Soft Matter

PAPER

Publ

ishe

d on

29

Oct

ober

201

3. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

03/0

3/20

14 1

8:11

:06.

View Article OnlineView Journal | View Issue

Department of Materials Science and Engin

Walnut Street, Philadelphia, PA 19104, USA

+1 (215) 573-2128; Tel: +1 (215) 898 9645

† Electronic supplementary informa10.1039/c3sm51640g

Cite this: Soft Matter, 2014, 10, 1392

Received 14th June 2013Accepted 4th October 2013

DOI: 10.1039/c3sm51640g

www.rsc.org/softmatter

1392 | Soft Matter, 2014, 10, 1392–1399

Buckling, symmetry breaking, and cavitation inperiodically micro-structured hydrogelmembranes†

Gaoxiang Wu, Yu Xia and Shu Yang*

We investigated swelling induced instabilities in porous membranes with a square array of micron-sized

circular holes prepared from a pH and temperature dual-responsive hydrogel, poly(2-hydroxyethyl

methacrylate-co-N-isopropylacrylamide-co-acrylic acid) (PHEMA-co-PNIPAAm-co-PAA). At room

temperature (25 �C), the hydrogel swelled to �1.5 to 8 times of its dried volume when pH was increased

from 2 to 7. Within this regime, we observed four distinctive morphologies of the hydrogel membrane,

including a “breathing” mode of the membrane having circular pore arrays, a buckled pore array of

alternating mutually orthogonal ellipses, twisted snap-shut pores forming “S” shaped slits, and cusps formed

in local regions that perturbed the 2D periodicity of the hydrogel membrane. Using a 3D confocal imaging

technique, we followed the post-buckling behaviors of the porous membranes and investigated the pattern

evolution process as a function of pH. Amplification of buckling and symmetry breaking were observed

when we increased the pH of the buffer solutions from pH 4.0 to 5.0, leading to the transition from an

achiral buckled state (pH 4.0) to a chiral twisted state (pH 5.0) driven by the compaction of the hydrogel

domains within the space to completely close the pores. When the pH of the aqueous environment was

further increased to 7, star-shaped patterns appeared randomly in the film, where the hydrogel domains

were compressed by the adjacent neighbors, thus resulting in out-of-plane deformation. Finally, we

demonstrated the temperature-dependent reversible switching of the hydrogel membrane among the

chiral twisted state, buckled state, and circular state via changing the temperature between 20 �C and 45 �C.

Introduction

So materials that can recongure their shape and structure inresponse to external stimuli are of great interest for potentialapplications inminiaturizeddevices,1–3micro-actuation,3–6opticalmanipulation,7,8 microuidic devices,9 and drug delivery.10 Thedesign of so materials with reversible, fast, and controllabledeformation present scientic and engineering challenges.

Mechanical instabilities in so materials, oen accompaniedby large deformation and shape change in response to relativelysmall external perturbations, provide new opportunities to designrecongurable materials with dramatic change of physical prop-erties. Recently, we and others have demonstrated spontaneouspattern transformation in periodic porous structures triggered byswelling, drying, polymerization and mechanical compressiontogether with shape memory effects over multiple lengthscales.11�24 The resulting photonic25,26 and phononic24 bandgapproperties, and mechanical behaviors (e.g. negative Poisson's

eering, University of Pennsylvania, 3231

. E-mail: shuyang@seas.upenn.edu; Fax:

tion (ESI) available. See DOI:

ratio effect17,21) are signicantly altereddue to the change of latticesymmetry, pore size or shape, and the volume lling fraction.

Among different approaches to induce instabilities in peri-odic porous structures, swelling induced buckling13,15,16,24 is ofparticular interest because of its simplicity, reversibility andversatility to generate a variety of recongurable structures indifferentmaterial systems from sub-micron tomillimeter scales.It is known that the choice of solvent and elasticity of the polymermembrane together with geometric features of the pattern (e.g.pore size, spacing and symmetry) will determine the volumechange, and the buckling behavior of the membrane, and thus,control the emergent patterns at equilibrium. So far, moststudies reported in the literature have focused on achieving asingle equilibrium state of the buckled structure by choosing asolvent with desired solvent–polymer network interactionparameter. However, many materials could be swollen to a largeextent and the swelling ratio can be ne-tuned continuously, forexample, by changing thepHor temperature of the environment.As a result, questions arise whether the morphology of thedeformed so membrane can be varied continuously by ne-tuning the swelling ratio of the polymer network. More speci-cally, how the swelling induced pattern transformation occurs,and whether the transformed pattern can be amplied andevolved into new and more complex morphologies.

This journal is © The Royal Society of Chemistry 2014

Paper Soft Matter

Publ

ishe

d on

29

Oct

ober

201

3. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

03/0

3/20

14 1

8:11

:06.

View Article Online

As a three-dimensional (3D) hydrophilic network of poly-mers, hydrogels offer a unique system to study swelling inducedpattern transformation andmorphology evolution. The swellingratio and the equilibrium state of the hydrogel pattern areresults of balancing the osmotic pressure generated fromexpansion and the elasticity of the polymer network, which canbe tailored by the chemistry and composition of the hydrogelnetworks, as well as external stimuli, including the choice ofsolvent,2 pH,27 temperature,28 and light.29 The poly(N-isopropylacrylamide) (PNIPAAm) hydrogel has a lower critical solutiontemperature (LCST)30 at �32 �C. Below the LCST, the gel isswollen, and above the LCST, the gel collapses out of water.When incorporating acid groups [e.g. carboxylic groups frompoly(acrylic acid) (PAA)] into the PNIPAAm network, the swellingratio and thus the pattern deformation behaviors could be ne-tuned by both temperature and pH. Such dual-responsivehydrogel networks have been shown to have a wide range oftunable swelling ratio (up to 20 times from the collapsednetwork to the fully swollen state).31 Another important factor toconsider is the response time of the hydrogel swelling. Since thecharacteristic swelling rate is inversely proportional to thesquare of the linear gel size,32 microstructured hydrogelmembranes are desired here to study pattern transformation.

Here, we report the swelling-induced buckling instability andthecorrespondingpost-bucklingbehaviors inpHand temperaturedual-responsive hydrogel membranes of micron sized pores in asquare array. The membranes were fabricated from poly(2-hydroxyethyl methacrylate-co-N-isopropylacrylamide-co-acrylicacid) (PHEMA-co-PNIPAAm-co-PAA) (43.7 : 40.8 : 14.5 mol%). Werst studied themorphologychangeof thehydrogelmembranesatdifferent pH values at room temperature (25 �C in our experi-ments). The bulk hydrogel swelled �1.5 to 8.0 times of its driedvolume when increasing pH from 2 to 7. Within this regime, weobserved four distinctive morphologies of the hydrogelmembrane, including breathing mode of the membrane havingcircular pore arrays, buckled pore arrays of alternating mutuallyorthogonal ellipses, twisted snap-shut pores forming “S” shapedslits, and cusps formed in local regions that perturbed the 2Dperiodicityof thehydrogelmembrane.Using3Dconfocal imaging,we followed the pattern evolution process as pH increased. Weobserved a symmetry breaking transition occurred between thebuckled state (pH 4, achiral) and the twisted state (pH 5, chiral),where the latter exhibited a mixed distribution of right- and le-handed chiral structures at pH 5. Further increase of the swellingratio at pH 7 resulted in compression of the adjacent swollenhydrogel domains, leading to the formation of cusps randomlydistributed in the lm. Finally, we demonstrated the temperature-dependent reversible switching of the hydrogelmembrane amongthe chiral twisted state, buckled state, and circular state viachanging the temperature between 20 �C and 45 �C.

ExperimentalSample preparation

A polydimethylsiloxane (PDMS, SYLGARD® 184 kit from DowCorning, prepolymer : crosslinker ¼ 10 : 1 wt/wt) pillar array(diameter D ¼ 10 mm, pitch P ¼ 15 mm, and depth H ¼ 20 mm)

This journal is © The Royal Society of Chemistry 2014

in a square lattice was cured at 65 �C for 3 h and used as thetemplate to fabricate PHEMA-co-PNIPAAm-co-PAA membranes(molar ratio, PHEMA/PNIPAAm/PAA ¼ 43.7/40.8/14.5) byreplica molding following the procedure reported earlier.13

For the hydrogel precursor, 1 g NIPAAm (Sigma-Aldrich) wasdissolved in 1 mL 2-hydroxyethyl methacrylate (HEMA, Sigma-Aldrich), followed by mixing with 200 mL acrylic acid (AA, Sigma-Aldrich) and 30 mL Darocur 1173 (Ciba Specialty Chemicals) asphotoinitiators. The mixture was then exposed to UV light (365nm, 97435 Oriel Flood Exposure Source, Newport Corp.) toobtain a partially polymerized precursor solution. In order toensure the homogeneity of photopolymerization, UV exposurewas carried out in an aliquot of 200 mJ cm�2 three times.Additional 20 mL Darocur 1173 and 40 mL ethylene glycoldimethacrylate (EGDMA, Sigma-Aldrich) were added to theprepolymer solution for the subsequent molding.

The glass substrates were rinsed with acetone, followed byoxygen plasma (Harrick Expanded Plasma Cleaner &PlasmaFlo™) for 1 h. These glass slides were then treated with 3-(trimethoxysilyl)propylmethacrylate (Sigma-Aldrich) via vapordeposition ina vacuumchamber for 2hasanadhesionpromoter.5 mL hydrogel precursor solutionwas cast onto the silanized glasssubstrate, and then thePDMSmoldwasplacedon top to allow theprecursor solution inltrated into the PDMS mold by capillarity.Aer exposure to UV light at a dosage of 1 J cm�2, the PDMSmoldwas peeled off from the hydrogel membrane.

Characterization of the hydrogel swelling

To characterize the swelling ratio of the hydrogel, we preparedthe bulk hydrogel rods using the same hydrogel precursorsolutions to fabricate the membranes in a plastic tube with adiameter of 2 mm. Aer UV curing at 1 J cm�2, the plastic tubewas stripped off and the solidied hydrogel was cut into smallpieces (�5 mm length). These hydrogel rods were immersedinto DI water for three days to remove the unreacted mono-mers and oligomers, and then into the pH buffer solutions ofdifferent pH values: 2, 3, 4, 4.3, 4.7, 5, 5.3, 5.7, 6, 6.4, and 7.The buffer solutions were prepared by mixing potassiumbiphthalate solution (0.05 M, Micro Essential Laboratory) witha certain amount of hydrochloric acid (0.05 M) or sodiumhydroxide (0.05 M) to achieve pH values of 2 to 7. The pHvalues of the prepared buffer solutions were conrmed using apH meter (Oakton Waterproof pHTestr 30). Aer swelling foranother three days to ensure the hydrogel rods reaching theequilibrium states, the weight (Ws) was recorded. The hydrogelrods were then dried in an oven at 65 �C for three days toobtain their dry weight (Wd). The mass swelling ratio wascalculated as:

QM ¼ Ws

Wd

(1)

The volumetric swelling ratio of the hydrogel was obtainedas:

QV ¼ ðQM � 1ÞrPrW

þ 1 (2)

Soft Matter, 2014, 10, 1392–1399 | 1393

Soft Matter Paper

Publ

ishe

d on

29

Oct

ober

201

3. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

03/0

3/20

14 1

8:11

:06.

View Article Online

where rP is the density of the polymer network and rW is thedensity of water. Since the value of rP is difficult to be directlymeasured, we estimated rP ¼ 1.02, based on the average densityof the monomers and the crosslinkers used in the hydrogelprecursor, assuming that there was only a minor volume changeduring the polymerization process. The measured pH depen-dent volumetric swelling ratio of the hydrogel is shown inFig. S1a.†

In a similar manner, we tested the thermal responsiveness ofthe hydrogel. In this case, the hydrogel rods were placed into pH4, 5, and 6 buffer solutions to reach equilibrium, and thenplaced into a water bath with a controlled temperature of 20, 25,30, 35, 40, and 45 �C, respectively. At each temperature, thehydrogel rods were equilibrated for at least 2 days, and theweight was measured. The mass and the volumetric swellingratio were calculated using the above equations. The hydrogelvolumetric swelling ratio as a function of temperature is shownin Fig. S1b.†

Observation of the morphology change of the hydrogelmembrane at different pH and temperature

To investigate the swelling induced buckling instability and thecorresponding post-buckling behaviors, the hydrogelmembranes were immersed into the buffer solutions ofdifferent pH values overnight before taking images under anoptical microscope (Olympus BX61 motorized microscope). Forrecording the transformation process, the samples were coveredwith cover glasses with a small amount of buffer solutiontrapped underneath, and were heated on a small hotplate(Mettler Toledo FP82HT Hotstage), which was tted into themicroscope, from 20 to 45 �C at a heating rate of 10 �C min�1.To image the equilibrium morphologies of the hydrogelmembranes at 20, 30 and 40 �C, the samples were heated insidea plastic container (Warner Instrument Corporation, Micro-Incubation System) with buffer solution under a microscope,which has a feedback loop that controlled the buffer solutiontemperature inside with an accuracy of �1 �C. In this case, thehydrogel membrane was equilibrated until there was noobvious structural change, before images were taken.

Imaging of the hydrogel membranes

To better observe the structural change of hydrogel membranesat different pH values and temperatures, we copolymerized asmall portion (0.05 wt%) of rhodamine-B (RB, Sigma-Aldrich)attached HEMA monomers into the hydrogel network, allowingfor the direct imaging using confocal microscopy. The attach-ment of RB to the HEMA monomers was based on the esteri-cation reaction between the carboxylic (–COOH) groups on RBand the hydroxyl (–OH) groups on HEMA following similarreactions reported in the literature.33 In brief, 1 g RB (2.26mmol), 1 ml HEMA (8.24 mmol), 0.512 g dicyclohexyl-carbodiimide (DCC, 2.49 mmol), and 0.304 g 4-dimethylami-nopyridine (DMAP, 2.49 mmol) were dissolved in 50 mldichloromethane. The reaction mixture was stirred at roomtemperature (i.e., 25 �C in all our experiments) overnight forreaction to complete. Aer the insoluble salts that formed

1394 | Soft Matter, 2014, 10, 1392–1399

during the reaction were removed by ltration, the product wasconcentrated and further puried by silica gel column chro-matography using an ethyl acetate–isopropanol mixture.

The morphologies of the uorescence labeled samples werecaptured under a laser scanning confocal microscope (Leica SP5Spectral Imaging Confocal) with a 20� water immersing lens(HCX APO L20�/1.00W; working distance ¼ 2 mm). The greenlaser of 543 nm wavelength was used to excite the uorescenceof the RB. The images taken by the microscope were then post-processed by using commercial soware (Volocity 6.2.1).

The 3-D reconstruction was obtained by carrying out a depthprole in z-scans at a step of 0.45 mm by confocal microscopy.The 3D model was then reconstructed using the visualizationpackage in the soware (Volocity 6.2.1). For the hydrogelmembranes at pH 2 and pH 4, the reconstructed structures areshown in Fig. S2.† The 3-D image of individual hydrogeldomains was obtained by cropping the 3-D reconstruction ofthe hydrogel membranes.

The hydrogel membranes in the dry state were imaged byscanning electron microscopy (SEM) using an FEI Strata DB235Focused Ion Beam in high vacuum mode with an accelerationvoltage of 5 kV.

Results and discussion

We are interested in swelling-induced recongurations of thehydrogel membranes, specically, how the deformed patternsevolve in response to the change of swelling ratio triggered by theexternal stimuli, such as pH and temperature. PHEMA basedhydrogels were chosen in our experiments because of the highmechanical strength in the dry state, which allows for faithfulreplica molding of micron sized features, and mechanicalstability in water, which is crucial for studies that require a largedegree of swelling without losing the integrity of the membranestructures. However, the pure PHEMA hydrogel does not swellmuch in water: the equilibrium linear expansion (ae) is �1.5when using 1 wt% ethyleneglycol dimethacrylate (EGDMA) as acrosslinker, and the degree of swelling is not tunable in responseto external stimuli.34 Previously, we fabricated the PHEMA andPHEMA-co-PNIPAAm membranes (pore diameter D ¼ 1 mm,pitch P ¼ 2 mm, and depth H ¼ 4 mm) in a square lattice. Duringswelling, the corresponding hydrogel membrane underwent a“breathing” mode deformation, where the pore size shrankslightly but remained circular in shape.20 Pattern transformationoccurred during drying driven by capillary forces. Here, PAA wasintroduced to the thermoresponsive PHEMA-co-PNIPAAmnetwork to increase the swellability of the hydrogel due to thehydrophilic nature of PAA andmake the network pH responsive.The porous hydrogel membranes with periodic circular holeswere fabricated by replica molding of the hydrogel precursorcomposedofHEMA,AA, andNIPAAmasmonomers, EGDMAasacrosslinker in a molar ratio of HEMA/NIPAAm/AA/EGDMA ¼43.7 : 40.8 : 14.5 : 1 from a PDMSmold (Fig. 1a–b), following theprocedure described in the literature.35 Fig. 1c shows SEMimages of the hydrogel membrane in the square lattice of thecircular pores with a pore diameter of 10 mm, inter-pore spacingof 15 mm and depth of 20 mm.

This journal is © The Royal Society of Chemistry 2014

Fig. 1 (a) Schematic illustration of the fabrication of the hydrogelmembrane by replica molding. (b) Chemical structures of the hydrogelprecursors composed of monomers, including HEMA, NIPAAm, AA,and crosslinker, and EGDMA. (c) SEM images of the fabricated hydrogelmembrane. Left: top view. Right: cross-sectional view.

Paper Soft Matter

Publ

ishe

d on

29

Oct

ober

201

3. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

03/0

3/20

14 1

8:11

:06.

View Article Online

When placed in buffer solution, the hydrogel swelled, andthe degree of swelling depended on both pH and temperature.We rst characterized the volumetric swelling ratio, QV (seedenition in the Experimental section), of the bulk hydrogelfrom pH 2 to 7 at room temperature (Fig. S1a†). The volume ofthe hydrogel increased with the increase of the pH value in thebuffer solution due to the enhanced degree of ionization of PAAin a more basic environment. At pH 2, which was well below thepKa (�4.7) of PAA, the swelling ratio was 1.5. At this pH, the acidgroups remained fully protonated; the polymer network wasrelatively hydrophobic due to the low ionization of the acidgroups, and the formation of the hydrogen bonding betweenthe protonated acids and the amide groups on PNIPAAm.36 Nearthe pKa of PAA in the range of pH 4.5 to 5.5, there was a sharpvolume change of the hydrogel, which was attributed to thelarge degree of ionization change of the weak acid groups. Theswelling ratio QV � 8.0 was obtained at pH 7, when most of theacid groups became dissociated. Further increase of pH did notaffect the swelling ratio due to the saturated ionization ofcarboxylic acid groups. Therefore, the hydrogel swellingbehaviors were not further investigated at pH > 7.

In a similar manner, we characterized the thermal respon-siveness of the hydrogel from 20 �C to 45 �C at pH 4, 5, and 6,respectively (Fig. S1b†). We observed a temperature dependentswelling similar to that from the random copolymers of PNI-PAAm and PAA.31 At pH 6 (>pKa of the PAA), the LCST behaviorof PNIPAAm in the copolymer network disappeared because theionized PAA components generated sufficient hydrophilicity toovercome the aggregation of the hydrophobic PNIPAAm chainswhen heated above the LCST.36 On the other hand, at pH 4 (<pKa

of the PAA), we found that the hydrogel networks exhibited low

This journal is © The Royal Society of Chemistry 2014

swelling capability even at 20 �C due to the low ionization of theacid groups. As a result, here we focused on the thermalswitching behavior of the hydrogel at a pH value near the pKa ofthe PAA (i.e. pH 5), where good swelling capability at lowtemperature (QV � 4.4 at 20 �C) and a clear transition to thecollapsed state at high temperature (QV � 1.4 at 45 �C) can besimultaneously achieved. It is worth noting that the swellingtransition in our hydrogel network was broad (25 �C to 40 �C),which might be attributed to the interruption of the hydrogenbonding due to the randomly copolymerized PNIPAAm withPHEMA and PAA.31 This broad transition is benecial, whichoffers a broad temperature window for ne-tuning the hydrogelswelling and thus the morphologies of the hydrogel membranesin response to the temperature change.

The hydrogel membrane was rst swollen in the pH 2 buffersolution, where the hydrogel exhibited the lowest degree ofswelling (QV � 1.5). As shown in Fig. 2a, the pores shranksomewhat but remained circular aer swelling, so-called“breathing” mode deformation.37 When the swelling ratio wasincreased to �1.7 at pH 4, the circular pores in a square arraywere compressed and buckled into a diamond-plate structure ofelliptical pores with long axis mutually orthogonal to eachother. The transformation was observed through the entiresample (�1 cm2). At the unit cell level, the pattern trans-formation of hydrogel membranes can be related to the rota-tions of the hydrogel domains in clockwise andcounterclockwise directions, and the bending of the inter-connected joints11 as indicated in the magnied optical images(Fig. 2b). We also examined the overall deformation of theindividual hydrogel domain during the transition via 3-Dconfocal microscopy (Fig. S2†). An obvious torsional deectionoccurred, noticed as the helical edges on the hydrogel domainsduring the transition to the buckled state (Fig. 2b).

We believe that increased swelling induced compression atthe joints between neighboring hydrogel domains triggeredsuch dramatic deformation of the hydrogel membrane. As seenin Fig. 2c, in a square lattice, the hydrogel domains diagonallybridge four neighboring holes (Fig 2c). The two adjacentdomains are connected by a thin joint (indicated by the reddashed circle). An isolated individual hydrogel domainundergoes conned swelling, where the bottom of the domainis restricted by the rigid substrate, while the top expands freelyin x, y, and z directions (Fig. 2d). In the case of swelling of thehydrogel membrane, the hydrogel domains are interconnected.Therefore, the lateral expansion of the top layer becomesrestricted by its neighbors, leading to a compressive forceexerted to the joints and the pores. At a low degree of swelling(pH 2), the compressive stress generated at the joint is smallcompared to the critical buckling threshold, thus, pores shrinkslightly, but remain circular.20 When compressive stressgenerated by osmotic pressure is increased above the bucklingthreshold (e.g. at pH 4), the joints are buckled, and the corre-sponding rotation of the hydrogel domains is triggered simul-taneously (Fig. 2e). Once one pore is buckled, the jointsconnecting the swollen hydrogel domains also contribute toexerting the stress to the neighboring joints, resulting in patterntransformation of the entire porous membrane, similar to what

Soft Matter, 2014, 10, 1392–1399 | 1395

Fig. 2 (a) Fluorescent images of the hydrogel membranes after being immersed and equilibrated in pH 2 and pH 4 buffer solutions at roomtemperature. (b) Fluorescent images of a single unit cell of the square lattice, showing detailed structural reconfiguration occurred during thepattern transformation process. (c–e) Illustration of the buckling instabilities in the hydrogel membrane. (c) Hydrogel domains and inter-connected joints. (d) 3D view of confined swelling of an individual hydrogel domain. (e) Structural change of the adjacent hydrogel domain. Theyellow and white arrows indicate the rotation directions of each domain.

Soft Matter Paper

Publ

ishe

d on

29

Oct

ober

201

3. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

03/0

3/20

14 1

8:11

:06.

View Article Online

we have previously observed in the case of swelling of the PDMSmembrane in toluene.13 Due to the strong connement of thehydrogel on the substrate (the bottom of the hydrogel domainscannot rotate), the rotational deformation on the top cannotpropagate vertically through the hydrogel domain. Therefore,gradual torsional deformation occurs to accommodate therotational misalignment between the top and bottom surfaces(Fig. 2e).

As we immersed the hydrogel membrane in a pH 5 buffersolution (QV � 4.3), we observed a secondary structural recon-guration process, which ended up with a state of collapsedpores twisted into an “S” shape (Fig. 3a). It is worth noting thatthis transition involved a symmetry breaking process, where thenal structure of the hydrogel membrane has either right or lehandedness (see insets in Fig. 3a). At the phase boundary ofthese regimes (shown as yellow dashed line), the rotationaldirections of hydrogel domains were not well-dened due to thecompetition from the regimes of opposite handedness.

To understand the morphological transition and why thehydrogel membrane underwent symmetry breaking, we took acloser look at the hydrogel membrane at pH 4.3 (Fig. 3b), anintermediate state (QV � 2.1) between the buckled state at pH 4and the twisted state at pH 5. Here, the deformation to thehydrogel membrane was accentuated (Fig. 3c), which causedthe circular pores being largely compressed to form an “8”shape, whereas the joints in the opposite side of the pore started

1396 | Soft Matter, 2014, 10, 1392–1399

to touch against each other (shown in the inset of Fig. 3b).Further swelling intensied the compression at the touchingjoints, and triggered symmetry breaking at pH 5, as shown atthe unit cell level (Fig. 3d). During this symmetry breakingprocess, the increased swelling of the hydrogel membrane wascompensated by rotating all the hydrogel domains in the samedirection (indicated by the solid arrows). By having such reor-ientation of the surface structure, the two touching joints wereable to slide against each other (see the white dashed arrows),which relieved the compression forces between the touchingjoints.

We also noticed that there were two possibilities of thesymmetry breaking process: clockwise or counterclockwiserotations, which explained the coexistence of both right- andle-handed chirality of the features. Meanwhile, symmetrybreaking could start from different nucleation sites and prop-agate, leading to the formation of phase boundaries where twoindependently evolved regimes of opposite handedness metwith each other.

The symmetry breaking process in the cellular solid has beenrecently reported by Kang et al.24 by swelling induced bucklingof the substrate-attached cellular solids with thin plates. Webelieve that the symmetry breaking process reported here issignicantly different from that reported by Kang et al., eventhough both processes involved the swelling induced buckling.The major difference is that the chirality reported by Kang et al.

This journal is © The Royal Society of Chemistry 2014

Fig. 3 Confocal microscopy images of the hydrogel membranes visualized from different planes (a) xy plane at pH 7. xy- (b) and xz- (c) planes atpH 5 and 7. Dotted lines in (b) correspond to the location where the cross-sectional images (viewed from xz planes) were taken, showing a cuspin the hydrogel film due to compression of the neighboring hydrogel domains. (d) Schematic illustration of creasing instability when increasingthe compressive stress in the swollen hydrogel membrane. The dots on the free surface of the adjacent hydrogel domains move toward center,forming a cusp.

Paper Soft Matter

Publ

ishe

d on

29

Oct

ober

201

3. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

03/0

3/20

14 1

8:11

:06.

View Article Online

was directed by the programed buckling modes, depending onthe plate aspect ratio (length/height), and the symmetrybreaking occurred simultaneously with the triggering of thebuckling instability. In our experiments, the buckling instabil-ities did not invoke the symmetry breaking process immedi-ately; the circular pores were initially buckled into the achiraldiamond-plate structure. Aer the hydrogel was further swollento a larger degree, the chirality was induced in a secondarybifurcation step. Such a two-step reconguration mechanismled to three states of distinctive pattern symmetries, rather thanswitching between the original structures and buckled one. Weexpect that our system can be further modied to generate adefect-free reconguration with single handedness by limitingnucleating sites, for example, wetting the sample from a singlepoint.24 Moreover, since the swelling ratio of the hydrogelmembrane was coupled to both the pH and thermal respon-siveness, later we will show that the chirality can be erased bydeswelling the hydrogel membrane at a temperature aboveLCST.

When the pH of the buffer solution was increased to 7, thecollapsed pores were further compressed by the highly swollenhydrogel, and star-shaped patterns appeared randomly in thelm (Fig. 4a), due to the out-of-plane deformation. A closer lookby confocal imaging at the depth prole of the hydrogelmembrane at pH 5 and pH 7 (see Fig. 4b and c) showed that thehydrogel surface in the cusp region was compressed by the

This journal is © The Royal Society of Chemistry 2014

neighboring hydrogel domains at pH 7 to make an offset in thez-direction in contrast to a rather uniform height in differentswollen regions at pH 5. Solvent swelling induced out-of-planedeformation, i.e. buckling and creasing has been reported by usand others in a thin lm of highly swollen gels conned on arigid substrate.34,38 However, we differentiated this type ofdeformation from wrinkling or creasing due to the fact that themembrane itself is not a continuous piece of gel: there are many“cuts” on the membrane that are the collapsed pores. These“cuts” act as local defects within the gels, where swellinginhomogeneity is amplied in the highly swollen gel, offeringlocal energy minimum by further releasing the in-planecompression between neighbouring hydrogel domains, andthus created out-of-plane deformation of the hydrogelmembrane.

Finally, we demonstrated the reversible switching of thehydrogel membrane between the twisted state and circular holearrays by applying the heating and cooling cycles between 20 �Cand 45 �C at pH 5 (see Movies S1 and S2 in the ESI†). Theswitching time was on the order of tens of seconds due to thefast diffusion of the water and ions across the micron-sizedhydrogel membrane. Since the volume of the hydrogel lmcould be varied continuously by temperature in a largetemperature window, we were able to dynamically tune the poresize, shape, and geometry. As seen in Fig. 5a, the hydrogelmembranes equilibrated at different temperatures exhibited

Soft Matter, 2014, 10, 1392–1399 | 1397

Fig. 4 Confocal microscopy images of the hydrogel membranes visualized from different planes (a) xy plane at pH 7. xy- (b) and xz- (c) planes atpH 5 and 7. Dotted lines in (b) correspond to the location where the cross-sectional images (viewed from xz planes) were taken, showing a cuspin the hydrogel film due to compression of the neighboring hydrogel domains.

Fig. 5 (a) Fluorescent images showing the morphologies of thehydrogel membranes equilibrate at 40 �C, 30 �C, and 20 �C, respec-tively, in pH 5 buffer solutions. (b) The volumetric swelling ratio of thehydrogel measured at 10 heating and cooling cycles between 20 �Cand 45 �C.

Soft Matter Paper

Publ

ishe

d on

29

Oct

ober

201

3. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

03/0

3/20

14 1

8:11

:06.

View Article Online

distinctive morphologies, including the chiral twisted structure(20 �C), diamond-plate structure (30 �C), and the originalcircular pore array (40 �C).

Moreover, we showed that the thermal response of thehydrogel membrane persisted at least 10 heating and coolingcycles, while the pattern transformation behavior did not alteraer multiple cycles, indicating that the temperature triggeredswitching of hydrogel membranes was quite robust. Incomparison, the hydrogel membrane swollen at pH 5 at roomtemperature (the twisted structure) could not be fully recoveredto the original circular pore array when immersed in a pH 2buffer solution even aer the membrane was equilibrated formore than a week. Instead, the membrane was partially recov-ered to the diamond-plate structure. We suspect that there maybe two reasons for the incomplete recovery of the membrane bythe deswelling at a lower pH: (1) the fast dehydration upon

1398 | Soft Matter, 2014, 10, 1392–1399

immersing into the pH 2 buffer might cause dramatic collapseof the hydrogel network, thus kinetically trapping the deformedstate of the hydrogel membrane due to the strong interactions(e.g. hydrogen bonding and hydrophobic interactions) betweenpolymer chains at the collapsed state and (2) a hysteresis that iscommonly observed in the pH responsive hydrogels since PAAbased gels have different swelling and deswelling curves,depending on the pH value and pH gradient.39,40 Here, thehydrogel membrane might not be able to reach the sameequilibrium state when deswelling from pH 5 to pH 2 as it wasinitially transferred to the pH 2 buffer solution.

Conclusion

In this report, we investigated swelling-induced pattern trans-formation and various morphologies of a pH and temperatureresponsive hydrogel membrane. We showed that patterntransformation can be triggered and evolved into differentreconguration states by controlling the swelling ratio of thehydrogel. At pH 2, the hydrogel membrane underwentbreathing mode deformation, and then became buckled into adiamond-plate structure withmutually perpendicular ellipses atpH 4. When further increasing the pH and swelling ratio, weobserved a symmetric breaking transition from pH 4.3 to pH 5,where the achiral buckled state of the hydrogel membrane wastransformed into a chiral twisted state, along with thecompaction of the hydrogel domains within the 2-D space. Sucha spontaneous generation of chirality in the absence of a chiralinduction has a profound impact on the understanding of theorigin of single chirality found in nature, and could lead topotential applications including optical devices, sensors, drugdelivery, and advanced structural components. At an evenhigher degree of swelling, the hydrogel membrane fully occu-pied the 2D space, the excessive swelling of the hydrogel wasfound to be accommodated by forming randomly distributedcusps on the hydrogel surface, due to the out-of-plane

This journal is © The Royal Society of Chemistry 2014

Paper Soft Matter

Publ

ishe

d on

29

Oct

ober

201

3. D

ownl

oade

d by

Uni

vers

ity o

f Pe

nnsy

lvan

ia L

ibra

ries

on

03/0

3/20

14 1

8:11

:06.

View Article Online

deformation. Finally, we demonstrated the temperature-dependent reversible switching of the hydrogel membraneamong the chiral twisted state, buckled state, and circular state.The study demonstrated here using the dual-responsivehydrogel membrane with tunable swelling ratio as a modelsystem to dynamically control the pore size and shape, patterngeometry, and chirality of periodic achiral structures inresponse to the external stimuli will offer new insights into thedesign of the next generation recongurable structures. Mean-while, since the entire transformation process is driven by thetunable swelling of the hydrogel, we expect that such a switch-ing process can be obtained by a wide range of external stimuli(e.g. light, electric eld, and biomolecules) that trigger volumephase transition of the hydrogel materials.

Acknowledgements

The research is funded in part by National Science Foundation(NSF) # CMMI-0900468, EFRI-1038215, and NSF/MRSEC, grant# DMR-1120901. Penn Regional Nanotech Facility (PRNF) isacknowledged for the access of SEM, and Penn Vet ImagingCore Facilities (PVIF) is acknowledged for the access of confocalmicroscopy, and Gordon Ruthel at PVIF for offering kind helpon operating the confocal microscope and soware Volocity.

References

1 X. He, M. Aizenberg, O. Kuksenok, L. D. Zarzar, A. Shastri,A. C. Balazs and J. Aizenberg, Nature, 2012, 487, 214.

2 A. Sidorenko, T. Krupenkin, A. Taylor, P. Fratzl andJ. Aizenberg, Science, 2007, 315, 487.

3 C. L. van Oosten, C. W. M. Bastiaansen and D. J. Broer, Nat.Mater., 2009, 8, 677.

4 P. J. Glazer, J. Leuven, H. An, S. G. Lemay and E. Mendes, Adv.Funct. Mater., 2013, 23, 2964.

5 J. le Digabel, N. Biais, J. Fresnais, J.-F. Berret, P. Hersen andB. Ladoux, Lab Chip, 2011, 11, 2630.

6 Z. L. Wu, M. Moshe, J. Greener, H. Therien-Aubin, Z. Nie,E. Sharon and E. Kumacheva, Nat. Commun., 2013, 4, 1586.

7 L. Dong, A. K. Agarwal, D. J. Beebe and H. Jiang,Nature, 2006,442, 551.

8 N. Shimamoto, Y. Tanaka, H. Mitomo, R. Kawamura, K. Ijiro,K. Sasaki and Y. Osada, Adv. Mater., 2012, 24, 5243.

9 D. J. Beebe, J. S. Moore, J. M. Bauer, Q. Yu, R. H. Liu,C. Devadoss and B.-H. Jo, Nature, 2000, 404, 588.

10 J. D. Ehrick, S. K. Deo, T. W. Browning, L. G. Bachas,M. J. Madou and S. Daunert, Nat. Mater., 2005, 4, 298.

11 T. Mullin, S. Deschanel, K. Bertoldi and M. C. Boyce, Phys.Rev. Lett., 2007, 99, 084301.

12 K. Bertoldi, M. C. Boyce, S. Deschanel, S. M. Prange andT. Mullin, J. Mech. Phys. Solids, 2008, 56, 2642.

This journal is © The Royal Society of Chemistry 2014

13 Y. Zhang, E. A. Matsumoto, A. Peter, P.-C. Lin, R. D. Kamienand S. Yang, Nano Lett., 2008, 8, 1192.

14 S. Singamaneni, K. Bertoldi, S. Chang, J.-H. Jang,E. L. Thomas, M. C. Boyce and V. V. Tsukruk, ACS Appl.Mater. Interfaces, 2008, 1, 42.

15 J.-H. Jang, C. Y. Koh, K. Bertoldi, M. C. Boyce andE. L. Thomas, Nano Lett., 2009, 9, 2113.

16 S. Singamaneni, K. Bertoldi, S. Chang, J.-H. Jang,S. L. Young, E. L. Thomas, M. C. Boyce and V. V. Tsukruk,Adv. Funct. Mater., 2009, 19, 1426.

17 K. Bertoldi, P. M. Reis, S. Willshaw and T. Mullin, Adv.Mater., 2010, 22, 361.

18 F. Goencue, S. Willshaw, J. Shim, J. Cusack, S. Luding,T. Mullin and K. Bertoldi, So Matter, 2011, 7, 2321.

19 J. Li, J. Shim, J. Deng, J. T. B. Overvelde, X. Zhu, K. Bertoldiand S. Yang, So Matter, 2012, 8, 10322.

20 X. Zhu, G. Wu, R. Dong, C.-M. Chen and S. Yang, SoMatter,2012, 8, 8088.

21 J. T. B. Overvelde, S. Shan and K. Bertoldi, Adv. Mater., 2012,24, 2337.

22 J. Shim, C. Perdigou, E. R. Chen, K. Bertoldi and P. M. Reis,Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 5978.

23 S. Willshaw and T. Mullin, So Matter, 2012, 8, 1747.24 S. H. Kang, S. Shan, W. L. Noorduin, M. Khan, J. Aizenberg

and K. Bertoldi, Adv. Mater., 2013, 25, 3380.25 X. L. Zhu, Y. Zhang, D. Chandra, S. C. Cheng, J. M. Kikkawa

and S. Yang, Appl. Phys. Lett., 2008, 93, 161911.26 D. Krishnan and H. T. Johnson, J. Mech. Phys. Solids, 2009,

57, 1500.27 D. Liu, C. W. M. Bastiaansen, J. M. J. den Toonder and

D. J. Broer, So Matter, 2013, 9, 588.28 J. Kim, J. Yoon and R. C. Hayward, Nat. Mater., 2010, 9, 159.29 J. Yoon, P. Bian, J. Kim, T. J. McCarthy and R. C. Hayward,

Angew. Chem., Inter. Ed., 2012, 51, 7146.30 M. Heskins and J. E. Guillet, J. Macromol. Sci., Part A: Pure

Appl.Chem., 1968, 2, 1441.31 H. Yu and D. W. Grainger, J. Appl. Polym. Sci., 1993, 49,

1553.32 T. Tanaka and D. J. Fillmore, J. Chem. Phys., 1979, 70, 1214.33 J. Yin, H. Hu, Y. Wu and S. Liu, Polym. Chem., 2011, 2, 363.34 M. Guvendiren, J. A. Burdick and S. Yang, So Matter, 2010,

6, 5795.35 D. Chandra, J. A. Taylor and S. Yang, SoMatter, 2008, 4, 979.36 G. H. Chen and A. S. Hoffman, Nature, 1995, 373, 49.37 S. Cai, K. Bertoldi, H. Wang and Z. Suo, So Matter, 2010, 6,

5770.38 V. Trujillo, J. Kim and R. C. Hayward, So Matter, 2008, 4,

564.39 A. Suzuki and H. Suzuki, J. Chem. Phys., 1995, 103, 4706.40 A. Richter, A. Bund, M. Keller and K.-F. Arndt, Sens.

Actuators, B, 2004, 99, 579.

Soft Matter, 2014, 10, 1392–1399 | 1399