Control of Polymers’ Amorphous-crystalline Transition for Hydrogel Bioelectronics Miniaturization and Multifunctional Integration

Bioelectronic devices made of soft elastic materials exhibit motion-adaptive properties suitable for brain-machine interfaces and for investigating complex neural circuits. While two-dimensional microfabrication strategies enable miniaturizing devices to access delicate nerve structures, creating 3D architecture for expansive implementation requires more accessible and scalable manufacturing approaches. Here we present a fabrication strategy through the control of metamorphic polymers’ amorphous-crystalline transition (COMPACT), for hydrogel bioelectronics with miniaturized fiber shape and multifunctional interrogation of neural circuits. By introducing multiple cross-linkers, acidification treatment, and oriented polymeric crystalline growth under deformation, we observed about an 80% diameter decrease in chemically cross-linked polyvinyl alcohol (PVA) hydrogel fibers, stably maintained in a fully hydrated state. We revealed that the addition of cross-linkers and acidification facilitated the oriented polymetric crystalline growth under mechanical stretching, which contributed to the desired hydrogel fiber diameter decrease. Our approach enabled the control of hydrogels’ properties, including refractive index (RI 1.37-1.40 at 480 nm), light transmission (> 96%), stretchability (95% - 111%), and elastic modulus (10-63 MPa). To exploit these properties, we fabricated step-index hydrogel optical probes with contrasting RIs and applied them in optogenetics and photometric recordings in the mouse brain region of the ventral tegmental area (VTA) with concurrent social behavioral assessment. To extend COMPACT hydrogel multifunctional scaffolds to assimilate conductive nanomaterials and integrate multiple components of optical waveguide and electrodes, we developed carbon nanotubes (CNTs)-PVA hydrogel microelectrodes for hindlimb muscle electromyographic and brain electrophysiological recordings of light-triggered neural activities in transgenic mice expressing Channelrhodopsin-2 (ChR2).

hydrogel shrinking behaviors in a desiccated state have been applied to densify patterned materials 1 6c), and increased autofluorescence (17.8% increase of 4 wt.% TEOS hydrogels compared to 0 1 wt. % TEOS hydrogels, excitation wavelength: 485 nm, excitation peak: 520 nm, Supplementary 2 Fig. 6d). The optimal TEOS content was chosen as 3 wt. %, which resulted in hydrogels with 1.54 3 ± 0.01 of refractive index (Fig. 2c), > 96% of transmittance ( (Fig. 2c, for 0.15 ± 0.02 mm thick 4 membranes), and 6.13 ± 0.16 relative fluorescent units (RFU)/mm of autofluorescence (for 0.15 ± 5 0.02 mm thick membranes. water: 3.70 RFU/mm, Supplementary Fig. 6d). 6 We then examined whether COMPACT hydrogels maintained tissue-like elasticity. COMPACT 7 hydrogel fibers exhibited relatively low elastic moduli while maintaining high stretchability 8 ( Fig.2d and Supplementray Fig. 7a-b). The optimized COMPACT hydrogel fiber (3 wt.% TEOS, 9 12 mM HCl acidification treatment and 200% stretching, diameter: 227 ± 18 μm) exhibited an 10 elastic modulus of 34.03 ± 7.38 MPa. Compared to silica fibers (~20GPa elastic modulus) 29 and 11 polymer fibers (~1GPa elastic modulus) 2,4 , COMPACT hydrogel fibers offer enhanced mechanical 12 matching to the nervous tissues (1-4 kPa) 30 and lead to less neural tissue damage from micro-13 motion involved in vivo studies 31 . 14 We then tested whether crystalline-enabled size reduction of COMPACT hydrogels can overcome 15 the intrinsic hydrogel swelling exhibited upon hydration and maintain structural stability in vivo, 16 we incubated COMPACT hydrogel fibers in ex vivo physiological conditions (pH: 6-8, 37 o C, 17 saline solution) and monitored fibers' dimension over time. We observed the shrinking percentage 18 maintained above 74% over 3 months ( Fig. 2e and Supplementary Fig. 8). Cytotoxicity tests with 19 human embryonic kidney cells (HEK293) exhibited no significant cell death in the presence of 20 COMPACT hydrogels ( Fig. 2f and Supplementary Fig. 9). 21 Step-index hydrogel optical fibers COMPACT hydrogels were first fabricated into step-index optical fibers (Supplementary Note 1 3). Increased RI contrast between optical core and cladding layers ensures light transmission and 2 the consequent photodetection sensitivity (Fig. 3a). Based on tunable refractive indices of 3 COMPACT hydrogels ( Fig. 2c and Supplementary Fig. 6 a-b), we designed step-index hydrogel 4 fibers with high-RI core (ncore=1.40) and low-RI cladding (ncladding=1.34). 5 Hydrogel fibers were connected to a silica segment embedded in an optical ferrule, which provides 6 a strong connection while preventing directly exposed hydrogel dehydration out of tissues and 7 light loss (Supplementary Note 3). We validated the function of RI-contrasting core-cladding 8 structures by comparing the light transmission between bare core fibers, step-index fibers with 9 plain cladding and those with light-protective cladding (Fig. 3b-c, and Supplementary Note 3-4). 10 The bare core fibers (diameter of 329 ± 17 μm) exhibited a relatively high attenuation (1.87 ± 0. 53 11 dB/cm) while introducing a thin low-RI cladding layer (thickness of 84 ± 4 μm on the surface of 12 372 ± 10 μm cores, ncladding=1.34) decreased the light transmission attenuation to 1.75 ± 0.08 dB/cm 13 ( Fig. 3c). A representative light-absorption nanomaterial 32,33 , reduced graphene oxide (rGO) was 14 loaded into low-RI cladding to further protect light leakage from fibers' lateral surface and 15 consequently reduced the light attenuation to 0.94 ± 0.25 dB/cm (core 339 ± 35 μm, cladding: 36 16 ± 11 μm of 5 wt.% PVA with 0.21 wt.% rGO) ( Fig. 3c and Supplementary Fig. 10). 17 To validate their functionality for in vivo optical interrogation, we tested COMPACT hydrogel 18 fibers with fiber photometric recording in the context of mouse social behaviors. Activation of 19 VTA region and its related circuits has been studied with various techniques, including 20 optogenetics 34 , electrical stimulation and chemogenctics 35 , related to social behaviors in mice 36 . 21 As a proof-of-concept application, we applied COMPACT hydrogel fibers to record mouse deep COMPACT optical fibers (580 ± 35 µm) in VTA after injecting of adeno-associated virus (AAV) 1 containing genetically encoded calcium indicator (hSyn::GCaMP6s) (Fig. 3e). A home-built fiber 2 photometry system (wavelengths: λisosbestic point=405 nm, λexcitation=470 nm, λemission=510 nm) based 3 on the previous design was used to collect GCaMP fluorescent change as a proxy to reflect the 4 neural activity 37 (Fig. 3g and Supplementary Fig. 13). We utilized the stiffness change of 5 hydrogel fibers from a desiccated state (stiff) to a hydrated state (soft) and implanted the hydrogel 6 fiber in the desiccated state with calibrated coordinates (Supplementary Figure. 11-12). After an 7 incubation period of 4 weeks for AAV virus expression, we subjected mice to a social behavioral 8 test with concurrent photometric recordings. Mouse social interactions were analyzed with 9 DeepLabCut markless pose estimation and a custom-developed MatLab algorithm (Fig. 3f). We 10 found that increased fluorescent intensity of GCaMP was correlated with mouse social interaction 11 epochs (Fig. 3h). 12

COMPACT multifunctional hydrogel neural probes 13
Hydrogel matrix can support various nanoscale materials to extend the functionalities while 14 maintaining desired mechanical properties 35 , 38 . To enrich hydrogel neural probes' modality for 15 electrical recordings, we incorporated conductive carbon nanotubes (CNTs, 12 ± 6 nm diameter) 16 into PVA hydrogel scaffolds during hydrogel cross-linking ( Fig. 4a and Fig. 4c and Supplementary Fig. 15) and 21 impedance was tunable with designed mold sizes and CNT loadings ( Fig. 4d and f). CNTs-PVA styrene (SEBS) (Supplementary Note 5 and Supplementary Fig. 16). To verify the stability of 1 CNTs-PVA hydrogel electrodes, we incubated them in PBS solutions and characterized the 2 impedance over 6 weeks (Fig.4e). No significant increase of impedance at 1kHz was found. 3 Then we deployed CNT-PVA hydrogel electrodes for electromyographic (EMG) recordings of 4 mouse hindlimb muscles in response to the pulsed blue light illumination. CNT-PVA hydrogel 5 electrodes detected hindlimb muscle electrical signals upon transdermal optical stimulation 6 (wavelength λ=473 nm, 200 mW/mm 2 , 0.5 Hz, pulse width 50ms) in Thy1::ChR2-EYFP mice, 7 which express photo-excitatory opsin, Channelrhodopsin 2 (ChR2), in the nervous system ( Fig.  8   4f). EMG signals exhibited repeatable amplitude and signal-to-noise ratios, which indicates the 9 reliability of CNT-PVA hydrogel microelectrodes. 10 When extending hydrogel miniaturization from bulk materials to interfaces, the COMPACT 11 strategy offers a new avenue for multiple components integration. Since RI-distinct core-cladding 12 structures ensure light transmission in optical cores, we introduced two CNT-PVA electrodes into 13 the cladding layers with a COMPACT hydrogel core (Fig. 4b). A hydrogel optoelectrical device 14 (optrode), is designed to enable optical modulation with simultaneous electrophysiological 15 recording (Supplementary Note 6). In Thy1::ChR2-EYFP mice, blue light pulses (λ=473 nm, 0.5 16 Hz, pulse width 50 ms, 10 mW/mm 2 ), delivered through the hydrogel optical core, consistently 17 activated ChR2-expressing neurons in VTA while the neural electrical signals were collected 18 through CNT-PVA electrodes (Fig. 4l). The optical evoked potentials were repeatedly captured 19 with correlation with the onset of light stimulation over two weeks post-implantation. 20

Discussion
In this study, we developed a set of hydrogel cross-linking chemistry and fiber-shaped device 1 microfabrication approaches through a bottom-up strategy of tuning polymers' amorphous-2 crystalline transition for hydrogel bioelectronics miniaturization and integration. COMPACT 3 provides an accessible, scalable, and controllable fabrication method for micro-structured hydrogel 4 fibers as small as 80 μm with consistently low asperity. These hydrogels provide a platform for 5 functionally augmented interfaces through loadings of additional nanomaterials. COMPACT 6 hydrogels can be further designed into step-index optical probes and optoelectronic devices 7 (optrodes) which are well-suited for neural modulation and recordings concurrent with behavioral 8 assays in mice. 9 Unlike established approaches to shrink hydrogels via desiccation, where collapse of polymer 10 chain during drying leads to reversible swelling upon hydration, COMPACT hydrogels' 11 polymetric nanocrystalline and enhanced interpolymer chain interactions maintained stable 12 folding in the hydrated state and therefore permit retained volumetric size reduction. Over 3 13 months of incubations under physiological temperature and osmolarity, the shrunk COMPACT 14 hydrogel fibers maintained the designed diameters within less than 1% variance (Fig. 2e), which 15 illustrates COMPACT bioelectronics' volumetric stability of their miniaturized size in vivo. In 16 contrast, COMPACT hydrogel fibers incubated at PVA dissolution temperature (100 o C) in water 17 for several hours resumed their pristine swollen size; this volume reversion demonstrates the 18 crystalline impact on size reduction through control of local free volume in hydrogel matrices. 19 This crystalline-dominated hydrogel miniaturization phenomenon can be extended to other semi-20 crystalline polymers at different material interfaces, where volumetric stability is important, such 21 as the proton-exchange membrane in packed fuel cells.
In COMPACT hydrogels, chemical cross-linkers and acidification treatment both contribute to the 1 retained volumetric decrease upon re-hydration while mechanical deformation induced the 2 orientated nanocrystalline growth. An increased number of chemical cross-linkers, TEOS (0 wt.% 3 to 4 wt.%, Fig. 1g), enhanced the anchoring of amorphous PVA chains through covalent cross-4 linking and prevent swelling in the hydrated state. Under the same cross-linking degree, 5 acidification treatment granted polymer chains enhanced interactions and suppressed crystallinity 6 ( Fig. 1j and Supplementary Fig. 1 and 7c). Nanocrystalline domains maintained the nanoscale 7 size (~3.5 nm) without compromising the transmittance in the visible range. Axial mechanical 8 deformation re-orientated nanocrystalline and created anisotropic nanostructures (Fig. 1k), which 9 enabled hydrogel fibers' desired decrease in diameter while causing a minimal effect on 10 crystallinity degree (Supplementary Fig. 1c) or nanocrystalline size (Fig. 1k). 11 Controllable hydrogel shrinking provides an effective methodology for miniaturization and 12 integration for neural probe fabrication. The molding and extrusion approaches offer a series of 13 precisely controlled hydrogel fiber diameters with structural homogeneity and low surface asperity 14 to avoid diffuse reflection at the hydrogel interfaces. COMPACT hydrogel fibers' tissue-like 15 mechanical properties exhibit improved immune response compared to stiff silica fibers ( Fig. 2d  16 and Supplementary Fig. 14). Although the mold sizes are commercially limited, COMPACT 17 procedures, including regulating polymer and crosslinker constituent content and fiber extensions 18 can expand the range of available fiber sizes. Successive rounds of molding with strong polymer 19 chain infiltration at the interfaces enable the design of multimodal microstructures, including core-20 cladding (30-80 µm) in step-index optical probes and electrode integration in the cladding layer of 21 optrodes. Currently, the number of integrated components, such as electrodes and microfluidic 22 channels, is limited by the coaxial alignment in the secondary molding step; the accessibility and throughput of multimodal fabrication can be further improved with guiding devices to facilitate 1 integration and alignment, or alternative coating approaches. 2 COMPACT strategy is generalizable for soft and stretchable bioelectronics. Polymer matrices 3 provide sufficient free volume for water access as well as nanomaterials' incorporation. High 4 aspect-ratio nanomaterials, such as silver nanowires and carbon nanotubes, can be effectively 5 entangled with polymer chains through cross-linking and condensation during acidification and 6 stretching. This procedure augments electrical conductivity while maintaining viscoelasticity. The 7 colloidal stability of nanomaterials in viscous polymer precursor solutions is important to create a 8 homogeneous composite after cross-linking to prevent phase separation and ensure stable electrical 9 conductivity. 10 Compared to other soft bioelectronics fabrication approaches, such as lithography and micro-11 printing, COMPACT technique offers scalable and efficient multimodal hydrogel fibers 12 manufacturing without the need for expensive and sophisticated facilities. COMPACT 13 multifunctional neural probes have been employed for bi-directional optical interrogation 14 concomitant with mouse social behaviors and electrical recordings of light-triggered neural 15 activity in mice. Extended functionalities, such as drug or viral vector delivery, can be further 16 achieved by integrating additional microfluidic channels in the cladding layer and retains light 17 transmission efficiency in the optical core. COMPACT multifunctional neural probes involve 18 independent components alignment and miniaturization steps, which potentiates the integration of 19 multiple components with various lengths to target multiple depths of tissue within single-step 20 implantation. This adaptability will increase the density of functional interfaces and overcome the 21 traditional limitation of fiber-shaped neural probes with single-target interfaces at the tip.
Control over semi-crystalline polymers' amorphous-crystalline transition creates a direct 1 fabrication methodology for elastic soft materials. Extending it to the manufacture of sophisticated 2 optoelectronic devices, the COMPACT strategy imparts a generalizable and modular platform for 3 hydrogel bioelectronics' miniaturization and integration, which consequently enables multimodal 4 interrogation of complex biological systems. TEOS emulsion while homogenizing at 12000 rpm using a portable homogenizer until a stable 20 emulsion was formed. The resulting emulsion was further homogenized using a high-speed 21 homogenizer (FSH2A lab). The mixed solutions were stirred in a water bath at 100 °C for 1 hour until transparent solutions were obtained, followed by an additional 12 hours of stirring at 60 °C. 1 The composition of all solutions used in this study is provided in Table 1. 2 Optical hydrogel probe fabrication. A step-index multimode silica fiber (core diameter 400 μm, 5 NA 0.5, Thorlabs FP400URT) was prepared by removing the protective coating using a fiber 6 stripping tool (Micro-strip, Micro Electronics, Inc). The stripped fiber was then divided into 13-7 mm segments using a diamond cutter. These fiber segments were inserted and extruded from one 8 end of an optical ferrule (bore diameter 400 μm, Thorlabs CFX440-10) with a length of 2.5 mm 9 and secured with EccoBond F adhesive (Loctite). Both ends of the silica fibers in the ferrules were 10 polished using a polish kit (Thorlabs D50-F, NRS913A, and CTG913). The light transmission of 11 all silica fibers and ferrules was tested by coupling with a 470 nm blue light-emitting diode (LED) 12 (Thorlabs M470F3) after polishing. To remove the plastic coatings on the extruded silica fibers, 13 they were treated with 2M sodium hydroxide solution (Sigma-Aldrich, 1064980500) for 2 hours 14 followed by an additional treatment with chloroform (Sigma-Aldrich, 472476) for 30 minutes. A 15 thin layer of 10 wt.% PVA was then coated on the extruded silica fibers via dip coating, and the 16 PVA-coated silica fibers were air-dried at room temperature for 12 hours and annealed at 100 °C 17 for 2 hours. A vacuum planetary mixer (Musashi ARV-310, 2000 rpm, and 16 kPa vacuum) was 18 utilized for the mixing and degassing of all solutions. For degassing and mixing, 100 μL of GA TEOS solution was also degassed and mixed for 1 minute. Subsequently, the above two solutions 1 were combined (weight ratio of 1:1) and mixed for another minute. The resulting PVA-TEOS-GA 2 solution was infused into silicone tubes (McMaster-Carr 5236k204, 80 mm in length), and the 3 optic ferrules were inserted into the silicone tubes, with the silica fiber end connected to the PVA 4 mixture. After curing at room temperature for 4 hours, the PVA-TEOS-GA fibers were demolded 5 using dichloromethane (DCM, Sigma-Aldrich, 270997, 99.8%) and washed with a large amount 6 of water to remove residual chemicals for two days. Ferrule-connected fibers were air-dried at 7 room temperature for 12 hours and annealed at 100 °C for 20 minutes. Finally, the hydrogel fibers 8 were rehydrated with MilliQ water before use. The compositions of all fabricated fibers are listed 9 in Table 2. , where k is a 9 dimensionless shape factor that varies based on the actual shape of the nanocrystalline domain (k 10 = 1, approximating the spherical shape of the nanocrystalline domains), λ is the wavelength of X-11 ray diffraction (λ = 1.54 Å), θ is the peak of the Bragg angle, and β is the full width at half 12 maximum (FWHM) of the WAXS peaks. The d-spacing between nanocrystalline domains was 13 calculated using = for the membranes were prepared using the same method as discussed previously. After spin-20 coating, the PVA membrane-coated Si wafers were allowed to cross-link and dry in the air for at 21 least 12 hours and then annealed at 100 °C for 20 minutes. The refractive index (RI) of the PVA 22 membrane-coated Si wafers was measured using an ellipsometer (J.A. Woollam RC2) in the range of 400 nm to 700 nm. The measurements were carried out on the membranes in their desiccated 1 states. A series of COMPACT hydrogel membranes (0-4 wt.% TEOS) were prepared using a 2 similar procedure as described above but using a rectangular mold (21.5 × 21.5 × 1 mm). The 3 membranes were demolded after cross-linking, dried at room temperature for 12 hours, and cut 4 into small sheets (2 × 2 mm). The sheets were then annealed at 100 °C for 20 minutes and reswelled 5 in MilliQ water for 1 hour. The RI of the membranes in their hydrated states was measured using 6 a refractometer (Sper Scientific 300034) with water used for calibration. techniques. Subsequently, 1 mL of PVA solution was added to each well and allowed to cross-link 12 and air dry for at least 12 hours, followed by annealing at 100 °C for 20 minutes. Rehydration of 13 the membranes was achieved by the addition of 100 μL of MilliQ water to each well. To obtain 14 transmittance spectra in the range of 400 nm to 700 nm, the 96-well plate was subjected to analysis 15 using a plate reader (Biotek Synergy 2). Autofluorescence measurements were acquired using 16 excitation/emission wavelengths of 470 nm/510 nm and 485 nm/520 nm, respectively. Membrane interval of cutting was adapted. Starting from the far end of the ferrule, the output power was 20 measured after each cut using a cutter. The attenuation coefficient (α) was calculated using the 21 in meters, respectively. P1 and P2 are the transmitted power readings before and after the cut, 1 respectively. 2 3 Dimension measurements of hydrogel fibers. Microscopic images of hydrogel fibers were 4 captured using a bright field mode microscope (AmScope) in MilliQ water. Three distinct regions 5 of each fiber, namely two ends and the middle part, were imaged. The diameter of each fiber was 6 measured using ImageJ software, with nine measurements taken for each fiber. The length of the 7 fibers was measured using a caliper, with three measurements taken for each fiber. Animals. All animal surgeries were reviewed and approved by the Committee on Animal Care at 2 the University of Massachusetts Amherst. Wild-type (C57BL/6J) mice and Thy1::ChR2-EYFP 3 mice were purchased from the Jackson Laboratory. Mice were given ad libitum access to food and 4 water and were housed at 24 °C ± 1 °C, with 50% relative humidity, and on a 12-h light/12-h dark 5 cycle. All experiments were conducted during the light cycle. 6 7 In vivo hydrogel optical fiber implantation into the mouse brain. C57BL/6J mice were 8 anesthetized using 1.0% isoflurane administered in a chamber and subsequently secured onto a 9 stereotactic frame (RWD Life Science) with a heating pad to maintain their body temperature. All 10 surgical procedures were conducted in sterile conditions with 1% isoflurane used to maintain 11 anesthesia. The Allen Brain Atlas was used to align the skull and determine the coordinates for 12 viral injection and fiber implantation, specifically targeting the ventral tegmental area (VTA) at 13 coordinates AP: -2.95 mm, ML: ± 0.50 mm, DV: -4.80 mm. An opening was made in the skull 14 using a micro drill (RWD Life Science) at the designated coordinates. A total of 600 nL of adeno-15 associated virus (AAV) carrying hSyn::GCaMP6s was injected into the target region via a micro 16 syringe and pump (World Precision Instruments, Micro 4). The viral injection device was held in 17 place in the VTA region for 15 minutes to facilitate virus diffusion. Following fiber probe insertion, 18 the probes were lifted by 0.1 mm to accommodate for the viral volume. Finally, the fiber probes 19 were secured to the skull using an adhesive (Parkell, C&B METABOND) and reinforced using 20 dental cement (Jet Set-4). The mice were monitored on the heating pad following removal of 21 isoflurane until they were fully awake. 22 In vivo optrode device implantation into the mouse brain. Thy1::ChR2-EYFP mice were 1 anesthetized with 1.0% isoflurane and placed on a stereotactic frame (RWD Life Science) 2 equipped with a heating pad to maintain body temperature. Surgery was conducted under sterile 3 conditions, and 1% isoflurane was continuously administered to maintain anesthesia. Allen Brain 4 Atlas was utilized to align the skull and establish optrode device coordinates (VTA, AP: -3.00 mm, Subsequently, a novel mouse was introduced to the social zone, and the test mouse was exposed 7 to the novel mouse and allowed to interact freely. Concurrently, GCaMP fluorescence changes 8 were recorded during social tests. A dark-vision camera was installed above the social chamber to 9 record video footage during the social tests. The time spent interacting and the distance of social 10 interaction were analyzed using customized algorithms for social interaction assessment with 11 DeepLabCut. The analyzed social interaction epochs were then correlated with GCaMP signals. The slides were left to dry in air at room temperature overnight before images were acquired using 9 a confocal microscope (Leica SP2). 10 11 Electromyography. EMG signals were recorded from the gastrocnemius muscle with one 12 reference needle electrode, one hydrogel working electrode (287 ± 14 μm) and one ground 13 electrode. A 473 nm laser (200 mW/mm 2 , 0.5 Hz, pulse width 50ms) was used for transdermal 14 optical stimulation. EMG data triggered by optogenetic activation were collected through a 15 DAM50 system. 16 17 In vivo electrophysiology. Electrophysiological recordings were performed by connecting the pin 18 connectors of optrode devices to a DAM50 recording system. Optical illumination was carried out 19 using a 473 nm laser connected to the implanted optrode devices via a ferrule-sleeve-ferrule 20 connecting system. The laser (10 mW/mm 2 ) was pulsed at a frequency of 0.5 Hz with a pulse width 21 of 50 ms during optical stimulation. Signals were sampled at 50 kHz and filtered between 1-1000 22 1 Code availability 2 The custom code used in this study is available from the corresponding author upon reasonable 3 request. 4