progress in polymer science - nc state university · progress in polymer science 52 (2016) 79–106...

YD

a

AA

KSAFP

C

0

Progress in Polymer Science 52 (2016) 79106

Contents lists available at ScienceDirect

Progress in Polymer Science

journa l h om epa ge: www.elsev ier .com/ locate /ppolysc i

2D or not 2D: Shape-programming polymer sheets

ing Liu, Jan Genzer , Michael D. Dickey

epartment of Chemical and Biomolecular Engineering, North Carolina State University, Raleigh, NC 27695-7905, USA

r t i c l e i n f o

rticle history:vailable online 23 October 2015

eywords:hape programmingctive materialsoldingolymer sheets

a b s t r a c t

This review summarizes progress toward programming two-dimensional (2D) polymersheets which respond to a variety of external stimuli to form three-dimensional (3D) shapesor topographical features on macroscopically planar sheets. Shape programming strategi-cally adds value or function to 2D sheets, films, or coatings that can be created inexpensively.2D substrates are common form factors that are compatible with ordinary 2D patterningtechniques (i.e., inkjet, photolithography, roll-to-roll printing) and may be stored, packed,and shipped efficiently. Polymer materials are attractive due to their flexibility, light weight,low price, and compatibility with high throughput processing. This review highlights strate-gies for triggering shape change in planar polymeric materials. The strategies are dividedinto four broad categories: (1) 2D substrates with latent topography programmed using

conventional microfabrication, (2) 2D substrates that form topography due to imposed orself-generated stress, (3) 2D substrates that form 3D shapes by out-of-plane bending, and(4) 2D substrates that use hinges to achieve out-of-plane folding. The review highlightsall strategies while focusing primarily on last two approaches.

2015 Elsevier Ltd. All rights reserved.

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 801.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 801.2. Why is shape programming from 2D to 3D interesting? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .81

2. Strategies for transforming 2D polymer sheets to 3D shapes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 812.1. Conventional fabrication (Strategy 1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 812.2. Induced surface topographies (Strategy 2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 822.3. Bending (Strategy 3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 842.4. Folding (Strategy 4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 892.5. Discussion for out-of-plane shape programming (Strategies 3 and 4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .932.6. Applications for out-of-plane shape programming (Strategies 3 and 4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

3. Conclusion, opportunities, and challenges for shape programmingAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Corresponding author. Tel.: +1 919 515 2069. Corresponding author. Tel.: +1 919 513 0273.

E-mail addresses: [email protected] (J. Genzer), [email protected] (M.D. Dic

http://dx.doi.org/10.1016/j.progpolymsci.2015.09.001079-6700/ 2015 Elsevier Ltd. All rights reserved.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .98. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

key).

dx.doi.org/10.1016/j.progpolymsci.2015.09.001http://www.sciencedirect.com/science/journal/00796700http://www.elsevier.com/locate/ppolyscihttp://crossmark.crossref.org/dialog/?doi=10.1016/j.progpolymsci.2015.09.001&domain=pdfmailto:[email protected]:[email protected]/10.1016/j.progpolymsci.2015.09.001

olymer S

demonstrates that a series of local folds can lead to globalbending.

80 Y. Liu et al. / Progress in P

1. Introduction

1.1. Overview

This review focuses on strategies for shape-programming polymeric materials that can convertplanar two-dimensional (2D) polymer sheets to three-dimensional (3D) shapes. Shape-programming changesthe physical geometry of materials in response to anexternal stimulus. The desire to create 3D shapes isstimulated by the premise that shape defines function ofmost materials. Employing 2D planar sheets to fabricate3D shapes brings several technological advantages andappeals: (1) 2D polymer sheets and coatings can bemass-produced inexpensively using high-throughputprocessing methods and find use in our daily livesincluding the substrates for printed electronics, laminatecoatings, and plastic films; (2) 2D polymer sheets arecompatible with most planar processes developed forthe semiconductor industry as well as other patterningtechniques, such as, screen printing, roll-to-roll processing,gravure printing, and inkjet printing; (3) planar sheetscan be stacked efficiently for storage, transport, or remotedeployment.

We divide shape-programming of 2D materials into fourbroad categories as illustrated in Fig. 1: (1) 2D substratesthat use additive or subtractive processes, such as lithog-raphy, to create topography; (2) 2D substrates that formtopography, e.g., wrinkles or creases, on their surface inresponse to an external stimulus; (3) 2D substrates that canform out-of-plane 3D shapes by bending, and (4) 2D sub-strates that use hinges to achieve out-of-plane folding.Unlike shape transformation, which describes the imme-diate alteration of the shape due to mechanical processing,shape programming involves a processing step (or a designstrategy) to change shape rationally and controllably at adesired time and defined location in response to an externalstimulus. It could be argued that Strategy 1 does not involvetrue shape programming steps since the shape change hap-

pens during processing and not in response to a subsequentexternal stimulus. However, we include it here for com-pleteness and because it can be used to create latent shapechange.

Fig. 1. Cross-sectional illustrations of strategies to generate 3D shapes byusing shape-programming of polymeric materials in four categories: (1)conventional fabrication; (2) induced surface topography on a macroscop-ically flat substrate; (3) 3D structures by bending; and (4) 3D structuresby folding. The dashed black line marks the original 2D substrate prior toshape change.

cience 52 (2016) 79106

The importance of shape programming in planar mate-rials unifies the four strategies depicted in Fig. 1 andpoints to some similarities among them. For example, whilesamples prepared by strategies 1 and 2 result in macroscop-ically planar substrates that support microscopic (typicallysub-millimeter) surface topography, strategies 3 and 4describe the formation of 3D out-of-plane shapes by bend-ing, folding, or rolling to produce dramatic macroscopicshape changes as well as microscopic ones, often in a facilemanner. Shape programming characterized by strategies3 and 4 can occur due to asymmetrical distortion of thematerial, expansion (e.g., thermal expansion, swelling),contraction (e.g., strain relaxation, de-swelling), or flowor deformation (e.g., displacement of material due to astress).

Fig. 2 differentiates the subtle differences betweenbending (strategy 3) and folding (strategy 4). Theseterms are sometimes used interchangeably since thereis not a universally agreed upon definition that differ-entiates them. Several recent publications discuss thedifference between bending and folding in shape pro-gramming materials [15]. Simply stated, folding involveslocalized deformation whereas bending is global. Lauff et al.provide a thorough summary of the literature on this topicand define bending as the distributed deformation of amaterial along the deflected area that creates curvature(i.e., distributed curvature) and folding as the localizeddeformation of a material along crease patterns to cre-ate new shapes (i.e., localized curvature along a crease)[1,2]. Fig. 2a shows a clear case of folding (where theratio of red to blue approaches zero) and Fig. 2d showsa clear case of bending (where the ratio of blue to redapproaches zero). Fig. 2b and c illustrate why it is diffi-cult to definitively distinguish bending from folding. Fig. 2e

Fig. 2. Illustration of the difference between folding and bending of sheets(in a cross-sectional view): (a) ideal case of folding with localized defor-mation; (b) in practice, folding is defined when the deformed region (red)is small relative to the sample size (blue); (c) in practice, bending is definedwhen the deformation region is commensurate with the sample size; (d)ideal case of bending with global deformation; (e) a series of local folds canlead to a global bend. Red colors denote the regions of deformation. (Forinterpretation of the references to color in this figure legend, the readeris referred to the web version of this article.)

olymer S

ghifgmaftwp4odlt

1i

amsangatfipsl

cnts2sahsaatdai[

tsfa[scm

Y. Liu et al. / Progress in P

Other excellent reviews cover aspects of shape pro-ramming by focusing on common materials (e.g.,ydrogels, elastomers) or highlighting adaptive compos-

tes [2,4,616]. In this review we provide a general roadmapor different strategies and materials used in shape pro-ramming. Specifically, we concentrate only on polymericaterials with an occasional mention of other materi-

ls that might be utilized in conjunction with polymersor shape programming. Polymeric materials are attrac-ive for shape programming due to their flexibility, lighteight, low price, and compatibility with high through-ut processing. We also focus primarily on Strategies 3 and; that is, conversion of 2D planar polymer sheets to out-f-plane structures by bending and folding. We limit ouriscussion of strategies 1 and 2 to a brief review with high-

ights of previously published reviews and references onhese topics.

.2. Why is shape programming from 2D to 3Dnteresting?

2D polymeric substrates may be manufactured industri-lly using low-cost, high-throughput processes, including,elt extrusion, injection molding, vacuum forming, pultru-

ion, or transfer molding [17,18]. In general, 2D substratesre compatible with many inexpensive 2D patterning tech-iques (e.g., lithography, screen printing, inkjet printing,ravure printing, roll-to-roll processing, and laser cutting)nd more sophisticated microfabrication and nanofabrica-ion techniques (e.g., spin casting, photolithography, thinlm deposition, etching, and other additive/subtractiverocesses including unconventional patterning techniquesuch as dip pen lithography, colloidal lithography, imprintithography, and directed self-assembly) [19].

Converting 2D sheets to 3D structures may be benefi-ial for a variety of purposes. Examples include, but areot limited to: (1) changing the shape of a polymer sheeto achieve a final product (e.g., packaging); (2) assemblingtructures (e.g., electronics or optics) that are fabricated inD to achieve final desired 3D shape; (3) reconfiguring thehape, and thus, function of a substrate for applications thatre both low-tech (e.g., responsive toys for children) andigh-tech (e.g., shape reconfigurable devices or non-foulingurfaces) [2024]; (4) deploying materials convenientlynd compactly in a flat 2D form and have a 3D shapessemble at the destination site; (5) tuning the proper-ies of a surface (e.g., wetting or optical properties) thatepend strongly on surface topography; and (6) implanting

structure in a compact form factor and have it assemblen situ (e.g., a boat in a bottle, or an implantable device)2527].

There are many compelling applications of 2D to 3Dransformations that have been proposed and demon-trated. For example, sensors constructed in 3D space allowor the measurement of signals from three independentxes to obtain accurate angular and orientation parameters

22]. Morphing structures such as Morphees (i.e., shape-hifting mobile devices) and MorePhone (i.e., a cellphoneurls upon a call) have been proposed for next-generationobile devices. [28] In addition, shape programming has

cience 52 (2016) 79106 81

also been applied in smart adhesives [29], anti-foulingcoatings [30,31], sensors [32,33].

2. Strategies for transforming 2D polymer sheets to3D shapes

Many materials strategies and stimuli can be utilizedfor shape programming polymers. Following the schemein Fig. 1, we first overview those four strategies to convert2D polymer sheets to 3D shapes and then narrow our focusto strategies 3 and 4.

2.1. Conventional fabrication (Strategy 1)

Additive or subtractive microfabrication can generatetopography on 2D substrates [19]. Of the four strategiesintroduced in Fig. 1, this approach is the most commonlyused because of its importance for the mass productionof electronics and computer chips. It could be argued thatthese techniques are not truly shape programming becausethe shape change usually takes place during processing,and not at some later time programmed into the material.However, conventional photolithography may be con-sidered a shape programming method since the latenttopography can be revealed only upon exposure to a liq-uid developer which may be thought of as an externalstimulus.

Fig. 3 illustrates the two general types of conventionallithographic methods, which expose polymeric films (e.g.,photoresists) to photons (e.g., UV light, X-ray) or particles(e.g., electrons and ions) to form relief structures in thefilm. Chemical changes in the films alter the solubility ofthe polymer in a liquid developer relative to the unexposedregions [34,35]. Positive photoresists become more solublein a developer solution after exposure to the incident beam,while negative photoresists become less soluble after expo-sure. The most common lithographic method utilizes amask to transmit patterns of photons onto selected regionson the substrate. Alternatively, a focused beam of pho-tons or electrons may be rastered across a substrate, butthese serial processes generally cannot achieve the highthroughput patterning realized using a mask. Lithographicmethods are dominant technologies in the fabrication ofmicroelectromechanical systems (MEMS), optics, and com-puter chips. These processes are inherently 2D and thusare poorly suited for patterning curved and out of planesurfaces. However, features created by lithography maybe patterned on a 2D surface and then assembled into 3Dshapes by, for example, bending or folding (discussed laterin this review) [3638].

In addition to conventional lithographic methods, thereare many additive techniques (e.g., sputtering and evap-oration [39,40]) or subtractive methods (e.g., etching,machining process [41,42]) for defining topography on asurface.

There are also numerous unconventional lithographicapproaches for introducing surface topography on 2D sub-

strates [34,43,44]. Examples include imprint lithography[4549], directed self-assembly [50], dip pen lithography[51], direct-write lithography [52], colloidal lithogra-phy [53,54], or soft lithography [55,56]. Self-assembly

82 Y. Liu et al. / Progress in Polymer Science 52 (2016) 79106

meric fi
Fig. 3. Schematic of conventional lithographic methods that expose polyare shown here for the sake of illustration.
processes (e.g., crystallization, phase separation, order-to-order transition, colloidal assembly) can also be employedto alter surface topography spontaneously by interaction-driven association of individual (often disordered) buildingblocks into organized arrays [34,5759]. There are alsonumerous low-cost, low-resolution patterning techniques,including, inkjet printing, screen printing, and roll-to-rollpatterning. In addition, 3D shapes can be realized by, forexample, stereolithography [6064] or ink-based direct-write methods [6568]. The examples given here are byno means comprehensive and are only given to suggest thewide range of patterning techniques available. The readeris referred to several monographs and references on uncon-ventional fabrication and printed electronics [6973].

In summary, most of these techniques are shape trans-forming rather than shape programming because thetopography change occurs directly during the patterningprocesses themselves rather than in response to an externalstimulus. For this reason, these conventional microfabrica-tion methods will not be discussed in more detail in thisreview despite their prevalence industrially and in labora-tory environments.

2.2. Induced surface topographies (Strategy 2)

There are many strategies to induce surface topogra-phy onto smooth 2D substrates. Unlike materials additionor subtraction (Strategy 1), the surface topographies

lms to patterns of photons, electrons, or ions. Only negative photoresists

discussed here form due to the inherent propertiesdesigned into the substrates.

Wrinkling (or buckling) is a prominent mechanismto form tunable periodic sinusoidal surface topography,which has been reviewed in detail previously [6,7480].Buckles typically form due to compressive forces acting ona thin rigid film attached to a soft substrate (Fig. 4a). Elas-tomers are usually used as the substrate. PDMS is employedoften for these applications due to its outstanding flexibil-ity and ability to be stretched [74,81]. Other materials, suchas viscoelastic polymers (e.g., polystyrene), hydrogels andpolymer brushes can also be used to form buckles [8286].The stress may be imposed externally (e.g., thermal expan-sion or mechanical strain) or by chemical modification ofthe surface of the polymeric substrate. For example, thebottom substrate may be heated and expanded uniformlyeither by an external heat source before the depositionof the top film or by thermal energy generated duringthe deposition. Upon cooling, compressive stress devel-ops in the deposited film, which competes with the tensilestress in the bottom substrate and results in the forma-tion of randomly oriented wrinkles (isotropic wrinkles)[82,8789]. Pre-stretched polymer sheets supporting arigid top film (e.g., metal, oxide, or highly crosslinked

surfaces) form wrinkles upon releasing strain [80,90,91].Imposing larger pre-strains (>10%) results in topographycomprising hierarchically-structured wrinkles, in whichsmaller wrinkles reside on top of larger ones (Fig. 4b).

Y. Liu et al. / Progress in Polymer Science 52 (2016) 79106 83

Fig. 4. (a) Schematic of the formation of wrinkles on an elastic substrate by pre-stretching; (b) Wavelengths of multiple generations of wrinkles can formon PDMS sheets by using large pre-strains [81], Copyright 2005. Adapted with permission from Nature; (c) optical microscopy images (top view) of wrinklesw t direct&

Wtt[siotri

dsoFhfkffms

bcgd[lalt[i

ith different orientation preference due to the applied strain in differen Sons Inc.

rinkles can also be induced by swelling bilayer struc-ures or by employing gradients in crosslink density. Theseechniques are applied commonly in hydrogel systems76,78,86,9295]. In addition, wrinkles can occur due to theurface modification (e.g., oxidation) of an elastomer dur-ng sputter deposition. The oxidation creates a stiff layern the surface of the elastomer and the mismatch betweenhe mechanical properties of such skin and the base mate-ial generates a compressive stress during deposition thatnduces buckling [96].

Buckles possess a characteristic wavelength thatepends on the mechanical properties and geometry of theystem. The in-plane organization of the wrinkles dependsn the direction of the imposed strain, as illustrated inig. 4c [76]. During the past several decades various modelsave been developed to study the mechanism for wrinkle

ormation [80,82,97102]. The geometry and size of wrin-les (including the wavelength and amplitude) has beenound to depend on the thickness of the top film and theoundation as well as the materials properties (e.g., Youngs

odulus, Poissons ratio) of the two layers and appliedtrain.

The ability to form wrinkled/buckled surfaces haseen used in various applications, including, anti-foulingoatings [31,103], microfluidic devices to direct the cellrowth [104], smart adhesive surfaces [105], opticalevices such as microlenses [106] and diffraction gratings87,107]. By making use of the buckling mechanism, buck-ed structures can be applied as a characterization tool tossess the modulus of materials employed as rigid thin

ayers [75,77,107,108]. The application of buckled struc-ures in light trapping materials has also been explored109112]. Hierarchical buckles are well-suited topograph-cal structures that enhance trapping of light with various

ions [76], Copyright 2010. Reproduced with permission from John Wiley

wavelengths since the buckle wavelengths range frommicro- to nano-scale [81].

Besides wrinkling (continuous undulations on the sur-face), other thin-film instabilities can form in response tostress, including, craters, creases, folds, localized ridges, ordelamination (non-continuous deep folds on the surface).The type of topography depends on the degree of adhe-sion and difference in modulus between the film and thesubstrate. These instabilities form self-organized topogra-phies such as wrinkles, creases, and buckles in response toa stimulus (Fig. 5) [90,113121]. The use of electric fieldsto induce film instabilities is another means of generat-ing functional surfaces and on-demand creases or folds[113,114].

Dewetting instabilities of polymer thin films (thickness100 nm) offer a low-cost and lithographic free approachfor patterning surfaces with tailored topographies. Molec-ular interactions (e.g., van der Waals interactions, residualstresses due to long chain entanglement) or surface het-erogeneities drive the spontaneous rupture of these thinfilms (e.g., spinodal dewetting) (Fig. 6a and b) [122130].Desired patterns can be generated by tuning physical orchemical heterogeneity on the substrate [122,123]. Sur-face patterns created by rational dewetting have beenapplied for optoelectronics and semiconductor devices[122124].

Dewetting of thicker films (thickness >100 nm) can beinduced by applying an external field, i.e., temperature,electric or magnetic field [124]. For example, electrohydro-dynamic instabilities utilize electric fields to overcome the

stabilizing forces of surface tension and thereby form sur-face topography (e.g., pillars) out of molten polymer filmsor planar liquids (Fig. 6cf) [131137]. In general, patternsformed by instabilities are difficult to control precisely over


Fig. 5. (a) Schematics and optical microscopy images from (b) top view and (c) side view for typical wrinkles (left column), creasing (middle column) andmission
delamination (right column) [118], Copyright 2013. Reproduced with per
large areas and usually result in features with characteristiclength scales that lack long range order.

Light-induced chemical transformation can also createstress that results in surface topography. For example, filmscontaining azobenzene moieties undergo a cistrans tran-sition in response to light and generate surface topographywhen exposed to patterned light (Fig. 7) [138]. It could beargued that this approach is a shape transformation ratherthan shape programming since the topography forms dur-ing light irradiation, but we include it since the topographyis dormant until stimulated by light, the response is pro-grammed within the composition of the molecules, andtopography forms in the material without human (mechan-ical) intervention. The mechanism for this response willbe discussed in more detail in Section 2.3. Moreover, thesurface topography can be triggered by a light-inducedMarangoni effect (i.e., mass transfer through the interfacedue to surface energy gradients). For example, the surfaceenergy of polystyrene increases due to dehydrogenationunder patterned UV irradiation so that the polymer in theliquid state flows from unexposed area to exposed area toform surface feature [139].

Formation of surface topographies on planar sub-strates can be also triggered by the shape memory effect[32,140,141]. Shape memory polymers, i.e., materials thatcan be programmed into a temporary shape that recover toa permanent shape upon heating, will be discussed in detail

later in this review. As shown in Fig. 8, surface topographycan form in these SMP coated with a metal film by releas-ing the local stress formed by indenting or embossing, orby releasing a uniform stress generated by pre-stretching

from the ASME.

the sample [142]. As another example, grating patterns canbe flattened into a stable temporary, flat substrate. The sur-face topography recovers by heating the samples above theglass transition temperature (Tg) of the polymer (Fig. 9)[140]. Light diffraction due to the grating pattern reappears,as evident from the apparent color change of the surface.Moreover, the surface topography recovered from a shapememory polymer can form a multi-use dry adhesive to aglass substrate that is strong and reversible [29].

Shape memory materials are also applied widely forforming 3D structures in a responsive fashion, which willbe elaborated in more detail in Section 2.4.

2.3. Bending (Strategy 3)

This section focuses on bending as a mechanism toform out-of-plane 3D structures. Expansion (e.g., swelling,thermal expansion and pneumatics) and contraction (e.g.,de-swelling, thermal shrinkage, and deflation) generatestresses necessary to convert 2D polymer sheets into 3Dshapes by bending.

Polymer gels (or polymer networks) are often usedfor bending because they can swell dramatically in thepresence of good solvents and contract when exposedto poor solvents while retaining their overall networkstructure. Polymer gels can be classified as either hydro-gels or organogels. Hydrogels are hydrophilic polymeric

networks that can absorb water and swell [143,144].Hydrogels are noted for their ability to swell or de-swellin response to environmental changes in aqueous environ-ments including pH, temperature, ionic strength, solvent


Fig. 6. (a) Schematic of the time evolution for spinodaldewetting (left) and corresponding experimental result (right) (unpublished data); (b) Opticalmicroscope image of dewetting of poly(methyl methacrylate) (PMMA) on a patterned PDMS substrate [130], Copyright 2007. Reproduced with permissionfrom Elsevier Ltd.; (c) Schematic of 3D structures induced by an electric field at polymer/air bilayer [132], Copyright 2000. Reproduced with permissionfrom Nature; (d) Schematic of 3D structures induced by electric field at polymer/polymer/air trilayer [136], Copyright 2006. Reproduced with permissionf ture forp ade of t .

tt

t

Fs

rom the American Chemical Society; (e) SEM images of cage-type strucermission from the American Chemical Society; (f) column structure mrilayers [137], Copyright 2002. Reproduced with permission from Nature

ype, or electrical potential [143150]. These environmen-al cues can be employed as stimuli to induce shape change.

Temperature is an appealing stimulus for alteringhe swelling behavior of hydrogels because it is easy

ig. 7. AFM image of surface topography on a polymer film containing photosurface-plasmon field [138], Copyright 2011. Reproduced with permission from t

med by electrohydrodynamics [136], Copyright 2006. Reproduced withPMMA while removing polystyrene with electric field on PMMA/PS/air

to control and reverse externally. While technically allpolymers undergo conformational changes in responseto temperature, in some cases changes of polymersolubility in water are very dramatic and lead to

ensitive azo groups after exposing to patterned UV light to generate ahe Royal Society of Chemistry.


y polym
Fig. 8. Illustration of sample preparation procedures on shape memorReproduced with permission from IOP Publishing.
large changes in swelling. Poly(N-isopropylacrylamide)(PNIPAAm) and its derivatives are popular temperature-responsive hydrogels due to a reversible coil-to-globuletransition that occurs upon exceeding the lower critical

solution temperature (LCST 32 C) [143,144,151154].Other hydrogels also exhibit LCST [155,156], for example,poly(N-vinylcaprolactam) or poly(N-vinylpyrrolidone).

Fig. 9. Surface topography due to shape memory effect in an acrylate-based SMP (clate) (PEGDMA) as crosslinker and 2,2-dimethoxy-2-phenylacetodphenone (inipermanent pattern molded by thermally embossing nanoimprint lithography at (c) recovered surface topography after heating at 120 C. The pattern has a pitwith permission from John Wiley & Sons Inc.

er to form surface topography (e.g., wrinkling) [142], Copyright 2011.

Hydrogels often contain charged (i.e., strong polyelec-trolytes) or chargeable (i.e., weak polyelectrolytes) groups.The presence of charges causes gels to swell in water.The concentration of chargeable groups in weak poly-

electrolytes depends on the nature of the polymer (i.e.,polycation vs. polyanion), pH, and ionic strength of thesolution. Changing the pH of the solution (for weak

ured from methyl methacrylate (MMA), poly(ethylene glycol dimethacry-tiator)). Schematics (top) and images (bottom) for different stages: (a)180 C; (b) 2D planar surface after programming with a flat mold; andch of 800 nm and a height of 180 nm [140], Copyright 2011. Adapted

olymer S

ptt

dooeama

irig

tTsmTebcc(tltlctpaHFpfbe(mgs

mpczlnsfbcidca[


olyelectrolytes) and/or ionic strength (for all polyelec-rolytes) alters the swelling behavior of the gel and areherefore useful stimuli for gel shape change [148,157,158].

The ability to control the degree of swelling ore-swelling in hydrogels using external stimuli offerspportunities to program shape change. Isotropic swellingr de-swelling results in uniform changes in volume; i.e.,ach dimension of the gel shrinks or expands proportion-lly. To induce out of plane motion, the generated stressust be asymmetric, which can be induced by employing

number of approaches.One approach to achieve asymmetry may be real-

zed by laminating bimorph structures with differentesponses to the same stimuli (i.e., temperature, solvent,onic strength) or by patterning stimuli-responsive hydro-els [23,154,159162].

The general theory of bilayer bending is derived fromhe model of a bi-metallic strips developed originally byimoshenko in 1925 [163]. A classic example of a bilayertructure is a thermal bimorph, featuring two layers ofaterials with different thermal expansion coefficients.

he bilayer curls in response to heat since one layerxpands more than the other. This concept may alsoe employed in polymeric systems; a laminate structureomposed of a layer of hydrophobic polymer (i.e., poly-aprolactone) and a layer of thermo-responsive hydrogele.g., PNIPAAm) can bend to generate a 3D structure byuning the solution temperature (Fig. 10a) [160]. Simi-arly, a closed micro-cage made of an SU8 epoxy (highhermal expansion coefficient) laminated with diamond-ike carbon (DLC) layer (low thermal expansion coefficient)an open upon heating [164]. In addition, the curva-ure of bilayer structures can be controlled by employingatterned polymers with various Youngs moduli on thective layer which is stimuli responsive (Fig. 10b) [165].umidity can also trigger the bending of bilayer structures.or example, a bilayer laminate comprising cross-linkedoly(acrylic acid) and a layer of polymer that absorbs waterrom humid air (e.g., poly(allylamine hydrochloride)) willend in response to humidity [166168]. Hydrogel bilay-rs with a copolymerized layer of N-isopropylacrylamideNIPAm) and acrylic acid (AAc) or poly-hydroxyl ethyl

ethacrylate (HEMA) show asymmetric swelling and trig-er 3D structures in solutions with different pH and ionictrengths [23].

Another approach to bending creates out-of-planeotion by patterning crosslinks in a hydrogel [169]. The

rocess utilizes halftone photolithography to photo-rosslink PNIPAAm copolymers containing pendant ben-ophenone units. The crosslinked regions of the gel swelless than the rest of the gel, which generates the stressecessary for out of plane deformation with nearly con-tant Gaussian curvature (Fig. 10c) [170]. Various 3D shapesorm by bending if the stress distribution in the gel cane controlled [171175]. For example, hydrogels can formomplex 3D helical structures using sheets with alternat-ng stripes of different chemical composition and thus

ifferent swelling response (Fig. 10d) [176]. In addition,hemical oscillations of ruthenium ions between reducednd oxidized states within gels can cause cyclic swelling177].

cience 52 (2016) 79106 87

Materials containing gradients of fillers or crosslinksthrough the depth of the sheet also bend. This asymme-try results in out-of-plane motion in response to a uniformstimulus (Fig. 11) [178183]. For example, rolling of aphoto-crosslinkable copolymer based on PNIPAAm can betriggered thermally in aqueous medium by patterning bothhigh and low swelling regions using UV light (Fig. 10c)[170,179].

Electric fields represent another attractive stimulusthat can moderate swelling of polymeric gels. Electricfields can disrupt the distribution of ions in gels andtherefore change the local osmotic pressure. Significanteffort in this field has been achieved during the past fewdecades following the pioneering work by Tanaka andcolleagues [92,184]. For example, electrical potential candrive cations from the electrolyte (e.g., sodium dodecyl-benzenesulfonate, Na+DBS) into a polymer and causeasymmetrical swelling [185,186]. More recently iono-printing was introduced to introduce local folding andglobal bending of flat hydrogel sheets. Ionoprinting usesmetal electrodes under anodic conditions to inject ionsinto a polyelectrolyte hydrogel (e.g., sodium polyacry-late) [147,187,188]. Divalent metal cations (e.g., Cu2+) bindwithin an anionic network to form crosslinks that generatesufficient mechanical stress to create a hinging responsein air (Fig. 12a). Whereas actuation of most gels relies onswelling/de-swelling, which is a slow process, ionoprintingresults in rapid actuation (a few seconds) without the needfor submersion in electrolyte. In addition, the ionoprintedlines are stiffer than the rest of the gel. These stiff lines cangenerate an anisotropic response to de-swelling of the gelin ethanol (i.e., the water diffuses out of the gel), as shownin Fig. 12b.

Organogels, i.e., crosslinked polymer networks thatswell in the presence of organic solvents, can alsobe utilized for shape programming [189191]. Forexample, many elastomers such as those based onpoly(dimethylsiloxane) (PDMS) [192] swell to differentextents based on the solubility of solvents in the networks.

The major appeal of swelling/de-swelling for controllingthe shape of a polymer sheet is its simplicity and abil-ity to operate in wet environments. Furthermore, manyhydrogels are biocompatible because they are largely com-posed of water. Swelling/de-swelling, however, suffersfrom least two notable drawbacks: (1) it requires a sol-vent medium, and (2) it is generally slow due to thereliance on diffusion of solvent into the polymer (althoughapproaches like ionoprinting can operate in air). Here, welimit the discussion of gels because there is a rich litera-ture describing strategies (temperature, solvent, electricalpotential, pH, etc.) for shape programmable gels; the exam-ples given here are just illustrative and not comprehensive[12,143,144,146,152,153,193196].

In addition to swelling, pneumatics can also induceout of plane motion from a 2D substrate via the volu-metric expansion and contraction achieved by applyingpressurized gas inside a cavity (e.g., a balloon) made of

elastomeric materials such as PDMS and Ecoflex (a com-mercial elastomer) to be utilized as grippers (Fig. 13)[24,197200]. To realize bending, asymmetrical inflationhas to be achieved using material and design strategies. A


Fig. 10. Examples of 3D structures induced by controlling the stress within sheets of polymer gels: (a) bending is triggered by a temperature change to athermal responsive PNIPAAm bilayer laminate [160], Copyright 2011. Adapted with permission from the Royal Society of Chemistry; (b) bilayer structurewith one passive layer (top layer, where E represents modulus) and one temperature-sensitive hydrogel (green layer) to induce the bending [165], Copyright2013. Adapted with permission from IOP Publishing; (c) 3D curved shapes are generated when asymmetrical swelling occurs due to highly crosslinked dotswithin a low crosslinked matrix [170], Copyright 2012. Reproduced with permission from the American Association for the Advancement of Science; (d)

ith chem
complex helical structure can be formed when a planar sheet patterned wwith permission from Nature.
downside of this method is the inconvenience of tether-ing of the inlets/outlets for the gas, although it is possibleto induce chemical reactions to rapidly create gas in situ[201].

Liquid crystal elastomers (LCEs) are popular shapeprogrammable polymers. Whereas there are few strate-gies for creating shape reversible SMPs, LCEs can actuate

reversibly, which is a salient feature. LCEs are hybridmaterials, which combine oriented liquid crystals andpolymeric elastomers. Usually, chemical synthesis cou-ples the liquid crystal units with polymer chains to

ically distinct regions of hydrogel swells [176], Copyright 2013. Adapted

form the LCE networks. Polymer chains in the networkexperience anisotropic conformation due to the crys-talline phase of liquid crystals, while the polymer chainsreturn to their coil conformation upon heating. There-fore, the shape change in LCEs can be triggered by atemperature jump (Fig. 14a) [202210]. LCEs incorporat-ing photoresponsive functional groups constitute another

large class of light-induced shape programmable materi-als (Fig. 14b) [202208,211]. Many excellent reviews andbooks have summarized the development in this field[204,212217].


F across ths . Scale bN

iPioml[pmtortwslsbi

enrc

Feaf

ig. 11. (a) Schematic to illustrate the light-induced crosslinking gradient olvents; (b) Fluorescence images of actuation of flower-shaped structuresature.

Light can also induce shape changes to polymers byncorporating photosensitive groups, such as azobenzene.hotoresponsive SMPs (or so-called light-activated SMPs)nvolve materials with photosensitive functional groupsr fillers in the polymer networks, such as photoiso-erizable molecules, i.e., azobenzenes, triphenylmethane

euco derivatives, and photoreactive molecules (Fig. 15ac)218225]. Triggered by appropriate UV wavelengths, thehotoinduced effects on the molecular level (e.g., photoiso-erization, ionic dissociation, or photodimerization) lead

o macroscopic volume changes (shrinking or swelling)f polymers. For example, the cistrans isomerizationeactions of azobenzene generate reversible conforma-ion changes at the molecular level in response to specificavelengths and, in turn, induce shape change macro-

copically. Azobenzene molecules may be introduced intoiquid crystal elastomers [205,206,226228] or photosen-itive polymer films [5,138,229,230]. Fig. 15d demonstratesending of the polymer sheet with azobenzene molecules

n response to light polarized at different angles.It is also possible to use electric fields to actuate

lastomers and dielectric polymers by a number of mecha-isms including the Maxwell stress. These topics have beeneviewed elsewhere [231236]. Likewise, piezoelectricsan convert voltage signals into displacement, but involve

ig. 12. (a) A hydrogel bends into a 3D coil by using multiple folds patterned vilectrode; (b) ionprint lines are stiffer than the rest of the gel, which causes gels tond release a small blue cube of PDMS (1 g) in ethanol and water, respectively.rom Nature.

e film thickness of SU-8 film, and resulting bending of the film in differentars are 100 m [178], Copyright 2011. Reproduced with permission from

generally ceramic materials (with the noted exception ofpoly(vinylidene fluoride), PVDF [237243]) (Fig. 16). Wethus do not review them in detail here but mention themfor thoroughness.

The examples we have provided within Strategy 3 relyprimarily on bulk changes in volume to induce bending.However, it is possible to localize those changes to just thesurface. Changes in surface topography can be induced byemploying polymer brushes that behave similarly to gelsby responding to external stimuli, such as, pH, tempera-ture, salt concentration, or type of solvent; those structureshave been reviewed before [153,245,246]. The conforma-tional change of brushes due to an external trigger canbe employed in controlling surface wettability [247,248],or adhesion properties [249,250]. The changes in surfacetopography due to brush expansion or contraction are rel-atively small compared to the other techniques discussedhere and we therefore limit our discussion of this topic.

2.4. Folding (Strategy 4)

While there is some phenomenological overlap with theexamples of bending in Strategy 3, Strategy 4 is distin-guished by its focus on folding as defined in Fig. 2.

a ionoprinting lines of copper ions in a hydrogel via the use of a copper swell asymmetrically. A consequent gel gripper is demonstrated to catch

The scale bar is 10 mm [187], Copyright 2013. Adapted with permission


Fig. 13. (a) Schematics of asymmetrical inflation of a 2D sheet by applying different thickness of elastomers (top row) or using composite materials withdifferent Youngs Modulus (Ea , Eb) [24], Copyright 2011. Adapted with permission from John Wiley & Sons Inc.; (b) Example of a paper-Ecoflex composite

ermissipermiss.

showing asymmetrical actuation [197], Copyright 2012. Adapted with pby pneumatic balloon actuator [199], Copyright 2007. Reproduced with Copyright 2011. Reproduced with permission from John Wiley & Sons Inc

Folding occurs commonly in nature (e.g., certain sensi-tive plants close their leaves in response to gentle touch;folding and unfolding of leaves and wings) and in theancient art of paper folding [251253]. Here, we discussfolding of polymeric materials using external stimuli in ahands-free manner; these approaches are often called self-folding. It is a deterministic self-assembly approach thatconverts a 2D template with pre-defined hinges into a 3Dshape in response to an external stimulus, as shown inFig. 17. Hinges are regions on the substrate that get acti-vated (i.e., fold) in response to an external stimulus [57].The need for reconfigurable devices, actuators, and sensorsmotivates research on folding utilizing different drivingforces and a variety of active (i.e., stimuli responsive) mate-rials [4,14,20,57,175,186,252,254256]. While strategiescan originate from or be inspired by biological systems, theapproaches discussed here will not include cases of respon-sive biomolecules or living organisms, although they can beapplied for folding processes as well [257264].

Folding can be achieved by at least three means: (1)defining locally responsive hinges fabricated using materi-als with a different composition from the rest of the sheet;(2) combining, laminating, and patterning non-responsivesheets with responsive sheets; (3) modifying physically anotherwise homogeneous sheet (e.g., shape memory poly-mers) in local regions to define hinges that fold upon

exposure to the external stimuli; (4) stimuli can be appliedto only the hinge region of a sheet or composite (e.g.,Joule-heating, focused laser) [20,254,265,266]. The follow-ing paragraphs describe these different means.

on from John Wiley & Sons Inc.; (c) Micro hand robotic system drivenion from the IEEE; (d) Demonstration of gripping an uncooked egg [24],

One approach to fold 2D substrates is to define hingesthat differ in chemical composition from the rest of thesheet. These hinged regions may be composed of a varietyof materials that actuate in response to an external stim-ulus. For example, polyimide shrinks at high temperatures(e.g., >200 C [165,166]) due to out-gassing of strongly-bound solvents (Fig. 18a). Likewise, pre-strained polymerfilms shrink upon heating and can be utilized as localhinges different from the bulk materials to induce fold-ing (Fig. 18b) [269]. Stimuli-responsive polymers patternedas hinges can also control folding and unfolding of non-responsive laminates (Fig. 18c) [38]. For instance, a polymer(e.g., poly(caprolactone) [270]) or molten metal (e.g., sol-der [36,271]) can be patterned as a hinge and causefolding of polymer sheets when heated due to surfacetension. Some gels swell in response to selected sol-vents and may therefore be used as hinges as well.For example, hydrophilic hinges based on polyurethane(PU)/2-hydroxyethyl methacrylate (HEMA) swell signifi-cantly in acetic acid for folding planar sheets of PDMS intoof 3D structures [175].

Another approach employed to fold planar sheetsapplies multilayer laminates comprised of responsive filmscombined with non-responsive, rigid films that locally limitdeformation. Fig. 19a shows a schematic of rigid plates heldtogether by a shrink film which can actuate in response to

an external stimulus. The folding angle can be controlled bythe gap distance, thickness and shrinkage (or swelling) ofthe active sheet [2,272277]. One implementation of thisstrategy consists of sandwiching a temperature responsive

Y. Liu et al. / Progress in Polymer S

Fig. 14. (a) 3D shape change of a liquid crystal elastomer film upon heat-ing [209], Copyright 2015. Reproduced with permission from AmericanAssociation for the Advancement of Science; (b) Bending of a cantilevermade of azo-liquid crystal network under the exposure of 442 nm lightwith different polarization angles [218], Copyright 2011. Reproduced withpermission from the Royal Society of Chemistry.

Fig. 15. Illustration of representative photoisomerizable molecules: (a) transcismethane leuco derivatives; (c) cinnamic acid group; and (d) photos demonstratorientation [220], Copyright 2006. Reproduced with permission from John Wiley

cience 52 (2016) 79106 91

hydrogel (e.g., PNIPAAm) between two glassy polymer lay-ers. Patterned openings in the top glassy layer allow thegel to swell only locally and therefore create a hingingresponse, which enables micro-scale origami including thecreation of the classic origami crane (Fig. 19b). A similarstrategy sandwiches a pre-strained thermoplastic betweentwo layers of structural panels. When heated, the ther-moplastic shrinks only in the gap between the panels(Fig. 19c). Folding occurs due to asymmetric openings inthe panels. Another example is bonding elastomeric poly-mers with a layer of magnetic material and fold can beinduced in response to external magnetic fields [278,279].This approach requires multiple (yet simple) fabricationsteps and results in folds that are very reproducible.

A third approach to fold planar sheets utilizes physicalpre-programming of chemically homogeneous substrates.For example, thermally-responsive SMPs can be locallypre-programmed to a temporary shape and return back toa permanent shape at T > Ttrans to achieve shape changingfrom 2D to 3D. The heat may be delivered by using eitheruniform or local stimuli, including focused light, Joule heat-ing, or thermal radiation [254,280286]. An illustrativeexample is shown in Fig. 20.

In general, shape memory materials have been usedextensively for 3D shape programming. These struc-tures demonstrate the so-called shape memory effect inresponse to various external stimuli. The shape memoryeffect describes a phenomenon wherein a material canreturn to its original shape from a temporary shape, definedby deforming the material and fixing it in a metastable

state [287]. The temporary shape does not have to be flat,although it can be, as shown in Fig. 20. Although the finalshape could be either a bend or a fold, we discuss it within

photoisomerization of azobenzenes; (b) ionic dissociation of triphenyl-ing bending of polymer sheet by applying polarized light with different

& Sons Inc.


Fig. 16. (a) A typical piezoelectric sheets sandwiched by electrodes. The dashed lines indicated the deformed shape in response to an electric field [244],Copyright 1991. Reproduced with permission from Elsevier Ltd.; (b) Schematic of piezoelectric PVDF films used as a pressure sensor due to the chargearising from deflection of the film [238], Copyright 2008. Reproduced with permission from Elsevier Ltd.

Fig. 18. (a) Schematic of shrinking polyimide fabricated as a hinge [268],Copyright 2006. Reproduced with permission from Springer; (b) 3Dstructure (right) generated from the template (left) of a patterned cop-per/Kapton sheet with pre-strained polymer strips as hinges that respondto temperature [269], Copyright 2011. Reproduced with permission fromthe International Society for Optics and Photonics; (c) Microgripper trig-gered by temperature or chemicals in the biological environment with apatterned responsive polymer (red, top layer) [38], Copyright 2009. Repro-

Fig. 17. Schematics of the comparison between a) conventional self-assembly and b) folding.

Strategy 4 because it provides a popular approach for fold-ing.

Swedish physicist Arne Olander discovered the shape-memory effect in the early 1930s in goldcadmium(AuCd) alloys, which can be plastically deformed uponcooling and can return to the original shape by heating[288]. Shape memory materials include not only metalalloys but also other materials such as ceramics, liquid crys-talline elastomers, polymers, and gels [202,203,289292].Shape memory alloys (SMA) and shape memory poly-mers (SMPs) are two of the most popular shape memorymaterials, which will only be briefly introduced here sincethey have been reviewed thoroughly in the literature[287289,293299].

Although SMAs are not necessarily polymeric, theymay be combined strategically with polymer sheets toprogram shape changes. The most common SMA is nickel-titanium alloy (Nitinol) [280,288,289,300]. Those metal

alloys demonstrate the shape-memory effect due to thethermally-driven reversible phase transition from a sym-metrical austenitic phase at high temperatures to anasymmetrical martensitic phase [280,287,289]. SMAs can

duced with permission from the National Academy of Sciences of the USA.(For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)

be patterned as thin films to deflect substrates into 3Dstructures [301305], or as actuating hinges that inducethe folding of 2D planes when heated (Fig. 21) [20,254].

SMPs were first developed commercially in 1980s andreceived much attention due to their light weight, flexi-bility, high transformation strain, and low actuation forcescompared to similar properties observed in metal alloys[280]. SMPs possess great potential in a variety of appli-

cations such as automatic switches or sensors, intelligentpackaging and tissue engineering, and have been reviewedthoroughly [281,282,289,294296,298,306315]. Both themolecular architecture of polymers and the programming

Y. Liu et al. / Progress in Polymer S

Fig. 19. (a) Folding of a sandwiched sheet with shrink film in the middlewhile heat resistant materials on the top and bottom layers. The arrowsshow the shrinking directions [274], Copyright 2014. Reproduced withpermission from the IEEE; (b) a fluorescence image of a folded flappingbird (bottom) made from the trilayer film (top) with photo-crosslinkabletemperature-responsive polymer (PNIPAM) in between [279], Copyright2015. Reproduced with permission from John Wiley & Sons Inc.; (c) Anicosahedron (bottom) made from the template (top) incorporated withpwp

ptcmbtattt

Fm2

re-strained polymers inside a trilayer laminate. The folding is actuatedith uniform heating in an oven [273], Copyright 2014. Adapted withermission from IOP Publishing.

rocess govern the shape memory behavior of SMPs. Aypical programming process uses a thermo-mechanicalycle (Fig. 22a) that starts with the polymer in the per-anent shape, deforms the polymer to a temporary shape

y applying an external force over the thermal transitionemperature (Ttrans) (e.g., Tg or melting temperature, Tm),nd cools the polymer to preserve that temporary struc-

ure. Upon heating above Ttrans, the polymer recovers tohe permanent shape (i.e., the recovered shape) becausehe polymer chains relax back to the entropically favorable

ig. 20. Photographs of a folded cube from a pre-programmed 2D shape-emory polymer sheet that folds in response to heat [282], Copyright

010. Reproduced with permission from John Wiley & Sons Inc.

cience 52 (2016) 79106 93

state. The shape memory effect is commonly characterizedby a cyclic thermo-mechanical strain-temperature-stressdiagram (Fig. 22b) [316]. The strain recovery rate, whichdescribes how much the polymer memorizes the perma-nent shape, and the strain fixity rate, which shows how wellthe polymer can be fixed at the temporary shape, quanti-fies the shape memory performance [287]. Most SMPs haveone temporary shape. Several researchers have developedclever ways to introduce more than one temporary shapeinto a SMP [317321].

As shown in Fig. 23, the ability for polymer to remem-ber its permanent shape relies on cross-links in the form ofeither covalent bonds (i.e., chemical cross-linking) or strongintermolecular interactions (e.g., chain entanglement orcrystalline domains). So-called switching segments fixthe temporary shape of the polymer network and preventthe flow of polymer chains below the transition tempera-ture. Although Fig. 23 summarizes switching approaches,the most common switching segments are frozen poly-mer chains at T < Tg for amorphous polymers or polymers incrystalline domains at T < Tm for semi-crystalline polymers[280,287,309,310].

Some macroscopically-homogeneous planar materialsadopt 3D shapes under uniform stimuli due to the gradientsof crosslink densities that form across the sheet (Fig. 24).

In addition, stimuli can be applied locally to thehinge region, by, for example, Joule-heating [20,254,322],focused laser light [265,266], or pneumatics [323].

Recently, our group reported on a simple approach toachieve folding using homogeneous 2D pre-strained poly-mer sheets with inkjet-printed hinges on a planar surface[324]. Pre-strained polymer sheets are a type of shapememory polymer that shrinks in-plane if heated uniformly.Black toner patterned on the surface provides localizedabsorption of light, which heats the underlying polymerto temperatures above the Tg to induce the local shrink-age of the polymer so that the planar sheet folds into a3D shape. Various 3D structures have been demonstratedusing this approach from simple folding to complex struc-tures (Fig. 25) and a simple geometric model predicts itsfolding angle based on strain relaxation [325]. It is also pos-sible to induce the folding by irradiating the samples withlight patterned in the shape of a hinge [266].

2.5. Discussion for out-of-plane shape programming(Strategies 3 and 4)

The approaches described in Strategies 3 and 4 oftenpresent trade-offs. For example, patterned hinges fre-quently require multiple fabrication steps but can resultin a range of complex 3D structures (Figs. 18 and 21). Incontrast, bilayer laminates are simple, but deform into alimited set of geometries and only produce sharp foldswhen combined with patterned, rigid sheets (Fig. 19).SMPs require pre-programming and are usually actuatedby heating (Fig. 20). Hydrogels are biocompatible but theiruse is usually limited to liquid environments and materi-

als that are often very soft. Thermal actuation is one of themost popular means to induce folding or bending due toits simplicity and the availability of thermal triggers (i.e.,light, resistive heating and thermal radiation), but the high


Fig. 21. (a) SMA actuators can be folded by applying Joule-heating (red arrows show the heating location). (b) Demonstration of folding a boat from a 2Dfolding tf the re

planar composite sheets patterned with SMA hinges which can cause the from the National Academy of Sciences of the USA. (For interpretation oversion of this article.)

temperatures required often limit the applicability of suchstructures. More generally, selecting materials for shapeprogramming often presents a trade-off between elasticstrain and modulus. Fig. 26 summarizes the mechanicalcharacteristics of a wide range of existing materials appliedfor reconfigurable and morphing structures [326328]. Thered dashed curve in the upper right corner of Fig. 26 depictsthe trade-off between strain and stiffness for those struc-tures. Materials that can be deformed significantly (i.e.,large strain) tend to have a lower modulus, which limitstheir utility. Polymers included in the diagram in Fig. 26are no exception; elastomers or gels possess a low modulusand can be extended to large strains. For many applications,

it is desirable to have large deformations and large modulito store mechanical energy. Development of materials thatmeet these criteria represents a great research challenge aswell as opportunity.

Fig. 22. (a) Schematic of a typical thermo-mechanical process for programming shred; and (b) A representative cyclic strain-temperature-stress diagram. In this spetemperature) and is (i) deformed (elongated by stretching), (ii) cooled and then of the sample is recovered by heating again [130], Copyright 2007. Reproduced wof the references to color in this figure legend, the reader is referred to the web v

hrough Joule-heating [20], Copyright 2010. Reproduced with permissionferences to color in this figure legend, the reader is referred to the web

Table 1 summarizes representative stimuli for fold-ing and bending of polymeric materials, length scales,and reversibility [4,57,255,288]. Among all parameterslisted in Table 1, reversibility is one of top considerationsfor realizing practical applications of out-of-plane actu-ation. Hydrogels, electroactive polymers, and LCEs haveexhibited various degrees of reversibility. Swelling/de-swelling of gels is a reversible process but it takes placetypically only in liquid environments, diffusion of liq-uid in and out of the gel limits the rate of actuation,and gels are typically soft and brittle. The actuation ofphotoresponsive shape memory polymers and liquid crys-tal elastomers with photosensitive groups is reversible

by switching the exposure wavelengths (usually in theUV range). Bilayers with differential thermal expansionbend reversibly over small displacements. In contrast,the folding of most SMPs is usually irreversible (i.e., the

ape-memory polymers. T < Ttrans is represented in blue while T > Ttrans is incific case, a heated sample starts from the asterisk (i.e., low stress, high

(iii) the stress applied for deformation is released, (iv) the original shapeith permission from the Royal Society of Chemistry. (For interpretation

ersion of this article.)

Y. Liu

et al.

/ Progress

in Polym

er Science

52 (2016)

79106

95Table 1Representative approaches for folding and bending of polymer sheets. The table is organized by the phenomena and stimuli. The term length scale here refers to the dimension of the hinges or curvature ofbending.

Phenomenon Stimulus Materials Length scale Advantages Disadvantages Reversibility Reference

Differential thermalexpansion (e.g.,bimorph)

Heat Bilayers with differentthermal expansioncoefficients (e.g.,polymer/metal)

mmm Simple and availablethermal actuation

Elevated temperaturerequired, curvaturedepends on degree ofmismatch; limitedcurvature induced

Yes [160,164,363365]

Volumetricexpansion /contraction

Solvents Gels mmm Simple triggering inliquid

Restricted to liquidenvironment

Yes [161,166,366]

Temperature Thermal-responsivepolymers (e.g.,temperature-responsivehydrogels)

mmm Simple thermalactuation

Asymmetry needed Yes [160,170,176,367,368]

Pressure(pneumatics)

Elastomers (e.g.,Ecoflex, PDMS);Parylene

mmcm Inexpensive, lightweight, high load; fastresponse time (ms sdepending on air flowrate)

Source/control of fluidpressure needed; requiresmismatch inflexibility/modulusbetween layers forasymmetrical bending

Yes [197199,369]

pH, ionicconcentration

Gels mmm Simple operation Restricted to liquidenvironment

Yes [144,148,152,370]

Electricfield/chemicalpotential

Piezoelectric polymers(e.g., PVDF);electroactivehydrogels; ionicpolymers

mmm Simple application ofelectric field orchemical potential

Requires liquidenvironment, usually slowresponse exceptpiezoelectric polymers

Yes [11,185187,237,238,371,372]

Surface tension Heat Solder, tin, polymers,etc.

nmmm Hinges can be sealed(locked) after folding

Elevated temperature tomelt metals

No [36,271]

Magnetism Magnetic fields Ferromagneticmaterials

nmmm Remote control bymagnetic field

Locking mechanismrequired to keep it inposition

Yes [181,373377]

Shape memory Heat Shape memorypolymers

mmm Simple thermalactuation

Programming required forSMPs or heat must bedelivered locally

No [254,281,282,287]

Light to inducemolecularchanges

Liquidcrystalelastomers and gels orphotosensitivepolymers

mmm Remote; roomtemperature

Usually requires UV light Yes [168,205,219,220,291,378380]

Light to deliverheat

Pre-strained shrinkfilms

mcm Simple printingapproach

Unproven at small lengthscales

Noa [254,269,324]

Shrinkage Heat Polyimide (PI) m Different angledisplacement can beachieved depending oncuring temperature

High temperature needed(250500 C); pattern PIhinge

Nob [267,268]

a Also includes approach as described in Fig. 25.b But can dynamically control angle by heating due to thermal expansion of polyimide [267].


SMPs [3
Fig. 23. Summary of molecular structures and switching mechanisms of Chemistry.
so-called one-way shape memory effect) due to the strainrelaxation that occurs. Likewise, actuation of pre-strainedpolymers is typically irreversible because the first actua-tion relaxes (often completely) the stored strain. Several

Fig. 24. (a) Schematic of a patterned hinge through a photomask due to photo-inhinges (right) [5], Copyright 2012. Reproduced with permission from the American(N-isopropylacrylamide-co-acrylic acid) (pNIPAM-AAc) and induced folding, and2014. Reproduced with permission from IOP Publishing.

10], Copyright 2013. Adapted with permission from the Royal Society of

recent reports discuss polymers with two-way shape mem-ory effects [310,329338]. The term two-way impliesthat the polymers can change shape reversibly multi-ple times simply by changing the stimulus. The two-way

duced stress relaxation in a polymer (left) and a closed box with multiple Institute of Physics; (b) Schematic of local crosslink gradient through poly

the demonstrated folding and unfolding of the gripper [272], Copyright


Fig. 25. Photographs of 3D structures formed by folding of pre-strained polymer sheets patterned with black hinges using a desktop printer. The blacki ingle lins 20 mm f 2. Repr

spttga

n

FometamlePP[t

nk absorbs light from a lamp to induce folding via localized heating. (a) Sides width hinge with of 1 mm and 12 mm spacing; (c) rectangular box (acet (10 mm 10 mm) and hinge widths of 2.0 mm) [324], Copyright 201

hape memory effect has been explored in semicrystallineolymers [310,329,330,332,336,337,339345]. In general,hese systems shrink due to chain relaxation at elevatedemperature, but revert to a different shape due to stressesenerated by the reformation of crystals at lower temper-

tures.

Response time for folding is an important parameterot listed in Table 1. Swelling/de-swelling in solvents is

ig. 26. Summary of existing active materials applied for reconfigurabler morphing structures show a trade-off between maximum strain andodulus. Red dashed curve on the top right part shows the limit of

xisting materials for these two materials properties. Abbreviations ofhe various compounds (added by the authors of this review, not theuthors of the original article): CFRP: carbonfiberreinforced poly-er; FSMA: ferromagnetic shape memory alloys; G-10: fiberglass epoxy

aminates; IPMC: ionic polymer metal composite; LCE: liquid crystallastomers; PAN:polyacrylonitrile; PMMA: poly(methyl methacrylate);VDF: poly(vinylidene fluoride); PZN-PT: (1 x)Pb(Zn1/3Nb2/3)xPbTiO3;ZT: lead zirconate titanate (PbZr0.52Ti0.48O3); SMA: shape memory alloys.327], Copyright 2010. Reproduced with permission from SAGE Publica-ions.

e with hinge width (w) of 1 mm; (b) three lines patterned on alternating 10 mm 10 mm, w = 1.5 mm); (d) tetrahedral box with a square bottom

oduced with permission from the Royal Society of Chemistry.

usually slow due to inherently slow process of diffusionwhich takes minutes to hours. Chemical changes inducedby light cause folding relatively slow (few minutes). Pneu-matic actuation occurs typically within seconds or less andcan be done reversibly. Actuation in magnetic field is alsorelatively fast (seconds), and can be done reversibly. Ther-mal actuation (including light irradiation for heating) canalso be fast (a few seconds).

Another challenge facing the field of out-of-plane actu-ation is the relatively simple 3D structures that have beendemonstrated to date, such as single folds and cubes. Thereare some more complex structures beginning to appear inthe literature [277,279]. The ability to provide sequentialfolding of hinges (rather than all hinges folding at once)may help increase the level of complexity.

Although most current research on converting 2D pla-nar sheets to 3D out-of-plane structures by folding doesnot use paper, it is often called origami engineeringwhich covers a wide range of topics including develop-ment of responsive materials, the study of algorithmsand mechanics for designing folded structures [346353],the establishment of physical models (e.g., to predict andunderstand compliant mechanisms and its link to origamifolding [354358]), the development of new design toolsfor fabrication [359], and the implementation of novelconcepts for structure or topography optimization (e.g.,hinge design and layout, folding sequence, reconfiguration)[360362].

In summary, currently there is no universal, idealapproach that provides reversibility, fast response, largeforces or torques, geometric and sequential control, andease of fabrication. Realizing systems that have most orall of these properties is a goal of materials researchersworking on shape programming.

2.6. Applications for out-of-plane shape programming

(Strategies 3 and 4)

Out-of-plane shape transformation from 2D to 3Dis an appealing strategy to generate 3D structures


Fig. 27. Additional examples of applications for out-of-plane shape programming. (a) A self-assembled robot by folding [277], Copyright 2014. ReproducedScience;oyable a

with permission from the American Association for the Advancement of 2011. Reproduced with permission from Nature; (c) Partially opened deplfrom SPIE Proceedings.

from 2D patterns for many applications spanningmicroscale to macroscale. Examples include con-tainers for drug delivery and biomedical devices,[25,38,255,281,381385]; actuators, grippers and robotics(Fig. 27a); [24,164,185,186,363,364,386] robotic micro-hand or minimally invasive surgery; [26,257,281,387390]optical sensors and devices; [22,32,265,391] 3D microflu-idic devices (Fig. 27b); [178,392,393] reconfigurabledevices or robots; [20,33,68,254,322,394396] adhesivesand interlocks [29,397] nanoinjector; [398] programmablelithography for nanostructures [399] smart packaging;deployable structures (Fig. 27c); [251,400404] andshape-shifting mobile devices [28].

3. Conclusion, opportunities, and challenges forshape programming

Shape-programming of materials from 2D sheets to3D shapes is appealing in many applications includingreconfigurable devices, responsive actuators, and assem-bly processes. This review motivates the advantages ofcreating 3D shapes from 2D sheets and introduces sev-eral representative strategies and mechanisms for shapeprogramming polymeric materials involving out-of-planedeformation, i.e., folding, bending, rolling, contraction,expansion, induced surface topography, and conventionallithographic processes. This review focuses primarilyon strategies for out-of-plane actuation, of polymericmaterials from 2D sheets to 3D shapes, as well as rel-evant programmable materials within the context offolding.

Although great progress has been made toward shapeprogrammable materials and relevant methods from 2Dto 3D, there many opportunities still exist for new shapeprogramming approaches, especially folding/bending, toevolve regarding design optimization and materials devel-opment.

In this context, novel strategies need to be devel-

oped that lower the complexity of fabrication of foldingsubstrates, without the loss of high fidelity (i.e., withfew defects such as over-folding or under-folding). Manycurrent strategies rely on multi-step microfabrication

(b) 3D dual-channel microfluidics. Scale bar is 500 m [178], Copyrightrray for a solar panel. [403], Copyright 2015. Reproduced with permission

processes. Second, the most popular stimulus is thermalactuation due to its ease of implementation. How-ever, other stimuli such as chemical and mechanicalresponses are also desired. Third, although efforts havebeen made to control and program the folding pathways,[20,254,265,405407] opportunities exist to better controlthe folding/unfolding pathways that may lead to sequentialor programmed folding to create more complex geometriesand more sophisticated applications, which is interest-ing for 4D assembly (i.e., 3D structures that changeshape with time) or printing where the motion or shapeis highly correlated with time [408,409]. Fourth, the rateof folding could be improved along with precise controlover folding dynamics. Employing strategies such as snap-buckling instabilities (e.g., those employed by the Venusfly trap [161,410,411]) may allow for more rapid foldingrelative to mechanisms that rely on polymer shrinkageor gel swelling, which are generally slow. In addition,most folding occurs slowly and often requires large inputsof energy, which represents new opportunities for opti-mization and improvement. Fifth, the actuation associatedwith many folding strategies does not generate significantforce (or torque), which limits the utility. Sixth, effort inmodeling and folding algorithms is needed to guide theoptimal design of folding/bending processes. Lastly, meth-ods based on kirigami (origami with cuts) are emerging aspromising methods for making 3D and out of plane struc-tures with greater design freedom for shape programming[352,353,412415].

Current materials and processes often do not meetthe needs of applications. For example, shape reversiblematerials are desired for practical applications that callfor reconfigurability. Many current polymers undergoirreversible shape change in response to a stimulus (e.g.,shape change usually represents a minimization of energy),while the approaches and materials that are reversible arelimited to a narrow range of external stimuli that are ofteninconvenient such as large changes in temperature or theneed to operate in a liquid environment (for controlling

pH, ionic concentration). Thus, there are opportunities todevelop new polymers or mechanisms for reversible shapechange.

olymer S

ratepda

A

tom

R

write lithography into photopolymer. Appl Opt 2007;46:295.


There are also opportunities for modelers and theo-eticians to provide guidance for optimized design. Mostpplications are case-dependent regarding materials selec-ion and stimuli utilization. The performance metrics (e.g.,nergy efficiency, performance versus fabrication com-lexity and economic costs, pathway, accuracy, fidelity,urability, packing ratio) should guide the design strategynd materials selection.

cknowledgements

We thank the National Science Foundation for suppor-ing this work under the grant no. 1240438. We also thankur collaborators in the NSF EFRI program and beyond forany fruitful interactions during the past few years.

eferences

[1] Lauff C, Simpson TW, Frecker M, Ounaies Z, Ahmed S, von Lock-ette P, Strzelec R, Sheridan R, Lien J-M. In: Differentiating BendingFrom Folding in Origami Engineering Using Active Materials. ASME;2014. V05BT08A040/1-12.

[2] Peraza-Hernandez EA, Hartl DJ, Malak Jr RJ, Lagoudas DC. Origami-inspired active structures: a synthesis and review. Smart MaterStruct 2014;23, 094001/1-094001/94001.

[3] Abbott AC, Buskohl PR, Joo JJ, Reich GW, Vaia RA. In: Characteriza-tion of Creases in Polymers for Adaptive Origami Structures. ASME;2014. V001T01A009/1-7.

[4] Ionov L. Soft microorigami: self-folding polymer films. Soft Matter2011;7:678691.

[5] Ryu J, DAmato M, Cui X, Long KN, Qi HJ, Dunn ML. Photo-origamibending and folding polymers with light. Appl Phys Lett2012;100:161908/1161908.

[6] Chen D, Yoon J, Chandra D, Crosby AJ, Hayward RC. Stimuli-responsive buckling mechanics of polymer films. J Polym Sci, B:Polym Phys 2014;52:144161.

[7] Kempaiah R, Nie Z. From nature to synthetic systems: shape trans-formation in soft materials. J Mater Chem, B 2014;2:235768.

[8] Gracias DH. Stimuli responsive self-folding using thin polymerfilms. Curr Opin Chem Eng 2013;2:1129.

[9] Vaia R, Baur J. Materials science: adaptive composites. Science2008;319:4201.

[10] Geryak R, Tsukruk VV. Reconfigurable and actuating structuresfrom soft materials. Soft Matter 2014;10:124663.

[11] Ionov L. Biomimetic hydrogel-based actuating systems. Adv FunctMater 2013;23:455570.

[12] Ionov L. Polymer origami: programming the folding with shape.E-Polym 2014;14:10914.

[13] Ionov L. Polymeric actuators. Langmuir 2015;31:501524.[14] Ionov L. Bioinspired microorigami by self-folding polymer films.

Macromol Chem Phys 2013;214:117883.[15] Ionov L. 3D Microfabrication using stimuli-responsive self-folding

polymer films. Polym Rev 2013;53:92107.[16] Ionov L. Hydrogel-based actuators: possibilities and limitations.

Mater Today 2014;17:494503.[17] Dealy JM, Wissbrun KF. Melt rheology and its role in plas-

tics processing: theory and applications. Dordrecht, Netherland:Springer; 1990. p. 665.

[18] Rosato DV. Plastic product material and process selection hand-book. New York, NY, USA: Elsevier; 2004. p. 618.

[19] Madou MJ. Fundamentals of microfabrication: the science of minia-turization. 2nd ed. Boca Raton, FL: CRC Press; 2002. p. 723.

[20] Hawkes E, An B, Benbernou NM, Tanaka H, Kim S, Demaine ED, RusD, Wood RJ. Programmable matter by folding. Proc Natl Acad SciUSA 2010;107:124415.

[21] Khan MR, Hayes GJ, So JH, Lazzi G, Dickey MD. A frequency shiftingliquid metal antenna with pressure responsiveness. Appl Phys Lett2011;99:013501/113501.

[22] Cho JH, Keung MD, Verellen N, Lagae L, Moshchalkov VV, Van

Dorpe P, Gracias DH. Nanoscale origami for 3D optics. Small2011;7:19438.

[23] Bassik N, Abebe BT, Laflin KE, Gracias DH. Photolithographi-cally patterned smart hydrogel based bilayer actuators. Polymer2010;51:60938.

cience 52 (2016) 79106 99

[24] Ilievski F, Mazzeo AD, Shepherd RF, Chen X, Whitesides GM. Softrobotics for chemists. Angew Chem Int Ed 2011;50:18905.

[25] Small W, Wilson TS, Benett WJ, Loge JM, Maitland DJ. Laser-activated shape memory polymer intravascular thrombectomydevice. Opt Express 2005;13:820413.

[26] Small IVW, Singhal P, Wilson TS, Maitland DJ. Biomedical applica-tions of thermally activated shape memory polymers. J Mater Chem2010;20:335666.

[27] Kuribayashi K, Tsuchiya K, You Z, Tomus D, Umemoto M, Ito T,Sasaki M. Self-deployable origami stent grafts as a biomedical appli-cation of Ni-rich TiNi shape memory alloy foil. Mater Sci Eng, A:Struct Mater 2006;419:1317.

[28] Roudaut A, Karnik A, Lchtefeld M, Subramanian S. Morphees:toward high shape resolution in self-actuated flexible mobiledevices. In: Bdker S, Brewster S, editors. Proc SIGCHI Conf HumFactors Comput. Syst. Paris, France: ACM Press; 2013. p. 593602.

[29] Eisenhaure JD, Xie T, Varghese S, Kim S. Microstructured shapememory polymer surfaces with reversible dry adhesion. ACS ApplMater Interfaces 2013;5:77147.

[30] Blewitt MJ, Willits RK. The effect of soluble peptide sequences onneurite extension on 2D collagen substrates and within 3D collagengels. Ann Biomed Eng 2007;35:215967.

[31] Efimenko K, Finlay J, Callow ME, Callow JA, Genzer J. Developmentand testing of hierarchically wrinkled coatings for marine antifoul-ing. ACS Appl Mater Interfaces 2009;1:103140.

[32] Xu H, Yu C, Wang S, Malyarchuk V, Xie T, Rogers JA. Deformable, pro-grammable, and shape-memorizing micro-optics. Adv Funct Mater2013;23:3299306.

[33] Hayes GJ, Liu Y, Genzer J, Lazzi G, Dickey MD. Self-folding origamimicrostrip antennas. IEEE Trans Antennas Propag 2014;62:54169.

[34] Xia Y, Rogers JA, Paul KE, Whitesides GM. Unconventional meth-ods for fabricating and patterning nanostructures. Chem Rev1999;99:182348.

[35] Reichmanis E, Houlihan FM, Nalamasu O, Neenan TX. Chemicallyamplified resists: chemistry and processes. Adv Mater Opt Electron1994;4:8393.

[36] Leong TG, Lester PA, Koh TL, Call EK, Gracias DH. Surface tension-driven self-folding polyhedra. Langmuir 2007;23:874751.

[37] Cho JH, Gracias DH. Self-assembly of lithographically patternednanoparticles. Nano Lett 2009;9:404952.

[38] Leong TG, Randall CL, Benson BR, Bassik N, Stern GM, Gracias DH.Tetherless thermobiochemically actuated microgrippers. Proc NatlAcad Sci USA 2009;106:7038.

[39] Kern W. Thin film processes II. Boston, MA: Academic Press; 1991.p. 866.

[40] Thornton JA. High rate thick film growth. Annu Rev Mater Sci1977;7:23960.

[41] Kovacs GTA, Maluf NI, Petersen KE. Bulk micromachining of silicon.Proc IEEE 1998;86:153651.

[42] Bustillo JM, Howe RT, Muller RS. Surface micromachining for micro-electromechanical systems. Proc IEEE 1998;86:155274.

[43] Gates BD, Xu Q, Love JC, Wolfe DB, Whitesides GM. Unconventionalnanofabrication. Annu Rev Mater Res 2004;34:33972.

[44] Rogers JA, Lee HH, editors. Unconventional nanopatterning tech-niques and applications. Hoboken, NJ: John Wiley & Sons Inc; 2009.p. 617.

[45] Guo LJ. Recent progress in nanoimprint technology and its applica-tions. J Phys Appl Phys 2004;37:R12341.

[46] Guo LJ. Nanoimprint lithography: methods and material require-ments. Adv Mater 2007;19:495513.

[47] Costner EA, Lin MW, Jen WL, Willson CG. Nanoimprint lithographymaterials development for semiconductor device fabrication. AnnuRev Mater Res 2009;39:15580.

[48] Chou SY. Nanoimprint lithography. J Vac Sci Technol B: Microelec-tron Nanometer Struct 1996;14:412933.

[49] Schift H. Nanoimprint lithography: 2D or not 2D? A review. ApplPhys, A 2015;121:41535.

[50] Ruiz R, Kang H, Detcheverry FA, Dobisz E, Kercher DS, AlbrechtTR, de Pablo JJ, Nealey PF. Density multiplication and improvedlithography by directed block copolymer assembly. Science2008;321:9369.

[51] Piner RD, Zhu J, Xu F, Hong S, Mirkin CA. Dip-pennanolithography.Science 1999;283:6613.

[52] Sullivan AC, Grabowski MW, McLeod RR. Three-dimensional direct-

[53] Yang SM, Jang SG, Choi DG, Kim S, Yu HK. Nanomachining by col-loidal lithography. Small 2006;2:45875.

[54] Zhang G, Wang D. Colloidal lithography-the art of nanochemicalpatterning. ChemAsian J 2009;4:23645.

http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1000http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1005http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1010http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http://refhub.elsevier.com/S0079-6700(15)00102-1/sbref1015http:/

progress in polymer science - nc state university · progress in polymer science 52 (2016) 79–106...

Documents