APPLIED PHYSICS LETTERS 92, 223108 共2008兲

Pulsed laser dewetting of patterned thin metal films: A means

of directed assembly
Philip D. Rack,1,2,a兲 Yingfeng Guan,1 Jason D. Fowlkes,2 Anatoli V. Melechko,2 and
Michael L. Simpson1,2
1
Department of Materials Science and Engineering, The University of Tennessee, Knoxville,
Tennessee 37996, USA
2
Center for Nanophase Materials Science, Oak Ridge National Laboratory, Oak Ridge,
Tennessee 37831, USA
共Received 25 April 2008; accepted 15 May 2008; published online 4 June 2008兲
Thin nickel films were patterned into various shapes and treated with a series of laser pulses. The
edges and vertices of the patterned shapes act as programable instabilities, which enable directed
assembly via dewetting when the laser energy density is above the melting threshold. The pattern
formations were monitored as a function of laser pulse and the retraction process was attributed
liquid dewetting and a subsequent resolidification. The calculated retraction velocity 共83 m / s兲 and
liquid lifetime 共12.3 ns兲 were consistent with the measured nickel retraction distances. The vertices
of the shapes had an initially larger retraction velocity which was attributed to an additional in-plane
curvature. © 2008 American Institute of Physics. 关DOI: 10.1063/1.2939436兴

One of the challenges of nanoscience and technology is
understanding and controlling bottom up directed assembly
of materials. A lot of work has been done studying the as-
sembly of continuous thin polymer1 and metal films,2 which
reveal interesting dewetting phenomenon via nucleation of
holes and Spinodal dewetting. Nucleation driven processes
are not spatially correlated; however Raleigh–Taylor or
surface wave instabilities give rise to spatially correlated
assembly.
While the break-up and pattern formation via dewetting
of continuous thin metal and polymer films has been studied
in detail, less work has been devoted to the dewetting and
pattern formation of confined or patterned thin films. The
edges of the patterned thin film give rise to programable
instabilities which can be useful for the directed assembly of
materials. One means that has been used to investigate
spatially confined wetting properties is via chemically pat-
terning the surface. For instance, Darhuber et al.3 litho-
graphically patterned the self assembled monolayer
octadecyl-trichlorosilane 共OTS兲 onto silicon oxide surfaces
to pattern hydrophobic 共OTS兲 and hydrophilic 共silicon oxide兲
regions. They subsequently investigated both experimental
and simulated shapes for various glycerol pattern geometries
and liquid volumes. Oriented gold nanoparticles have been
grown via solid state dewetting of a gold film deposited onto
arrays of inverted silicon pyramid shaped pits.4 Choi et al.5
used a similar approach to pattern gold nanoparticles, and
subsequently synthesized silicon nanowires from the orga-
nized gold nanoparticles via a vapor-solid-liquid growth
process. Of particular interest to the work presented here,
Habenicht et al.6 recently investigated dewetting induced
nanoparticle “jumping” which resulted from the inertia of the
dewetting fronts of pulsed laser irradiated nanoscale gold
triangles. In this work, the potential surface energy associ-
ated with the patterned gold triangles was correlated to the
a兲
Author to whom correspondence should be addressed. Tel.: 865-974-5344.
FAX: 865-974-4115. Electronic mail: prack@utk.edu.
velocity of the ejected nanoparticle which had an effective
conversion efficiency of ⬃20%.
In this paper we demonstrate the directed assembly of
patterned thin nickel films via nanosecond pulsed laser pro-
cessing. The short liquid lifetimes offer a unique way to
monitor the time dependence of the dewetting process and
the subsequent pattern formation. Thin nickel films 共30 nm兲
were dc sputter deposited from a nickel foil 共3 in. diameter,
30 W, 3 mTorr Ar sputtering gas兲 onto electron beam lithog-
raphy patterned polymethylmethacrylate 共or PMMA兲 coated
共60 nm兲 silicon substrates. Thin nickel patterns of a variety
of sizes of circles, squares, and triangles were achieved by a
conventional lift-off process. No attempt was made to re-
move the native silicon oxide film. Subsequently the thin
patterned features were simultaneously laser treated with a
series of ⬃25 ns 共full width at half maximum兲 248 nm
wavelength laser pulses with an energy density of
420 mJ/ cm2, which was previously determined to be above
the melting threshold of nickel.7 To evaluate the dewetting
pattern evolution, plane-view scanning electron microscope
images were taken after pulses 1, 2, 3, 5, and 10, respec-
tively, and tilted images were taken after pulse 10 to image
the final patterns.
Figure 1 shows a series of scanning electron micro-
graphs of a patterned circle 共radius= 10.6 ␮m兲, square 共edge
length= 9.7 ␮m兲, and an equilateral triangle 共edge length
= 12.3 ␮m兲 as a function of the number of laser pulses,
which reveal some very intriguing dewetting characteristics.
Figure 2 is a plot of the cumulative retraction 共shrinkage兲 of
the center-to-edge distance for the circle, the vertex of the
square, the vertex of the triangle, and the edge center of the
square and triangle. For all the features, the general trends in
the amount of lateral contraction per pulse or the growth rate
共proportional to the slopes in Fig. 2兲 is initially higher over
the pulses 1 and 2 and then begins to slow down and in some
cases stop by the tenth pulse.
The inset in Fig. 2 is a high contrast image of the circle
after 3 pulses which reveals that each pulse has a concentric
signature bright region followed by a dark region. Compar-

0003-6951/2008/92共22兲/223108/3/$23.00 92, 223108-1 © 2008 American Institute of Physics

223108-2 Rack et al.
FIG. 1. Scanning electron micrographs of pulsed laser treated thin nickel
patterns. The top images are the initial thin film circle, square and triangle.
Subsequent images in each column are after 1, 2, 3, 5, and 10 pulses. The
bottom image is a tilted view of the pattern after 10 laser pulses 共the dashed
lines on the top square and triangle illustrate an axis of the lateral contrac-
tion from the vertices and the solid lines demonstrate the axes from the
center of the edges兲.
ing these two regions and the slight topology that forms at
the bright/dark interface 共revealed in the tilted images兲, the
outer bright ring appears to be due to the liquid retraction,
and the dark ring is due to the solidification of the liquid. By
analyzing the pulse 1 ring of the scanning electron micro-
graph in Fig. 2, the retraction during the liquid phase was
estimated to be ⬃580 nm, and the estimated retraction due to
solidification was ⬃150 nm. Considering the density change
from the liquid 共7.89 g / cm3兲 and the solid 共8.91 g / cm3兲 and
assuming the solidification retraction is isotropic the esti-
mated retraction during the nickel solidification was
FIG. 2. Measured edge-to-center retraction distances as a function of the
number of laser pulses for the circle, the edge center and vertex of the
square, and the edge center and vertex of the triangle 共see dashed and solid
lines in Fig. 1 for clarification兲. Inset is an enhanced contrast image of the
circle after 3 pulses demonstrating the two-ring signature for each laser
pulse.
Appl. Phys. Lett. 92, 223108 共2008兲
FIG. 3. Simulated temperature as a function of time of the 30 nm nickel
surface temperature for the 25 ns, 420 mJ/ cm2 laser pulse demonstrating a
⬃12.3 ns liquid lifetime.
⬃175 nm, in good agreement with the observed values. The
contrast change in the scanning electron micrographs are as-
sumed to be due to some silicide formation at the Ni–Si
interface which we have previously observed in some solid
state nickel dewetting studies.8
Dewetting phenomena can vary significantly depending
on whether the film is in a solid or liquid state. To understand
the dewetting patterns, the laser induced time-temperature
profiles of the 30 nm nickel films were simulated using
algorithms similar to those outlined in Ref. 9 and described
in Ref. 10 in more detail. Figure 3 illustrates the surface
nickel temperature as a function of time for a 30 nm film
irradiated with a 25 ns laser pulse of 248 nm wavelength at
420 mJ/ cm2. For simplicity, bulk properties for nickel were
assumed in the thermal modeling. As demonstrated by the
plot, the nickel film melts and has a liquid lifetime of
⬃12.3 ns. Recently, Favazza et al.11 compared the growth of
instabilities in the solid state versus the liquid state in pulsed
laser treated metal thin films. It was concluded that the cur-
vature induced transport via surface and volume diffusion,
evaporation, and stress in the solid state was inconsequential
relative to liquid state transport governed by capillary and
viscous dissipation as described by the Navier–Stokes equa-
tions and simplified via the lubrication approximation.12
Following the conclusions of Favazza et al. and
Habenicht et al., we have estimated the liquid retraction
velocity 关v = 共2S␳−1d−1兲0.5兴 of the patterned nickel films to be
83 m / s, where S is the spreading parameter, ␳ is the liquid
nickel density 共7.8 g / cm3兲, and d is the film thickness. The
spreading parameter was calculated from literature values
of the surface tensions 共␥兲 of Si-air 共0.75 J / m2兲, Si–Ni
共1.714 J / m2兲 共Ref. 13兲, and liquid nickel–air 共1.78 J / m2兲.14
To confirm the validity of these parameters, the contact angle
was measured on several small laser treated nickel disks and
the measured contact angle was between 108° and 115°,
which compares favorably to the calculated value of 122° per
Young’s equation for the liquid nickel-solid silicon-air sys-
tem. To compare the measured retraction distances demon-
strated in Fig. 2, the calculated retraction velocity was mul-
tiplied by the simulated liquid lifetime which yields a
retraction distance of ⬃1000 nm/pulse, which is in agree-
ment with the range in the measured retraction distances over
223108-3 Rack et al.
the initial 2–3 pulses. The decrease in the retraction velocity
at beyond the second and third pulse is likely due to thick-
ening of the dewetting front which decreases the curvature
and increases viscous dissipation. The thickening is exacer-
bated at the vertices as it accumulates material from two
fronts.
The observation that the triangle and square vertices re-
tract more than the center of the edges is an interesting ob-
servation. If the entire perimeter of the square and triangle
retracted toward the center at the same velocity, curvature
would evolve with the triangle curvature being greater than
the square, consistent with what is observed 共particularly evi-
dent after pulse 3 in Fig. 1兲. The enhanced velocity of the
vertices is speculated to be due to in-plane curvatures.1 For
the patterned features, there are two curvatures to consider,
namely, 共1兲 the orthogonal or out-of-plane curvature which is
initially proportional to the film thickness and 共2兲 the in-
plane curvature associated with the vertices of the triangle
and square, the radius of the circle and is effectively zero at
the center of the triangle and square edge lengths. An esti-
mate of the in-plane radii of curvatures for the triangle and
square vertices were measured to be 230 and 720 nm, re-
spectively, from the as-deposited scanning electron micro-
graphs. These are, of course, much smaller than the circle
radius of 5300 nm, and ⬁ for the edge centers. Applying the
retraction equation above and replacing the thickness term
共d兲 with the measured in-plane radii of curvatures, an esti-
mate of the velocity associated with the in-plane curvature
was determined to be ⬃30 and 17 m / s for the triangle and
square vertices, respectively. Thus, the observed increase in
the retraction distances of the vertices as compared to the
edges and the circle is consistent with the additional in-plane
curvature.
By inspecting the images in Fig. 1 and comparing them
to the retraction distances/rates demonstrated in Fig. 2, sev-
eral final trends can be understood. Initially, based on the
common out-of-plane curvatures 共film thickness兲 and the
negligible in-plane curvatures, it is understandable that the
edges of the triangle and square and the circle all have com-
parable retraction distances over the first few pulses. The
circle retraction rate slows down more so than the edges
because the material accumulation for the circle is greater
than an edge front. The triangle edge effectively stopped af-
ter the fifth pulse because it coalesced into three lines. Thus,
comparing this to the square border which is still retracting
at the tenth pulse, is an artifact of the size.
Appl. Phys. Lett. 92, 223108 共2008兲
In summary, pulse laser treating patterned nickel thin
films has been studied as a means to direct the assembly of
thin film materials. The patterned edges act as programable
instabilities and, when laser heated, assemble in predictable
ways. The laser energy density was beyond the melt thresh-
old for the nickel films thus the liquid fronts have been cor-
related to the dewetting of the films during the short liquid
lifetime. The lateral retraction and pattern formation was cor-
related to a two step process: 共1兲 initially, the surface tension
drives the flow of the melted nickel films and 共2兲 a smaller
contraction associated with the density difference between
the liquid and solid when the liquid film solidifies. The en-
hanced retraction rate of the vertices of the triangle and
square were correlated to the additional in-plane curvature.
The reduced retraction rates at subsequent pulses were attrib-
uted to thickening of the front which reduces the curvature
and enhances viscous dissipation.
A.V.M. and M.L.S. acknowledge support from the
Material Sciences and Engineering Division Program of the
DOE Office of Science. P.D.R. and J.D.F. acknowledge that a
portion of this research was conducted at the Center for
Nanophase Materials Sciences, which is sponsored at Oak
Ridge National Laboratory by the Division of Scientific User
Facilities, U.S. Department of Energy.
1
J. Becker, G. Grun, R. Seemann, H. Mantz, K. Jacobs, K. R. Mecke, and
R. Blossey, Nat. Mater. 2, 59 共2003兲.
2
S. Herminghaus, K. Jacobs, K. Mecke, J. Bischof, A. Fery, M. Ibn-Elhaj,
and S. Schlagowski, Science 282, 916 共1998兲.
3
A. A. Darhuber, S. M. Troian, S. M. Miller, and S. Wagner, J. Appl. Phys.
87, 7768 共2000兲.
4
A. L. Giermann and C. V. Thompson, Appl. Phys. Lett. 86, 121903
共2005兲.
5
W. K. Choi, T. H. Liew, H. G. Chew, F. Zheng, C. V. Thompson, Y. Wang,
M. H. Hong, X. D. Wang, L. Li, and J. Yun, Small 4, 330 共2008兲.
6
A. Habenicht, M. Olapinski, F. Burmeister, P. Leiderer, and J. Boneberg,
Science 309, 2043 共2005兲.
7
Y. F. Guan, R. C. Pearce, A. V. Melechko, D. K. Hensley, M. L. Simpson,
and P. D. Rack, Nanotechnology 19, 235604 共2008兲.
8
S. J. Randolph, J. D. Fowlkes, A. V. Melechko, K. L. Klein, H. M. Meyer,
M. L. Simpson, and P. D. Rack, Nanotechnology 18, 465304 共2007兲.
9
J. Trice, D. Thomas, C. Favazza, R. Sureshkumar, and R. Kalyanaraman,
Phys. Rev. B 75, 235439 共2007兲.
10
A. J. Pedraza, Y. F. Guan, J. D. Fowlkes, and D. A. Smith, J. Vac. Sci.
Technol. B 22, 2823 共2004兲.
11
C. Favazza, R. Kalyanaraman, and R. Sureshkumar, Nanotechnology 19,
072107 共2007兲.
12
J. A. Diez and L. Kondic, Phys. Fluids 19, 072107 共2007兲.
13
M. I. Vasiliu and V. N. Eremenko, Sov. Powder Metall. Met. Ceram 4, 238
共1965兲.
14
M. Chen, C. Yang, and Z. Y. Guo, Mater. Sci. Eng., A 292, 203 共2000兲.