Soft Matter

PAPER

Publ

ishe

d on

27

June

201

3. D

ownl

oade

d by

UN

IVE

RSI

TA

T G

IESS

EN

on

29/1

0/20

14 1

6:01

:03.

View Article OnlineView Journal | View Issue

aDepartment of Materials Science and Engi

Raleigh, NC 27606, USA. E-mail: mdcasper

+1 919 513 4660bDepartment of Chemical and Biomolecu

University, Raleigh, NC 27606, USA

Cite this: Soft Matter, 2013, 9, 7797

Received 8th April 2013Accepted 26th June 2013

DOI: 10.1039/c3sm50966d

www.rsc.org/softmatter

This journal is ª The Royal Society of

Surface wrinkling by chemical modification ofpoly(dimethylsiloxane)-based networks duringsputtering

Michelle D. Casper,*a Arif O. Gozen,a Michael D. Dickey,b Jan Genzerb

and Jon-Paul Mariaa

Wrinkling is an important mechanical phenomenon that generates periodic topographical patterns across

a surface. This paper presents experimental evidence that surface wrinkles, which form consequent to thin

film magnetron sputtering of either indium tin oxide (ITO) or aluminum on poly(dimethylsiloxane)

networks (PDMS-N) made from a commercial Sylgard-184 kit, result from chemical modification of the

PDMS-N surface as opposed to extrinsic thermomechanical stresses originating from differential thermal

expansion. X-ray photoelectron spectroscopy results reveal that the PDMS-N surface becomes depleted

in carbon and concurrently enriched in oxygen relative to silicon due to sputtering. This silica-like

surface layer possesses intrinsic compressive stress that leads to wrinkle formation during the first

z5 seconds of sputtering. The wrinkles maintain their periodicity irrespective of the thickness of the ITO

film formed during subsequent deposition. Furthermore, upon removal of the ITO layer, the wrinkles

persist with their periodicity unchanged. A narrow sputtering pressure window between 2 and

12 mTorr generates wrinkles. Pressures below this range cannot sustain a radio frequency plasma, while

pressures above this range provide sufficient thermalization of kinetic energy as to eliminate the

energetic bombardment that modifies the PDMS-N. This study provides a new understanding of the

origins of wrinkling in sputtered films on polymeric substrates and creates opportunities to manipulate

the topography produced by spontaneous surface wrinkling.

Introduction

Electronic and photonic materials rely routinely on topo-graphically patterned micro- and nano-structures to endowfunctionality and/or enhance performance.1 In systemscomposed of hard and so multi-layers, spontaneous surfacewrinkling can be transformed from an unwanted mechanicalinstability to a naturally occurring opportunity that producescorrugated surfaces. Wrinkling occurs by subjecting a bilayercomprising a stiff skin resting on and adhering strongly to adeformable substrate to stress. This stress can be releasedthrough the bending and deformation of the layers to formregularly spaced corrugations. By controlling the directionalityof the compressive stress, wrinkling can be used to producedifferent types of periodic surfaces, ranging from randomly-oriented to well-aligned wrinkles, as a consequence of therelaxation of isotropic and anisotropic stresses, respectively.2,3

neering, North Carolina State University,

@ncsu.edu; Fax: +1 919 515 3419; Tel:

lar Engineering, North Carolina State

Chemistry 2013

In the case of uniaxial stress, both compressive stress as well asstress in elongation can lead to wrinkling.

In many cases, wrinkled surfaces are formed by depositing arigid lm (e.g., vapor deposited metal or metal oxides) onto acompliant polymeric substrate, such as a poly(dimethylsilox-ane) network (PDMS-N).2,4–11 Alternatively, the rigid lm may begenerated by chemically modifying the surface of a PDMS-Nsubstrate (e.g., oxidation via exposure to oxygen plasma).10,12–18

Ultimately, the wrinkling instability occurs by subjecting thebilayer system to some form of extrinsic or intrinsic stress, i.e.,thermal expansion mismatch strain due to heating ormechanical stretching, or swelling of the substrate by exposureto a solvent.19–21 While the stress can result from a variety ofsources, the common requirement is surface compression.

Here we report a study of surface wrinkling in a systemconsisting of an indium thin oxide (ITO) lm deposited on aPDMS-N substrate. This system forms wrinkles (1) in situ duringthe rst 5–10 seconds of sputter deposition, (2) with charac-teristic wavelengths independent of lm thickness, and (3) withtopography that perseveres aer removing the deposited ITOlm. These results are in contrast to multiple studies thatattribute wrinkles formed by physical vapor deposition of rigidlms on elastomeric substrates to the stresses arising from

Soft Matter, 2013, 9, 7797–7803 | 7797

Soft Matter Paper

Publ

ishe

d on

27

June

201

3. D

ownl

oade

d by

UN

IVE

RSI

TA

T G

IESS

EN

on

29/1

0/20

14 1

6:01

:03.

View Article Online

thermal expansion mismatch due to local heating. The presentresults illustrate that chemical modication of the substrate issufficient to cause in situ buckling of PDMS-N. We explorebuckling in the ITO/PDMS-N system because of opportunities tocreate transparent conductive surfaces featuring topography,although the results appear to extend to other sputtered systemssuch as Al/PDMS-N.

We present a self-consistent set of data, which shows that (1)magnetron sputtering plasmas can rapidly and substantiallymodify the chemical composition of PDMS-N surfaces in veryshort times, (2) intrinsic compressive stresses accompany thismodication, and (3) the initial formation of this layer regulateswrinkling onset and ultimate wavelength. We determine thechemical changes that occur during surface modication andestablish the window of sputtering conditions in which it occursby pressure-dependent energetic bombardment of the PDMS-Nsurface and the growing ITO lm.

Experimental procedurePreparation of poly(dimethylsiloxane) network substrates

Poly(dimethylsiloxane) networks (PDMS-N) were prepared froma commercially available elastomer kit. Specically, Dow Cor-ning's Sylgard-184. Sylgard-184A and Sylgard-184B were mixedin a 10 : 1 ratio by mass, according to the recipe suggested bythe manufacturer. Gentle vacuum (z100 Pa) removed thetrapped air bubbles prior to casting the mixture into a atpolystyrene dish. An oven cured the PDMS at 60 �C for at least120 min. A scalpel cut the crosslinked, yet exible PDMS-Nsubstrates (z1 mm thick) into 2.5 cm � 2.5 cm pieces. Thesubstrates were placed on a quartz plate and heated under highvacuum (2 � 10�6 Torr) at 120 �C for 3 hours to aid in outgas-sing any low-molecular mass components from the network.Substrates were stored in a covered Petri dish at ambientconditions until use. A 3.75 cm � 3.75 cm glass slide supportedthe PDMS-N during ITO deposition.

An additional set of thinner PDMS-N samples was preparedby spin coating uncured PDMS-N onto 2.5 cm� 2.5 cm pieces ofsilicon wafers. Varying the spin coating speed from 5000 to7000 rpm and the spin coating duration from 1 to 10 minproduced samples with thicknesses spanning 1 to 35 mm. Anoven cured these lms at 65 �C for at least 120 minutes. Aprolometer (Veeco Dektak 150 Surface Prolometer) measuredthe thickness of the lms.

Fig. 1 Optical microscope image of a buckled ITO film on a PDMS-N substrate inwhich 40 nm ITO has been deposited at 4 mTorr Ar. The inset shows the FFTof theimage, from which the characteristic wavelength can be determined.

Deposition of indium tin oxide top lms

A radio frequency (RF) magnetron sputtering chamber preparedITO lms using a commercially available sputtering target ofindium tin oxide (In2O3/SnO2, 90/10 wt%, 99.99% purity, 2 inchdiameter, Kurt J. Lesker Company). We deposited the lms atroom temperature in a load-locked instrument with a basepressure of 2.0 � 10�8 Torr, a sputter power of 30 W, a target-to-substrate distance of 8.5 cm, and a pure Ar sputtering ambient.These conditions resulted in a deposition rate ofz5 nm min�1

such that a 40 nm thick ITO lm could be prepared inz8 minutes of deposition. The ITO target was not exposed to

7798 | Soft Matter, 2013, 9, 7797–7803

atmosphere between depositions and ITO was pre-sputteredprior to lm growth to ensure consistent lms. A 40 nm thickITO lm was used initially because it provided a sheet resistancethat was suitable for a number of photovoltaic and light-emitting applications.

Results and discussion

Fig. 1 depicts an optical microscope image of a representativewrinkled surface of ITO on PDMS-N (40 nm ITO on 1 mm thickPDMS-N). The surface is populated by regularly spaced, butrandomly oriented wrinkles exhibiting a characteristic wave-length of z1.2 mm, which was determined through fast Fouriertransforms (FFT) of optical microscope images.

The formation of random wrinkles necessitates a compres-sively stressed ITO/PDMS-N laminate. A routine way to generatecompression is to stretch an elastomeric substrate prior todeposition and subsequently release the strain. The PDMS-N inFig. 1, however, was not stretched. In the absence of externallyapplied strain, a compressive stress may arise due to thermalexpansion mismatch between the ITO lm and the PDMS-Nsubstrate. For example, metal layers deposited on PDMS-N atelevated temperatures will wrinkle upon cooling to minimizecompressive surface forces that evolve because PDMS-N has amuch larger coefficient of thermal expansion (CTE) than themetal. The PDMS-N in Fig. 1, however, was not intentionallyheated.

Wrinkles are known to form during plasma oxidation ofPDMS-N. A common explanation is based on the surface ofPDMS-N heating during deposition, and upon cooling (i.e.,when the plasma process ends) wrinkles evolve to minimizestress due to the CTE mismatch between the lm and thesubstrate. It is important to note that this explanation does notaccount for all observations of wrinkling transitions in thinlm/elastomer stacks.18,22 In an attempt to elucidate the oper-ative mechanism, we performed a series of experiments aimedat determining the onset of wrinkling.

This journal is ª The Royal Society of Chemistry 2013

Paper Soft Matter

Publ

ishe

d on

27

June

201

3. D

ownl

oade

d by

UN

IVE

RSI

TA

T G

IESS

EN

on

29/1

0/20

14 1

6:01

:03.

View Article Online

Kinetics of wrinkle formation

If wrinkles are generated by a thermal mechanism, they shouldform at the end of the sputtering process (i.e., when thesubstrate begins to cool). We observed the onset of wrinkleformation by visually inspecting the substrate through awindow in the sputtering chamber in real time, and by exposingPDMS-N surfaces to the magnetron plasma at 5 second inter-vals. Interestingly, wrinkles form at ITO deposition timesbetween 5 and 10 seconds, causing the surface to appear hazy.At a growth rate of z5 nm min�1 this time interval will providebetween 0.4 and 0.8 nm of material. ITO layers will not coalesceat these thicknesses, thus eliminating their ability to apply abiaxial stress and to participate in buckle formation. We havealso seen similar behavior during deposition of aluminium. Inaddition, because wrinkling occurs rapidly during sputtering,the temperature of the surface is modest and is heating grad-ually rather than cooling. These observations suggest that thewrinkling mechanism is not due to thermal expansionmismatch of the ITO and PDMS-N. It also suggests the lm itself(i.e., ITO or Al) is not important, but rather, there is somechemical modication of the substrate during the rst fewmoments of sputtering that induces wrinkles.

Effect of ITO lm thickness

Because wrinkles appear during the rst few seconds of sput-tering, we sought to understand whether additional depositionwould change the periodicity or, instead, if the wavelength is setduring the initial sputtering stage. We deposited ITO lms 40–300 nm thick (8 to 60 minutes of deposition at 30 W RF power)onto 1 mm thick PDMS-N substrates on silicon wafers. The datain Fig. 2 show that the wavelength does not vary with thethickness of the ITO top lm. We also deposited ITO (100 nmthick) on PDMS-N (1 mm thick) at various sputtering powers of30, 70, and 90W RF and 4mTorr Ar. Because the deposition ratevaries with sputter power, we varied the sputtering times tomaintain a constant thickness of the ITO top lm. In all cases,the sample surfaces exhibited random buckles with a charac-teristic wavelength of z1 mm, similar to Fig. 1.

Fig. 2 Characteristic buckle wavelength as a function of ITO top film thicknesswith a constant substrate thickness. Estimated error is within the symbol size.

This journal is ª The Royal Society of Chemistry 2013

At rst consideration, the results in Fig. 2 seemingly disagreewith established theoretical models and experimental reportsthat predict the wavelength to depend strongly on top lmthickness.2,13,23,24 In systems where the substrate is heatedbefore top lm deposition, the top lm thickness is establishedprior to cooling and wrinkling; thus its entire thicknesscontributes to the mechanical response.2 Similarly, if thesubstrate is heated prior to exposing its surface to oxygenplasma, then the thickness of the oxidized surface of thesubstrate will affect wrinkle wavelength as stress develops uponcooling.17

However, if we consider the hypothesis that wrinkles resultsolely from initial PDMS-N modication at modest tempera-tures while the system is gradually warming, this result isself-consistent: additional ITO simply deposits on top of pre-existing wrinkles. If this hypothesis is true, not only should thewrinkles form very early during the sputtering process, but theyshould persist aer removing the ITO, provided the modiedsurface remains stable.

Wrinkles aer removing ITO

We removed the ITO top lm by etching for 5 minutes withhydrochloric acid (HCl, Fisher Scientic, trace metal grade).Brief HCl exposure does not alter PDMS-N yet it etches awayITO.20,25 Films were then rinsed in deionized water and driedwith nitrogen gas. Resistivity measurements conrmed theremoval of the ITO, and AFM shows the wrinkles remain presenton the sample and maintain their original characteristic wave-length, regardless of sputtering power (Fig. 3). This resultfurther supports the conclusion that wrinkles result fromsurface modication of the PDMS-N substrate during the rstfew seconds of sputtering, and that the ITO coating adopts thepre-existing buckle texture.

Chemical modication by sputtering

The persistence of wrinkles in the absence of the top lm withno change in wavelength suggests the sputtering process

Fig. 3 AFM image of the PDMS-N surface clearly demonstrating the presence ofbuckles after the removal of 40 nm ITO by etching.

Soft Matter, 2013, 9, 7797–7803 | 7799

Fig. 4 XPS survey scans of the surfaces of (a) the PDMS-N control, and PDMS-Nthat was stripped of ITO sputtered at (b) 30 W RF, (c) 70 W RF, and (d) 90 W RF.

Table 1 Atomic percent as determined by XPS

C (%) O (%) Si (%) In (%) Sn (%) Ca (%)

PDMS-N 45.43 30.08 24.48 0 0 030 W 33.85 40.78 24.89 0.23 0.08 0.1770 W 36.71 37.99 24.60 0.41 0.08 0.2190 W 33.23 42.34 24.35 0 0.08 0

Soft Matter Paper

Publ

ishe

d on

27

June

201

3. D

ownl

oade

d by

UN

IVE

RSI

TA

T G

IESS

EN

on

29/1

0/20

14 1

6:01

:03.

View Article Online

chemically modies the PDMS-N surface during the initialstages of sputtering and creates compressive stresses that resultin buckles. The stability of polymers in oxidizing environmentshas been investigated previously; it is well established thatargon ion (Ar+) bombardment, oxygen plasma treatment, andultraviolet (UV) or ultraviolet/ozone (UVO) exposure lead tomodication of the surface of PDMS.10,17,24,26–28 For example,oxygen plasma treatment removes preferentially methyl groups,leading to the formation of additional Si–O–Si cross-links inPMDS and hydroxylation of the PDMS surface, while Ar+

bombardment creates crosslinks between silicon atoms.10

During sputtering of ITO, the plasma that is generatedcomprises a partially ionized gas consisting of ions such as Ar+,O*2, O

�, and UV photons.When exposed to such environments PDMS forms a silica-

like surface layer.2,27,28 For example, one study utilizing oxygenplasma exposure reports the formation of a 130–160 nm thicksmooth, oxidized SiOx-rich surface layer aer 80–180 secondsexposure, in which silicon bonds to three or four oxygenatoms.27 Similarly, argon plasma exposure forms a 30 nm thickoxygen rich layer.24 In the case of UVO exposure, PDMS-Nsurface modication (chain scission, recombination anddensication) extends 5–10 nm below the surface, while forlonger exposure a silica-like surface layer formed with a densityapproaching 50% of the theoretical silica value.28 These reportsdetermined that active oxygen species are integral to themodication process, as UV treatments alone did not yieldextensive chemical modication.28

Consequently, there are important implications for sputter-ing oxides on PDMS-N given the presence of energetic species inall sputter plasmas, particularly at the initial stages of lmdeposition. A modied or silica-like layer formed in situ mustimpact wrinkle formation.

To explore this possibility, we used X-ray photoelectronspectroscopy (XPS, Riber LAS-3000 with MgKa excitation of1254 eV) to compare the surfaces of (1) PDMS-N stripped of40 nm ITO that was sputtered at 30, 70 and 90 W RF, (2) virginPDMS-N, and (3) virgin PDMS-N exposed only to the HCletchant. We established an energy calibration for the XPSmeasurements by referencing to adventitious carbon (C-1s lineat a binding energy of 285 eV). During measurements, thebackground pressure within the analysis chamber was z10�10

Torr, the take-off angle was z75� from the surface, the X-rayincidence angle was z20�, and the X-ray source to analyzerangle was z55�. With this setup, the XPS probing depth isapproximately 8.7 nm.

Fig. 4 displays XPS survey scans of each sample across thebinding energy range of 0 to 1200 eV, revealing prominentsignals that are attributed to O-1s, O-2s, C-1s, Si-2s, and Si-2pbinding energies. In some cases, minor peaks were alsoobserved and correspond to In-3d, In-3p, and Sn-3d. Table 1summarizes the atomic percents determined from these surveyscans.

Both control samples (PDMS-N and PDMS-N exposed brieyonly to HCl) show surface compositions that are similar to whatis theoretically expected based on the chemical formula ofPDMS (50 at% C, 25 at% O, and 25 at% Si). The deviations from

7800 | Soft Matter, 2013, 9, 7797–7803

pure PDMS are similar to several literature reports (47.1 at% C,25.1 at% O, and 27.7 at% Si)29 and can be understood consid-ering that Sylgard-184 is not pure PDMS, rather it is a compositecontaining a substantial fraction of silica llers (30–60 wt%),resins, and other components. The entire set of surfacecompositions is represented graphically in Fig. 5 for clarity.Though not shown, the plasma-exposed samples exhibited Si 2pand O 1s peak shis of approximately +0.3 and 0.4 eV, respec-tively. The increase in binding energy is consistent with theincreased electronegativity of higher oxygen coordination and amore SiO2-like chemical environment. Similar peak shis havebeen reported as a consequence of plasma exposure.10,29,30

The surface composition of the PDMS-N substrates, whenconsidered relative to Si (which will not change since it is notvolatile and it is not a component of the ITO plasma), shows anaverage 24% decrease in C and an average 35% increase in O,which can be determined from the values shown in Table 1.These data show that more oxygen was added than carbonremoved. In addition, traces of In and Sn were found aeretching. We propose that in addition to the surface modica-tion by oxidation, the PDMS-N surface was implanted by In,and Sn from the plasma during sputtering. We furtherhypothesize that the additional z10 atomic% of materialobserved aer deposition results in surface swelling, and thuscompressive stress. This newly formed surface/substrate stack

This journal is ª The Royal Society of Chemistry 2013

Fig. 5 Atomic percent as determined by XPS for the PDMS-N control and forfilms in which 100 nm ITO was deposited at 30, 70, and 90 W RF and thenremoved by etching.

Fig. 6 Characteristic buckle wavelength predicted for a 10 nm thick silica-liketop film and for 10 nm and 40 nm thick ITO top films compared to bucklewavelengths observed in experiment (estimated error is within the symbol size).

Paper Soft Matter

Publ

ishe

d on

27

June

201

3. D

ownl

oade

d by

UN

IVE

RSI

TA

T G

IESS

EN

on

29/1

0/20

14 1

6:01

:03.

View Article Online

accommodates subsequently this volume expansion by abuckling transition.

The formation of a silica-like surface has been reportedpreviously upon exposure of PDMS to UVO, argon plasma, Ar+

bombardment, or oxygen plasma, but the formation of wrinkleswas attributed to stresses arising from thermal expansion. Ourreport is the rst to describe such modication and wrinklingin situ to thin lm deposition and relate its importance to themechanism of wrinkling.

Modeling wrinkle periodicity

We sought to model the characteristic wavelength of the wrin-kles with the understanding that they originate from a silicalayer on the surface of the PDMS-N. We evaluated the wave-length (l) in the context of the classic model that describeswavelength as a function of thickness (t):

l ¼ tfilm

2p

313

! Efilmð1� nsubstrate

2ÞEsubstrateð1� nfilm2Þ

!13

where l is the wrinkle wavelength, t is the thickness, n is Pois-son's ratio, and E is Young's modulus.31

We employed the model to predict separately the wavelengthassuming properties for ITO and silica overlayers. Fig. 6summarizes the results for constant thickness ITO lms (10 and40 nm) on a variable thickness PDMS-N substrate assumingElm ¼ 118 GPa, tlm ¼ 10 nm or 40 nm, nlm ¼ 0.35, andEsubstrate ¼ 1.5 MPa. The model signicantly overestimates thewavelength at all ITO thicknesses. This result is consistent withthe understanding that the ITO lm is a minor participant tothe wrinkling.

In addition, a more complex model that minimizes the totalfree energy of the top lm bending and substrate deformationwith respect to characteristic buckle wavelength was used topredict wavelength and gave nearly identical results.32,33

This journal is ª The Royal Society of Chemistry 2013

Fig. 6 also contains model predictions for a 10 nm silica-likelayer on the same PDMS-N substrates where Elm ¼ 40 GPa,nlm ¼ 0.17, tlm ¼ 10 nm, and Esubstrate ¼ 1.5 MPa. This theoryagrees with our experimental observations as it predicts awavelength of z1 mm. Importantly, these predictions are alsoconsistent with the ITO top lm thickness series, since buckleperiodicity does not depend on the ITO thickness. Conse-quently, this model supports our observations of the formationof a silica-like layer during sputtering of ITO on PDMS-N, andsuggests a thickness on the order of z10 nm.

Intrinsic stress from sputtering

The studies presented thus far suggest that the compressivesurface stress originates from the modied surface layer of thePDMS-N. However, the stress in a sputtered lm is known todepend strongly on deposition pressure, and the potentialcontributions from as-deposited, intrinsic stresses must beconsidered. To tune the as-deposited stress, we investigated aspectrum of deposition energetics by varying the argon pressureduring sputtering. This technique is effective in regulating thelanding energy of sputtered ions by controlling the number ofgas-phase collisions, and thus the degree to which kineticenergy of the plasma species is thermalized in the gas phase. Todo so, we prepared 40 nm thick ITO layers on 1 mm thick slabsof PDMS-N at pressures between 2 mTorr and 12 mTorr argon,while maintaining constant power (30 W RF) and argon owrate (9 sccm Ar). Fig. 7 displays characteristic buckle wave-lengths as a function of Ar deposition pressure. Randomlyoriented wrinkles form only at pressures between 2 and12 mTorr, and there is no signicant relationship betweenwavelength and pressure. However, at deposition pressuresgreater than 12 mTorr, the surface forms cracks without buck-ling, while at pressures below 2 mTorr, the plasma cannot besustained, and ITO cannot be sputtered.

Soft Matter, 2013, 9, 7797–7803 | 7801

Fig. 7 Characteristic buckle wavelength as a function of deposition pressure. It isapparent that a deposition pressure based window exists in which bucklingoccurs.

Soft Matter Paper

Publ

ishe

d on

27

June

201

3. D

ownl

oade

d by

UN

IVE

RSI

TA

T G

IESS

EN

on

29/1

0/20

14 1

6:01

:03.

View Article Online

While we have proposed a mechanism for buckle formationvia bombardment, it is necessary to understand this pressurewindow of buckling and lm cracking within the same context.By considering how the plasma interacts with the growing lmaer PDMS-N modication, we can establish a self-consistentexplanation. At low pressures, gas phase scattering is limited, andenergetic impingement results in forward sputtering (or atompeening) of the growinglm. Peening contains enough energy forimplantation into a growing lm and for dislodging alreadydeposited atoms to higher energy positions that are strainedwithrespect to the equilibrium lattice.34 This effect produces densi-cation and inmany cases compressive stress, which can generallybe supported without catastrophic mechanical failure. Wepropose that these energetic species are responsible for modi-fying the surface of PDMS-N. As pressures are increased beyond10 mTorr, the kinetic energy of plasma species is thermalizedprior to deposition. The surface damage to PDMS-N is reduced,and adatoms have less overall mobility aer landing. Under suchconditions surface modications that produce buckling nolonger occur, and tensile stresses are expected to evolve followingthe grain boundary relaxation model.34 In brittle oxides, i.e., ITO,such stresses lead to cracking at modest levels, particularly fordepositions on substrates like PDMS-N, which offer limitedstructural support. The transition from compressive to tensilestress can be quite sharp. In one well-documented example, anearly 4 GPa swing from compressive to tensile lm stressoccurred over a 5 mTorr pressure range in sputtered molyb-denum and other metals.35–39 It is important to note thatintrinsic stress in thermally evaporated lms is usually tensile,34

and thus wrinkling in thermally evaporated systems typicallyoccurs from external compressive stress. This behavior is repor-ted for metal evaporation on PDMS-N at elevated temperatures,and wrinkle formation upon post-deposition cooling.2

These pressure-dependent bombardment-based mecha-nisms can also explain lm cracking, because deposition

7802 | Soft Matter, 2013, 9, 7797–7803

pressures that mitigate energetic bombardment should reduceadatom mobility to levels that allow tensile stress to accumu-late, simultaneously eliminating substrate damage that chem-ically alters the PDMS-N to instigate a wrinkle transition.Finally, it is necessary to consider the role of intrinsiccompressive lm stresses on wrinkle formation, as in principle,these can provide a contributing inuence. To do so, we returnto experiments measuring the deposition time required to formwrinkles. As discussed previously, these measurements showbuckle formation at times between 5 seconds and 10 seconds. Ata growth rate of z5 nm min�1 this time interval will providebetween 0.4 and 0.8 nm of material. ITO layers will not coalesceat these thicknesses, thus eliminating their ability to apply abiaxial stress and to participate in buckle formation.

Conclusions

We present evidence that buckles formed by sputter depositionform due to stress arising from chemical modication of thePDMS-N substrate rather than thermal stress. Buckles aretypically attributed to the compressive forces that result when alm deposited on an elastomeric substrate cools from elevatedto room temperature. Here, a number of factors suggest a newmechanism: (1) the buckles form in the rst 5–10 seconds ofdeposition at which time the temperature of the surface shouldbe modest and heating gradually rather than cooling and (2)the wavelength of the buckles is independent of the thicknessof the ITO layer and persists when acid chemically removes theITO lm. XPS analysis shows that the modied surface ofPDMS-N has lower carbon content and higher oxygen contentrelative to the parent material. Taken in entirety, these resultssuggest that the chemical modication of the surface of thePDMS-N creates compressive stresses that induce wrinkles.These wrinkles establish the nal wavelength, and subsequentITO deposition only increases the lm thickness withoutchanging the wavelength.

These results demonstrate the important effects of sputter-ing on controlling intrinsic top lm stress such that sponta-neous wrinkling can occur. Furthermore, this work leads to aclearer understanding of the origins of wrinkling in sputteredtop lms on polymeric substrates, which is promising as astrategy for controlled texturing and patterning of surfaces bywrinkling.

Acknowledgements

This work was supported by the Army Research Office (GrantNo. W911QY-10-C-0027). JG acknowledges partial support fromthe Office of Naval Research (Grant No. N000141210642), andMDD acknowledges partial support from the NSF ResearchTriangle MRSEC (DMR-1121107).

Notes and references

1 J. Genzer and J. Groenewold, So Matter, 2006, 2, 310–323.2 N. Bowden, S. Brittain, A. G. Evans, J. W. Hutchinson andG. M. Whitesides, Nature, 1998, 393, 146–149.

This journal is ª The Royal Society of Chemistry 2013

Paper Soft Matter

Publ

ishe

d on

27

June

201

3. D

ownl

oade

d by

UN

IVE

RSI

TA

T G

IESS

EN

on

29/1

0/20

14 1

6:01

:03.

View Article Online

3 K. Emenko, M. Rackaitis, E. Manias, A. Vaziri,L. Mahadevan and J. Genzer, Nat. Mater., 2005, 4, 293–297.

4 G. C. Martin, T. T. Su, I. H. Loh, E. Balizer, S. T. Kowel andP. Kornreich, J. Appl. Phys., 1982, 53, 797–799.

5 X. Chen and J. W. Hutchinson, J. Appl. Mech., 2004, 71, 597–603.

6 T. Okayasu, H.-L. Zhang, D. G. Bucknall and G. A. D. Briggs,Adv. Funct. Mater., 2004, 14, 1081–1088.

7 H. Vandeparre, J. Leopoldes, C. Poulard, S. Desprez,G. Derue, C. Gay and P. Damman, Phys. Rev. Lett., 2007, 99,188302.

8 H. Plank, R. Gunter, U. Scherf and E. J. W. List, So Matter,2007, 3, 713–717.

9 F. Iacopi, S. H. Brongersma and K. Maex, Appl. Phys. Lett.,2003, 82, 1380–1382.

10 P. Bodo and J.-E. Sundgren, Thin Solid Films, 1986, 136, 147–159.

11 A. L. Volynskii, S. Bazhenov, O. V. Lebedeva and N. F. Bakeev,J. Mater. Sci., 2000, 35, 547–554.

12 D. B. H. Chua, H. T. Ng and S. F. Y. Li, Appl. Phys. Lett., 2000,76, 721–723.

13 D. C. Hyun and U. Jeong, J. Appl. Polym. Sci., 2009, 112, 2683–2690.

14 A. Chiche, C. M. Stafford and J. T. Cabral, So Matter, 2008,4, 2360–2364.

15 J. Y. Chung, J. P. Youngblood and C. M. Stafford, SoMatter,2007, 3, 1163–1169.

16 M.-W. Moon, S. H. Lee, J.-Y. Sun, K. H. Oh, A. Vaziri andJ. W. Hutchinson, Proc. Natl. Acad. Sci. U. S. A., 2007, 104,1130–1133.

17 N. Bowden, W. T. S. Huck, K. E. Paul and G. M. Whitesides,Appl. Phys. Lett., 1999, 75, 2557–2559.

18 J.-Y. Park, H. Chae, C.-H. Chung, S. J. Sim, J. Park, H. H. Leeand P. J. Yoo, So Matter, 2010, 6, 677–684.

19 T. Tanaka, S.-T. Sun, Y. Hirokawa, S. Katayama, J. Kucera,Y. Hirose and T. Amiya, Nature, 1987, 325, 796–798.

20 T.-K. Shih, J.-R. Ho, C.-F. Chen,W.-T. Whang and C.-C. Chen,Appl. Surf. Sci., 2007, 253, 9381–9386.

This journal is ª The Royal Society of Chemistry 2013

21 J. Y. Chung, A. J. Nolte and C. M. Stafford, Adv. Mater., 2009,21, 1358–1362.

22 J. Kim and H. H. Lee, J. Polym. Sci., Part B: Polym. Phys., 2001,39, 1122–1128.

23 R. Huang, C. M. Stafford and B. D. Vogt, J. Aerosp. Eng., 2007,38–44.

24 H. T. Evensen, H. Jiang, K. W. Gotrik, F. Denes andR. W. Carpick, Nano Lett., 2009, 9, 2884–2890.

25 J. N. Lee, C. Park and G. M. Whitesides, Anal. Chem., 2003,75, 6544–6554.

26 J. R. Hollahan and G. L. Carlson, J. Appl. Polym. Sci., 1970, 14,2499–2508.

27 H. Hillborg, J. F. Ankner, U. W. Gedde, G. D. Smith,H. K. Yasuda and K.Wikstrom, Polymer, 2000, 41, 6851–6863.

28 K. Emenko, W. E. Wallace and J. Genzer, J. Colloid InterfaceSci., 2002, 254, 306–315.

29 B. Schnyder, T. Lippert, R. Kotz, A. Wokaun, V.-M. Graubnerand O. Nuyken, Surf. Sci., 2003, 532–535, 1067–1071.

30 D. Eon, L. de Poucques, M. C. Peignon, C. Cardinaud,G. Turban, A. Tserepi, G. Cordoyiannis, E. S. Valamontes,I. Raptis and E. Gogolides, J. Phys.: Conf. Ser., 2002, 61–62,901–906.

31 H. G. Allen, Analysis and Design of Structural Sandwich Panels,Pergamon, New York, 1969.

32 J. Groenewold, Physica A, 2001, 298, 32–45.33 P. J. Yoo, K. Y. Suh, S. Y. Park and H. H. Lee, Adv. Mater.,

2002, 14, 1383–1387.34 F. M. d'Heurle and J. M. E. Harper, Thin Solid Films, 1989,

171, 81–92.35 D. W. Hoffman and J. A. Thornton, Thin Solid Films, 1977, 40,

355–363.36 D. W. Hoffman and J. A. Thornton, Thin Solid Films, 1977, 45,

387–396.37 D. W. Hoffman and J. A. Thornton, J. Vac. Sci. Technol., 1979,

17, 380–383.38 D. W. Hoffman and J. A. Thornton, J. Vac. Sci. Technol., 1982,

20, 355–358.39 J. A. Thornton, Thin Solid Films, 1989, 171, 5–31.

Soft Matter, 2013, 9, 7797–7803 | 7803