Sideways and stable crack propagation in asilicone elastomerSeunghyun Leea and Matt Pharra,1

aDepartment of Mechanical Engineering, Texas A&M University, College Station, TX 77843

Edited by John W. Hutchinson, Harvard University, Cambridge, MA, and approved March 26, 2019 (received for review November 30, 2018)

We have discovered a peculiar form of fracture that occurs in ahighly stretchable silicone elastomer (Smooth-On Ecoflex 00–30).Under certain conditions, cracks propagate in a direction perpen-dicular to the initial precut and in the direction of the applied load.In other words, the crack deviates from the standard trajectoryand instead propagates perpendicular to that trajectory. The crackarrests stably, and thus the material ahead of the crack front con-tinues to sustain load, thereby enabling enormous stretchabilities.We call this phenomenon “sideways” and stable cracking. To ex-plain this behavior, we first perform finite-element simulationsthat demonstrate a propensity for sideways cracking, even in anisotropic material. The simulations also highlight the importanceof crack-tip blunting on the formation of sideways cracks. Next,we provide a hypothesis on the origin of sideways cracking thatrelates to microstructural anisotropy (in a nominally isotropic elas-tomer). To substantiate this hypothesis, we transversely prestretchsamples to various extents before fracture testing, as to determinethe influence of microstructural arrangement (chain alignment andstrain-induced crystallization) on fracture energy. We also performmicrostructural characterization that indicates that significantchain alignment and strain-induced crystallization indeed occurin this material upon stretching. We conclude by characterizinghow a number of loading conditions, such as sample geometryand strain rate, affect this phenomenon. Overall, this paper pro-vides fundamental mechanical insight into basic phenomena asso-ciated with fracture of elastomers.

fracture | elastomers | soft materials

Soft materials are vital in flexible electronics, soft robotics,drug delivery, microfluidics, bioengineering, and adaptive

optics, among others (1–12). Applications in these areas oftenrequire large stretchability, flexibility, and toughness due to re-peated stretching, bending, and twisting during operation. Assuch, elastomers have emerged as a material class of choice inthese areas due to their low cost, large stretchabilities, andchemical tunabilities. In particular, chemical modifications haveenabled certain classes of elastomers to stretch from hundreds tothousands of percent before fracture (13–16). Owing to theirlarge stretchabilities, fracture of elastomers largely differs fromother material classes. For instance, elastomers often exhibitcrack-tip blunting and/or rate-dependent fracture toughnesses(17–20). Likewise, the fracture toughness of elastomers highlydepends on the size of initial flaws (21), and the initial crack-tipsharpness (22). Moreover, cracks in elastomers sometimes de-viate from “standard” paths, particularly in the initial stages oftheir growth (23). For instance, Hamed and Park (24) studiedfracture of carbon-black–filled natural rubber. Near the tip of anedge crack, they observed small auxiliary cracks that initiallyturned in somewhat stochastic directions (including some per-pendicular relative to the initial edge crack), before eventuallyturning back to the standard direction across the specimen (i.e.,back to the direction of the edge crack). Additionally, Green-smith (25) studied carbon-black–filled compounds and foundthat under certain conditions, the cracks deviated from thestandard direction by either curving or bifurcating into twocracks. However, upon further loading, a second crack emanated

from this region, again in the standard direction. This behavior isoften referred to as “stick–slip” or “knotty” tearing (25, 26).Overall, experimental observations suggest that fracture is fun-damentally different in elastomers compared with other materials,thus compelling further studies aimed at providing basic under-standing of their fracture.To this end, this paper investigates fracture of a highly

stretchable silicone elastomer (Smooth-On Ecoflex 00–30). De-spite this specific elastomer’s use in various research fields(particularly that of flexible/stretchable electronics) (27–32), itsfracture characteristics and basic polymer properties have notbeen systematically studied. During fracture testing, we find anintriguing phenomenon: Under certain conditions, cracks prop-agate in a direction perpendicular to the initial precut and in thedirection of the applied load. In other words, the crack deviatesfrom the standard trajectory and instead propagates in a di-rection perpendicular to that trajectory. We call this phenome-non “sideways” and stable cracking. Compared with previousstudies from literature, the behavior we have observed does notoccur intermittently [e.g., as in the case of bifurcation, stick–slip,or knotty tearing (25, 26, 33, 34)] or solely in the initial stages offracture [as in the Hamed and Park study (24)]; instead, sidewayscrack growth is fully stable in the elastomer described herein.The length of the sideways crack slowly increases with increasingload, and the crack arrests if the loading stops. The crack neverturns back to the standard direction or shows any bifurcation.Thus, as a practical ramification of this behavior, the materialahead of the crack front continues to sustain load, thereby en-abling enormous stretchabilities. To explain this phenomenon,

Significance

Soft materials exhibit fundamentally different fracture char-acteristics from other materials, and as such represent an openand fascinating research area. Herein, we have discovered aform of fracture in soft elastomers that we call “sidewayscracking” in which cracks propagate perpendicular to their“standard” trajectory. These sideways cracks stably arrest,thereby allowing the material ahead of the crack to continue tosustain large loads. As such, understanding this phenomenonmay enable the engineering of highly robust and stretchablematerials. To explain this behavior, we perform mechanics-based analyses and substantiate a hypothesis that this be-havior stems from structural rearrangement of polymer chainsduring stretching. Overall, this paper provides fundamentalmechanical insight into basic phenomena associated withfracture of elastomers.

Author contributions: S.L. and M.P. designed research; S.L. performed research; S.L. andM.P. analyzed data; and S.L. and M.P. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1To whom correspondence should be addressed. Email: m-pharr@tamu.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1820424116/-/DCSupplemental.

Published online April 19, 2019.

www.pnas.org/cgi/doi/10.1073/pnas.1820424116 PNAS | May 7, 2019 | vol. 116 | no. 19 | 9251–9256

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Apr

il 12

, 202

0

we first perform finite-element simulations that demonstrate thepropensity for sideways cracking, which highlight the importanceof crack-tip blunting. We then provide a hypothesis on the originof sideways cracking related to microstructural anisotropy thatdevelops during stretching (in a nominally isotropic elastomer).To substantiate this hypothesis, we transversely prestretch sam-ples to various extents before fracture testing to determine theinfluence of microstructural arrangement (chain alignment andstrain-induced crystallization) on fracture energy. We also per-form microstructural characterization that indicates that signifi-cant chain alignment and strain-induced crystallization occur inthis material upon stretching. We conclude by characterizinghow a number of loading conditions, such as sample geometryand loading rate, affect this phenomenon.

Results and DiscussionDescription and Implications of Sideways Cracking. We have dis-covered that a peculiar form of fracture occurs in particularsilicone elastomers. This fracture is fundamentally distinct fromtypically found in other materials such as ceramics, metals, brittlepolymers, and even other commonly used elastomers. Specifi-cally, under certain conditions, cracks propagate in a directionperpendicular to the initial precut and in the direction of theapplied load. The length of the sideways crack slowly increaseswith increasing load, and the crack arrests if the loading stops.The crack never turns back to the standard direction or showsany bifurcation. Fig. 1 shows snapshots of this phenomenon (thecorresponding movie is shown in Movie S1) in the siliconeelastomer studied herein (Smooth-On Ecoflex 00–30); we callthis phenomenon sideways and stable cracking (23, 35, 36).Beyond being scientifically intriguing, these results have im-

portant ramifications from a practical perspective. Specifically,since the sideways cracks stably arrest, the uncracked region ofthe material ahead of the crack front can continue to sustain

load, thereby enabling significantly larger stretchabilities. Fig. 2illustrates this point; despite the relatively early onset of sidewayscrack propagation in a precut specimen (red curve at around λ =1.7), the specimen continues to sustain loads to much largerstretches, in a manner reflective of a pristine specimen (with noprecut, blue curve). In fact, if the stresses in Fig. 2 are replottedto account only for uncracked areas (i.e., that of the regionahead of the crack), the red and blue curves are virtually iden-tical. Furthermore, the uncracked region ahead of the crack tipcompletely recovers after a cycle of loading/unloading, i.e., de-formation continues to be elastic ahead of the crack tip (Fig. 3).

Mechanics Analysis of the Origin of Sideways Cracking. The phe-nomenon of sideways crack growth is likely unintuitive. In par-ticular, in most fracture mechanics analysis, a large stressconcentration exists at the crack tip that drives the crack in theforward (standard) direction. However, elastomers are quitecompliant and stretchable; as a result, crack-tip blunting readilyoccurs. Blunting reduces the stress concentration at the crack tip;correspondingly, fundamentally different fracture behavior canoccur in these systems. Furthermore, crack-tip blunting in elas-tomers usually does not involve highly irreversible deformationand thus does not dissipate much energy in and of itself. Instead,the main mechanism for dissipating energy is through releasingelastic strain energy during crack propagation. With these effectsin mind, we aim to provide a general explanation of the me-chanics that can lead to sideways crack propagation.To do so, we employed three-dimensional finite-element

analysis simulations of fracture in these systems. Specifically,we used the commercial software ABAQUS to subject a siliconeelastomer with a precut to a fixed level of stretch, from which thestored elastic energy, U, was computed. An eight-node linearbrick, hybrid (C3D8H) was applied as the element type, andnonlinear geometric effects were included to allow for finitedeformation. The simulations implemented a Mooney–Rivlinconstitutive model to represent a highly stretchable hyperelasticmaterial (31). We also performed a separate analysis thatimplemented a linear elastic constitutive model and found sim-ilar results (both qualitatively and quantitatively) to those de-tailed below. Thus, the particular form of the constitutive modeldoes not seem to affect our results. The initial crack size is c0,and we extended a crack in steps of Δc = c0/40 during each step.A uniform grid of elements was employed in all directions, andthe size of the grid is equal to one step of the crack growth size,Δc. The energy release rate was calculated by computing thestrain energy at each step of crack growth and using the re-lationship: G = −∂U/∂A = (Ui − Ui−1)/(Ai − Ai−1), where the Uirepresent the internal energies when a crack is a certain length(at step i), and the Ai represent the area of the cracked region atthe same step, i.e., the quantity (Ai − Ai−1) represents the area

ε = 0%

ε = 300%

ε = 200%

ε = 120%

Fig. 1. Photos of crack propagation in a silicone elastomer (Ecoflex 00–30)that demonstrates sideways cracking. The specimen is precut in the hori-zontal direction and subjected to a load in the vertical direction. Duringstretching, sideways cracking occurs: the crack turns perpendicular to thestandard (horizontal) direction. This sideways crack is stable, and the speci-men continues to sustain load, reaching large strains before failure(>>300%). The red dashed line shows the initial position of the precut, andthe yellow dashed line shows the position of the stable sideways propaga-tion. The blue boxed region shows a zoomed-in view of the crack tip.

0

20

40

60

80

100

120

1 1.5 2 2.5 3 3.5 4 4.5

)aPk( σ

λ

No pre-cut

Pre-cut (sideways

crack growth)

Onset of sideways crack propagation

Fig. 2. Stress–stretch curves of a silicone elastomer and correspondingfracture morphologies (thickness B = 5 mm, stretch rate _λ = 0.1/min).

9252 | www.pnas.org/cgi/doi/10.1073/pnas.1820424116 Lee and Pharr

Dow

nloa

ded

by g

uest

on

Apr

il 12

, 202

0

swept out by the crack during step i (37). To analyze the effect ofblunting on the propensity for crack propagation, simulationswere performed at different stretch levels: λ = 1.1 to represent arelatively sharp crack tip and λ = 2 to represent a blunted cracktip. λ = 2 also corresponds to a strain level representative ofcrack initiation in our precut silicone elastomer samples (Fig. 2).From the computed strain energy at each step, we calculate theenergy release rate when the crack grows in the sideways di-rection, Gs, and when the crack grows in the forward direction,Gf. Fig. 4 shows the simulations to compute the crack drivingforce in the forward and sideways directions at λ = 2 as a func-tion of the freshly created crack length, dc. At small crack lengths(corresponding to crack initiation), the crack driving force in thesideways direction, Gs, is around 50% that of crack driving forcein the forward direction, Gf. This result suggests that the crackdriving force in the forward direction is indeed larger than that inthe sideways direction, as one might intuitively expect. However,the relative value of the energy release rate associated withforward cracking is not orders of magnitude larger than that ofsideways cracking. Correspondingly, if the resistance to crackpropagation (i.e., the fracture energy) in the sideways direction islower than ∼50% of resistance to crack propagation in the for-ward direction, then a sideways crack should form. In short, it isnot unreasonable to expect that sideways crack propagationcould occur. Additionally, Fig. 5 shows the effect of crack-tipblunting on the propensity for sideways cracking. Gs/Gf forlargely blunted specimens (λ = 2) is around 48% when cs/cf issmall, and increases up to 60%, before eventually decreasingagain. By comparison, the largely unblunted specimen (λ = 1.1)shows smaller relative values of Gs/Gf. These results suggest thata sharp crack (e.g., in a material with low stretchability) requiresa higher degree of anisotropy of the fracture energy to inducesideways crack propagation.

Anisotropic Fracture Energy in a Nominally Isotropic SiliconeElastomer. The finite-element simulations suggest that only amild degree of anisotropy of fracture energy is required to in-duce sideways cracking. To this end, we performed experimentsto determine if the fracture energy of this silicone elastomer isindeed anisotropic. At first glance, it may not be obvious why its

fracture energy should be anisotropic. Namely, fabrication in-volves a seemingly random process (with no preferred orienta-tion) so we expect that polymer chains orient randomly, whichshould lead to isotropic mechanical properties. However, uponstretching of elastomers, their chains often align and can evenform crystalline structures (so-called strain-induced crystalliza-tion) in the direction of stretching (Fig. 6) (38, 39). We hy-pothesize that a crack aligned and propagating in a directionperpendicular to the alignment of polymer chains or crystallizeddomains (e.g., Fig. 6B) has a higher resistance to propagation;simply, it would have to break more chains/domains per unitlength of propagation. Correspondingly, we hypothesize that theresistance to fracture in directions parallel to the alignment (e.g.,propagation of a crack between aligned chains, as in Fig. 6C) isrelatively low, e.g., consider a crack propagating horizontally inFig. 6B compared with one propagating vertically in Fig. 6C. Ifthis were indeed the case, it would generally explain the phe-nomenon of sideways cracking.To this end, we performed comparative experiments of sam-

ples that were prestretched in the direction of the precut. Fig. 7shows the experimental setup for these tests. Samples andloading conditions (thin samples at high strain rates; see detailsin Effects of Sample Geometry and Loading Rate on the Propensityfor Sideways Cracking) were implemented that only inducedforward crack propagation to allow a fair comparison of thefracture energies. In these experiments, we applied differentprestretch levels to control the degree and direction of anisot-ropy (due to chain alignment and/or strain-induced crystalliza-tion), Fig. 7. We expect that prestretching the specimen in thedirection of the precut will align the polymer chains in that

Fig. 3. Photos of sideways crack propagation from a precut specimen. (A)During uniaxial stretching (in the vertical direction), a crack propagates inthe direction of the applied load. (B) An image of the crack after unloading,showing the stable propagation of a sideways crack.

Forward

Sideways

Forward Sidewaysdcf = 0 mm dcs = 0 mm

dcf = 2.54 mm dcs = 2.54 mm

(kPa)220

0

0100200300400500600

0 2 4 6 8 10 12 14

G (J

/m2 )

dc (mm)Fig. 4. Finite-element simulations comparing energy release rates for for-ward and sideways cracking. (Top) Finite-element simulations of the maxi-mum principal stress distribution in a silicone elastomer with a precut at λ =2.0 for a crack that has propagated by dc = 0 and dc = c0/10 (2.54 mm) in theforward direction (left images) and the sideways direction (right images).(Bottom) Energy release rate, G, for forward crack growth and sidewayscrack growth as a function of the length of newly created crack, dc.

Lee and Pharr PNAS | May 7, 2019 | vol. 116 | no. 19 | 9253

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Apr

il 12

, 202

0

direction (Fig. 7D). As a result, when a crack in this specimenadvances forward a unit length, it will break fewer polymerchains/domains compared with a specimen that is not pre-stretched, e.g., compare Fig. 7E to Fig. 7C. Correspondingly, weexpect that the prestretched specimen will exhibit a lower valueof fracture energy for forward cracking. Fig. 7A shows the resultsof tests performed at three different levels of prestretch, λp. Boththe strengths and the fracture strains of the samples decreasedwith increasing prestretch. In these experiments, the fractureenergy of the specimen is related to the area under stress–stretchcurves. The relative area under the curves for λp = 1.5 and λp =1.14 in Fig. 7A are 10% and 44%, respectively, of the area underthe curve at λ = 1. Thus, these results indeed indicate thatsamples with larger alignments of polymer chains in the directionperpendicular to the crack (e.g., Fig. 7C, λp = 1) demonstratemuch larger values of fracture energy for forward cracking (i.e.,they have much larger resistance to fracture), thereby supportingour hypothesis.To provide further evidence as to whether polymer chain

alignment and/or strain-induced crystallization occurred in thesespecimens, we performed a number of characterization studies.First, we implemented photoelasticity, which can provide evi-dence of polymer chain alignment (in a general sense) by usingthe alignment’s resulting influence on the optical properties ofthe material. SI Appendix, Fig. S1 shows the birefringence photosat a stretch of λ = 4. In SI Appendix, Fig. S1, Left, the polarizerand analyzer were set at angles of 0 and 90°, respectively, relativeto the stretching direction. In SI Appendix, Fig. S1, Right, thepolarizer and analyzer were rotated to 45 and 135°, respectively,relative to the stretching direction. SI Appendix, Fig. S1, Leftshows a quite dark color, whereas the right image shows a bright-white color, which indicates that chains do indeed align in thedirection of the applied load during stretching. We also utilizedX-ray diffraction (XRD) to probe microstructural changes dur-ing stretching of this elastomer. In SI Appendix, Fig. S2, the XRDpatterns of Ecoflex change during stretching from that repre-sentative of an amorphous material to that representative of asemicrystalline material. It can be seen that as the stretch in-creases, more crystallization occurs in the elastomer, as evi-denced by the substantial increase in intensity of the peaks near2θ = 22.5° and 2θ = 34° (SI Appendix, Fig. S2). It is also in-teresting to note in this figure that the elastomer apparently fullyrecovers back to its initial amorphous state upon release ofthe stretch. Differential scanning calorimetry (DSC) (in theunstretched state) also indicated an exothermic peak at −64.5 °Cthat corresponds to the onset of crystallization during cooling, SIAppendix, Fig. S3. Likewise, an endothermic peak of the siliconeelastomer occurs at −41.6 °C during heating, thereby indicating

the opposite phase transformation (amorphization), SI Appendix,Fig. S3. The glass transition temperature (Tg) of this siliconeelastomer was not observed over a temperature range from−150 ° to 30 °C, which implies that Tg of this silicone elastomer islower than −150 °C. Finally, to gain some basic information asrelated to polymer properties of this material, we performedmatrix-assisted laser desorption ionization time-of-flight massspectrometry (MALDI-TOF), which indicated that both pre-cursors of the silicone elastomer (“part A” and “part B”) consistof the base polymer with additives (SI Appendix, Fig. S4). Thediffering patterns for parts A and B in the low m/z (mass-to-charge ratio) range of the spectra indicate that each precursorcontains different types and amounts of additives. SI Appendix,Fig. S4, Top Insets show the isotopic distributions with a centerpeak at m/z = 3,006 and m/z = 3,155 for parts A and B, re-spectively. These values indicate the number averaged molecularweight of the polymers in the precursors in units of g/mol (sincethe charge of the cation in this mass spectroscopy is almost al-ways +1, the m/z corresponds to the mass of the measured cat-ion). Both spectra show repeat spacings of m/z = 74 (see SIAppendix, Fig. S4, Bottom Insets). This value is equivalent to themolecular weight of the repeat unit of polydimethylsiloxane(C2H6OSi)n, a well-known silicone elastomer. Combining thesetwo results implies that each base polymer consists of ∼40 and∼42 repeat units for parts A and B, respectively (ignoring endgroups). Overall, these experiments provide further credence tothe proposed microstructural origins of the anisotropic fractureenergy outlined in this section (Fig. 7). In particular, due to theincreasing levels of chain alignment and crystallization with in-creasing stretch level, it is likely that the microstructure near thecrack tip is highly spatially inhomogeneous, due to the largestress gradients near the tip (Fig. 8). Namely, a region in front ofthe crack tip (i.e., in the forward crack sense as depicted by theinset on the right) has relatively large stresses, which induce highlevels of chain alignment and crystallinity. By comparison, a re-gion “above” the crack tip (i.e., in the sideways crack sense asdepicted by the inset on the left) has less alignment and crys-tallinity. As such, the fracture toughness is highly anisotropicnear the crack tip.

Effects of Sample Geometry and Loading Rate on the Propensity forSideways Cracking. To examine the propensity for sideways crackingunder various sample and loading conditions, we performed tests

Forward cracking

Sideways cracking

L

Rigid plate

c0HL

Silicone elastomer

Lcf

B

λH

λHc0

cs

L

C

λH

A

Fig. 6. Schematics, including hypothesized microstructures, of a siliconeelastomer under a pure-shear test geometry. In the microstructural sche-matics, the red dots denote cross-links, and the orange boxed regions rep-resent crystalline domains. (A) Initial configuration of a sample with a precutof length c0 (25.4 mm). (B) Sample when the crack propagates in the forwarddirection. cf is the crack length in the forward direction (which includes c0).(C) Sample when the crack propagates in the sideways direction. cs is thefreshly created crack length in the sideways direction.

Forward Sideways

dcf = 2.54 mm dcs = 2.54 mm

(kPa)22

0

0

20

40

60

80

100

0 5 10 15 20 25 30 35

Gs/G

f (%

)

cs/cf (%)

(kPa)220

0

λ = 1.1

λ = 2

λ = 1.1

λ = 2

λ = 1.1

λ = 2

Fig. 5. Finite-element simulations showing the effects of crack-tip blunting.(Left) Finite-element simulations of the maximum principal stress distribu-tion in a silicone elastomer with a precut at λ= 1.1 (Top) and λ= 2.0 (Bottom)for a crack that has propagated by dc = c0/10 (2.54 mm) in the forward di-rection (left images) and the sideways direction (right images). (Right) En-ergy release rate for sideways crack growth relative to that of forward crackgrowth as a function of the ratio of the sideways crack length cs to theforward crack length cf.

9254 | www.pnas.org/cgi/doi/10.1073/pnas.1820424116 Lee and Pharr

Dow

nloa

ded

by g

uest

on

Apr

il 12

, 202

0

with samples under the so-called pure-shear geometry, as shown inFig. 6. Specimens were stretched in the vertical direction (along H)under constant stretch rates (0.1/min, 1/min, and 10/min) for sam-ples of various thickness, B, (∼1, 3, and 5 mm). SI Appendix, Fig. S5shows that the stress–stretch response of the as-fabricated elasto-mer does not demonstrate any significant rate effects over thetested range of stretch rates. By contrast, we found that the fracturebehavior of precut specimens strongly depends on the stretch rate(test data shown in SI Appendix, Fig. S6A). Namely, Fig. 9 showsthat forward crack propagation generally occurred at larger stretchrates, whereas sideways cracking generally occurred at smallerstretch rates. One potential explanation for this behavior is that theprocesses associated with forward and sideways cracking may in-volve different microstructural mechanisms. For instance, forwardcracking may primarily involve breaking covalently bonded polymerchains and crystalline domains, whereas sideways cracking maylargely involve viscous-type separation between individual chains(e.g., as in Fig. 8). The distinct microstructural mechanisms asso-ciated with fracture will have different timescales associated withthem, which thereby may produce rate-sensitive fracture. Anotherpotential explanation is that strain-induced crystallization (as evi-denced in SI Appendix, Fig. S2) takes time. As a result, developinghighly anisotropic microstructures required for significant aniso-tropic fracture toughness takes time. As such, loading at high ratesmay not allow enough time for anisotropic microstructure/tough-ness to develop, thereby only leading to forward cracking.Additionally, Fig. 9 shows that forward crack propagation

generally occurred in thinner specimens, whereas sidewayscracking generally occurred in thicker specimens. One potentialexplanation for this behavior is related to the distribution ofdefects in the material. Thicker samples generally have moredefects/heterogeneities, which could produce configurationsnear the crack tip that are conducive to sideways cracking.However, we should point out that we believe sideways crackingis not merely a geometric “artifact;” that is, it does not occursolely due to boundary effects. To substantiate this claim, weperformed further experiments in an entirely different loadinggeometry, that of “simple extension” (SI Appendix, Fig. S7).

Under otherwise similar conditions, both simple extension and“pure shear” tests produced sideways cracks (SI Appendix, Fig.S7), thereby indicating that sideways cracking can occur under awide range of geometries.

Summary and ConclusionsWe have identified an intriguing form of fracture in a siliconeelastomer in which the crack turns and propagates stably in adirection perpendicular to the initial precut. The length of thesideways crack slowly increases with increasing load, and thecrack arrests if the loading stops. The crack never turns back tothe standard direction or shows any bifurcation. We call thisbehavior sideways and stable cracking. As a result of sidewayscracking, the material ahead of the crack tip can continue tosustain load, thereby enabling enormous stretchabilities. To ex-plain this phenomenon, we performed finite-element simula-tions, which demonstrated that the crack driving force in the“forward” direction is not orders of magnitude larger than thatof the sideways direction. Likewise, these simulations show thatlarge crack-tip blunting increases the propensity for sidewayscracking to occur. These simulations suggest that a moderatelevel of anisotropy in fracture energy may produce sidewayscracking. To this end, we performed fracture testing of samplestransversely prestretched to various extents, which indeed in-dicate a large anisotropy in the fracture energy upon stretching.We also performed microstructural characterization throughphotoelasticity, XRD, and DSC, which indicate that significantchain alignment and strain-induced crystallization indeed occurin this material upon stretching. Finally, we systematically charac-terized how a number of loading conditions (sample geometry andloading rate) affect this phenomenon. Generally, the propensity for

Fig. 8. Hypothesized inhomogeneous distribution of microstructure at twolocations of interest near a crack tip due to large stress gradients near the tip.

D E

CB

0102030405060

1 1.1 1.2 1.3 1.4 1.5 1.6

σ (k

Pa)

Rigid plates

Silicone elastomer

F

Pre-stretch

F

Pre-stretch Pre-stretch Pre-stretch

F

F

A

Fig. 7. Experiments demonstrating the effects of polymer chain alignmentand strain-induced crystallization on fracture energy. In the microstructuralschematics, the red dots denote cross-links, and the orange boxed regionsrepresent crystalline domains. (A) Stress–stretch curves and (B–E) schematicsof silicone elastomer specimens during fracture testing to examine pre-stretch effects. (Insets) Schematics of representative hypothesized chainorientations and crystalline domains at the given state: (B) original state ofthe specimen, (C) specimen with vertically applied load, (D) specimen pre-stretched in the horizontal direction before applying load, and (E) specimenprestretched in the horizontal direction with a vertically applied load.Samples with large prestretches show significantly smaller areas under thestress–stretch curve, thereby indicating significantly smaller fracture ener-gies for cracks propagating in the horizontal (forward) direction.

0.01

0.1

1

10

100

0 1 2 3 4 5 6

( etar hctertS

min

-1)

Thickness (mm)

ForwardSideways

Fig. 9. Scatter plot showing the propensity for sideways crack propagationin an elastomer of various thickness and subjected to various stretch rates.Generally, sideways cracks are more prevalent in thicker samples and underlower stretch rates.

Lee and Pharr PNAS | May 7, 2019 | vol. 116 | no. 19 | 9255

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Apr

il 12

, 202

0

sideways cracking increases with sample thickness and decreaseswith loading rate. We hypothesize that these observations are re-lated to microstructural features associated with fracture, includingthe distribution of defects and the rate dependence of varyingmicrostructural processes. As such, we hope that this study willmotivate future studies of how specific microstructural featurescontribute to peculiar fracture behavior in elastomers and softmaterials in general.

Materials and MethodsFabrication of test specimens involved mixing Smooth-On Ecoflex 00–30 in a1:1 ratio (part A:part B), followed by stirring for 3 min. Casting into moldsformed the shape of the test specimens, and placing in a vacuum for 10 minremoved bubbles. Curing at room temperature for at least 4 h produced thefinal specimens. Gluing the cured specimens directly to acrylic plates helpedmitigated stresses induced by gripping and helped prevent slipping duringmechanical testing. For precut specimens, cutting with razor blades pro-duced sharp tips in the specimens.

For all of the experiments, tensile tests used an Instron model 5943 with a1-kN load cell. The stretch along the H direction wasmeasured as λ = (H+ΔH)/H,where ΔH is the cross-head displacement. We defined the stretch at rupture forprecut samples as the moment at which the crack started to propagate, asdetermined by corresponding movie footage during the experiments. In thecase of the samples without a precut, fracture did not occur (failure occurred atvery large stretches due to slipping between the sample and the grips).

To investigate the anisotropy of the fracture energy (Fig. 7), we adoptedthree different prestretch values (λ = 1, 1.14, and 1.5) on ∼3-mm-thickspecimens. The direction of the prestretch is parallel to that of the initial

crack. We subsequently applied loads in the direction perpendicular to initialcrack at a stretch rate of 1/min.

We adopted tests in a pure shear geometry (23) (Figs. 6 and 9) to in-vestigate the effect of sample geometry and loading rate on the propensityfor sideways cracking. The tensile tester stretched specimens in the verticaldirection (along H) under a constant stretch-rate (0.1/min, 1/min, and 10/min)for samples of various thickness, B (∼1, 3, and 5 mm). For each testing con-dition, we tested three samples with a precut as well as three samples with-out a precut (as fabricated).

We performed XRD on cured (solid) Ecoflex 00–30. Ecoflex samples withdifferent stretch levels of λ = 1 (initial state), 1.7, 3.7, and 1 (unstretched)were analyzed by a Bruker-AXS D8 advanced Bragg–Brentano X-ray dif-fractometer. XRD patterns were obtained using Cu radiation (λ = 1.54 Å)over 2θ from 5 ̊ to 45 ̊ at room temperature.

We also performed DSC and thermogravimetric analysis on cured (solid)Ecoflex. Heat flow and weight changes associated with material transitionswere measured as function of time and temperature with a TA instrumentsQ2000 and SDT Q600 over a temperature range from −150 to 30 ̊C.

The platinum catalyzed silicone elastomer used in this study (Smooth-onEcoflex 00–30) consists of two precursors (part A and part B). The mass tocharge ratio of each precursor was measured by MALDI-TOF. This techniqueproduced mass spectra of the precursors to determine the molecular weightof the repeat units.

ACKNOWLEDGMENTS. We thank Anastasia Muliana, Arun Srinivasa, and AlanNeedleman for a number of useful discussions pertaining to this work. We alsoacknowledge helpful suggestions from an anonymous reviewer of thismanuscript. M.P. acknowledges funding from the mechanical engineeringdepartment at Texas A&M University and the Texas A&M EngineeringExperiment Station.

1. Valdastri P, et al. (2006) Integration of a miniaturised triaxial force sensor in a mini-mally invasive surgical tool. IEEE Trans Biomed Eng 53:2397–2400.

2. Kim D-H, et al. (2011) Epidermal electronics. Science 333:838–843.3. Takei K, et al. (2010) Nanowire active-matrix circuitry for low-voltage macroscale

artificial skin. Nat Mater 9:821–826.4. Mannsfeld SC, et al. (2010) Highly sensitive flexible pressure sensors with micro-

structured rubber dielectric layers. Nat Mater 9:859–864.5. Ramuz M, Tee BCK, Tok JBH, Bao Z (2012) Transparent, optical, pressure-sensitive

artificial skin for large-area stretchable electronics. Adv Mater 24:3223–3227.6. Sokolov AN, Tee BC, Bettinger CJ, Tok JB-H, Bao Z (2012) Chemical and engineering

approaches to enable organic field-effect transistors for electronic skin applications.Acc Chem Res 45:361–371.

7. Cai S, Lou Y, Ganguly P, Robisson A, Suo Z (2010) Force generated by a swellingelastomer subject to constraint. J Appl Phys 107:103535.

8. Schwartz G, et al. (2013) Flexible polymer transistors with high pressure sensitivity forapplication in electronic skin and health monitoring. Nat Commun 4:1859.

9. Sekitani T, Zschieschang U, Klauk H, Someya T (2010) Flexible organic transistors andcircuits with extreme bending stability. Nat Mater 9:1015–1022.

10. Dong L, Jiang H (2007) Autonomous microfluidics with stimuli-responsive hydrogels.Soft Matter 3:1223–1230.

11. Beebe DJ, et al. (2000) Functional hydrogel structures for autonomous flow controlinside microfluidic channels. Nature 404:588–590.

12. Pang C, et al. (2012) A flexible and highly sensitive strain-gauge sensor using re-versible interlocking of nanofibres. Nat Mater 11:795–801.

13. Frogley MD, Ravich D, Wagner HD (2003) Mechanical properties of carbonnanoparticle-reinforced elastomers. Compos Sci Technol 63:1647–1654.

14. Wang Y, Ameer GA, Sheppard BJ, Langer R (2002) A tough biodegradable elastomer.Nat Biotechnol 20:602–606.

15. Li C-H, et al. (2016) A highly stretchable autonomous self-healing elastomer. NatChem 8:618–624.

16. Liu M, Sun J, Sun Y, Bock C, Chen Q (2009) Thickness-dependent mechanical prop-erties of polydimethylsiloxane membranes. J Micromech Microeng 19:035028.

17. Creton C, Ciccotti M (2016) Fracture and adhesion of soft materials: A review. RepProg Phys 79:046601.

18. Beiermann BA, et al. (2014) The effect of polymer chain alignment and relaxation onforce‐induced chemical reactions in an elastomer. Adv Funct Mater 24:1529–1537.

19. Hui C-Y, Jagota A, Bennison S, Londono J (2003) Crack blunting and the strength ofsoft elastic solids. Proc R Soc Lond A 459:1489–1516.

20. Pharr M, Sun J-Y, Suo Z (2012) Rupture of a highly stretchable acrylic dielectric elas-tomer. J Appl Phys 111:104114.

21. Chen C, Wang Z, Suo Z (2017) Flaw sensitivity of highly stretchable materials. Extreme

Mech Lett 10:50–57.22. Feichter C, Major Z, Lang R (2006) Deformation analysis of notched rubber specimens.

Strain 42:299–304.23. Gent A, Razzaghi-Kashani M, Hamed G (2003) Why do cracks turn sideways? Rubber

Chem Technol 76:122–131.24. Hamed G, Park B (1999) The mechanism of carbon black reinforcement of SBR and NR

vulcanizates. Rubber Chem Technol 72:946–959.25. Greensmith H (1956) Rupture of rubber. IV. Tear properties of vulcanizates containing

carbon black. J Polym Sci 21:175–187.26. De D, Gent A (1996) Tear strength of carbon-black-filled compounds. Rubber Chem

Technol 69:834–850.27. Hu H, et al. (2018) Stretchable ultrasonic transducer arrays for three-dimensional

imaging on complex surfaces. Sci Adv 4:eaar3979.28. Park M, et al. (2012) Highly stretchable electric circuits from a composite material of

silver nanoparticles and elastomeric fibres. Nat Nanotechnol 7:803–809.29. Xu S, et al. (2013) Stretchable batteries with self-similar serpentine interconnects and

integrated wireless recharging systems. Nat Commun 4:1543.30. Ying M, et al. (2012) Silicon nanomembranes for fingertip electronics. Nanotechnology

23:344004.31. Pan T, et al. (2017) Experimental and theoretical studies of serpentine interconnects

on ultrathin elastomers for stretchable electronics. Adv Funct Mater 27:1702589.32. Liu Y, Pharr M, Salvatore GA (2017) Lab-on-skin: A review of flexible and stretchable

electronics for wearable health monitoring. ACS Nano 11:9614–9635.33. Busse W (1934) Tear resistance and structure of rubber. Ind Eng Chem 26:1194–1199.34. Hamed G, Al-Sheneper A (2003) Effect of carbon black concentration on cut growth in

NR vulcanizates. Rubber Chem Technol 76:436–460.35. Caimmi F, Calabrò R, Briatico-Vangosa F, Marano C, Rink M (2015) Toughness of

natural rubber compounds under biaxial loading. Eng Fract Mech 149:250–261.36. Marano C, Calabrò R, Rink M (2010) Effect of molecular orientation on the fracture

behavior of carbon black‐filled natural rubber compounds. J Polym Sci B Polym Phys

48:1509–1515.37. Hardin G, Zhang Y, Fincher C, Pharr M (2017) Interfacial fracture of nanowire elec-

trodes of lithium-ion batteries. JOM 69:1519–1523.38. Trabelsi S, Albouy P-A, Rault J (2002) Stress-induced crystallization around a crack tip

in natural rubber. Macromolecules 35:10054–10061.39. Tosaka M, et al. (2013) Strain-and temperature-induced polymorphism of poly

(dimethylsiloxane). Colloid Polym Sci 291:2719–2724.

9256 | www.pnas.org/cgi/doi/10.1073/pnas.1820424116 Lee and Pharr

Dow

nloa

ded

by g

uest

on

Apr

il 12

, 202

0