an enhanced fluorescence
of the plume show that there is a correlation between the
brightness of the plume and the amount of ablated material.
The sequence of Fig. 8 is fully consistent with the
proposition that the cones, despite being constantly ab-
lated by the laser, grow by the overwhelming influx of
silicon-rich material transported from the microholes
onto the molten tips. The evolution of the height profiles
presented here as well as the joint development of mi-
crocones and microholes followed by SEM11 are also
consistent with the picture presented above.
The rate of evaporation of silicon can be approxi-
mately calculated by assuming that ablation takes place
as the irradiated surface is rapidly heated resulting in a
solid–gas interface progressing into the target material.25
Melting is ignored as a part of the physical process that
leads to ablation. This assumption is suitable since the
To understand the causes of the extraordinarily large
ablation in the microholes, the total intensity of laser
radiation in a cavity was calculated taking into account
FIG. 13. (a) Profile of a typical microhole and path of a ray reflecting inside the cavity. (b) Ratio of local laser fluence to nominal fluence as a
function of linear distance along microhole surface.
J. Mater. Res., Vol. 17, No. 5, May 2002
1011
http://journals.cambridge.org
Downloaded: 16 Mar 2015
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
heat of vaporization is several times larger than the heat
of melting and therefore dominates the energy balance.
For silicon, the heat of melting is 50 kJ/mol while the
heat of vaporization is 359 kJ/mol. Also, no atmosphere
or plume–surface interactions are considered. In this
model, the velocity and temperature of the solid–gas
front is governed by the energy absorption and heat trans-
fer characteristics of the target. The Drude–Zener theory
of optical absorption is used which is based on a linear
relation between absorptivity, A, and temperature, Ts.
The following two relations can be obtained:25
This model leads to Eq. (1) and fitted values of maxi-
mum front distance xf ס 46 ␮m and drag coefficient
␤ ס 47.5 ␮s−1.
Geohegan20 observed a similar behavior for the abla-
tion of YBa2Cu3O7−x in 100 mtorr of oxygen with a
value of xf ס 3.0 cm and ␤ ס 0.36 ␮s−1. The initial
velocity of the plume
␷ = ␤xf
,
(6)
0
was in this case 1 cm/␮s. The initial velocity of the
plume due to ablation/etching of silicon in our case is
2.4 cm/␮s indicating that the pressure of the gases pro-
duced during etching is very high. However our drag
coefficient, ␤, is 150 times larger than in the case of
ablation of YBa2Cu3O7−x. This difference reflects the
fact that the background gas pressure is almost 4000
times larger than in the YBa2Cu3O7−x experiments.
In summary, we have shown that the growth of the
cones is closely correlated with the evolution of the laser-
generated plume. This correlation is strong evidence that
the silicon-rich material that is removed by an ablation-
etching process from the microholes feeds the growth of
the cones near them. The interplay between the deepen-
ing of microholes and growing of cones was revealed,
showing a continuing evolution leading to a stage where
holes are so deep that insufficient material reaches the
cones to overcome the always ongoing ablation process.
Then, for a sufficiently large number of pulses, the cones
recede beneath the initial surface. It is shown that, at the
bottom of the microholes, ablation is greatly enhanced
due to multiple reflections and focusing of the laser beam.
AI = V ␳ L + c T ͒
,
(2)
͑
s
v
p s
−MLv
V = V exp
,
(3)
ͩ ͪ
s
0
kbTs
where I is the laser fluence, ␳ is the density of silicon, Lv
is the latent heat of vaporization, cp is the heat capacity,
V0 is the speed of sound in the material, M is the molar
mass, and kb is Boltzman’s constant. The equations are
then solved simultaneously to find Vs and Ts, the evapo-
ration front velocity and temperature. At a laser fluence
of 12 J/cm2 (the estimated intensity at the bottom of the
microholes), Vs is found to be 38.2 m/s. A rough estimate
of the time required to establish steady state evaporation,
tc, is given by
␬
tc =
,
(4)
Vs2cp␳
where ␬ is the thermal conductivity. For the same high
laser fluence, this equation yields tc ס 5.8 ns. If it is
assumed that most of the evaporation takes place after the
establishment of a steady-state front and until the end
of the 25 ns pulse, this model gives an ablation rate of
0.6 ␮m/pulse for a laser fluence of 12 J/cm2. This rate is
500 times higher than the ablation rate of a flat surface in
the presence of SF6.
The advance of the plume front as a function of time
can only be measured with the ICCD camera is approxi-
mately the first 100 ns because the ambient gas quickly
quenches the laser-induced fluorescent light. The pres-
sure at which these measurements were performed was
thousands of times higher than in other reported meas-
urements of plume expansion.20,21 At a pressure of
0.5 bar the plume front reaches 60 ␮m before the back-
ground gas quenches the excitations producing the fluo-
rescent light. We have found that at these very early
stages of plume expansion, the drag-force model fits
better than the shock model originally described by
Zel’dovich and Raizer.26 In the drag-force model it is
assumed that the ablation products are moving as a whole
and that the background gas produces a viscous decelera-
tion of the plume proportional to the expansion velocity,
ACKNOWLEDGMENTS
This research was sponsored by the National Science
Foundation Grant No. DMR-9901238 and by the Oak
Ridge National Laboratory, managed by UT-Battelle,
LLC, for the U.S. Department of Energy under Contract
DE-AC05-00OR22725.
REFERENCES
1. R. Kelly and A. Miotello, Mechanisms of Pulsed-Laser Sputtering,
in Pulsed Laser Deposition of Thin Films, edited by D.B. Chrisey
and G.K. Hubler (Wiley, New York, 1994), pp. 55–87.
2. A.J. Pedraza, J.D. Fowlkes, and D.H. Lowndes, Appl. Phys. Lett.
77, 3018 (2000).
3. A.J. Pedraza, J.D. Fowlkes, and D.H. Lowndes, Appl. Phys. A 69,
S731 (1999).
4. D.H. Lowndes, J.D. Fowlkes, and A.J. Pedraza, Appl. Sur. Sci.
154, 647 (2000).
5. A.J. Pedraza, D. Fowlkes, and D.H. Lowndes, Appl. Phys. Lett.
74, 2322 (1999).
6. V.V. Voronov, S.I. Dolgaev, S.V. Lavrishchev, A.A. Lyalin,
A.V. Simakin, and G.A. Shafeev, Phys. Vib. 7, 131 (1999).
d␷
= −␤␷
.
(5)
dt
1012
J. Mater. Res., Vol. 17, No. 5, May 2002
http://journals.cambridge.org
Downloaded: 16 Mar 2015
IP address: 138.251.14.35

S. Jesse et al.: Etching-enhanced ablation and the formation of a microstructure in silicon by laser irradiation in an SF6 atmosphere
7. V.V. Voronov, S.I. Dolgaev, S.V. Lavrishchev, A.A. Lyalin,
A.V. Simakin, and G.A. Shafeev, Quantum Electron. 30, 710
(2000).
17. E.I. Givargizov, Curr. Top. Mater. Sci. 1, 79 (1978).
18. D. Bauerle, Laser Processing and Chemistry, 2nd ed. (Springer,
Berlin, Germany, 1996), p. 253.
8. F. Sanchez, J.L. Morenza, and V. Trtik, Appl. Phys. Lett. 75, 3302
(1999).
19. A.J. Pedraza, S. Jesse, Y. Guan, and J.D. Fowlkes, J. Mater. Res.
16, 3599 (2001).
9. F. Sa´nchez, J.L. Morenza, R. Aguiar, J.C. Delgado, and
M. Varela, Appl. Phys. Lett. 69, 620 (1996).
10. F. Sa´nchez, J.L. Morenza, R. Aguiar, J.C. Delgado, and
M. Varela, Appl. Phys. A 66, 83 (1998).
11. J.D. Fowlkes, A.J. Pedraza, and D.H. Lowndes, Appl. Phys. Lett.
77, 1629 (2000).
12. A.J. Pedraza, J.D. Fowlkes, S. Jesse, C. Mao, and D.H. Lowndes,
Appl. Sur. Sci. 168, 251–257 (2000).
20. D.B. Geohegan, Mechanisms of Pulsed-Laser Sputtering, in
Pulsed Laser Deposition of Thin Films, edited by D.B. Chrisey
and G.K. Hubler (Wiley, New York, 1994), pp. 115–165.
21. D.B. Geohegan, Appl. Phys. Lett. 60, 2732 (1992).
22. G. Bloem and L.J. Giling, Curr. Top. Mater. Sci. 1, 147 (1978).
23. S. De Unamuno and E. Fogarassy, Appl. Surf. Sci. 36, 1 (1989).
24. R.F. Wood and G.E. Jellison, Jr., in Semiconductors and Semimet-
als, edited by R.F. Wood, C.W. White, and R.T. Young (Aca-
demic Press, Orlando, FL, 1984), Vol. 23, p. 165.
25. S.I. Anisimov, V.A. Khokhlov, Instabilities in Laser-Matter In-
teraction (CRC Press, Boca Raton, FL, 1995), p. 17.
26. Ya.B. Zel’dovich, and Yu.P. Raizer, in Physics of Shocks Waves
and High Temperature Hydrodynamics Phenomena (Academic
Press, New York, 1966), Vol. 1, p. 94.
13. T-H. Her, R.J. Finlay, C. Wu, S. Deliwala, and E. Mazur, Appl.
Phys. Lett. 73, 1673 (1998).
14. S.R. Foltyn, Mechanisms of Pulsed-Laser Sputtering, in Pulsed
Laser Deposition of Thim Films, edited by D.B. Chrisey and
G.K. Hubler (Wiley, New York, 1994), pp. 89–113.
15. J.P. Hirth and J.M. Pound, Prod. Mater. Sci. 11, 107 (1963).
16. R.S. Wagner, in Whisker Technology, edited by A.P. Levitt
(Wiley, New York, 1970), p. 47.
J. Mater. Res., Vol. 17, No. 5, May 2002
Downloaded: 16 Mar 2015
1013
http://journals.cambridge.org
IP address: 138.251.14.35