Mechanically tunable conductive interpenetrating network hydrogels that mimic the elastic moduli of biological tissue — TEI-tagged view

Vivian R. FeigpersName, Helen TranpersName, Minah LeepersName, Zhenan BaopersName (2018date). “Mechanically tunable conductive interpenetrating network hydrogels that mimic the elastic moduli of biological tissue.” Nature Communicationstitle·j.

Object of study — conducting interpenetrating networkterm·dom hydrogelsterm·dom of gelled PEDOT:PSSterm and a secondary polymer network

Research question — how to keep electronic performance in conducting-polymer hydrogelsterm·dom with useful mechanical properties

Contribution — a two-network method that overcomes the trade-off between conductivityterm·key and tissue-like mechanics

Full text — tagged

The reference list is not part of the text we hold for this paper. Our copy was extracted from the publisher's page and the bibliography did not survive that extraction, so the article below ends before its references. Nothing has been removed by us.

Mechanically tunable conductive interpenetratingterm·key network hydrogelsterm·dom that mimic the elastic moduliterm·dom of biological tissue Abstract Conductive and stretchable materials that match the elastic moduliterm·dom of biological tissue (0.5–500 kPa) are desired for enhanced interfacial and mechanical stability. Compared with inorganic and dry polymeric conductors, hydrogelsterm·dom made with conducting polymersterm·dom are promising soft electrode materials due to their high water content. Nevertheless, most conducting polymer-based hydrogelsterm·dom sacrifice electronic performance to obtain useful mechanical properties. Here we report a method that overcomes this limitation using two interpenetratingterm·key hydrogelterm·dom networks, one of which is formed by the gelationterm·key of the conducting polymerterm·dom PEDOT:PSSterm. Due to the connectivity of the PEDOT:PSSterm network, conductivities up to 23 S m −1 are achieved, a record for stretchable PEDOT:PSS-based hydrogelsterm·dom. Meanwhile, the low concentration of PEDOT:PSSterm enables orthogonal control over the composite mechanical properties using a secondary polymer network. We demonstrate tunability of the elastic modulus over three biologically relevant orders of magnitude without compromising stretchabilityterm·key ( > 100%) or conductivityterm·key ( > 10 S m − 1 ).

Introduction Electronics with tissue-like properties enable a natural integration between electronic functionality and the biological world. Such integration is increasingly important for treating numerous medical diseases like Parkinson’s, for which electrical deep-brain stimulation has proven to be highly effective 1 . Moreover, electronic interfaces with biological tissue provide valuable scientific and diagnostic insights into complex medical phenomena, ranging from atrial fibrillation 2 to Alzheimer’s 3 , 4 . However, conventional conductive materials have difficulty forming long-term stable and conformal interfaces with many biological tissues due to severe mechanical mismatchterm·dom between the two material types. For example, brain tissue typically has elastic moduliterm·dom < 1 kPa 5 , whereas common electrode materials like silicon and tungsten have moduli of 50 GPa and 130 GPa, respectively 6 . Even conducting polymersterm·dom (CPsterm·dom), which have superior flexibility compared to inorganic conductors, typically have moduli in the 1 GPa range 6 . Furthermore, interfacing materials should be stretchable enough to maintain conformal contact with dynamic tissue surfaces. For example, the surface of the brain has an intrinsic displacement of over 10 μm from respiration alone 6 . Mechanical mismatchterm·dom of tissue-interfacing materials can lead to reduced efficacyterm·dom of both recording and stimulation, which may be further exacerbated by immunological responses that lead to scarring 6 – 8 . Thus, there is a strong demand for materials that can be rationally designed to possess mechanical properties that mimic tissue, without compromising their electronic performance. Ideally, such materials should be able to be tuned to form mechanically compliant interfaces with a wide range of biological tissues, from ultra-soft tissues such as the brain (0.5–1 kPa) 5 to stiffer tissues such as the skin 9 and certain regions of the heart (100–500 kPa) 10 .

To mimic the mechanical properties of these tissues, hydrogelsterm·dom are a promising class of synthetic materialsterm·dom due to their high water content (70–99 wt%) that is similar to tissue 11 . Electrically conducting hydrogelsterm·dom made with CPsterm·dom, including poly(3,4-ethylenedioxythiophene) polystyrene sulfonateterm (PEDOT:PSSterm) 12 – 15 , polyanilineterm (PANIterm) 16 – 18 , and polypyrroleterm (PPyterm) 19 – 22 have been reported, although there are few examples of mechanically robust gels that can stretch to conform to dynamic tissue surfaces. Among these CPsterm·dom, PEDOT:PSSterm is advantageous because of its biocompatibility 23 , 24 , and because it is commercially available in its doped form with reproducibly high conductivityterm·key, thus avoiding problems with batch to batch variation from in situ polymerization 25 and the use of alternative dopants that are less effective than PSS 26 . Reported examples of stretchable conductive hydrogelsterm·dom based on PEDOT:PSSterm either rely on in situ polymerization of EDOT within an inert hydrogelterm·dom matrix 13 , 14 , 27 , or involve blending PEDOT:PSSterm with hydrogel-forming precursors 15 . However, these strategies suffer from low conductivityterm·key (0.01–2.2 S m −1 ) with relatively high PEDOT:PSSterm content up to 30 wt%, which makes it difficult to selectively tune the mechanical properties of the gel 13 – 15 .

We hypothesize that the low performance of these gels arises from limited control of the CP network morphology, resulting in largely disconnected aggregates that require high loadings of CP to meet the percolation thresholdterm·dom for electronic conduction. Accordingly, we hypothesize that using gelationterm·key to ensure CP network connectivity should result in improved conductivityterm·key. Recently, it has been reported that PEDOT:PSSterm can form gels directly from aqueous solution, either by increasing the ionic strengthterm·dom 28 , increasing concentration 28 , or lowering pH 12 . Although these gels behave like solids, on their own they are highly brittle and difficult to handle. Still, the ability to form gels directly from commercially available PEDOT:PSSterm solutions presents a unique opportunity to control CP network connectivity to improve electronic conductivityterm·key.

In this work, we demonstrate a method for making conducting interpenetrating networksterm·dom (C-IPNsterm·dom) by infiltrating a loosely-crosslinked PEDOT:PSSterm gel with the precursors for a secondary polymer network (Fig. 1a ). The controlled gelationterm·key of commercially available PEDOT:PSSterm enables us to achieve record-high conductivities up to 23 S m −1 , while the fact that PEDOT:PSSterm can gel at a low solids content of 1.1 wt% enables us to orthogonally control the mechanical properties of the C-IPNterm·dom by tuning the secondary network properties. Using this method, we demonstrate that we can fabricate gels with ultra-soft moduli over three biologically relevant orders of magnitude (8–374 kPa) without compromising conductivityterm·key ( > 10 S m −1 ) or stretchabilityterm·key ( > 100%). Fig. 1 Fabrication and structure of C-IPNterm·dom hydrogelsterm·dom. a Process for fabricating C-IPNterm·dom hydrogelsterm·dom. First, PEDOT:PSSterm hydrogelsterm·dom are formed from aqueous solutions of PEDOT:PSSterm by using ionic liquid to screen the electrostatic repulsions between PEDOT:PSSterm microgels, enabling them to aggregate into a macroscopically connected network. Second, the PEDOT:PSSterm hydrogelterm·dom is infiltrated with acrylic acid, bisacrylamide, and an azo-initiator. Finally, the polyacrylic acidterm network is formed by polymerizing the monomers in water at 70 °C. b Various large shapes made by casting the PEDOT:PSSterm/acrylic acid mixture into different silicone soap molds. Scale bars are 1 cm. c , d C-IPNterm·dom can be micropatterned into pyramidal structures with features as small as 10 μm by casting into silicon molds. Scale bar is 200 μm c and 100 μm d . e Cross-sectional SEM image of freeze-dried C-IPNterm·dom showing that the final gel is homogeneous and porous. Scale bar is 100 μm. f FTIR spectra of PEDOT:PSSterm, PAActerm, and C-IPNterm·dom, showing a clear presence of PAActerm within the C-IPNterm·dom composite Results Synthesis and characterization of PEDOT:PSSterm hydrogelsterm·dom To synthesize C-IPNterm·dom, PEDOT:PSSterm hydrogelsterm·dom were first formed from commercial aqueous dispersions of PEDOT:PSSterm with a low polymer content of approximately 1.1 wt% (11 mg mL −1 ). Recently, Leaf and Muthukumar 28 proposed that aqueous PEDOT:PSSterm solution consists of large microgel particles which internally resemble a semi-dilute polyelectrolyte mesh and which have a low overlap concentration of 1 mg mL −1 . The authors found that increasing ionic strengthterm·dom can cause PEDOT:PSSterm solution to phase change into a solid-like gel, resulting in dramatically different rheological properties without changing the internal scattering profile of the microgels. As the primary interaction between microgel particles was found to be electrostatic in nature, the authors concluded that increased ionic strengthterm·dom helps screen the electrostatic repulsions between particles and thus allows them to form physical crosslinks between each other through π–π stacking of PEDOT. When a sufficient number of physical crosslinks form between particles, a macroscopically connected gel results 28 . Based on these findings, we hypothesized that the macroscopically connected microgel particles would lead to an increase in the number of connected pathways available for charge conduction, and should thus be a way to form highly conductive PEDOT:PSSterm hydrogelsterm·dom at a relatively low concentration.

To form PEDOT:PSSterm gels, we increased ionic strengthterm·dom of the solution using the ionic liquid 4-(3-butyl-1-imidazolio)-1-butanesulfonic acid triflate. A previous publication from our group found that this ionic liquid is an excellent dopant for PEDOT:PSSterm thin films, and that it interacts strongly enough with PEDOT:PSSterm to induce a significant change in the thin film morphology 29 . Based on these observations, we surmised that the same ionic liquid should also sufficiently interact electrostatically with PEDOT:PSSterm to push the solution past its gel phase boundary. Indeed, mixing the ionic liquid with PEDOT:PSSterm resulted in a solid-like gel, as indicated by the fact that the gel’s storage modulusterm·mth ( G ’) exceed its loss modulus ( G ”) in the frequency range of 0.1–100 Hz (Fig. 2a ). Furthermore, the gel stiffness, given by G ’, increased with ionic strengthterm·dom (Fig. 2b ), which is consistent with the proposed mechanism wherein electrostatic screeningterm·dom induces physical crosslinkingterm·dom. Within this rubbery regime, G ’ increased slightly with frequency, reflecting the dynamic nature of the physical crosslinks present in the gel (Fig. 2b ). These rheological properties are consistent with those observed by Leaf and Muthukumar 28 , who used both small salts and larger ionic liquids such as 1-ethyl-3-methylimidazolium tetracyanoborate (EMIM:TCB) to induce PEDOT:PSSterm gelationterm·key. Fig. 2 Gelationterm·key of PEDOT:PSSterm. a , b PEDOT:PSSterm forms a gel a after mixing it with ionic liquid. The gel strengthterm·dom, given by storage modulusterm·mth ( G ’), increases with increasing ionic strengthterm·dom b . c Mixing PEDOT:PSSterm with either ionic liquid or CuCl 2 will cause the solution to gel, although the gelationterm·key occurs much more rapidly with CuCl 2 . With ionic liquid, the PEDOT:PSSterm mixture only starts to become more solid-like ( G ’ > G ”) after 13 min. By contrast, the mixture with CuCl 2 gels so quickly that G ’ already exceeds G ” by the start of the rheology measurement Interestingly, we found that the gelation rateterm·dom, which is an important consideration for processability, could be tuned by the selection of the molecular additive. To investigate this, we mixed PEDOT:PSSterm with both our ionic liquid and a smaller metal salt (CuCl 2 ) at the same overall ionic strengthterm·dom, and loaded the mixture onto a parallel plate rheometer oscillating at a constant strain rate of 1% and a frequency of 1 Hz. With the ionic liquid mixture, the crossover point at which G ’ > G ” occurs after 15 min, before which the mixture can flow readily. By comparison, mixing PEDOT:PSSterm with CuCl 2 at an equivalent ionic strengthterm·dom resulted in nearly immediate gelationterm·key: by the time the solution was loaded onto the rheometer, G ’ already exceeded G ” (Fig. 2c ). The widened window of liquid-like properties for the PEDOT:PSSterm/ionic liquid mixture enables us to more easily process it into different shapes (Fig. 1b ). We could even process it into pyramidal structures with 10 μm resolution by casting the mixture into pre-fabricated silicon molds (Fig. 1c, d ). To our knowledge, this is the first discussion of the kinetics surrounding the gelationterm·key behavior of PEDOT:PSSterm solutions.

Synthesis and characterization of C-IPNterm·dom gels Although the PEDOT:PSSterm hydrogelsterm·dom behave similar to solids, they are also highly brittle. To improve mechanical properties, we selected polyacrylic acidterm (PAActerm) as a secondary polymer network due to its biocompatibility 30 , 31 and high concentration of hydrogen bonding, which we hypothesized could further mechanically reinforce the C-IPNterm·dom. In addition, aqueous PEDOT:PSSterm solution has been shown to gel when exposed to strong acids, and we surmised that infiltrating the PEDOT:PSSterm hydrogelterm·dom with acidic monomers would result in additional reinforcement of the PEDOT:PSSterm matrix 12 . Finally, PAActerm hydrogelsterm·dom can be easily tuned by varying the concentration of monomers in water and the ratio of bisacrylamide to acrylic acid, which controls the degree of covalent crosslinking (Supplementary Figure 1 ).

To infiltrate the PEDOT:PSSterm gels with PAActerm monomers, the gels were soaked in an aqueous solution of acrylic acid, bisacrylamide, and a thermal radical polymerization initiator that activates above 60 °C. As expected, the acidity of the acrylic acid caused the PEDOT:PSSterm gels to shrink slightly during the exchange process (Supplementary Figure 2b ), which qualitatively strengthened the PEDOT:PSSterm/acrylic acid composite even before the acrylic acid was polymerized. Finally, the acrylic acid network was polymerized within the PEDOT:PSSterm gel by placing the gel in a sealed container in an oven at 70 °C for 30 min. The thermal initiator was selected instead of a UV initiator due to its relatively fast kinetics and insensitivity to oxygen, and because it allows acrylic acid to be polymerized within the bulk PEDOT:PSSterm gel, which is opaque due to its electronically conducting nature.

To confirm the presence of the secondary network inside the C-IPNterm·dom, Fourier-transform infrared spectroscopy was performed on C-IPNterm·dom and compared with PEDOT:PSSterm and PAActerm hydrogelsterm·dom (Fig. 1f ). Peaks attributed to carboxylic acid and hydrogen bonding between carboxylic acid groups can be identified in both PAActerm and C-IPNterm·dom samples, indicating that the PAActerm successfully was incorporated. In addition, scanning electron microscopic (SEM) imaging on the cross-section of a lyophilized C-IPNterm·dom sample confirms that the gel is homogeneous and porous (Fig. 1e ). Finally, X-ray photoelectron spectroscopyterm·mth (XPSterm·mth) measurements on PEDOT:PSSterm and C-IPNterm·dom suggest that the PEDOT:PSSterm network is chemically unaffected by the presence of PAActerm. In XPSterm·mth spectra of PEDOT:PSSterm, two S2p peaks correspond to the presence of sulphur in PEDOT and in PSS, respectively. The relative size of these two peaks is proportional to the ratio of PEDOT to PSS, whereas the locations of the peaks correspond to the binding energy of the sulphur in its two respective chemical environments 29 , 32 . No significant change in peak area or position was observed in the PEDOT:PSSterm after PAActerm was introduced (Supplementary Figure 2a ).

Although ionic liquid is used to induce PEDOT:PSSterm gelationterm·key, we wanted to remove it from the final C-IPNterm·dom gel because of its cytotoxicity. To facilitate diffusion of ionic liquid out of the gel, we soaked the PEDOT:PSSterm gels in deionized (DI) water and exchanged the water multiple times before infiltrating the PEDOT:PSSterm with acrylic acid. For our system, fluorine is a unique marker for the anionic component of the ionic liquid, yet no fluorine could be detected by XPSterm·mth on the C-IPNterm·dom. To confirm that the cationic component is also removed, we also performed 1 H-nuclear magnetic resonance (NMR) spectroscopy on the solution obtained from washing the PEDOT:PSSterm gels with DI water and compared it with a spectrum of the neat ionic liquid (Supplementary Figure 3 ). The cation is clearly present in the wash solution. The broad peaks around 6.5–7.5 p.p.m. correspond to aromatic rings which are unique to PSS, suggesting that some PSS is also removed from washing the gel. The slight shifts in the peak at 6.2 p.p.m., which is attributed to the cation hydroxyl, may be due to the presence of PSS, since the rate of proton exchange between the hydroxyl group and solvent is highly dependent on acidic environment. Taken together, the XPSterm·mth and NMR data suggest that the ionic liquid can be easily removed from washing, which is expected given the highly porous nature of the hydrogelterm·dom.

The final C-IPNterm·dom gels continue to swell when immersed in water, equilibrating at a final weight that depends on the gel formulation. The formulation with highest solid content (C-IPNterm·dom 1) swells to approximately two times its original weight, whereas the formulation with the lowest solid content (C-IPNterm·dom 5) swells to nearly seven times its original weight (Supplementary Figure 8a ). The gels can also be dried and re-swelled successfully in water. Interestingly, the final weight of the gels after re-swelling exceeds the equilibrium weight of as-synthesized C-IPNterm·dom gels immersed in water, with the re-swelled weight ratio also dependent on the gel formulation. The formulation with highest solid content (C-IPNterm·dom 1) has a final re-swollen weight that is 1.3 times its equilibrium swelled weight prior to drying, whereas the formulation with lowest solid content (C-IPNterm·dom 5) has a final re-swollen weight that is nearly 6 times the equilibrium swelled weight (Supplementary Figure 8a ). This observation suggests that drying the C-IPNterm·dom hydrogelsterm·dom results in a reconfiguration of the polymer morphology that changes its swelling ability. Drying the hydrogelsterm·dom may encourage π–π stacking interactions between PEDOT:PSSterm that collapse the conducting polymerterm·dom network into aggregates, leaving larger continuous void fractions for water to penetrate during the re-swelling process. This hypothesis is supported by the fact that PEDOT:PSS-only hydrogelsterm·dom made directly from mixing with ionic liquid also re-swell to 1.8 ± 0.5 times their original weight after drying. Importantly, the re-swelled gels retain the same high stretchabilityterm·key and low modulus of the as-made gels (Supplementary Figure 8b ).

To characterize the mechanical properties of C-IPNterm·dom hydrogelsterm·dom, stress strain measurements were recorded under tensile elongation. Based on the formulation, the C-IPNterm·dom hydrogelsterm·dom have a wide range of elastic moduliterm·dom ranging from 8 kPa to nearly 400 kPa (Table 1 ). Despite this wide range of stiffness, all C-IPNterm·dom formulations could be stretched to over 100% strain before breaking (Fig. 3a, b ). One reason for the excellent stretchabilityterm·key may be the high water content, which provides many degrees of freedom for polymer chains in the C-IPNterm·dom to reorient in response to mechanical strain. In addition, the presence of hydrogen bonding from the PAActerm network may provide sacrificial, dynamic bonds that can break to dissipate the stress from mechanical deformation, obviating the need to dissipate energy from fracture 33 , 34 . Table 1 Different formulations of C-IPNterm·dom and their corresponding electronic and mechanical properties C-IPNterm·dom formulation AAc wt% Bis/AAc wt ratio Modulus (kPa) Strain at break (%) Conductivityterm·key (S m −1 ) 1 50 0.02 374 121 23.1 ± 5.6 2 33 0.01 175 163 23.7 ± 4.5 3 20 0.01 99 191 20.4 ± 4.7 4 20 0.002 23 399 12.7 ± 4.5 5 11 0.002 8 338 13.5 ± 2.3 Note that even though the modulus varies over three orders of magnitude, the conductivityterm·key in all cases stays above 10 S m − 1 . Fig. 3 Mechanical and strain-dependent properties of C-IPNterm·dom hydrogelsterm·dom. a Tensile elongation curves of different C-IPNterm·dom formulations, showing that all formulations can be stretched to over 100% despite large differences in the elastic moduliterm·dom, which is given by the initial slopeterm·mth of the stress/strain curve. b Picture of a C-IPNterm·dom 4 gel being stretched to 250%. c Cyclic stress/strain tensile data for C-IPNterm·dom 2 exhibiting minimal hysteresisterm·dom, reflecting its excellent elastic properties. d Change in resistance, expressed as a ratio between resistance ( R ) and initial resistance ( R o ), across a C-IPNterm·dom 2 gel as it is cycled reversibly between 0 and 100% strain for 10 cycles. Despite the large changes in tensile strain, the resistance stays fairly constant near its initial value. e Due to the largely strain independent conductivityterm·key of the gels, it is able to keep an LED lit even after being stretched to 50% strain Highly elastic conductive gels could also be fabricated using this method (Fig. 3c ). This is particularly noteworthy, as elasticity has been a challenging property to incorporate into stretchable conducting polymeric materials. For instance, plasticizers have been used in literature to enhance the stretchabilityterm·key of dry PEDOT:PSSterm, but these materials still yield at relatively low strains < 20%, potentially limiting their usability as stretchable conductors in high-strain applications 29 . For instance, compared with the highly stretchable PEDOT:PSSterm/ionic liquid conductor developed in our group (Supplementary Figure 4 ), C-IPNterm·dom gels exhibit considerably less hysteresisterm·dom, while cycling the strain at the same conditions. The ability to orthogonally control mechanical properties using the PAActerm network therefore enables us to take advantage of the elasticity of certain PAActerm hydrogelterm·dom formulations to impart elasticity onto the conductive composite.

Electrical conductivityterm·key of C-IPNterm·dom gels was calculated from measuring the resistance of the gels using a four-point probe methodterm·mth. For all formulations, the conductivityterm·key was greater than 10 S m −1 . The highest conductivityterm·key was 23 ± 5.6 S m −1 and conductivityterm·key tended to decrease as a function of acrylic acid concentration in the PAActerm hydrogelterm·dom precursor solution (Table 1 ). The fact that the electrical conductivityterm·key could be maintained above 10 S m −1 despite large variations in elastic moduliterm·dom is likely due to the ability for aqueous PEDOT:PSSterm solutions to form connected CP networks at a low weight concentration relative to acrylic acid. Indeed, the weight percent of PEDOT:PSSterm in the final gel is significantly lower than previously reported stretchable PEDOT:PSSterm hydrogelsterm·dom (Supplementary Table 1 ). Although connectivity of conducting domains is critical to achieving high electrical conduction 35 , mechanical properties of composites, such as elastic modulus, are largely dictated by the relative amount of each component 36 , 37 . Thus, a composite material that is predominately made up of a soft material will be soft, even if it contains a small amount of rigid domains. Analogously, because of the dilute microgel structure of PEDOT:PSSterm in aqueous solutions, the relatively small amount of PEDOT:PSSterm needed to form a connected conductive pathway in C-IPNterm·dom enables its mechanical properties to be nearly orthogonally controlled by the non-conductive PAActerm network. This represents a clear advantage over other conductive, stretchable hydrogelsterm·dom that have been synthesized by polymerizing EDOT within an electrically inert hydrogelterm·dom matrix, wherein decent conductivities are only achieved with concentrations as high as 30 wt%.

Although the conductivityterm·key of C-IPNterm·dom stays within the same order of magnitude despite significantly different elastic moduliterm·dom, slight differences in conductivities can be attributed to differences in acrylic acid concentration, which is consistent with the different degrees of shrinking visually observed during the PAActerm infiltration step. The fact that the conductivityterm·key does not vary significantly suggests that these differences in PEDOT:PSSterm gel strengthterm·dom are small compared to enhancement in conductivityterm·key that comes from forming a connected conductive network. This finding makes sense given that, in conductive composites, the most dramatic increase in conductivityterm·key typically occurs at the percolation thresholdterm·dom 38 , 39 .

Given that the acidic monomers apparently interact with PEDOT:PSSterm, our infiltration method is required for them to effectively penetrate the PEDOT:PSSterm gel form a homogeneous blend. By contrast, directly blending PEDOT:PSSterm with acrylic acid resulted in rapid phase separation, probably due to the acidity of the monomers (Supplementary Figure 6a ). Although the mixture could be polymerized to form a soft gel with moderate stretchabilityterm·key (Supplementary Figure 6b ), the conductivityterm·key of the blend was significantly lower, at 0.21 S m −1 . As expected, this value is in the same range of conductivities reported in the recent work of Wu et al. 15 , who directly blended PEDOT:PSSterm with the monomers used to make mechanically reinforcing inert secondary networks (0.2–2.2 S m −1 ) 15 . In addition, the modulus and strain at break of PEDOT:PSSterm/PAActerm hydrogelsterm·dom made by blending were lower than for hydrogelsterm·dom made with PAActerm only, probably because a significant amount of monomer is lost when it crashes out with PEDOT:PSSterm upon blending (Supplementary Figure 6c ). Thus, we believe that our infiltration method enables a greater degree of orthogonal control over the selection of the two interpenetratingterm·key networks in the conductive gel. This feature also enables us to incorporate alternative secondary networks, besides PAActerm, into C-IPNterm·dom. For instance, composite hydrogelsterm·dom made by infiltrating PEDOT:PSSterm gels with polyacrylamideterm (PAAmterm) precursors exhibited mechanical properties that closely resembled gels made with PAAmterm alone (Supplementary Figure 7 ).

Next, we characterized the change in electrical behavior under mechanical deformation by measuring the resistance across C-IPNterm·dom while cycling back and forth between 0 and 50% strain (Fig. 3d ). The electrical properties of the gel were surprisingly robust within this region, with the resistance staying approximately at its starting value throughout all 10 cycles. To verify that a high contact resistance did not mask the resistance change from the two-point test, four-point resistance measurements were also performed on the C-IPNterm·dom after being stretched manually to multiple strains. Despite an apparent contact resistance contribution of 17 Ω from the two-point measurement, the four-point stretching measurement confirmed that the resistance of the gels stayed largely unchanged even at large strains over 60% (Supplementary Figure 5 ). The robust conductivityterm·key may be attributed to both the connected nature of the PEDOT:PSSterm gel, as well as the ability for C-IPNterm·dom to respond to deformation by re-orienting within the high water content environment or breaking hydrogen bonds in the PAActerm network. These alternate dissipation mechanisms may enable the entangled colloidal PEDOT:PSSterm mesh to stay intact rather than disrupt percolation.

For bio-interfacing applications, the presence of ions in physiological solution can additionally contribute to the conductivityterm·key of C-IPNterm·dom electrodes, particularly given the high liquid content of the hydrogelsterm·dom. To decouple ionic and electronic conductivities, we saturated C-IPNterm·dom gels in 1 × phosphate-buffered salineterm (PBSterm) solution and characterized their impedanceterm·key using electrochemical impedance spectroscopyterm·mth (EISterm·mth). The data were fit to the equivalent circuit modelterm·mth depicted in Fig. 4a , where R c represents ohmic resistance of the cell assembly, R e represents electronic resistance, R i represents ionic resistance, and CPE dl and CPE g are constant phase elements corresponding to the double-layer capacitance arising from ionic conduction and the geometric capacitance of the gel, respectively 40 – 43 . The distorted semicircular shape of the Nyquist plot suggests the presence of comparable ionic and electronic conductivities (Fig. 4b, d ) 44 . The parameters extracted from fitting the data to the equivalent circuit modelterm·mth confirms that the ionic and electronic resistances are within the same order of magnitude (Fig. 4e ). To validate this model, we measured the DC voltage as a function of time for a constant applied current of 5 mA on the C-IPNterm·dom 2 sample (Fig. 4c ) 45 . The plateau value of the voltage, 0.0247 V, corresponds to the electronic contribution to the resistance only, and the R e value extrapolated by this method (4.94 Ω) is comparable to the value from fitting the AC data (4.97 Ω) to within < 1% error. Finally, we characterized the impact of different C-IPNterm·dom formulations on mixed conductivityterm·key by performing the EISterm·mth measurement on three formulations of varying density, with C-IPNterm·dom 1 being the most dense and C-IPNterm·dom 3 being the least dense (Fig. 4d ). As the density of the gels decreases, the relative contribution of ionic resistance to the overall impedanceterm·key decreases as well (Fig. 4e ). This is reasonable given that the hydrogelterm·dom’s density is directly proportional to the electronic conductivityterm·key (Table 1 ), whereas it is inversely proportional to the total volume of electrolyte, which facilitates ion transport, in the hydrogelterm·dom. Fig. 4 Characterization of dual ionic and electronic conductivityterm·key. a Equivalent circuit modelterm·mth representing the bulk C-IPNterm·dom hydrogelterm·dom. R e represents electronic resistance, R i represents ionic resistance, CPE dl represents the double-layer capacitive phase element (CPE), whereas CPE g represents the geometric CPE. CPE elements are used to account for inhomogeneous or imperfect capacitance, and are represented by the parameters Q and α , where Q is a pseudocapacitance value and α represents its deviation from ideal capacitive behavior. The true capacitance ( C ) can be calculated from these parameters by the relationship C = Q ω max α-1 , where ω max represents the frequency at which the imaginary component reaches a maximum 43 . R c represents the total ohmic resistance of the cell assembly. b Nyquist plot obtained from performing electrochemical impedance spectroscopyterm·mth (EISterm·mth) through a bulk C-IPNterm·dom 2 gel, overlaid with the plot predicted from the equivalent circuit modelterm·mth. Impedanceterm·key was measured between 500 mHz and 7 MHz, with higher real components of the impedanceterm·key obtained at lower frequencies. c When a constant DC current of 5 mA is applied through the C-IPNterm·dom 2 gel, the voltage across the gel plateaus to a value of 0.0247 V. This value can be used to calculate an electronic resistance that is comparable to the value of R e extracted from the model. d Overlay of Nyquist plots obtained for three C-IPNterm·dom formulations, where C-IPNterm·dom 1 is the stiffest and densest, and C-IPNterm·dom 3 is the softest and least dense. Impedanceterm·key was measured between 500 mHz and 7 MHz, with higher real components of the impedanceterm·key obtained at lower frequencies. e Values for all relevant parameters extracted for the three C-IPNterm·dom formulations by fitting their EISterm·mth data with the equivalent circuit modelterm·mth. As the gel stiffness and density increase, the relative ionic resistance within the gel increases as well Discussion We have presented a method for fabricating highly conductive hydrogelsterm·dom with dual electronic and ionic conductivityterm·key and highly tunable mechanical properties that mimic biological tissue. This combination of properties makes C-IPNterm·dom hydrogelsterm·dom promising for integration into wearable and implantable devices, for which the ability to couple high electronic conductivityterm·key with low modulus and high stretchabilityterm·key is particularly desired at soft biological interfaces such as the brain. In addition, the ability to tune our gel’s mechanical properties without compromising its conductivityterm·key makes it an attractive materials platform for tissue engineering and cell culture, since it is well known that cells are highly responsive to the mechanical properties of their surrounding environment 46 , 47 . C-IPNterm·dom hydrogelsterm·dom offer a route to enable electrical stimulation and recording while preserving the appropriate three-dimensional architecture and matched mechanical properties needed to mimic human tissue in vitro and to support cell viability.

In the future, C-IPNterm·dom can be integrated into more application-specific devices by taking advantage of the fact that it can be easily molded into different shapes and geometries for different target applications. Although there may be a lower resolution limit given by the approximate PEDOT:PSSterm microgel size of 250 nm 28 , sub-micrometer patterning is unnecessary for many biomedical applications. For instance, deep-brain stimulation electrodes typically have surface areas on the order of 1 mm 2 48 . By contrast, we have demonstrated size resolution down to 10 μm using a simple mold casting method. Furthermore, the fact that C-IPNterm·dom can be dried and re-swelled means that it can be compatibly integrated with other materials that may require dry or non-aqueous environments for processing. With their processability, mechanical tunability, and excellent electronic properties, C-IPNterm·dom gels are a highly versatile electronic material for future bio-interfacing applications.

Methods Materials All chemicals were purchased from Sigma-Aldrich. PEDOT:PSSterm Orgacon ICP 1050 was provided by Agfa as a surfactant-free aqueous dispersion with 1.1 wt% solid content. The PEDOT:PSSterm dispersion was filtered through a 1.0 μm Nylon filter to remove any large agglomerates prior to use. The ionic liquid 4-(3-Butyl-1-imidazolio)-1-butanesulfonic acid triflate (19597 Aldrich, CAS 439937-63-0) was used to induce gelationterm·key of PEDOT:PSSterm. The thermal initiator used to initiate radical polymerization of the PAActerm precursors was 2,2’-Azobis(2-methylpropionamidine) dihydrochloride (440914 Aldrich, CAS 2997-92-4), a water-soluble azo-initiator that can be initiated above 60 °C. As-purchased acrylic acid (147230 Aldrich, CAS 79-10-7) was run through a basic alumina plug to remove MEHQ inhibitor. N , N ’-methylenebis(acrylamide) (146072 Sigma-Aldrich, CAS 110-26-9) was used as the crosslinking monomer.

Synthesis of C-IPNterm·dom hydrogelsterm·dom To induce gelationterm·key of PEDOT:PSSterm, ionic liquid was slowly added to the filtered PEDOT:PSSterm dispersion while stirring. Subsequently, the PEDOT:PSSterm/ionic liquid mixtures were poured into molds, briefly degassed, and then sealed and placed in a 70 °C oven for 12 h to ensure complete gelationterm·key. Next, the PEDOT:PSSterm hydrogelsterm·dom were immersed in DI water, which was exchanged a total of three times over the course of 24 h. Then, the washed PEDOT:PSSterm hydrogelsterm·dom were immersed in the PAActerm precursor solution, which was also exchanged a total of three times over the course of 24 h. Finally, the soaked PEDOT:PSSterm hydrogelsterm·dom were sealed and placed in a 70 °C oven to polymerize the PAActerm network for 30 min to form C-IPNterm·dom. After this step, the C-IPNterm·dom was thoroughly rinsed in water to remove any unreacted acrylic acid.

Microscopy SEM images of C-IPNterm·dom hydrogelsterm·dom were obtained lyophilizing the hydrogelsterm·dom. Images were obtained using an FEI XL30 Sirion SEM.

Characterization of mechanical properties Rheological characterization of PEDOT:PSSterm hydrogelsterm·dom was performed using a TA Instruments ARES-G2 rheometer. For frequency sweeps, a 25 mm parallel plate geometry was used with a 1 mm gap size at a constant temperature of 37 °C and strain of 1%. For the gelationterm·key time measurement, a 40 mm parallel plate was used instead to improve the signal quality, as the samples were liquid-like. Measurements were taken at 25 °C at 1 Hz and a strain rate of 1%. Tensile elongation measurements of PAActerm and C-IPNterm·dom hydrogelsterm·dom were performed using an Instron 5565 at a strain rate of 10% per minute. For cycling tests, samples were strained at a rate of 50% per minute and allowed to equilibrate for 5 min between each stretching cycle.

Characterization of electronic properties Conductivityterm·key was calculated by measuring resistance using a standard four-point probe head. Reported values reflect an average over a minimum of three to five measurements obtained for each condition. To continuously measure the resistance while stretching, samples were attached to a homemade stretching station attached to a Keithley multimeter by electrical leads at both ends. Samples were cycled to various strains at a constant strain rate of 10%/min. To measure the resistance change using a 4-point probe method, samples were manually stretched, with initial and final lengths noted using a ruler. At each strain, the resistance of the sample was measured using a standard four-point probe head.

EISterm·mth measurements were conducted using a Bio-Logic VSP potentiostat. A punching tool was used to make gel samples with a constant cross-sectional area. The samples were sandwiched between two platinum electrodes within a Swagelok cell and surrounded by 1 × PBSterm, to ensure samples had a consistent degree of electrolyte saturation. AC impedanceterm·key measurements were obtained between 500 mHz and 7 MHz at an open-circuit potential of 20 mV amplitude. The impedanceterm·key data were fit using the Zfit tool from Bio-Logic’s EC-Lab software. For DC measurements, a constant current of 5 mA was applied through the sample.

Chemical characterization NMR spectra were recorded on a Varian mercury console spectrometer at 400 MHz at room temperature. Chemical shifts for 1 H-NMR spectra were reported in parts per million and were referenced to residual protonated solvent (CD 3 ) 2 SO: δ 2.50). Infrared spectroscopy was performed on a Nicolet iS50 FTIR Spectrometer with a diamond ATR crystal and DTGS detector. XPSterm·mth was performed using a PHI Versaprobe III Scanning XPSterm·mth Microprobe.

Data availability All relevant data is available upon request from the authors.

Electronic supplementary material Supplementary Information Peer Review File