You are on page 1of 9

Langmuir 2003, 19, 5475-5483 5475

Silicone Transfer during Microcontact Printing


Karin Glasmästar,* Julie Gold, Ann-Sofie Andersson, Duncan S. Sutherland, and
Bengt Kasemo
Department of Applied Physics, Chalmers University of Technology,
SE-412 96 Göteborg, Sweden

Received September 16, 2002. In Final Form: April 14, 2003

Microcontact printing (µCP) is a widely used method to make miniaturized patterns on surfaces. In this
work, the issue of the possible transfer of stamp material from the stamp to the substrate during stamping
was addressed. Poly(dimethylsiloxane) was used to stamp Milli-Q water or buffer on substrates of SiOx,
TiO2, and Au. X-ray photoelectron spectroscopy (XPS) and time-of-flight secondary ion mass spectrometry
(ToF-SIMS) were used to measure and characterize the substrate before and after stamping to detect the
possible transfer of stamp-related material to the substrates. Both the XPS and the ToF-SIMS analyses
show that silicone-related material is transferred from flat stamps and that even more material is transferred
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

from patterned stamps. Interestingly, a UV/ozone treatment (essentially oxidation of the surface) of the
stamps before inking and stamping significantly reduces the silicone transfer. Two application examples
are used to illustrate the importance of silicone transfer to the substrates during µCP: water condensation
patterns and supported lipid bilayer formation.
Downloaded via SEJONG UNIV on March 22, 2024 at 01:34:05 (UTC).

Introduction layers function as a diffusion barrier and are reported to


Microcontact printing (µCP), or soft lithography, was be quite brittle10,11 compared to bulk PDMS.
introduced by Kumar et al. in the early 1990s1,2 as a means The use of hydrophobic inks such as alkanethiols in
for fast and low-cost pattern transfer onto gold films µCP do not normally require any pretreatment of the
supported on silicon substrates. The technique has since stamps.2,12 However, if the ink is hydrophilic, for example,
become widely used to pattern surfaces for a variety of water-soluble proteins,13-15 phospholipids,16 or amines,17
research and technical applications.1-6 During µCP, an the stamp i often oxidized.13,17 The oxidized stamp is
elastomeric stamp, having relief structures covered (inked) claimed to have a higher protein adsorption capability13
with molecules, is brought into contact with a solid and to give higher coupling yields when stamping amines17
substrate and then removed, leaving a pattern of the “ink” than the untreated stamp.
in the areas where the stamp was in contact with the A matter of concern, in some situations where µCP is
substrate. The conformal contact between the stamp and used, is the risk of the transfer of molecules or fragments
the surface is ascertained by the interfacial forces and from the stamp material to the patterned surface. In the
flexibility of the stamp material. Depending on the case of PDMS, these fragments could be low-molecular-
intended application, the ink used for printing varies from, weight residues, molecules, molecular fragments, etc. Such
for example, alkanethiols1-3 to proteins,4-6 and the transfer could compromise the end result by contaminating
substrate varies from glass5,6 to metal surfaces.1-4,7 The the surface, changing the wetting and adhesion properties.
most common surfaces are gold, glass/SiO2, and silver. There are surprisingly few studies of these aspects in the
Despite the broad variation of inks and substrates, there literature. A few publications report PDMS contamination
is one common feature: the stamp is made of an elastomer, on the printed substrates,18-20 but these studies were not
usually poly(dimethylsiloxane) (PDMS). aiming at studying the transfer process as such or a
PDMS is a hydrophobic polymer known to be very solution to the problem. In one of these publications,18 a
dynamic in terms of the mobility of the polymer chains at slightly different process than the standard µCP procedure
the surface.8 This mobility enables hydrophobic or more was used (1-6 bar overpressure during the printing of
hydrophilic side groups to be exposed at the surface thiols). PDMS was detected, by X-ray photoelectron
depending on the environment. The surface of PDMS can spectroscopy (XPS) and infrared reflection absorption
be modified by UV/ozone treatment, creating a thin (20-
30-nm) glassy layer on the surface.9 Such thin glasslike (9) Ouyang, M.; Yuan, C.; Muisener, R. J.; Boulares, A.; Koberstein,
J. T. Chem. Mater. 2000, 12, 1591.
(10) Chaudhury, M. K.; Whitesides, G. M. Science 1992, 255, 1230.
* To whom correspondence should be addressed. E-mail: (11) Hillborg, H.; Gedde, U. W. Polymer 1998, 39, 1991.
karing@fy.chalmers.se. (12) Delamarche, E.; Schmid, H.; Bietsch, A.; Larsen, N. B.; Rothuizen,
(1) Kumar, A.; Whitesides, G. M. Appl. Phys. Lett. 1993, 63, 2002. H.; Michel, B.; Biebuyck, H. J. Phys. Chem. B 1998, 102, 3324.
(2) Kumar, A.; Biebuyck, H. A.; Whitesides, G. M. Langmuir 1994, (13) St. John, P. M.; Davis, R.; Cady, N.; Czajka, J.; Batt, C. A.;
10, 1498. Craighead, H. G. Anal. Chem. 1998, 70, 1108.
(3) Biebuyck, H. A.; Larsen, N. B.; Delamarche, E.; Michel, B. IBM (14) Kam, L.; Boxer, S. G. J. Biomed. Mater. Res. 2001, 55, 487.
J. Res. Dev. 1997, 41, 159. (15) James, C. D.; Davis, R. C.; Kam, L.; Craighead, H. G.; Isaacson,
(4) Tan, J. L.; Tien, J.; Chen, C. S. Langmuir 2002, 18, 519. M.; Turner, J. N.; Shain, W. Langmuir 1998, 14, 741.
(5) Wheeler, B. C.; Branch, D. W.; Corey, J. M.; Weyhenmeyer, J. A.; (16) Hovis, J. S.; Boxer, S. G. Langmuir 2001, 17, 3400.
Brewer, G. J. Proceedings of the 19th International Conference-IEEE/ (17) Lahiri, J.; Ostuni, E.; Whitesides, G. M. Langmuir 1999, 15,
EMBS; Chicago, IL, 1997. 2055.
(6) Kung, L. A.; Kam, L.; Hovis, J. S.; Boxer, S. G. Langmuir 2000, (18) Böhm, I.; Lampert, A.; Buck, M.; Eisert, F.; Grunze, M. Appl.
16, 6773. Surf. Sci. 1999, 141, 237.
(7) Xia, Y.; Kim, E.; Whitesides, G. M. J. Electrochem. Soc. 1996, (19) Yang, Z. P.; Belu, A. M.; Liebmann-Vinson, A.; Sugg, H.; Chilkoti,
143, 1070. A. Langmuir 2000, 16, 7482.
(8) Mark, J. E.; Allock, H. R.; West, R. In Inorganic polymers; Prentice (20) Jud, P. P. Semester thesis. ETH Zürich, Zürich, Switzerland,
Hall Inc.: Englewood Cliffs, NJ, 1992, 141. 2001.

10.1021/la026558x CCC: $25.00 © 2003 American Chemical Society


Published on Web 05/20/2003
5476 Langmuir, Vol. 19, No. 13, 2003 Glasmästar et al.

spectroscopy, as a contamination on the printed surface.


In another report, Yang et al.19 printed biotinylated amino-
terminated oligoethyleneglycol on activated poly(ethylene
terephthalate) (PET) surfaces, using stamps treated by
oxygen plasma for 10 min. They found silicone in the
regions where the stamp had been in contact with the
substrate (by time-of-flight secondary ion mass spectrom-
etry, ToF-SIMS). In the third report, nonoxidized stamps
were used to print poly(ethylene glycol)-peptide deriva-
tives with flat stamps on TiO2, where the transfer of an
Si-containing species to the substrate was shown with
XPS.20
To the best of our knowledge, there is only one previous
study in the literature, by Graham et al., that examines
in any detail the issue of PDMS being transferred to the
substrate during the µCP procedure.21 They demonstrate
PDMS contamination on gold by XPS [Si(2s)] and Figure 1. Schematic of the procedure of µCP used in this work.
ToF-SIMS (positive and negative spectra) after the µCP A stamp is cast from a master. The stamp is soaked in ink
of thiols. The study was carried out using flat stamps. (Milli-Q or PBS) for 1 min, rinsed (Milli-Q only or PBS followed
The authors propose a 1-week-long cleaning process for by Milli-Q), and dried in a stream of N2. The stamp is “rolled”
the stamps to lower the amount of transfer to below the over the sample and left in contact for 3 min before it is rolled
detection limit of XPS. This treatment was suggested to off. No external pressure is applied.
remove low-molecular-weight fragments. coatings were produced by first covering the surface with a thin
There are several pending questions regarding the adhesion layer of Ti, followed by the evaporation of 30 or 40 nm
possible transfer of molecules/fragments from PDMS of Au. Uncoated silicon wafer substrates (with thin native oxide
stamps to the substrate. To what extent does the amount layers of ∼1-2 nm) were used as the model SiO2 substrates,
of the transferred contaminant material depend on the hereafter referred to as SiOx. The individual samples were
released from the wafer by breaking along the partially cut lines.
pretreatment of the stamp or material onto which the Before use, all the samples were blown with nitrogen gas to
pattern is stamped? Does the type of ink molecules used remove particulate contaminations, which was followed by
or size of the pattern affect the amount being transferred? exposure to UV/ozone for 5 min to remove hydrocarbon con-
Does the transfer influence the quality of the stamped taminations.
patterns? Such questions, some of them arising during an µCP. Stamps for µCP were formed by curing PDMS (Sylgard
application of µCP in our laboratory, motivated us to make 184, Dow Corning, Brussels, or Rhodorsil RTV 1556 A+B, Sikema
a more detailed study of this issue, focusing on the role AB, Stockholm) against a master (room temperature, >24 h).
of the treatment of the stamp preceding inking and The Sylgard 184 was chosen because it is the most common PDMS
stamping. Questions regarding the influence of the type in the µCP community. The master for the flat stamps was an
Au-coated silicon wafer. Patterned masters (lines or squares)
of ink molecules (e.g., competitive adsorption) were not were made by standard photolithography techniques. The line
addressed in this work. patterns were 5-µm-wide lines (in contact during stamping)
We used XPS and ToF-SIMS to measure the amount separated by 15-µm-wide spaces. The square patterns were a
and character of PDMS transfer from flat and patterned grid of 10-µm-wide, elevated lines. The squares between the lines
stamps to substrate surfaces of gold, TiO2, and SiOx. The were 10 by 10 µm and were not in contact with the surface during
µCP process was performed using water or buffer as the stamping.
ink and without external pressure. We explored whether The µCP procedure used is illustrated in Figure 1. The stamps
a UV/ozone pretreatment of the PDMS stamp could reduce were used either as prepared or after UV/ozone treatment for 10
min. Because the prime interest in this work was to investigate
the amount of transferred material. The choice of SiOx as
the possible transfer of material from the PDMS stamp during
one of the substrates was motivated by the many ap- the patterning of water-soluble molecules, such as proteins, the
plications of µCP on glass substrates. TiO2 and gold ink chosen should be representative of the bulk solutions used
surfaces were chosen for two reasons. First, there was a in such µCP processes. The two model “inks” were Milli-Q water
desire to have more than one surface (chemistry) for and phosphate buffer saline (PBS, Sigma-Aldrich).
comparative purposes, and second, XPS detection of Si- XPS Analysis. All the samples for the XPS analysis were
containing molecules (transferred from PDMS) is much mounted in the vacuum chamber within 20 min after preparation
easier on a surface not containing Si as an element. (i.e., after stamping or surface pretreatment). All the XPS spectra
were recorded using monochromatic Al KR radiation, with a take-
off angle of 45° from the sample surface normal (PHI 5500C
Materials and Methods MultiTechnique system). The survey spectra were recorded at
Surface Preparation. All substrates for µCP were prepared a pass energy of 187.85 eV, and detailed spectra of the individual
from (100)-oriented polished silicon wafers (Siltronix, Archamps, elements were recorded at 5.85 eV. The survey spectra from two
France). Wafers protected by a sacrificial poly(methyl meth- spots on each sample were used for the calculation of their atomic
acrylate) (PMMA) layer were partially cut using a diamond saw composition at the surface, and one detailed spectrum was
(Loadpoint Microace 3+). PMMA was then removed by sequential recorded to obtain chemical-shift information (at least two
ultrasonic agitation in acetone, 2-propanol, and Milli-Q water. samples). The X-ray-beam spot diameter was about 1 mm.
Thin films of TiO2 and Au were deposited on some of the wafers In one experiment, flat stamps (untreated or UV/ozone
by electron-beam-induced thermal evaporation (AVAC HVC-600). pretreated) were covered with a thin Al foil (approximately 13-
The TiO2 surfaces were prepared by first evaporating a 40-nm- µm thick), which had a square (5 × 5 mm) hole. A gold sample
thick film, which was then oxidized to saturation by active oxygen was positioned on top of the foil for 3 min, ensuring noncontact
treatment (O2 RF plasma: 0.5 Torr; 200 W; 120 s).22 The gold between the stamp and the surface but allowing possible vapors
from the PDMS stamp to reach the gold surface. This experiment
(21) Graham, D. J.; Price, D. D.; Ratner, B. D. Langmuir 2002, 18,
was performed to control any transfer of molecules from the
1518. PDMS stamp to the substrate via the gas phase, that is, from
(22) Aronsson, B. O.; Lausmaa, J.; Kasemo, B. J. Biomed. Mater. the regions of the stamp that are not in direct contact with the
Res. 1997, 35, 49. substrate.
Silicone Transfer Langmuir, Vol. 19, No. 13, 2003 5477

Table 1. Atomic Composition by XPS Survey Spectra of


PDMS Stampsa
PDMS sample C(1s) O(1s) Si(2p)
untreated 42.8 ( 0.3 31.4 ( 1.0 25.8 ( 0.7
treated 36.4 ( 0.1 38.5 ( 0.5 25.1 ( 0.6
a Mean values are noted in percent plus or minus the standard

deviation from two data points per sample type. PDMS was Sylgard
184 elastomer, untreated or UV/ozone-treated (10 min).

nitrobenz-2-oxa-1,3-diazol-4-yl)amino]dodecanoyl-1-hexadecanoyl-
sn-glycero-3-phosphocholine (NBD-HPC; Molecular Probes,
U.S.A.). Vesicles produced by this method have an average
diameter of 40-60 nm (by dynamic light scattering). Glass
coverslips (Merck) were cleaned by sequential ultrasonic agitation
in acetone, 2-propanol, and Milli-Q water, dried, and UV/ozone-
treated for 10 min. The cleaned coverslips were then stamped
and subsequently incubated in a solution of vesicles (30 µL in 1
mL of buffer) for 10 min, rinsed with buffer, and imaged using
an Axioplan 2 Imaging epifluorescence microscope (Zeiss).
Figure 2. High-resolution XPS Si(2p) spectra from PDMS Results
stamps (Sylgard 184 elastomer), untreated or UV/ozone-treated.
The curves are displaced along the y axis for clarity and are XPS Analyses of Stamps. To be able to trace the
normalized with respect to the total signal. The spectra were contaminations coming from PDMS used in this study,
recorded using an electron gun to compensate for the charging XPS analyses of untreated and UV/ozone-treated flat
and were later shifted to C(1s) ) 284.4 eV for a charging PDMS stamps were performed (Sylgard 184 elastomer).
reference. The arrow points at a shoulder, indicating the
presence of more oxidized species. Figure 2 shows the high-resolution XPS spectra of the
Si(2p) region for untreated and UV/ozone-treated PDMS.
The XPS analysis of PDMS (Sylgard 184 elastomer) stamps For both untreated and UV/ozone-treated PDMS, the Si-
were performed using equipment and settings the same as those (2p) peak is asymmetric. This is much more pronounced
described previously, with the addition of an electron flood gun for the untreated stamp. The most dominant peak at 101.9
to compensate for the charging. The electron gun was adjusted eV of the untreated PDMS corresponds well to the
so that the O(1s) signal of PDMS was at about 532 eV. A result reference spectra for PDMS (101.8 eV).25 The shoulder at
of the charging neutralization process was that the total intensity a higher BE, at the arrow in Figure 2, indicates the
varied between the samples. Therefore, we chose to normalize presence of more oxidized species. After UV/ozone treat-
the curves of Figure 2 with respect to the total signal.
The spectra shown in Figures 2 and 3 were corrected for
ment, the peak at 101.9 eV is shifted toward a higher BE
charging effects by shifting the peaks as follows: PDMS spectra (102.5 eV), which is in agreement with the literature.9,26,27
were shifted so that the binding energy (BE) for C(1s) was 284.4 However, the peak is not shifted completely to the SiO2
eV; spectra from the Au samples were shifted to Au(4f7/2) ) 84.0 value (103.3 eV),28 indicating that the surfaces of the UV/
eV; spectra from the SiOx samples were shifted to Si(2p3/2) ) 99.3 ozone-treated stamps have a surface chemistry somewhere
eV; and spectra from the TiO2 samples were shifted to Ti(2p3/2) between PDMS and SiO2. The atomic composition of the
) 458.8 eV. stamp surface (Table 1) shows a decrease in carbon and
ToF-SIMS Analysis. The ToF-SIMS analyses were performed an increase in oxygen content upon UV/ozone treatment.
using a ToF-SIMS IV (ION-TOF GmbH) system in static mode,23 XPS Analyses Using Flat Stamps. The first set of
that is, the ion dose was below 1013 ions/cm2. Gallium ions (25
stamping experiments was made with untreated and UV/
kV; 1 pA) were used as the primary beam. Both high mass
resolution spectra (m/∆m ) 7000) and high lateral resolution ozone-pretreated stamps, using Milli-Q water and buffer
images (<1 µm) were recorded. Flat stamps and stamps with solutions, to stamp SiOx, TiO2, and gold surfaces. Parts
line patterns, with and without UV/ozone pretreatment, were A-F of Figure 3 present, for each surface type, the XPS
used to stamp Milli-Q water on Au substrates. The analyses spectra taken before stamping, after stamping with
started at about 80 min after preparation. untreated stamps, and after stamping with UV/ozone
Light Microscopy. Dark-field light microscopy was used to pretreated stamps, using Milli-Q water as the ink. The
inspect certain aspects of the quality of the stamped patterns. main observations are that (i) the transfer of material
The microscope used was a Zeiss Axiotech microscope equipped occurs from untreated stamps on all the surface types
with a digital camera. Water condensation patterns were made
examined and (ii) a pretreatment of the stamp with UV/
by placing the sample on an ice-cooled piece of copper, under the
microscope objective, in an environment of humid air and by ozone lowers the amount of PDMS (Sylgard 184, Dow
letting water condense on the sample surface. Corning) transferred from flat stamps to substrates of
Another useful control system, sensitive to the chemical state Au, SiOx, and TiO2.
of the surface, that has been extensively studied in our group, The XPS analysis of PDMS transfer to the SiOx substrate
is the formation of supported phospholipid bilayers by vesicle is complicated by the fact that the substrate itself has a
adsorption and rupture (see, e.g., ref 24). A lipid bilayer will strong Si signal. The transfer from untreated PDMS
form on a clean glass surface if it is exposed to lipid vesicles. stamps is, however, still detectable by the Si(2p) signal
Fluorescently labeled unilamellar lipid vesicles were made by in the XPS spectra (Figure 3A), although a quantification
extrusion (11 × 100 nm followed by 11 × 30 nm polycarbonate is not possible without extensive (and often unreliable)
membranes, Whatman, U.S.A.) of a 5 mg/mL lipid mixture in a
Tris buffer of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine
(25) Beamson, G.; Briggs, D. High resolution XPS of organic polymers;
(POPC; Avanti Polar Lipids, AL) and 0.5% (w/w) of 2-[12-(7- John Wiley & Sons: Chichester, U.K., 1992.
(26) Bar, G.; Delineau, L.; Hafele, A.; Whangbo, M. H. Polymer 2001,
(23) Vickerman, J. C., Briggs, D., Eds.; ToF-SIMS: Surface analysis 42, 3627.
by mass spectrometry; IM Publications and SurfaceSpectra Limited: (27) Murakami, T.; Kuroda, S.; Osawa, Z. J. Colloid Interface Sci.
Huddersfield, U.K., 2001. 1998, 202, 37.
(24) Keller, C. A.; Glasmästar, K.; Zhdanov, V. P.; Kasemo, B. Phys. (28) Chastain, J., Ed.; Handbook of X-ray Photoelectron Spectroscopy;
Rev. Lett. 2000, 84, 5443. Perkin-Elmer Corporation: Eden Prairie, MN, 1992.
5478 Langmuir, Vol. 19, No. 13, 2003 Glasmästar et al.

Figure 3. High-resolution XPS spectra from samples not stamped or stamped with untreated or UV/ozone-treated flat stamps
(Sylgard 184 elastomer) using Milli-Q water as the ink. All the spectra from the SiOx samples were shifted to a BE for Si(2p3/2)
of 99.3 eV, the TiO2 samples to a BE for Ti(2p3/2) of 458.8 eV, and the Au samples to a BE for Au(4f7/2) of 84.0 eV. For clarity, the
curves are displaced on the y axis. (A) Si(2p) spectra of SiOx samples. (B) C(1s) spectra of SiOx samples. (C) Si(2p) spectra of TiO2
samples. (D) O(1s) spectra of TiO2 samples. (E) Si(2s) spectra of Au samples. (F) O(1s) spectra of Au samples.

curve fitting. Inorganic silica (SiO2) has a peak at 103.3 BE (∼102 eV), which is not observed on the samples where
eV,28 which is clearly visible on all the SiOx samples. UV/ozone-treated stamps were used. The XPS spectra from
However, for the samples that were stamped with an the untreated PDMS stamps themselves (Figure 2) show
untreated stamp, this peak shows a shoulder at a lower a nice correlation with this shoulder observed at 102 eV
Silicone Transfer Langmuir, Vol. 19, No. 13, 2003 5479

Table 2. Atomic Composition by XPS Survey Spectra of the Samples Stamped with Flat PDMS Stampsa
substrateb stamp C(1s) O(1s) Si(2p/2s)c Au(4f) Ti(2p)
Au untreated stamp 29.3 ( 5.7 22.3 ( 2.4 6.2 ( 1.6 41.5 ( 3.4
treated stamp 34.5 (5.5 14.0 ( 2.0 0.2 ( 0.4 50.3 ( 4.9
not stamped 32.3 ( 8.8 16.2 ( 3.8 0 48.8 ( 4.8
TiO2 untreated stamp 24.3 ( 4.8 50.7 ( 3.7 7.4 ( 3.4 15.6 ( 3.4
treated stamp 15.0 ( 0.9 58.1 ( 0.6 3.6 (0.5 20.7 ( 1.1
not stamped 21.7 ( 7.4 55.5 ( 4.3 1.5 ( 0.1 19.1 ( 2.9
SiOx untreated stamp 16.9 ( 2.1 38.8 ( 0.7 44.3 ( 1.4
treated stamp 13.9 ( 2.0 40.3 ( 0.5 45.6 ( 1.4
not stamped 12.9 ( 0.6 40.9 ( 1.1 46.1 (0.7
a Mean values are noted in percent plus or minus the standard deviation of at least four data points per sample type. b For some of the

Au samples, a contamination of Cu was found (max 4%), probably a result of a contamination of the evaporation source. All the Ti samples
had fluorine contamination (max 2%) from the oxygen plasma equipment. c The Si(2p) peak was used for all but the Au substrates, where
the Si(2s) peak was used. The ink used was Milli-Q water; the contact time was 3 min. The stamps were made of Sylgard 184 elastomer,
untreated or UV/ozone-treated (10 min).

Table 3. Atomic Composition by XPS Survey Spectra of


on the samples stamped with untreated stamps. Hence, the Au Samples Stamped with Patterned Stampsa
PDMS material is transferred from untreated stamps to
SiOx substrates during µCP. stamp C(1s) O(1s) Si(2s) Au(4f)
Figure 3B shows the C(1s) spectra for the different SiOx untreated stamp 30.5 ( 2.2 23.2 ( 1.8 10.0 ( 1.2 35.9 ( 1.4
samples. The peak at 285.5 eV, most likely arising from UV/O3-treated stamp 27.0 ( 1.9 19.3 ( 1.5 3.1 ( 2.2 49.8 ( 1.8
ambient hydrocarbons, is clearly shifted to a lower BE a Mean values are noted in percent plus or minus the standard

(285.1 eV) for the samples stamped with untreated stamps deviation. To be compared to when flat stamps were used, see Table
as compared to that of the samples that were not stamped 2. Patterned stamps were 5-µm-wide lines separated by 15-µm
or stamped with UV/ozone-treated stamps. This is prob- spaces. The ink used was Milli-Q water; the contact time was 3
ably a mixed peak of atmospheric hydrocarbons and carbon min. The stamps were made of Sylgard 184 elastomer, untreated
or UV/ozone-treated (10 min).
from PDMS, the latter having a peak at 284.4 eV.25
For the stamped TiO2 samples, the results were similar
to the ones described for the SiOx samples previously. An deposited on the surface as a result of the ozone treatment.
intensity change of the Si(2p) peak (Figure 3C) was (Our BE values are slightly higher than those reported
observed on the samples stamped with untreated stamps. by Krozer and Rodahl.) The O(1s) signal from the samples
The Si(2p) peak (Figure 3C) shows that some Si-containing stamped with untreated PDMS results in a dominant peak
species are also present on the sample that was not at a higher BE (531.8 eV), which can be associated with
stamped. The latter probably originates from wafer dust the presence of PDMS (532 eV).25
(remember that Ti was deposited on a Si wafer) produced When the observations in Figure 3A-F are summarized,
when snapping the chips or maybe from the reactive ion the most important result is that the UV/ozone-treated
etcher used to oxidize the TiO2 samples. The atomic stamps transfer less PDMS-derived material than the
compositions, as were calculated from the survey spectra untreated ones. This result is further supported by the
(Table 2), show a significantly higher Si content for the following description of the ToF-SIMS results. If we
stamped samples than that for the not stamped ones. compare the different surface types, we find that the UV/
Furthermore, the UV/ozone treatment of the stamp seems ozone pretreatment is more efficient in lowering the
to reduce the transfer of Si-containing species to the TiO2 transfer to gold surfaces than that to TiO2 surfaces (Table
surface. 2). Unfortunately, it is hard to perform a quantification
of the transfer to SiOx surfaces because the silicon peaks
The “normal” shape of the O(1s) peak for TiO2 surfaces
from PDMS and the substrate overlap.
is observed for the sample that was not stamped (Figure
That PDMS sticks less to Au than to TiO2 substrates
3D). The tail at a higher BE is assigned to the organic
is actually supported by observations from the casting
compounds bound to the surface.29 An additional peak at
process: PDMS cured on masters of plain SiOx or TiO2-
532.3 eV is found for the samples stamped with the
coated wafers was extremely hard to release. In contrast,
untreated stamps, which corresponds to the O(1s) peak
when Au-coated wafers were used as masters, PDMS was
for PDMS (532.4 eV, not shown). Also, the fact that the
quite easily released after the curing. Hence, it appears
intensity of the TiO2 oxygen peak is reduced when
that the direct chemical interaction between PDMS and
stamping with an untreated stamp supports the conclusion
the substrate is stronger for the two oxide surfaces
of a transferred adlayer.
compared to that of gold.
For the Au samples, the Si(2p) signal cannot be used
All the data presented here are from the µCP of Milli-Q
to identify the PDMS transfer because it is hidden in the
water using the Sylgard 184 elastomer. Trends the same
large background of the Au(4f) peak. However, the Si(2s)
as those described previously were found when stamping
signal (Figure 3E and Table 2) clearly shows that
PBS and also when using another brand of PDMS that is
Si-containing species are transferred to the gold surface
less commonly used for µCP: Rhodorsil RTV 1556 (data
from untreated stamps whereas the UV/ozone-treated
not shown).
stamps transfer only trace amounts (if any). A shape
XPS Analyses Using Patterned Stamps. Types of
change and shift of the O(1s) signal were also observed
experiments the same as those described previously were
(Figure 3F). For not-stamped, UV/ozone-treated Au, a
performed using the gold samples and patterned stamps
double peak is observed in the BE range of 529-531 eV.
having 5-µm-wide lines separated by 15-µm-wide spaces
According to the literature,30 these peaks correspond to
(depth 10 µm). The elemental concentration from the XPS
chemisorbed oxygen (529.1 eV) and gold oxide (530.1 eV)
survey spectra of gold surfaces stamped with patterned
(29) Lausmaa, J.; Kasemo, B.; Mattsson, H. Appl. Surf. Sci. 1990,
stamps are shown in Table 3. These stamps seem to
44, 133. transfer more Si-containing species to gold substrates than
(30) Krozer, A.; Rodahl, M. J. Vac. Sci. Technol., A 1997, 15, 1704. the flat stamps do. This conclusion is drawn from the
5480 Langmuir, Vol. 19, No. 13, 2003 Glasmästar et al.

Figure 4. ToF-SIMS images for selected fragments characteristic of PDMS. The chemical formula of each fragment is marked
under the respective image. (A) A gold surface stamped with an untreated PDMS stamp (Sylgard 184; Milli-Q water as the ink).
The stamp was in contact with the gold surface at 5-µm-wide lines separated by 15-µm-wide spaces. The scan size was 101 × 101
µm2. (B) Same as part A except that the stamp in this case was UV/ozone-treated for 10 min before stamping. The scan size was
101 × 101 µm2.

Table 4. ToF-SIMS Signals from the Au Samples gold samples. The Au (m/z ) 197+) signal is used as an
Stamped with Flat or Patterned Stamps intensity reference, and we focus on the signals charac-
flat stamp patterned stamp teristic for PDMS in the low-mass-range spectra,31 that
UV/O3 UV/O3
is, m/z ) 73+, 147+, 207+, 221+, and 281+. The contact
signal untreated treated untreated treated area between the stamp and the gold surface consists of
5-µm-wide, elevated lines, separated by 15-µm-wide
gold/total 249 814 89 487 spaces. A Au substrate stamped with an untreated stamp
ion count
and analyzed by ToF-SIMS in image mode shows the
a The numbers given are averages over the whole image (101 ×
expected pattern (Figure 4A). Most of the silicone transfer
101 µm). The lower the gold signal, the more adsorbates on the has occurred in the regions where the stamp was in contact
surface. For comparison, the gold/total ion count signal is 917 for with the surface during stamping (i.e., the 5-µm-wide lines
a sample that was not stamped. Patterned stamps were 5-µm-wide
lines separated by 15-µm spaces. The ink used was Milli-Q water; are the brightest). However, when the stamp was pre-
the contact time was 3 min. The stamps were made of Sylgard 184 treated with UV/ozone, the ToF-SIMS image shows more
elastomer, untreated or UV/ozone-treated (10 min). intensity from the PDMS-related fragments (relative to
the Au signal) in the areas of noncontact (see Figure 4B),
comparison of the data in Table 3 with the corresponding which is surprising.
values for flat stamps on gold in Table 2. Apart from the The surface topography influences the probability of
peak intensities, no qualitative differences (such as ionization of the surface molecules when imaging with
differences in the relative peak heights or peak shapes) ToF-SIMS.23 To rule out the possibility of such effects as
were observed between the spectra from surfaces stamped the cause of the unexpected ToF-SIMS images obtained
with flat or patterned stamps. with UV/ozone-treated stamps, tapping mode atomic force
There is a possibility that silicone monomers/fragments microscopy was used to image a gold sample stamped with
were transferred through the vapor phase in the regions patterned, untreated PDMS stamps (lines). No lines were
where the patterned stamp was not in contact with the detected in the height image (0-10 nm), but lines of
surface during stamping. To test this hypothesis, the XPS dimensions the same as those of the contact area were
analyses were performed for the gold samples exposed to clearly seen in the phase contrast image (not shown). The
vapors from the untreated or UV/ozone-treated flat stamps latter observation suggests that there is a chemical pattern
(see Materials and Methods for details). No silicon was on the surface.32
detected in either case [Si(2s); pass energy 5.85 eV; The most important conclusions from the ToF-SIMS
acquisition time 17 min over 10 eV]. analysis are: (i) It supports the conclusion from the XPS
ToF-SIMS Analyses Using Flat Stamps. The ToF- results that UV/ozone pretreatment of the stamp is
SIMS analyses of the gold samples, stamped with flat efficient in lowering the PDMS transfer from the stamp
stamps, showed a trend the same as that in the XPS to the substrate. (ii) When PDMS transfer occurs from
analyses: less PDMS is transferred if the stamp is UV/ozone-treated stamps, transferred PDMS is not located
pretreated with UV/ozone than if it is untreated. The only in the areas where the stamp and substrate were in
intensity of the gold peak (m/z ) 197+ normalized to the direct contact. (iii) There is more transfer from patterned
total ion intensity) is lower for the samples stamped with than from flat stamps.
untreated stamps than those stamped with UV/ozone- These samples were also investigated by dark-field
treated stamps; see Table 4. This suggests that more microscopy. We found that on both the TiO2 and the SiOx
material is transferred to the sample from an untreated surfaces, PDMS transferred from patterned, untreated
stamp than from a UV/ozone-treated stamp, that is, the stamps could be seen as brighter lines with dark-field
lower the gold intensity, the thicker the layer of adsorbed microscopy. Nothing could be seen when the stamp was
molecules. There is also a lower gold signal after stamping UV/ozone-treated before stamping or on the Au surfaces
with a patterned stamp compared to that of a flat one, that were stamped with untreated stamps.
indicating that more material is transferred from a
patterned stamp than from a flat. (31) Dong, X.; Proctor, A.; Hercules, D. M. Macromolecules 1997, 30,
63.
ToF-SIMS Analyses Using Patterned Stamps. Pat- (32) Pang, G. K. H.; Baba-Kishi, K. Z.; Patel, A. Ultramicroscopy
terned stamps were used to stamp Milli-Q water on the 2000, 81, 35.
Silicone Transfer Langmuir, Vol. 19, No. 13, 2003 5481

Discussion
µCP Using Flat Stamps. There is clear evidence from
the present study for a transferred overlayer of molecules
derived from PDMS, on substrates subjected to µCP with
untreated PDMS stamps (both flat and patterned), when
buffer or Milli-Q water is used as the ink. This conclusion
is based on the XPS and ToF-SIMS data of the present
study.
Surprisingly few studies in the literature have aimed
at elucidating the problem of transferral of stamp material.
Graham et al.21 recently published the only (to our
knowledge) more detailed study on this issue. They
compared the self-assembled monolayers on gold made
by either adsorption from ethanolic solution or µCP with
flat stamps. They found that PDMS is transferred to the
surface by the µCP process (using XPS, ToF-SIMS, and
mulitvariate analyses). The authors present an extensive
cleaning protocol for the PDMS stamps, including extrac-
tion in hexane and sonication in an ethanol/water mixture
to avoid the transfer. The cleaning protocol takes about
Figure 5. (A) Dark-field light microscopy (20× objective) of a 1 week to perform according to Graham et al.21 Addition-
gold sample stamped with a UV/ozone-treated stamp (Sylgard ally, they blotted the stamp on a cleaned piece of silicon
184; Milli-Q water as the ink) having a 10 by 10 µm grid pattern
(the squares were not in contact). Water was subsequently
before stamping and claim that this was an effective way
allowed to condense on the surface, producing this pattern. (B) to remove residual silicone that had migrated to the surface
Fluorescence microscopy image of NBD-labeled phospholipid during cleaning. By using this cleaning protocol, they were
vesicles backfilled onto a glass surface “pre-stamped” with an able to lower the PDMS transfer from flat stamps during
untreated PDMS stamp (Sylgard 184; Milli-Q water as the ink) printing in ethanol to levels below that of XPS detection,
of a design the same as that in part A. The upper left corner although PDMS was still found by ToF-SIMS analysis.
shows a black area where the sample dried during the
experiment and, thus, is free of lipid. Comparing our data for the samples stamped with
untreated, flat stamps with the corresponding data for
Condensation Figures and Bilayer Formation. To samples stamped with UV/ozone-treated stamps (Figure
test for differences in chemistry between the underlying 3A-F and Table 2), we find that 10 min of UV/ozone
substrate and any transferred material, water from humid treatment reduces the transfer significantly. We, thus,
air was allowed to condense on the gold samples stamped show that the simple and fast UV/ozone treatment of the
with patterned, UV/ozone-treated or untreated stamps stamp, for the present conditions, significantly reduces
(with Milli-Q water as the “ink”). The pattern of the stamp the amount of PDMS transferred during µCP to levels at
was a square grid of 10-µm-wide, elevated lines. The or below that of XPS detection. The same trends were
squares between the lines were 10 by 10 µm and were not found for a second brand of PDMS elastomer: Rhodorsil
in contact with the surface during stamping. Upon RTV 1556 (only XPS data).
condensation, clear patterns were formed. The details of µCP Using Patterned Stamps. The combined results
the patterns varied from run to run because the conden- from XPS and ToF-SIMS indicate that PDMS is trans-
sation is very sensitive to minor variations in the ferred both in the noncontact and in the contact regions
condensation conditions, primarily the humidity level and between the untreated stamp and the gold substrate
temperature. One condensation pattern is shown in Figure (Tables 2-4). This is despite only one-fourth of the stamped
5A for a UV/ozone-treated stamp. A clear pattern of the surface being in contact with the patterned stamp, which
expected dimensions was observed, demonstrating a intuitively would suggest less transfer. The XPS values
contrast in hydrophobicity/hydrophilicity between the are mean values over the analyzed area and, thus, a mean
contact/noncontact areas. The patterns were also seen with over the areas that were in contact with the stamp and
untreated stamps (not shown). the areas of noncontact. (These experiments were only
In another “application example”, fluorescently labeled, performed on the gold samples.) It is likely that the
extruded unilamellar vesicles of POPC were adsorbed on mechanism of transfer of PDMS is different in the areas
glass surfaces “pre-stamped”, as in the condensation of contact versus the areas of noncontact.
experiment previously described, using the same types of As for flat stamps, the transfer of PDMS from patterned
stamps, that is, untreated or UV/ozone-treated. It is known stamps is reduced by a UV/ozone pretreatment of the
that lipid bilayers are formed by the adsorption and stamp. However, the overall reduction of PDMS transfer
subsequent rupture of vesicles on clean glass surfaces24 is smaller for patterned stamps compared to flat stamps
and that they can be studied using fluorescence imaging (Tables 2-3). The distribution of the transferred PDMS
and fluorescence recovery after photobleaching.33 The first on the stamped surface, which for UV/ozone-treated
test results show that the fluorescent patterns formed on stamps is predominantly detected in the regions of
the surface when the stamp was both untreated (Figure noncontact, may indicate that the majority of the material
5B) and UV/ozone-treated (not shown). Homogeneous is not transferred across the contact interface. Some
fluorescence indicates that the lipid film on the surface possible mechanisms for the transfer of PDMS onto the
is continuous and of the same thickness everywhere. On portions of the surface not in contact with the stamp are
the basis of the intensity variations, one could attribute shown schematically in Figure 6. One mechanism that
the bright areas to bilayers and the less bright areas to may involve volatile PDMS fragments from the “ceiling”
monolayers. above the areas of noncontact being transported across
the vapor phase during the stamping process (option 1 in
(33) Boxer, S. G. Curr. Opin. Chem. Biol. 2000, 4, 704. Figure 6) seems to be ruled out by the XPS experiment
5482 Langmuir, Vol. 19, No. 13, 2003 Glasmästar et al.

resembles ours, but the fact that we get a higher ToF-


SIMS intensity on the noncontact areas seems to be a
contradiction. However, the differences in the substrate
and printing conditions (contact time, solvent, etc.) might
explain why we get a different result. It has recently been
shown that the hydrophilicity of the stamp and surface
are important for the transfer of proteins from the stamp
to the surface (mixed thiols).4 It is also possible that
including molecules in the “ink” significantly changes the
Figure 6. Schematic illustration of the possible mechanisms situation. Maybe the molecules cover the walls or the
of transport of PDMS onto the noncontact areas. (1) Volatile “ceiling” of the spaces of the features on the patterned
fragments from the “ceiling” (contradicted by the XPS control stamps, preventing PDMS fragments to come out in the
experiment). (2) Diffusion from the contact areas (should be space of noncontact. Another possibility is that oxygen
seen as fuzzy edges in ToF-SIMS images for both untreated plasma is more efficient than UV/ozone (10 min) in
and UV/ozone-treated but is not). (3) Transfer from the side
walls caused by the mechanical stresses during printing. PDMS
producing a continuous and protective SiOx coating on
material diffuses (or flows) during the printing process along the stamp surface. On the other hand, oxygen plasma of
the wall and out in the space between the contact areas. PDMS is said to produce low-molecular-weight frag-
ments27 on the surface, and it is possible that the UV/
of using a flat stamp, distanced from the gold surface by ozone-treated PDMS has a different surface composition.
a thin Al foil (∼13-µm thick), which showed that this However, the XPS results of SiOx films formed from UV/
transfer mechanism is negligible (no Si detected by XPS). ozone treatment9 or oxygen plasma treatment27 are very
Another option is lateral diffusion, or even flow during similar.
the µCP process, of the fragments from the contact area Influence of Material Properties on the Silicone
out into the space between the protruding features of the Transfer. The UV/ozone treatment reduces the PDMS
stamp (option 2 in Figure 6). Such flow/diffusion should, transfer. The likely mechanism is that the reactive oxygen
however, make the images of ToF-SIMS have “fuzzy” edges radicals formed by the UV light have two effects on the
of the lines, which is not the case (Figure 4). The third surface of the stamp: (i) they oxidize the hydrocarbon
and, as we see it, most realistic mechanism could involve groups that could contaminate the stamped surface and
the escape of the PDMS-related fragments from the side (ii) they reduce the hydrogen and carbon content in the
walls of the structures during the printing process and surface region and cause a larger number of Si-O bonds.
their transport down into the area of noncontact (option This glassy, SiOx-like surface is likely to reduce the PDMS
3 in Figure 6). This escape of material may be caused by transfer to the stamped substrate.
the increased forces (mechanical or capillary forces) at Another factor that might influence the silicone transfer
the edges of the structures, as compared to the case of flat to the substrate is the way PDMS is cured. There are
stamps. Especially in the case of UV/ozone-treated stamps, many different curing conditions for the Sylgard 184
this mechanism seems likely. The thin SiOx layer formed elastomer reported in the µCP literature, ranging from
on the surface by UV/ozone treatment is quite brittle, and room temperature to 80 °C for various lengths of time. We
cracks are likely to form in the layer upon mechanical chose to cure PDMS in room temperature for at least 24
stress. Crack formation in thin SiOx films on PET has h. Because the curing conditions influence some material
been noted when they are subjected to tensile strain, and properties of the stamp, it is possible that the amount of
the number of cracks increases with an increased applied transferred material during stamping is also influenced.
strain.34 The blotting of the stamps performed by Graham However, among the three previous studies similar to the
et al.21 could be a good complement to the UV/ozone present study, one does not mention the temperature
treatment of the stamps. (overnight in a vacuum desiccator),21 and the other two
One possible cause of the apparently inversed TOF- used elevated temperatures (1-2 h a 60 °C19 or 70 °C for
SIMS images (Figure 4) is the so-called enhanced mono- at least 72 h).20 In all three cases, the transfer of silicone-
layer effect, or matrix effects,23,35 which under some related material was detected.
circumstances can give significantly higher ion intensities What we have presented here is the worst case scenario
at the thickness of around one monolayer than that for of silicone transfer during µCP. It is possible that, for
thicker films. This means that brighter areas of the image example, protein molecules from the ink could function
can actually have a lower coverage than darker areas. as a physical barrier to the adsorption of the silicone
Based on this phenomenon, we suggest that Figure 4B is fragments on the substrate. Furthermore, even if the
actually showing the correct contrast; that is, there is PDMS residues do adsorb, they might be displaced by the
more PDMS in the noncontact areas of a sample stamped proteins by competitive adsorption. Most likely, the
with a UV/ozone-treated stamp. transfer is system-dependent and smaller biomolecules
A recent article of Yang et al.19 mentions the transfer in the ink, such as biotin or poly(lysine) derivatives, do
of silicone in the areas of contact between the stamp and not prevent the silicone transfer.19,20
the substrate during µCP and shows a ToF-SIMS image
where there is a higher ion intensity in the contact areas. Conclusions
The substrate for µCP was in this case an activated PET
surface, the stamps were oxygen plasma-treated PDMS, PDMS material is transferred from the stamp to the
and the “ink” was a biotin-amine in ethanol (contact time stamped surface during µCP with flat and patterned
10 min). When it is assumed that UV/ozone treatment stamps (with water or buffer as the ink). UV/ozone
and oxygen plasma treatment of PDMS gives a similar treatment of the PDMS stamp lowers the transfer
surface chemistry of the stamps, the study of Yang et al. significantly upon stamping on SiOx, TiO2, and Au
substrates. More PDMS is transferred from patterned
(34) Hsueh, C. H. J. Am. Ceram. Soc. 2001, 84, 2955.
stamps than from flat stamps. The water condensation
(35) Benninghoven, A.; Lange, W.; Jirikowsky, M.; Holtkamp, D. Surf. pattern and vesicle adsorption experiments (Figure 5A,B)
Sci. Lett. 1982, 123, L721. clearly show that molecules from PDMS affect the stamped
Silicone Transfer Langmuir, Vol. 19, No. 13, 2003 5483

pattern for both treated and untreated, patterned stamps. in Borås, Sweden, for running the ToF-SIMS and for
The transfer of PDMS to the surface during µCP is relevant fruitful discussions about the results. We also thank Sofia
to and calls for attention by those who are using this Svedhem for making the fluorescently labeled vesicles
method in applications where control of the surface and Sarunas Petronis for producing the masters. Financial
chemistry is of importance for the application. support for this study was provided by the Swedish
Foundation for Strategic Research through the Screening
of Tissue Integrated Materials project.
Acknowledgment. The authors thank Peter Sjövall
at the SP Swedish National Testing and Research Institute LA026558X

You might also like