Photolithographic structuring of stretchable conductors and sub-kPa pressure sensors

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2011 J. Micromech. Microeng. 21 115010

(http://iopscience.iop.org/0960-1317/21/11/115010)

Download details:

IP Address: 193.61.144.153

The article was downloaded on 10/10/2011 at 08:53

Please note that terms and conditions apply.

View the table of contents for this issue, or go to the journal homepage for more

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING JOURNAL OF MICROMECHANICS AND MICROENGINEERING

J. Micromech. Microeng. 21 (2011) 115010 (12pp) doi:10.1088/0960-1317/21/11/115010

Photolithographic structuring ofstretchable conductors and sub-kPapressure sensorsC L Tuinea-Bobe1, P Lemoine2, M U Manzoor2, M Tweedie2, R A D’Sa2,C Gehin3 and E Wallace4

1 MNT, School of Engineering, Design and Technology, University of Bradford, UK2 NIBEC, University of Ulster at Jordanstown, Northern Ireland, UK3 INL, INSA Lyon, Lyon, France4 SESRI, University of Ulster at Jordanstown, Northern Ireland, UK

E-mail: p.lemoine@ulster.ac.uk

Received 19 May 2011, in final form 4 August 2011Published 7 October 2011Online at stacks.iop.org/JMM/21/115010

AbstractThis paper presents a novel method to prepare stretchable conductors and pressure sensorsbased on the gold/polydimethylsiloxane (PDMS) system. The gold films were sputtered ontostructured PDMS surfaces produced with a photolithographic surface treatment with the aimof reducing tensile strains in the gold film. Scanning electron microscopy (SEM) and atomicforce microscopy analyses showed that these 3D patterns reduce cracks and delaminations inthe gold film. Electrical measurements indicate that the patterns also protect the films againstrepeated tensile cycling, although the un-patterned samples remained conducting as well afterthe completion of 120 cycles. The extrapolated resistivity value of the patterned sample(4.5 × 10−5 �cm) compares well with previously published data. SEM micrographs indicatethat the pattern features deflect the cracks and therefore toughen the gold film. However, x-rayphotoelectron spectroscopy and contact angle analyses indicate that the patterning process alsoslightly modifies the surface chemistry. This patterning method was used to prepare capacitivestrain gauges with pressure sensitivity (�Z/Z)/P of 0.14 kPa−1 in the sub-kPa regime. Suchstretchable and potentially conformal low-pressure sensors have not been produced before andcould prove advantageous for many smart fabric applications.

(Some figures in this article are in colour only in the electronic version)

1. Introduction

Current flextronics systems available commercially are basedon thermo-plastics like polyimide, can only bear strains of afew per cent and hence can flex but not stretch. Having a trulystretchable electronic circuit would be advantageous for manylow-pressure sensing applications where the sensing elementmust be integrated into fabrics or textile components.

Stretchable conductors have been prepared withconducting polymers [1], doped elastomers [2] or more simplymetal films on elastomeric substrates, like gold on PDMS;this last option has shown the best results to date in terms ofconduction, tensile behaviour and stability [3–5].

The challenge nonetheless is that the metal films are lacedwith cracks. This results from the large thermo-mechanicaldissimilarities between the two materials and the ensuingstrains, either built-in from the post-metal deposition coolingor from applied stresses. A significant work has been directedat modelling and testing optimized electrode layouts, suchas the horseshoe design, to minimize in-plane strains [6].However, to date, there has been scarce effort to understand andcontrol perpendicular strains and optimize the metal/elastomerinterface.

In this paper, a novel approach is presented which usesphotolithography to define bi-sinusoidal surface structures onthe elastomeric substrate before depositing the gold layer. Thisresults in strain compensation and crack deflection in the two

0960-1317/11/115010+12$33.00 1 © 2011 IOP Publishing Ltd Printed in the UK & the USA

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

(a) (b)

(c)

Figure 1. (a) Schematic description of the 3D surface structuring process, shown as a transverse cross-section of the patterned surface,(b) principle of operation of the maskless lithography system, (c) example of the mask pattern with typical dimensions of the irradiatedsample area for a 3 pixel mask.

planar directions. Others used a similar method for uni-axialcompensation [7]; obviously, its generalization to a bi-axialgeometry makes it more applicable to stretchable conductorapplications. These improved conductors were used to builda prototype compliant and a conformal pressure sensor withsub-kPa sensitivity.

2. Experimental details

2.1. Material and device preparation

The PDMS (CH3[Si(CH3)2O]nSi(CH3)3) sheets were preparedwith a two-component Sylgard 184 resin (10% hardener and90% base), cured at room temperature for 24 h in Perspexcasting moulds, cleaned with acetone and dried in nitrogen.The produced sheets are resilient; they can resist a very largenumber of stretchings at high strain (several 100%), and theyare also transparent and smooth (Ra ∼ 5 nm for 2 × 2 μm2

AFM images) with typically ∼1 mm thickness.The process steps to form the stretchable conductors are

shown in figure 1(a). A screening study was conductedto optimize the various process parameters (type of SU-8resist and thickness, spin coating speed, soft bake, exposuretime, post-exposure bake, development time, rinse and dry,hard bake). In the first step, a film of the UV sensitivematerial is applied via spin coating at 3000 rpm which gives adifferent thickness of SU-8, depending on the viscosity of thephotoresist (e.g. for SU-8 50–40 μm, SU-8 3025–25 μm). Asoft bake is also carried out at 65 and 95 ◦C on a hot plate,time varying with the type of SU-8 photoresist, as specifiedin the SU-8 material data sheet (figure 1(a), 1). During thesecond step, the 3D surface photo-structuring is carried outusing either a maskless system (SF 100) or a mask aligner(EVG 620) used in conjunction with a chrome on quartz mask

(5 μm features and 5 μm gaps from JD Phototools Ltd.)For both systems, the resist was SU-8 and the illuminationsource was the i-line (wavelength 365 nm) of a mercury arclamp with sufficient pre-heating to obtain stable irradiation.The maskless system (SF 100—Intelligent MicroPatterning,LLC Mercury Vapour Lamp) is based on a micro-mirrorarray (the smart filter—figure 1(b)). In this study, only twogrey levels bitmap masks were generated using the PAINTsoftware from Microsoft Windows. Each pixel of the maskaddresses an individual mirror with, for instance, a blackpixel corresponding to the light being deflected away fromthe sample plane. In this way, the pattern of the mask canbe replicated onto the sample. The patternable area andultimate resolution of the system depend respectively on theoptical arrangement and the size and number of micro-mirrors.Typically, this results in 200 mm2 area and 40 μm resolutionfor both systems. The EVG 620 is a conventional mask alignersystem used in a soft contact mode with much higher resolution(<5 μm) and a larger patternable area, ∼8000 mm2 (exposuretime between 10 and 60 s). For both lithographic systems, thisphoto-exposure is followed by a post-bake, performed as wellat 65 and 95 ◦C. The third step, the development is an essentialone. We observed that the application of the developer for shorttimes (∼30 s) by spin coating, hence avoiding immersion andsubsequent isopropanol rinsing, results in transparent couponsand better results. It also includes a hard bake, i.e. bakingthe sample at 150 ◦C for 10 min on a hot plate. The fourth,fifth and sixth steps of the stretchable conductor preparationprocess are, respectively, the PDMS casting, PDMS peelingand metal deposition. The aims of this process are

(1) to increase the surface length in the two biaxial directions,to reduce surface strain, as demonstrated previously foruniaxial studies [7];

2

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

Figure 2. Schematic diagram showing the fabrication steps of the low-pressure capacitive strain gauge using stretchable conductors. Steps 1to 9 are top views, step 10 represents a transversal cross-section of the central region of the device, and a photograph of the bottom electrodeis also shown.

(2) to obtain features as rounded as possible to diminish highstrains around sharp edges.

In this regard, the sinusoidal waveforms shown infigure 1(a) are not gold film ripples due to thermal stresson cooling but instead represent an ordered pattern purposelyrounded. In practice, the departure from a square profile (i.e.the original mask) will depend on the process parameters. Forthe SF 100 lithography, this rounding is obtained by usingdefocused conditions (in 0.25 mm steps) which give a bi-sinusoidal pattern. For the EVG 620 system, the rounding isobtained by controlling the mask/resist overlay distance, theexposure time and the post-bake which result in reflow.

In the process above, the post-exposure bake and hardbake are designed to cross-link both exposed and un-exposedSU-8 to produce stable structures. It certainly does this in thebulk but not necessarily so at the surface level. For instance,following development, low molecular weight residues mayremain on the surface. In addition, PDMS peeled fromSU-8 and glass may present different surface chemistries.Hence to study these effects, a single-feature photolithographicmask was prepared to define two ‘large’ exposed and un-exposed areas (several cm2 each). The surface of thesesamples was studied with contact angle measurements and x-ray photoelectron spectroscopy (XPS) analysis, as discussedbelow.

The gold films were deposited with a magnetron sputtersystem using a titanium adhesion layer (∼5 nm Ti and∼25 nm Au), as described previously [8, 9]. The temperature

variation in the chamber was monitored with temperature strips(TMC Hallcrest temperature indicating labels, scale range A,B). During deposition, the temperature increased to ∼60 ◦C.The deposition was always carried out on the PDMS surfaceexposed to the mould (glass, exposed SU-8 or un-exposed SU-8) to limit air contamination. The metal film thicknesses werecalibrated from the growth rates measured with AFM usinga resist pen lift-off technique [10]. A thin-resist pen traceis deposited on the substrate prior to the metallic deposition.After deposition, the sample is sonicated in acetone for 5 min.This results in a metal-free substrate strip with sharp metaledges, and the metal thickness can be measured with an AFMcross-section.

Finally, these stretchable conductors were used to preparea low-pressure capacitive strain gauge. The full process,shown in figure 2, can be summarized as follows. Thebottom PDMS slab was cured on an SU-8 substrate withan active area patterned as an egg box, with four Ag-coatedpolyimide (PI) strips contacting the egg-box area from the fouropposite sides. The PI/PDMS adhesion was strong allowingpeeling the cured PDMS/PI stripes ensemble as one cohesivestructure (figure 2-1). Two of these contacts (top electrodes)were masked (figure 2-2) and the device was gold coated(figure 2-3). All four contacts were masked (figure 2-4), anda 2:1 PDMS-toluene mixture was spin coated at 5000 rpm toproduce an 8 μm thick dielectric layer with a flat top surfaceand a sinusoidal bottom one (figure 2-5). Insignificant weightlosses (<0.1 wt.%) were measured after 24 h, indicating that

3

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

this layer was stable and not leaching out trapped solvent.The two bottom contacts were masked (figure 2-6) and thestructure was again gold coated (figure 2-7). Finally, maskingthe four contacts (figure 2-8), a final PDMS spin-coating stepwas carried out to embed the sensor in a protective elastomerictop layer (figure 2-9). Figure 2-10 shows a transversal cross-section of the central sensor region (i.e. active area). Thisdevice configuration permitted us to test the ac electricalbehaviour of the top electrode, the bottom electrode andthe device capacitance using a Solartron impedance/phaseanalyser operating with a 3 mV rms input signal and calibratedwith an 82 pF capacitor. The presented results correspond tothe 10 kHz frequency, although Bode and Nyquist plots werealso obtained for the full frequency testing range (1–103 kHz).The Solartron was also used to test the electrical behaviourof the silver-epoxy contacts used to connect the devices tothe instrument. These tests were carried out for variouspressures (up to 30 kPa), and applied with calibrated weightsover the active area of the sensor. In all, three devices wereprepared: one without patterning (U), one patterned withoutPDMS embedding (P) and one with patterning and with PDMSembedding (PE). As the bulk of this investigation is morefocused on the development and analysis of a novel processthan the production of devices, only one specimen of the finaldevice was prepared, although a number of other devices wereproduced albeit with different process parameters, all devicessatisfactorily operating as capacitive strain gauges. Moreover,for each sample, at least five electrical measurements weretaken, giving standard deviation error bars, as shown in therelevant graphs.

2.2. Material analysis

Tensile tests were performed on an Instron 3344 using ‘dogbone’ coupons (ISO 527-2:1996, 20 mm gauge length). ThesePDMS coupons had a tensile modulus of ∼1.2 MPa withstrain at failure in the excess of 100%. The morphology ofthe surfaces was examined with optical microscopy, contactprofilometry (CP) (Veeco Dektak 8 profiler), atomic forcemicroscopy (AFM, Veeco DI3100 SPM system used in tappingmode) and scanning electron microscopy (SEM, FEI Quanta3D system or a Hitachi S3200N system). For AFM, thepossibility of tip blunting was checked by imaging periodicallya standard Nioprobe sample to monitor any eventual changein tip apex geometry. CP was used to examine largefeatures over long scan lengths (∼100 μm and more), whereasAFM microscopy permitted to look at the smaller features.Generally, the tip semi-apex angle was smaller than the surfaceslopes so that the patterns were correctly measured. TheSEM micrographs were obtained mainly at low kV or in lowvacuum (∼20 Pa). In addition, an Oxford Instrument EnergyDispersive x-ray (EDX) spectrometer was also used to analysethe composition of the samples.

DI water contact angle (θ ) measurements were carried outat room temperature (KSV Instruments Ltd, Model CAM 200using 7 μl DI water). The reported θ values represent theaverage of the left- and right-hand angles measured from threedrops deposited onto the samples directly after peeling fromthe mould.

The electrical measurements of the stretchable conductorswere performed during tensile cycling. The tensile test wasperformed for the patterned and un-patterned samples usingInstron 3344 tensile testing system. This permitted bettercontrol of mechanical parameters such as load, displacement,loading rate and displacement rate. The test consisted inacquiring slow loading electrical and mechanical data up to35% strain. The unloading was not machine controlled andtherefore was made abruptly. The silver-epoxy electricalcontacts were placed on each side of the sample perpendicularto the extension direction cured for 24 h at room temperatureand connected to the electrical measuring apparatus (Turnbilldc source with current and voltage across the sample measuredindependently). Generally, these contacts were satisfactory(<0.5 �). The cycling consisted of acquiring 120 loadingcycles at low crosshead speed (strain rate ∼10−3 s−1) upto 35% strain while recording the variations in the electricalresistance upon extension. The 35% strain value was chosendue to the project aim; pressure ulcer prevention where seatingarea deformations are expected to be <20% [11].

An XPS system (Kratos Axis Ultra DLD spectrometer)equipped with a monochromatic AlKα x-ray (1486.6 eV) wasused to analyse a number of PDMS coupons, as discussedabove. These samples were analysed immediately afterpeeling. The chamber pressure was maintained at < 5 ×10−8 Torr and surface charging was compensated with asecondary electron gun. Wide energy survey scans (WESS)(0–1300 eV) with a step size of 0.5 eV were used to identifyall the elements present in the sample, and specific spectralregion of interest was scanned with 0.05 eV step size. Whendeconvoluting the XPS peaks, the energy scale was calibratedwith the 284.4 eV low-energy component of the C1s peak,representative of the siloxane group of PDMS.

3. Results and discussion

3.1. SF 100 maskless system processing

Initially, a viscous photoresist (SU-8 50) was used. It gavethick resist layers (∼20 μm) and high aspect ratio features,as shown in the inset of figure 3(a). The process results in aconcentration and organization of the gold film ripples betweenthe pattern features. It also proved difficult to prepare roundedfeatures and cracks developed in the proximity of sharp edges,as shown by the arrows in figure 3(b).

Therefore thinner resists (SU-8 2000.5 and SU-8 3025)were adopted which yielded patterned gold/PDMS surfaceswith fewer cracks and no delaminations. SEM micrographsof the un-patterned and patterned gold/PDMS samples areshown, respectively in figures 4(a) and (b). The un-patternedsample displays the usual arrays of cracks, shown with blackarrows in figure 4(a). Two high-magnification insets areshown in figure 4(a), showing film rippling and contact points(small white arrows) at the jagged edges of compressed cracks,with charging being apparent on the bare PDMS substrate.Ripples are also shown in the inset of figure 4(b) for thepatterned sample. Close to the edge, these ripples are alignedand running perpendicular to it, while in the centre of the

4

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

(a) (b)

Figure 3. (a) SU-8 features, (b) SEM image of the gold on PDMS obtained after casting on the SU-8.

(a) (b)

(c)

(d )

Figure 4. SEM micrographs of gold film on (a) un-patterned and (b) patterned PDMS using a low aspect ratio process (SU-8 2000.5). Thecracks are shown with black arrows. The insets show ripples and contact points at the jagged edges of compressed cracks (white arrows) aswell as charging on the PDMS; (c) the distinction between ripples and cracks can be seen from the analysis of an AFM image; (d) forcecurves showing surface slopes calculating on loading between the two dotted lines crossing the loading segment. These slopes are ∼480 nmV−1 inside the crack and ∼70 nm V−1 outside the crack, respectively (the blue trace is loading, the red trace is unloading). In order tocharacterize a material from the force curve, the cantilever approached the sample from position A when is free and jumped to contact inposition B. The slope of the deflection C provides information on the stiffness of the surface/cantilever system. The adhesion D providesinformation on the interaction between the probe and sample surface as the probe is trying to break free.

samples, they consist of a 3D random pattern of corrugations,as shown in the AFM micrograph figure 5(a). These effectsdue to thermal stress on cooling are well established [12]

but the distinction between cracks and ripples is not alwaysstraightforward and was asserted in this study as follows.Under the AFM tip load, the gold deforms minimally when

5

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

(a) (b)

(c) (d )

Figure 5. (a) AFM analysis of the un-patterned sample, (b) AFM analysis of the patterned sample, (c) profile of the un-patterned sampleresulting from the AFM analysis software, (d) profile of the patterned sample resulting from the AFM analysis software: left in the area ofthe pattern, right in between the pattern.

compared to the PDMS, hence delaminations and cracks canbe identified analysing the elasticity of a loading contactwith an AFM force curve, as shown in the figures 4(c) and(d). In the present case, the loading elastic compliances are∼70 and 480 nm V−1 for the region outside and inside thecrack, respectively. Alternatively, EDX spot analysis or SEMcharging was also used to detect the bare PDMS substrate.

Figure 5 shows 80 μm AFM scans of the un-patternedand patterned samples. Percentage increases in surface lengthwere measured in the two perpendicular directions. For thepatterned sample, these measurements were carried out acrossthe patterns (εpp

x and εppy ) and in the intermediate, i.e. un-

patterned, film regions (εupx and ε

upy ), as shown in figure 5.

For the un-patterned sample, the ε value was calculated,firstly, along the x- and y-axes (εuu

x and εuuy ). It was also

calculated as εb from the calculated surface area difference,using the following formula:

εb = 1

100

(√Asurf

Aproj− 1

), (1)

where Asurf is the area of the surface and Aproj is its projectiononto the x-y plane. Both areas can be obtained from thestatistical analysis of the AFM images. This latter value εb

represents the biaxial strain relief. For the patterned samples,the same procedure was adopted; however, the εx and εy

values were calculated for two series of cross-sections: onepositioned through the pattern ε

ppx and ε

ppy , and one positioned

in the area between the patterns (εupx and ε

upy ). These ε

values are not measured in-built strains which may haveother components due to thermal stress on cooling and other

deposition stresses. Nonetheless, they are important indicatorsof the efficiency of the pattern to protect the film against tensilestress. In effect, these ε values represent the tensile strain reliefwhen the film is subjected to an applied tensile strain ε, as thiswould only induce bending stresses.

An equivalent biaxial value (εb) is defined as(ε2x + ε2

y

)1/2.

This analysis shows that there is a significant differencebetween the εb values of the two samples: 0.5% forthe un-patterned sample, 1.2% on the patterned area and0.9% on the un-patterned area for the patterned sample.The patterning seems to constrain the gold film, producingincreased wrinkling in the areas adjacent to the patterns.Overall, this analysis indicates that the patterning brings 0.4–0.7% of tensile strain compensation. As failure strains forgold are often measured below 1% [13], this could representa significant improvement. However, the contribution fromsurface chemistry cannot be ruled out. During cure, the PDMSinteracts with the ambient (air) and the mould surfaces (glass,exposed SU-8 and unexposed SU-8) and these effects maymodify its surface stoichiometry and functionalities. This isinvestigated in the following section.

The XPS spectra (C 1s, O 1s and Si 2p peaks) wereobtained for four PDMS surfaces, cured on glass, the airsurface, exposed SU-8 (cross-linked) and an unexposed SU-8.Table 1 gathers the results of the peak fitting together withthe measured contact angles, displayed in square brackets.Looking at the atomic percentages, one notes Si deficientstoichiometries, probably because a hydrophobic materiallike PDMS is easily contaminated by hydrocarbons from theambient. In that regard, one notes that the surface exposed to

6

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

Table 1. Results from the XPS analysis showing the binding energy (BE in eV), full width at half maximum (FWHM in eV) and area (in %)of the various peak components for the C1s, O1s and Si2p peaks. The atomic percentages are also shown (in % ± 1%). The values in squarebrackets are the contact angles measured in degrees.

Contact material C O Si

[contact angle value] C O Si BE FWHM Area% BE FWHM Area% BE FWHM Area%

Glass [91 ± 5] 48 30 22 284.4 1.0 75 532.3 1.2 77 101.8 1.2 53285.0 1.5 25 533.0 1.7 23 103.1 1.8 47

Air [125 ± 1] 45 34 21 284.4 1.0 70 532.4 1.2 72 101.8 1.3 55284.9 1.3 30 532.9 1.6 28 103.2 1.6 45

Exposed SU-8 [104 ± 2] 47 31 22 284.4 0.9 61 532.4 1.2 76 101.8 1.2 50284.9 1.3 39 533.3 2.0 24 103.1 1.8 50

Unexposed SU-8 [115 ± 2] 46 30 24 284.4 0.8 44 532.3 1.1 51 101.9 1.2 53284.9 1.2 56 532.7 1.4 49 103.2 1.7 47

air for the full duration of the cure (24 h) shows the smallest Sipercentage, despite prolonged pumping to ultra-high vacuumand, incidentally, also the highest contact angle. However,the compositions are very similar for the four samples. Thetable also displays the deconvolutions of these XPS peaks.The pristine siloxane group can be identified by the main C1speak at 284.4 eV, assigned to the H3–C–Si–O carbon atoms,the main O1s peak at 532.2–532.3 eV, assigned to the Si–O–Si oxygen atoms and the Si2p component around 101.8 eVassigned to (C)2–Si–(O)2 silicon atoms. The samples curedon the un-exposed SU-8 surfaces show higher proportion ofcarbon atoms with slightly higher binding energy (BE). Thischemical shift (<1eV) is not large enough to correspondto an oxygen functionality (C–O or C=O, usually above286 eV) and may be indicative of more C–C or C–H bondsthan C–Si bonds. This sample also displays a small changein the O1s spectrum, again with a larger high BE component.This increase in BE (<1 eV) is too small to be assigned to C–Oor C=O functionalities. Moreover, the Si2p peak shows nodiscernable changes. Overall, these results indicate that PDMSsamples peeled from the unexposed SU-8 display very slightdifferences in surface chemistry with respect to the other threesamples. It should be emphasized that these are very smallvariations which are at the limit of the accuracy of the fittingroutine. On a polymer like PDMS, the XPS sampling depthis between 5 and 10 nm, hence not particularly sensitive tochange in surface chemistry as measured, for instance, bycontact angle analysis. Indeed, table 1 indicates that thePDMS peeled from the un-exposed SU-8 has a larger contactangle (115◦ ± 2◦) than PDMS peeled from either the exposedSU-8 (104◦ ± 2◦) or the glass (91◦ ± 5◦), the case of thecontaminated air surface having been already discussed. Thiscould indicate that the development of the un-exposed SU-8leaves a residue which is not fully stabilized within the SU-8 network formed by the post-exposure bake and the hardbake. This could have a direct bearing on the adhesion ofthe Ti and gold layers and, hence, on the behaviour of theun-patterned and patterned gold/PDMS samples. In fact,the effect of the patterning may be more significant. Thisis because the photolithographic process was designed toproduce rounded features; across the pattern features, there is acontinuous variation from full exposure to no exposure. Thisprobably leaves intermediate molecular weight compounds,

neither insoluble nor fully washable, even more likely to leavesurface residues on the glass. A possible remediation strategywould be to use dielectric barrier discharge in air [14] toincrease active oxygen functionalities (C–O and C=O) andoptimize adhesion; however, this is beyond the scope of thisinvestigation.

Examining now the result of tensile cycling experiments(0–35% strain) (figure 6), we find that, after 120 cycles, newcracks have appeared on both samples but more so for theun-patterned one. The tensile test was performed on an un-patterned sample and on a patterned sample. In this case,the patterning was done with the SF100 system using SU-8 2000.5. The tensile stress locally orientates the cracksperpendicular to the tensile direction and the ripples parallelto the tensile direction. Indeed, in the patterned sample,the cracks are also shorter and seem to be deflected by thepattern. The corresponding electrical and mechanical dataare shown in figure 7. Ri (0% strain) and Rf (35% strain)are the resistances at the initial and final portion of theloading segment, respectively. Initially the patterned samplehas smaller Ri and Rf values (15 and 641 �) than the un-patterned sample (385 and 1715 �). Hence, the patterningimproves conduction in the metal film. Purely from thegeometrical changes imposed by the 35% tensile strain andthe lateral Poisson’s contraction (ν = 0.45), one would expecta resistance increase of (1 + ε)/(1 − νε)2, that is, Rf ∼ 1.9 Ri .For both samples, Rf � 1.9 Ri . This is because, post-deposition, these micro-cracked films are under compression,and hence there are electrical contact points and lines betweengold ligaments, as discussed before (figure 4). These ligamentsare the small gold islands, delimited by cracks, which remainconductive. Generally, these ligaments are perpendicular tothe tensile direction. In the previous study [9], statisticalanalysis of SEM images was used to show that regularcrack features are observed on gold/PDMS films, suggestingthat the size and characteristics of these film ligaments arealso reproducible, for given preparation and applied strainconditions.

During tensile loading, the applied strain is much largerthan the strain relief induced by patterning (<1%); here thefilms are not only bent but also stretched and this results inout-of-plane deflection of gold ligaments and crack wideningas described by Mandlik et al [15]. The contacts across the

7

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

(a) (b)

(c) (d )

Figure 6. (a) SEM of the un-patterned sample after tensile test, (b) SEM of the patterned sample after tensile, (c) AFM of the un-patternedsample after tensile test, (d) AFM of the patterned sample after tensile test. The black arrows show the tensile direction, the white onesinterruptions in the crack due to the pattern features.

cracks disappear and conduction relies only on current flowbetween ligaments, hence the dramatic increase in resistancefrom Ri to Rf .

As the tensile experiments progress, the R(ε) trendchanges markedly for the un-patterned sample; after 12 cycles,the resistance of the sample does not recover to the originalvalue and this Ri value progressively increases through thetensile cycles. This effect is much smaller for the patternedsample. From cycle number 1 to 120, the �Ri/Ri signal is270% and 27% for the un-patterned and patterned samples,respectively. In fact, figure 7 demonstrates that the patternedsample resists remarkably well this cycling experiment withvery little change in behaviour. Hence the patterning alsoimproves tensile resistance. This observation is consistent withthe crack patterns seen in figure 6. At completion, the tensileexperiment left the un-patterned sample with resistances ofRi = 1425 � and Rf = 1744 � and the patterned samplewith Ri = 19 � and Rf = 638 �. The increase in theresistance with the applied strain corresponds to the crackdevelopment. The un-patterned samples do not recover tothe initial resistance value due to the permanent opening ofthe cracks. The crack density and dimension are higher

Table 2. Resistivity values calculated for the patterned andun-patterned samples.

Sample ρ i (� cm) ρf (� cm)

Un-patterned cycle 1 1.16E-03 2.57E-03Un-patterned cycle 120 4.28E-03 2.62E-03Patterned cycle 1 4.50E-05 9.62E-04Patterned cycle 120 5.70E-05 9.57E-04

for the un-patterned sample (figure 6(b)). The equivalentresistivity values for the un-patterned and patterned samplesare presented in table 2. In these calculations, the full samplewidth was 20 mm, ρi = Ri ·w·t/L and ρf = Rf ·w·t/L, with w

being much larger than the 2 mm electrode width. In practice,the current lines probably do not fill the full width but followa circular area; hence our resistivity values are probably over-estimated. Despite this, the equivalent ohmic resistivity of theinitial patterned surface (4.5 × 10−5 �·cm) is in line with therecent literature values, for instance, Graz et al [16]. This isencouraging and this result seems to be partly due to the abilityof the pattern to deflect cracks.

8

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

0

500

1000

1500

2000

0 20 40 60 80 100 120

cycle number

resis

tan

ce

, o

hm

0

0.1

0.2

0.3

0.4

stra

in

0

500

1000

1500

2000

0 20 40 60 80 100 120

cycle number

resis

tan

ce

, o

hm

0.0

0.1

0.2

0.3

0.4

stra

in

0.0

0.2

0.4

0

1000

2000

114 117 120

0.0

0.2

0.4

0

1000

2000

0 3 6

a b

0

0.2

0.4

0

400

800

114 117 120

0

0.2

0.4

0

400

800

0 3 6

ba

Figure 7. Variation of the resistance during the tensile cycles for: top un-patterned sample, bottom patterned sample (pink—the strain,blue—resistance). Insets: (a) detail of the cycles 0–6, (b) detail of the cycles 114–120.

3.2. EVG 620 chromium glass processing

To further exploit this effect, patterned surfaces were preparedusing the EVG 620 mask aligner system with a high-resolutionphotomask and a resist of intermediate viscosity (SU-8 3025).This gave patterns with higher resolution and aspect ratioyet without the detrimental cracked edge effect observed infigure 4 for SU-8 50. Figure 8 shows AFM and opticalmicrographs of the mask as well as AFM and SEMmicrographs of the resulting PDMS egg-box structure.Figure 9 shows the resulting gold/PDMS pattern. The AFMcross-sectional profile gives strain relief ε ∼ 9.45% and anaspect ratio λ/A = 3. Again, the pattern protects the film withmany fewer cracks on the patterned area than on the un-patterned one (figure 9(b)). Moreover, the cracks seen in thepatterned area are clearly deflected by the features (figures 9(d)and 6(b)). In this sample, no delamination were observed. Adelamination is defined as a total removal of the metallic filmover a certain area, for instance, as shown in the bottom insetof figure 4(a) for the un-patterned sample. The increased cracklength seen in figure 9(d) corresponds to a toughening effect.Mandlik et al [15] showed that nano-patterning of a gold layeron PDMS decreased the electrical resistance of the stretchedsample. He also found that this effect was stronger whenthe pattern was misaligned with the tensile direction. Theseauthors provide no explanation for the observed effect. In lightof what is seen in figure 9, we believe that the effect observed

(a) (b)

(c) (d )

Figure 8. SU-8 3025 process. Photomask: (a) AFM and (b) opticalimage. PDMS cast on the SU-8 egg box: (c) AFM and (d) glancingSEM images.

by Mandlik [15] is also due to crack deflection by patternfeatures; with an aligned pattern, the cracks can extend over

9

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

(a)

(b)

(c) (d )

Figure 9. (a) AFM of the SU-8 3025 after exposure and profile of the features, (b) optical image of the patterned and un-patternedAU/PDMS sample, (c) SEM of the Au/PDMS, (d) SEM Au/PDMS sample showing deflection of the crack by the pattern.

-14

-9

-4

1

0 0.2 0.4 0.6 0.8 1pressure, kPa

ΔZ/Z

, %

1.E-02

1.E+00

1.E+02

1.E+04

0 5 10 15 20pressure, kPa

impe

danc

e m

odul

us,

k Ω

patternedun-patterned

-14

-9

-4

1

ΔZ/Z

, %

1.E-02

1.E+00

1.E+02

1.E+04

0 5 10 15 20pressure, kPa

impe

danc

e m

odul

us,

k Ω

patternedun-patterned

(a) (b)

Figure 10. (a) Impedance modulus variation with the applied pressure for the patterned and un-patterned bottom electrodes and(b) impedance modulus of the capacitive patterned device in the sub-kPa regime.

large lengths, while with a misaligned pattern, they meet thepattern feature and hence are shorter. A chequer pattern wouldrepresent an improvement; an optimal pattern would be onewith randomly oriented features of high aspect ratio in order toslow down cracks coming from any random directions.

3.3. The pressure sensor

These promising results were exploited by using the techniqueto prepare prototype conformal pressure sensors, as describedabove and as seen in figure 2: un-patterned (U), patterned onSU-8 3025 without embedding (P) and patterned on SU-8 3025with embedding (PE). The testing was carried out by placingweights on top of the device, thus resulting in compressionand stretching of the device.

At the macro-level, these devices are thin plate capacitorswith high aspect ratio (length/thickness), hence displayingminimal fringing effects. At the micron level, the gold flakes

and thickness are both of similar magnitude (low aspect ratio);however the small width of the cracks (sub-micron) meansthat each gold flake acts as a guard electrode [16] for itsneighbours and, therefore, in this case again, the devices act asthin plate capacitors. Firstly, it was found that the three devices(U, P and PE) behaved as expected with electrical continuity inthe top and bottom electrodes and pure capacitive behaviourfor the sensor. Moreover, these sensors were mechanicallyresilient; the electrical characteristics of the electrodes anddevice were the same before and after pressurization. Theelectric behaviour of the bottom electrodes is shown infigure 10(a). The pattern protects the bottom electrode(P sample), its impedance modulus remains below 1 k� forthe entire pressure range whereas that of the un-patternedbottom electrode (U sample) rises to 5.5 M� above 2 kPa.The Nyquist (Z′ versus Z′′) and Bode plots (frequency versus/Z/) (figure 11) also show that this corresponds to an increaseof the capacitive impedance whereas the patterned electrode

10

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

Figure 11. Behaviour of the bottom conductor of the patterned sample (left) and the un-patterned sample (right) at 0.25 and 9 kPa.

(figure 11) remains resistive up to 30 kPa. Figure 11 shows thechange in phase from 0 (corresponding to the resistive regime)to −100 (corresponding to the capacitive regime) for the un-patterned sample, whereas, in the case of the patterned sample,the phase remains close to zero upon compression. TheNyquist plots give corroborating results; the 9 kPa compressedun-patterned sample displays a Nyquist graph parallel with theY-axis in the negative region which corresponds to a capacitivebehaviour. This does not happen for the patterned sample.These results are most likely due to the opening of cracksin the gold film which becomes discontinuous [17]. The topelectrode is significantly protected by the embedded PDMSlayer. Around 1 kPa, the impedance of the U and P topelectrodes becomes capacitive and rises sharply to ∼2 M�.For the PE top electrode, the impedance remains resistive,in the k� regime until 6 kPa. These sharp increases ofimpedance are caused by the friction between the weightand the gold surface and the lateral Poisson stresses resultingfrom the elastomer, as experienced, for instance by bondedrubber discs [18]. This combined action produces tensile andshear stresses on the gold surface and possibly the twisting ofgold ligaments. The top PDMS layer obviously mitigates thiseffect.

Figure 10(b) also shows the impedance modulus of thePE capacitive device (top to bottom contacts) in the sub-kPa regime; it displays a linear behaviour with small relativeerrors (<1%). However, at higher pressure the behaviour isnonlinear, possibly due to gold ligament shearing in the topelectrode as discussed above. In that sense, the sensor is partlypiezo-resistive. Despite this, the sub-kPa pressure sensitivity(�Z/Z)/P is 0.14 kPa−1 and compares well with sensitivityvalues published in many previous studies: 0.05 kPa−1 [19],0.008 kPa−1 [20], 0.005 kPa−1 [21], 0.02 kPa−1 [22]. Arecent study [23] of the capacitive pressure response of a

microstructured PDMS film gave a larger sensitivity value of0.55 kPa−1; however, here again the behaviour was nonlinearand the device contained a thermoplastic top electrode, aPET substrate and hence was not stretchable. Indeed, theprototype prepared in this study is the only truly stretchabledevice operating in this low-pressure regime and should permitthe elaboration of conformal pressure sensors, if tested athigher pressure than probed in this study. This potentialconformality represents a real advantage for a number ofbioengineering applications such as the elaboration of artificialskin or interfacial measurement systems for the early detectionof pressure ulcers [24].

4. Conclusion

The use of photolithographic 3D surface structuring has beeninvestigated to prepare resilient stretchable conductors andpressure sensors based on the gold/PDMS system. Therationale for such a process is to increase the specific surfacearea of the metal film and hence bring bi-axial tensile strainrelief and improve tensile behaviour. Microscopic analysisshows fewer cracks on the patterned surfaces with patternfeatures deflecting the cracks in the gold film. An interestingdevelopment, beyond the scope of this study, would be toprepare randomly oriented patterns with high aspect ratiofeatures, which would toughen the film in any random biaxialdirection. This study also demonstrates that the patternsprotect the films against repeated tensile cycling, althoughboth samples remained conducting after the completion of120 cycles. The extrapolated resistivity value of the patternedsample (4.5 × 10−5 � cm) compares well with previouslypublished data. However, XPS and contact angle analysesindicate that the patterning process slightly modifies the

11

J. Micromech. Microeng. 21 (2011) 115010 C L Tuinea-Bobe et al

surface chemistry. This method was used to prepare capacitivestrain gauges with pressure sensitivity of 0.14 kPa−1 in thesub-kPa regime. Such stretchable and potentially conformallow-pressure sensors have not been produced before and couldprove advantageous for smart fabric and sensing applications.

Acknowledgment

The main author would like to acknowledge the EU forfinancial assistance through the Marie Curie WARTHEtraining network.

References

[1] Brady S, Diamond D and Lau K T 2005 Inherently conductingpolymer modified polyurethane smart foam for pressuresensing Sensors Actuators A 119 398–404

[2] Shiga T, Akane Okada A and Kurauchi T 1993Electroviscoelastic effect of polymer blends consisting ofsilicone elastomer and semiconducting polymer particlesMacromolecules 26 6958–63

[3] Lacour S P, Li T, Chan D, Wagner S and Suo Z 2006Mechanisms of reversible stretchability of thin metal filmson elastomeric substrates Appl. Phys. Lett. 88 204103

[4] Lacour S P, Wagner S, Huang Z and Suo Z 2003 Stretchablegold conductors on elastomeric substrates Appl. Phys. Lett.82 2404–6

[5] Lacour S P, Jones J, Wagner S, Li T and Suo Z 2005Stretchable interconnects for elastic electronic surfacesProc. IEEE 93 1459–67

[6] Gonzalez M, Axisa F, Vanden Bulcke M, Brosteaux D,Vandevelde B and Vanfleteren J 2008 Design of metalinterconnects for stretchable electronic circuitsMicroelectron. Reliab. 48 825–32

[7] Wagner S, Lacour S P, Jones J, Hsu P I, Sturm J C, Li Tand Suo Z 2004 Electronic skin: architecture andcomponents Physica 25 326–34

[8] Tuinea-Bobe C L, McGourty G, Wallace E, Gehin C, Dixon Dand Lemoine P 2009 A stretchable and conformal pressuresensor operating at sub-kPa pressure Flex-StretchElectronics II, 2nd Int. Workshop on Flexible andStretchable Electronics

[9] Manzoor M U, Tuinea-Bobe C L, McKavanagh F, Byrne C P,Dixon D, Maguire P D and Lemoine P 2011 Amorphouscarbon interlayers for gold on elastomer stretchableconductors J. Phys. D: Appl. Phys. 44 245301

[10] Lemoine P, Quinn J P, Maguire P D and McLaughlin J A D2006 Measuring the thickness of ultra-thin diamond-likecarbon films Carbon 44 2617–24

[11] Bertocci G E, Szobota S, Ha D R and Van Roosmalen L 2000Development of frontal impact crashworthy wheelchairseating design criteria using computer simulationJ. Rehabil. Res. Dev. 37 565–72

[12] Bowden N, Brittain S, Evans A G, Hutchinson J Wand Whitesides G M 1998 Spontaneous formation ofordered structures in thin films of metals supported on anelastomeric polymer Nature 393 146–9

[13] Espinosa H D and Prorok BC 2003 Size effects on themechanical behavior of gold thin films J. Mater. Sci.38 4125–8

[14] Morent R et al 2007 Adhesion enhancement by a dielectricbarrier discharge of PDMS used for flexible and stretchableelectronics J. Phys. D: Appl. Phys. 40 7392–401

[15] Mandlik P and Lacour S P 2006 Fully elastic interconnects onnanopatterned elastomeric substrates IEEE Electron DeviceLett. 27 650–2

[16] Graz I M, Cotton D P J and Lacour S P 2009 Extended cyclicuniaxial loading of stretchable gold thin-films onelastomeric substrates Appl. Phys. Lett. 94 071902

[17] Greason W D 2004 On the development of passive/activeguard electrode systems for human body electrostaticdischarge (ESD) control J. Electrost. 62 155–66

[18] Pinarbasi S, Mengi Y and Akyuz U 2008 Compression of solidand annular circular discs bonded to rigid surfaces Int. J.Solids Struct. 45 4543–61

[19] Someya T, Sekitani T, Iba S, Kato Y, Kawaguchi Hand Sakurai T 2004 A large-area, flexible pressure sensormatrix with organic field-effect transistors for artificial skinapplications Proc. Natl Acad. Sci. 101 9966–70

[20] Shimojo M, Namiki A, Ishikawa M, Makino R and Mabuchi K2004 A tactile sensor sheet using pressure conductiverubber with electrical-wires stitched method IEEE SensorsJ. 4 589–96

[21] Lee H K, Chang S I and Yoon E 2006 A flexible polymertactile sensor: fabrication and modular expandability forlarge area deployment J. Microelectromech. Syst. 15 1681–6

[22] Shirinov A V and Schomburg W K 2008 Pressure sensor froma PVDF film Sensors Actuators A 142 48–55

[23] Mannsfeld S C B, Tee B C K, Stoltenberg R M, Chen C V HH, Barman S, Muir B V O, Sokolov A N, Reese C andBao Z 2010 Highly sensitive flexible pressure sensors withmicrostructured rubber dielectric layers Nat. Mater. 9 859–64

[24] Meffre R, Gehin C and Dittmar A 2007 MAPI: active interfacepressure sensor integrated into a seat Conf. Proc. IEEEEngineering in Medicine and Biology Society pp 1358–61

12