Tuning the Physically Induced Crystallinity of Microfabricated Bioresorbable Guides for Insertion of Flexible Neural Implants

Devices that safely interface with the brain are critical to advancing neuroengineering. Thin and flexible neural implants show great promise alongside established silicon technologies. They therefore require a physical stiffener to allow their insertion into brain tissue. Bioresorbable polymer shanks are novel transient guides enabling accurate implantation using biocompatible materials that will be absorbed by the body over time. The development of materials with optimized stiffness and degradation is needed to provide minimally invasive probes with precise insertion capability under surgical conditions. A microfabrication protocol for the patterning of polyvinyl alcohol and its physical cross‐linking is presented, resulting in insertion guides with precise shapes and tunable degradation and stiffness. The results demonstrate a remarkable improvement in batch fabricating micro‐scale neural shanks with designed crystallinity. It results in their prolonged degradation time, evaluated in agarose gel, and remarkably improved penetrability due to the increase in mechanical stiffness. In vitro and in vivo studies support the high acceptability of this combination in interfacing with neural cells and tissue. This work represents a novel approach to the material and process engineering of bioresorbable polymers for developing fully organic and safe implants.


Introduction
[3] Currently, available probes are based on silicon because it offers high-resolution electronic device fabrication.[6][7] The development of flexible and plasticbased microelectrodes promises to tackle this problem through the generation of ultra-thin flexible iMEAs based on organic and soft materials.0][11][12][13] Their performance in brain-machine interfaces has been demonstrated for over a decade.Although several device prototypes made of plastic electronics are mechanically minimally invasive and trace the future of neural implants, they are unable to reliably penetrate brain tissue.Two techniques have been explored to avoid the buckling of flexible MEAs and to precisely place the probes in the desired brain region.[16][17] This method suffers from tissue damage during both the insertion and retraction of the guide. [18]Another relatively safe alternative is the use of bioresorbable stiffeners, which can dissolve after insertion.Studies show that this type of guide has better outcomes in terms of mechanical and biological failures than non-resorbable stiffeners.[21][22][23][24] The remaining limitations in the bioresorbable shank approach consist of their restrained to the fabrication process miniaturization, only too fast or slow degradation time window possibility, and implantability, which is defined by the material's intrinsic mechanical stiffness, reducing the choice of the materials.
Miniaturization refers to limitations from a manufacturing process perspective.The actual methodology in bioresorbable polymer processing is to fabricate each shank individually and assemble them with flexible devices within a PDMS mold. [19]he mold fabrication process itself requires photolithography or 3D printing steps.The resolution of the mold fabrication defines the final shape of the devices.The molding process can also cause bending and damage to the probes during the blade coating and release steps.Cointe et al. proposed a more robust technique where silk fibroin is integrated into the flexible probe fabrication process. [20]This method allows the batch fabrication of neural implants together with the biodegradable shank.This co-fabrication process of the iMEA device and shank combines the resolution of the electronic features with the layer of the bioresorbable polymer as a starting substrate for fabrication.This presents complementary challenges that limit the final resolution of the overall device.In addition, several solvents are applied during device fabrication, which may not be compatible with various bioresorbable materials.Therefore, the versatility of the fabrication process is affected by the material specifics and process resolution.The choice of materials therefore defines its degradation profile and determines the device application.For neural implantation, surgical protocols require a slow rather than fast insertion procedure.Bioresorbability is thus a key parameter but is limited by the processable materials availability.[27] Although primary experiments showed promising data, PVA degrades within 30 s, an extremely short time window for its manipulation.In contrast, silk fibroin shanks degrade in the order of days to months, depending on the degree of polymer crystallinity.This long degradation time [20] may affect cellular responses from a long-term tissue inflammation perspective.Indeed, the development of neural probes requires a subtle balance between soft tissue specificity and a robust device design that can efficiently interface with the brain.The stiffness of the shank is another parameter that plays a key role in the location of the iMEA in a specific brain region.Often, the lack of rigidity in plastic-based implants is compensated by increasing their thickness.This approach causes significant damage to neural tissue during insertion and goes against the device's miniaturization concept toward minimally invasive interfaces.The optimal solution could be to use a highly crystalline material that provides the required stiffness at a low thickness with a degradation time that can be adapted to the specifics of the implantation protocols.
In this work, we introduce a scalable microfabrication process to advance the use of biodegradable shank with tailored degradation rates and stiffness for neuroengineering.In this context, we micropattern PVA and tune its degradation time and stiffness through the physical crosslinking procedure.Our study results in high-resolution and scalable shank fabrication with improved mechanics and prolonged degradation.We believe that this novel insertion guide will enable safe and long-term implantation of flexible iMEAs and improve the current state of the art in brain surgery protocols.

Physical Cross-Linking of PVA
The key property of PVA is its biodegradability simply in environmental conditions.It is also a highly processable synthetic polymer that is low cost, biocompatible, and is produced by hydrolysis, alcoholysis, or aminolysis of poly (vinyl acetate). [27]The presence of reactive hydroxyl groups in its backbone provides several opportunities for physical or chemical cross-linking reactions to alter the degree of PVA structuring into a semi-crystalline matrix. [26]The chemical cross-linking of PVA is done by using aldehyde molecules such as glutaraldehyde, [28,29] glyoxal, [30] borate-containing species, [31,32] which limit their applications in the bioengineering field due to potential toxicity issues. [26,33]Alternatively, a physical cross-linking process via repeated freezethawing (F-T) cycles induces crystallite formation within the polymer, forming PVA-based hydrogel-like structures.As illustrated in (Figure 1A), during the F-T cycles, the PVA solution is exposed to liquid nitrogen (LN 2 ), which induces its freezing and subsequent thawing at room temperature.Two complementary mechanisms are involved in the gelation of PVA during this process: a) the separation of PVA and water by ice formation and b) the growth of PVA crystallization by forming crosslinking points between the PVA chains.The former creates PVA-rich domains in close contact, initiating the hydrogen bonding between PVA chains (Figure 1A-2).Crystallized PVA (c-PVA) domains are linked together by inter-and intra-hydrogen bonds, resulting in a strong hydrogel-like network (Figure 1A-3).]34] This process is largely carried out on pre-formed samples but can also be applied to the thin films formation supported by a glass substrate, as described below.

Thermal Characteristics of PVA Crystallinity Enhancement
Structural changes in polymers following the F-T process are strongly reflected in their thermal characteristics.In this work, we have used the largely adopted DSC technique to characterize the thermal profiles of PVA over successive F-T cycles.The melting point (T m ) and the heat of melting (ΔH m ) are measured in order to evaluate the degree of crystallinity.The thermal profiles of PVA and c-PVA with 1 to 4 F-T cycles are shown in Figure 1B.The curves show sharp peaks between 195 and 200 °C with a good correlation to the number of cycles of the F-T process, corresponding to the T m of crystalline domains in the PVA.The values of the heat flow amplitude follow the same increasing tendency with the F-T cycle number.Additionally, we observe a deviation from the linear progression of the peak sizes between cycles 2 and 3, suggesting significant structural changes initiated after two cycles of the F-T process.This variation is also remarkable in the etching rate reduction presented below and can be attributed to the significant toughening of the material.The crystallinity is calculated by dividing the ΔH m values of each specimen by the ΔH m value of 100% crystallized PVA, as found in the literature.Therefore, the value of ΔH m is 22.0 ± 1.3, 27.1 ± 1.8, 36.0 ± 1.4, 47.2 ± 2.3, and 55.4 ± 1.7 J g −1 , corresponding to 13.7 ± 0.9%, 16.3 ± 1.3%, 20.6 ± 1%, 25.4 ± 1.6%, and 28.4 ± 1.2% of crystallinity for PVA, c-PVA with 1 to 4 F-T cycles, respectively.These results are in agreement with the literature.Kim et al. reported the ΔH m of 101.70 J g −1 for 10 F-T cycles in the same condition, corresponding to 41% crystallinity, resulting in stiffness enhancement, additionally supported by complementary structural characterization techniques. [25]Adelina et al. [26] showed that PVA concentration, molecular weight, and its degree of hydrolysis, as well as the temperature and duration of the F-T process, have a strong influence on the gelation degree, mechanical properties, degree of crystallinity, and the ability for the water uptake by the polymer.A low degree of crystallization results in incomplete gelation with materials exhibiting sticky and elastic properties with rapid degradation in water.Thus, improved mechanical properties are expected in materials with higher crystallinity, resulting in a slower degradation.

Micropatterning of PVA and C-PVA
The adoption of the F-T process at the microfabrication scale is demonstrated here through photolithography-based process de-velopment on PVA and c-PVA substrates.This enables the micromanufacturing of bioresorbable shanks with a defined degree of crystallinity by controlling the F-Y cycles number.This enables the fabrication of shanks with tunable degradation time and stiffness, as discussed above.Here, glass substrates with a thin acetate cellulose layer are spin-coated with a PVA solution that undergoes the F-T cycles.The substrates are then coated with a polymeric masking layer of parylene-C , which makes the sample compatible with the wet processing chemistry during the photolithography steps and prevents degradation of the underlying PVA layer (see schematics in Figure 1C).This material is widely used in the microfabrication of bioelectronic devices due to its chemical neutrality, which allows creating substrates or encapsulation layers for electronic devices with high biocompatibility and conformability.Pa-C is formed by a low-temperature, dry process.A photoresist is then applied to the substrate, followed by the traditional steps of baking, exposure, and development, defining the shank micro-design on the surface of the sample.The samples are then patterned using the reactive ion etching (RIE) technique.This process etches a Pa-C film with well-defined vertical sidewalls.As shown in Figure 1D, the SEM images reveal a fussy structure of the patterned c-PVA, caused by the F-T process and the etching step.As the F-T process induces structural changes in the PVA, this is also reflected in its etchability.The physical cross-linking increases the crystallinity of the PVA film, which leads to a considerable reduction in the etching rate as the material becomes tougher.This value decreases from 1000 ± 100 nm min −1 , for the pristine PVA, to 300 ± 100 nm min −1 , for the crosslinked PVA at 4 F-T cycles (Figure 1D).These values are within the etching rates reported for polymer substrates used in microfabrication processes.For example, using the same etching process parameters, Pa-C is etched at a rate of 250 nm min −1 . [9,19]ilk fibroin, a natural polymer that is also used as an implantation guide, etches at a rate between 400 and 1000 nm min −1 , depending on the ratio of CF 4 to O 2 gases. [20]The final release of the PVA shank from the substrate, after the etching step, is done by a flood etching of the entire substrate to remove the residual Pa-C masking layer and its immersion in an acetone bath to dissolve the acetate cellulose layer, which enables a simple detachment of the PVA shanks, shown in Figure 1F.Hence, the success of this microfabrication process relies on the Pa-C layer adoption as a masking layer and the ability to selectively etch PVA and c-PVA materials.This process can be extended to the other types of bio-polymers with the same capabilities as silk, cellulose, gelatine, PLGA, etc.The resolution of the process is mainly defined by the photolithography tool and the Pa-C patterning via etching.In our case, we have successfully reached 10 μm guide width resolution.Therefore, the minimum resolution of fabricated patterns depends on the PVA layer's characteristics and its control over the entire process in relation to the application specifics.Here, our target is flexible neural MEAs with critical probe dimensions in the range of 50 to 100 μm to match the flexible Pa-C-based probe.The guide dimensions also depend on specific form factors and the required mechanical robustness, induced by the F-T process and manually performed.This leads to variability in material properties, which is reflected in the precision of the optimized etch rate determination.With the high degree of crystallinity of the PVA, the guides became brittle and difficult to manually handle during the final assembly step.The aligned lamination technique is a way to deal with assembly in a controlled or automated way and to reach high-resolution microscale processing.The process will require that the PaC probe and c-PVA guide have similar layouts on the fabrication wafers and include specific alignment marks.Therefore, the final process optimization will depend on the implant geometry, the fabrication of a PVA guide with these geometrical specificities, and the evaluation of its implantability, which can be tuned by the F-T controlled cycling.Additional fine-tuning of the shank thickness and the F-T cycling number will be required to benchmark its bioresorbability in the brain.

Effect of F-T Cycles on the PVA Degradation Time
The efficiency of PVA as a bioresorbable shank has been demonstrated previously, showing that the rapid degradation time of about 30 s limits the surgical procedure to adopt a fast implantation procedure. [19]Figure 2A shows the results of the PVA thickness and the effect of the cross-linking process on its degradation.Here, the shanks are implanted in an agarose gel used as a brain in vitro model.By increasing the thickness of the PVA shanks from 25 to 120 μm, the degradation time ranges from 15 s up to 3 min.These values were compared with c-PVA.The 80 μm shanks of c-PVA, fabricated with different F-T cycles, were implanted and we observed an increase in the degradation time window up to 15 min at four cycles.As expected due to the higher level of crystallinity in the c-PVA changed.Pristine PVA (p-PVA) with a thickness of 80 μm degrades in 2 min, starting from the tip and spreading over the entire length of the shank, following the insertion path.This is shown in Figure 2B, which presents the photographs taken during the degradation window in the agarose gel.This behavior changes in c-PVA prepared with a maximum of four F-T cycles.As illustrated in Figure 2C, the c-PVA shanks absorb the water and then swell during the first 5 min, starting from the edges toward the center of the shank.Once the shank has completely absorbed the sufficient water content, it forms a hydrogel-like structure, clearly visible in the images.This suggests that the degradation will begin after 15 min as at this time point the c-PVA which is outside of the agarose gel separates from the inserted part.The swelling degradation profile of c-PVA confirms the successful crosslinking process of the PVA after the microfabrication, as its degree of crystallinity is estimated to increase from 14% to 28 ± 1.2%.This structural modification actually delays its dissolution time.The crosslinking reaction at various F-T cycle number defines the degree of the PVA molecular network interconnection and enables the modulation of the dissolution time window.As a key indication, such a rise in degradation time by the F-T process makes possible the miniaturization of the thickness of the shanks, or at the same thickness, the extension of the surgery time almost five times.Additional parameters can influence this phenomenon, as discussed in Section 2.2.

Improved Mechanics by Modulation of Crystallinity
As shown earlier, the crosslinking process after F-T cycles increases the crystallinity of the material, which impacts its Young's modulus and therefore stiffness. [25,35]In Figure 3A, we present the visual appearance of the pristine PVA and 4 F-T cycles c-PVA samples, which were molded in the size-defined form-template for mechanical evaluations.After the sample preparation, the PVA curves as it has a flexible nature, where the c-PVA retains its flat and planar shape.These samples then undergo a compression test to measure the buckling force and angle, essential for the implantation performance evaluation.We observe that the PVA folds entirely at the applied force of 2 N (Figure 3B) whereas c-PVA does not bend but breaks at the force of 15 N (Figure 3C).These results confirm the increase in stiffness and rigidity of the c-PVA.Additional Young's modulus evaluations allowed us to quantify its increase from 0.9 ± 0.1 to 28 ± 2 kPa for PVA and c-PVA, respectively.The (Figure 3D) shows an exponential increase in the modulus with the addition of the F-T cycles applied to the PVA.As defined by the thermal evaluations, the crystalline level of the PVA increased by a factor of two.This results in a mechanical stiffness enhancement by a factor of 30.
The significant stiffness improvement in c-PVA is further quantified using an insertion test with a universal mechanical testing machine into a PDMS membrane, mimicking rat dura mater as the closest animal model. [8]Most of the surgical protocols for implanting neural probes require the removal of the dura mater.This membrane is relatively thick and limits the penetration of low mechanical modulus polymer-based devices. [8]This makes surgery and animal recovery even more challenging as it raises the risk of bleeding and infection.Herein, we measure the penetrability of shanks in PDMS film with a thickness of 50 μm as an evaluation model. [36]As shown in Figure 3E, the tip of c-PVA shanks with 2, 3, and 4 F-T cycles placed above the PDMS film.The retraction machine pushes the shank down toward the film by gradually increasing the penetration force.The c-PVA with 2 F-T cycles, as well as pristine PVA (not shown here), do not penetrate the film but bend at the interface.The applied force is increased until it bends the shank by 90°(indicated by a  small pick around 40 mN at 2.2 mm displacement), which results in an overshoot of the force due to the shank getting in touch with the PDMS membrane (red curve).Similar measurements using the c-PVA with 3 and 4 F-T cycle shows good penetration at 130-140 and 35-60 mN, respectively.Such mechanical performance improvement between 2 and 3 cycles of F-T shanks is also highlighted in the Figure 1B, where significant thermal profile difference is observed, indicating structural modifications in the three F-T cycles material.The visual confirmation of this is presented in Video S1, Supporting Information.These forces are indicative, as they do not include the membrane deformation observed in the video.As the dissolution evaluations show in Figure 2, the PVA and c-PVA are able to be inserted into the agarose gel, mimicking the brain tissue.Therefore, c-PVA with a sufficient degree of crystallinity enables the fabrication of stiffer shanks that can penetrate the dura mater layer, providing a promising opportunity for the development of less invasive surgical protocols in neuroengineering.

Implantability Evaluations of the Flexible Probes with PVA Shank
As the agarose gel with 1% water content allows for in vitro evaluation of the implantability of neural implants, we assembled the shank with the flexible MEA device.Figure 4A shows the top-, 30°and side views of this assembly.This combination takes place in an acetone bath, where the two parts are brought together by their alignment, one on top of the other, in the liquid.The physical bonding happens after the removal of the acetone and the fast drying of the device.Upon insertion in the agarose gel, the PVA device degrades quickly within a window of 2 min, leaving the flexible probe stable (Figure 4B).The c-PVA shank absorbs water and forces the flexible device down for the first 5 to 15 min (Figure 4C).This time window corresponds to the swelling of the c-PVA as shown in Figure 2, top view.After 30 min, the probe returns to a straight position.This movement can be critical when targeting a small, specific part of the brain.We believe this probe displacement is exaggerated, as we expect c-PVA swelling followed by degradation will be slower in an animal model.The cellular network and blood vessels will strongly alter this movement.
While bioresorbable shanks are being explored by several groups as stiffeners for flexible probes, the effect of their degradation on potential device displacement has been neglected.Further in vivo investigations will be necessary to understand the probe displacement possibilities in the brain tissue.

In Vitro and In Vivo Assessment of the PVA-Assisted Flexible Probes in Neural Interfacing
The dissolution profiles of PVA suggest the presence of different degradation kinetics.The performance of the MEA electrodes during this process is investigated by electrochemical impedance spectroscopy (EIS) in an agarose gel formulated with the PBS solution, which supports an electrolytic contact in a three-electrode setup that includes an Ag/AgCl reference electrode, a large platinum counter electrode, and the MEA's electrode is used as the working.The initial impedance of the 50 μm diameter working electrode, measured at 1 kHz, is around 8 kΩ.The electrodes are then inserted into the gel, which mimics brain tissue, with the PVA shank placed to cover the working electrode.Figure 5a illustrates impedance values recorded during 48 hours from electrodes covered with p-PVA and c-PVA (with 4F-T cycles).This study is designed to evaluate the electrode's performance alteration due to the PVA's dissolution.The measured impedance values are initially extremely high at the beginning as the PVA forms a dielectric layer.They then gradually decrease to the initial electrode's impedance.The time window of the dissolution process is different for p-PVA and c-PVA, 5 min and 2 h, respectively.Such a time offset for c-PVA correlates with the slower dissolution as seen in Figure 4C.However, visually it was only attributed to a time period of 30 min.These measurements reveal a slower swelling profile of the c-PVA in PBS-formulated agarose gel, which supports the discussions in the previous section and indicates a necessity for a minimum 2 h time period to avoid any electrical variations at the recording electrodes.Therefore, this time is defined by the number of F-T cycles of the PVA.Cytotoxicity tests are then performed in primary rat cortical neuron cell culture prior to evaluation in an animal model.Figure 5b shows the fluorescence response of the PrestoBlue assay as an indication of cell viability through their ability to reduce the dye reagent content in the presence of the bioresorbable PVA shank.After 24 h, the rates of resazurin reduction by metabolically active cells as measured by fluorescence in the control (127.43 ± 1.36) and in a well containing the PVA shank (130.92 ± 2.37) are comparable.A slightly higher fluorescence is observed after 48 h, indicating an increase in cell viability in the case of PVA shank compared to the control (164.1 ± 3.74 vs 139.74 ± 1.42).Similarly, after 7 days of culture, the increase in fluorescence was 153.38 ± 3.11 and 194.79 ± 3.12 for the control and PVA, respectively.The differences are statistically significant for 48 h and D7 comparisons of the PVA shank with the control (p < 0.001).At each time point, cells were also inspected with an optical microscope to ensure healthy cell morphology and to support this positive cell growth progression.These results confirm the cytocompatibility of PVA in the interfacing primary neural cells.They were allowed to proceed with in vivo evaluations to demonstrate the probe implantability and electrophysiological recordings with flexible MEA electrodes.Two epileptic male mice were implanted with PVA-assisted Pa-C flexible electrodes above the DG region, targeting the right hippocampus and cortex areas.During in vivo recordings, we focused on local field potential activity including fast ripples (FRs) with a specific frequency band between 250 and 600 Hz.Recordings were taken on days 2, 4, 7, and 9 in freely moving animals, as shown in Figure 5d.FRs are short-duration, non-stationary transient events of low amplitude that are often mixed with background activity and other pathological events.Our results show successful recording of such low-amplitude activities in both brain regions with a stable signal energy content over 9 days.The signals are comparable to the routinely used gold wire study in the same animal model performed by our group. [37]As previously noted, the dissolution of the PVA shanks, presented here, should initially occur in the first 2 h after implantation.Our recordings confirm that the PVA shank dissolution does not affect the ability to record electrophysiological activities after this period and that the electrodes robustly perform over 9 days of observation.These preliminary results clearly support the idea that PVA shanks have sufficient mechanical strength and compatibly to be used as implantable guides.In order to assess the inflammatory and toxicological aspects in long-term evalua-tions, histological studies should determine the benefits of physical cross-linking of PVA as a safe insertion guide.Additionally, the degradation profile and adsorption pathway of the residual components of the PVA in the brain should be studied in depth, as speculated for the other organs.

Conclusion
The future clinical application of flexible implantable MEAs as minimally invasive deep brain-machine interfaces is highly dependent on their precise, robust, and biocompatible insertion.To date, the most promising technique is the use of bioresorbable shanks, as they degrade after implantation, thus reducing longterm damage to the brain tissue by the implant.The chemical and physical properties of bioresorbable materials, based on synthetic or natural polymers, determine their applicability in this field.Manufacturing techniques, stiffness, and dissolution rate are the remaining challenges for their large-scale production and integration.PVA is one of the most widely used materials for biointerfacing, with a great ability to physically induce its structural crystallinity through processing.Physical crosslinking of PVA through freeze-thawing cycles increases its crystallinity up to 28% while maintaining its etchability for micropatterning.This enables the development of a scalable photolithography-based microfabrication process that produces a sharp and well-defined microstructure of the shanks.The rapid degradation and low mechanical stiffness of PVA are overcome by modulating the degree of crystallinity, allowing the dissolution time to be delayed from 2 to up to 30 min for the insertion protocol.As demonstrated by the in vitro electrode's impedance assessment, PVA dissolution takes a much longer time after it is implanted than we observe visually.It also increases the Young's modulus from 0.9 to 28 kPa.As observed in the agarose gel insertion tests, such physical changes in PVA properties cause the flexible implant movement over the shank's degradation as it takes on the behavior of a hydrogel through its molecular network crosslinking.This phenomenon, as well as the effects of the material's bio-degradation and their consequent neurotoxicity, require further in vivo investigations.Since the crosslinking reactions of PVA are initiated by a physical freeze-thawing process without any other chemicals, and its biodegradation is slower than in its original form, we expect that the immune response to the degraded products should be correspondingly reduced.The biological tissue will be exposed to a lower dose of degraded PVA at a given time than in pristine PVA.The successful implantation and the evidence of the ability to record neural signals in mice models over 9 days of implantation reinforce this conclusion.Therefore, c-PVA demonstrates excellent suitability for brain-guided insertion and its processability by microfabrication for future scalability.This concept can be extended to various types of bioresorbable synthetic and natural polymers applied in transient electronics, neuroengineering, and miniaturized drug delivery systems.

Experimental Section
PVA Freeze-Thawing Process: PVA (polyvinyl alcohol, Sigma-Aldrich, M w 13 000-23 000, 87-89% hydrolyzed) solution was prepared by dissolving 20 g of polymer powder in 80 g of water at 70 °C.The PVA solution was deposited on pre-processed substrates for microfabrication or poured into a template mold for characterization.PVA samples were immersed in liquid nitrogen for 5 min and thawed at room temperature for 2 h.The freeze-thawing (F-T) process was repeated for up to four cycles in this work.
Shank Microfabrication: The precise manufacturing of PVA shanks was achieved by photolithography.First, the shape of the shanks was designed in AutoCAD 2021 Software.A 1 μm thick sacrificial layer was obtained by spin-coating on the clean glass slide of cellulose acetate (Sigma-Aldrich, average M n 30 000), followed by drop-casting PVA solution and drying at room temperature for pristine PVA (PVA) or followed by F-T process for c-PVA.All films were made with the same volume of PVA.Then, the samples were coated with 1 μm of Pa-C (Parylene C, Labcoater 2 machine, SCS).The samples were then spin-coated with AZ10XT photoresist (Microchemicals), baked, UV-light exposed (MJB4 contact aligner, SUSS Microtech), and developed in AZ developer.This process was adjusted according to the resist thickness.Next, the samples were etched in the reactive ion etching chamber (Oxford instrument, Plasmalab 80+) at 200 W, 10 °C using a mixture of O 2 (60 sccm) and CF 4 (10 sccm) gases.The etching step was measured with a mechanical profilometer (Dektak XT-S).After PVA pattering, the samples were rinsed with acetone to remove photoresist residues and etched again to remove the masking Pa-C layer.Finally, the samples were immersed in an acetone bath overnight to dissolve the cellulose acetate sacrificial layer and release the microfabricated shanks.
PVA and C-PVA Shanks Degradation Test: The degradation time of the PVA shanks was measured in the agarose gel (1% w/w in water) which mimics the brain tissue.The videos of PVA and c-PVA shanks dissolution were recorded by a numerical microscope (Keyence VHX-7000) during insertion and degradation.This allowed for recording of exact time in relation to the thickness and type of the shanks.
Shank Insertion Test: The artificial dura mater was made of polydimethylsiloxane (PDMS), as described in the literature. [8]First, elastomer and curing agent with the ratio of 10:1 (PDMS, Sylgard 184, Elseworth adhesives) were mixed together.The PDMS was spin-coated on a glass slide at a speed of 1000 rpm for a duration of 60 s and dried for 5 min at 100 °C.A compression test was carried out on the universal testing machine (Instron-3365).This test measured the insertion force of the shank into PDMS film with a thickness of 50 μm thickness.Using a 10 N load cell, the samples were inserted into the PDMS film at a speed rate of 1 mm min −1 .Force and displacement were measured while testing.
Estimation of Crystallinity in C-PVA: The crystallinity of the original and F-T PVA shanks (c-PVA) was measured by differential scanning calorimetry (DSC technique, TA Instrument-DSC Q200).Samples of 10 mg of dried film were sealed in an aluminum pan.The specimen was heated from 25 to 250 °C with 5 °C min −1 rate.The melting temperature (T m ) and heat (ΔH m ) were calculated from the melting peak on the DSC curve.Crystallinity was calculated by dividing ΔH m of each sample by the ΔH m 100% crystalline PVA, determined from the literature. [25]lexible Implant Fabrication Process: Devices were microfabricated according to the previously reported protocols. [19]Briefly, 2 μm of Pa-C was deposited onto clean glass wafers (Parylene C, Labcoater 2, SCS) as a flexible substrate.Gold electrodes were patterned using a lift-off process with S1813 photoresist following the resist datasheet.The UV irradiation dose was adjusted to the resist thickness (MJB4 contact aligner, SUSS Microtech).After the development, 10 nm of chromium and 120 nm of gold were evaporated on the wafers (Boc Edwards thermal evaporator).The wafer was then immersed in acetone for 4 h and rinsed with isopropanol.Next, another 2 μm of Pa-C insulation layer was deposited together with the adhesion promoter 3-(trimethoxysilyl) propyl methacrylate (A-174 Silane).All wafers were then spin-coated with AZ10XT photoresist, baked, exposed to the UV light, and developed in AZ developer as indicated in the resist datasheet.The samples were etched in the reactive ion etching chamber (Oxford instrument, Plasmalab 80+) at 200 W, 10 °C using O 2 (50 sccm) gas and rinsed with isopropanol to open the electrode interconnection and recording sites.This step was repeated to define the implant's outline.To detach the samples, all wafers were soaked in DI water overnight.Finally, the samples were gently removed from the water and collected on Kapton film for further processing. [38]lexible Implant and PVA Shank Assembly Process: After microfabrication of the flexible implant and PVA-based shanks, their assembly was done in a bath of a poor solvent of PVA placed under an optical magnifier.PVA cannot be dissolved by acetone.The two parts were immersed in the batch and manually aligned, one on top of the other.The assembly was then physically held together and removed from the bath.The rapid evaporation of the acetone allowed the sample to dry rapidly, promoting the physical bonding of the implant with the shank.
Electrochemical Impedance Spectroscopy (EIS) in Agarose Gel: Microfabricated electrodes were combined with the PVA shank as described in the previous section.Here, the shank was placed on top of the electrodes.To maintain the shank and device together before implantation, a 5% w/w solution of PLGA in acetone was prepared, and gently brushcoated on the probe.The impedance measurements were performed in a three-electrode cell set-up (Metrohm Autolab B.V.).Platinum coil and Ag/AgCl electrodes were used as the counter and reference electrodes, respectively.The impedance of the working electrodes, which were microfabricated electrodes with a diameter of 50 microns, was measured under the constant voltage of 0.01 V versus the reference electrode with sinusoidal waveform between 1 Hz and 10 kHz in PBS-PBS-formulated agarose gel.The gel was made by mixing 1 g of agar with the 99 g of PBS solution (electrolyte) and heated up until the powder was fully dissolved, then placed in the fridge to solidify.The initial average electrode's impedance was around 8 kΩ before the combination with PVA shanks.The measurements were taken during 48 h from p-PVA and c-PVA (4 F-T cycles) covered electrodes and values at 1 kHz were collected in the table for comparison.
Cell Culture of Primary Rat Cortical Neurons: Primary rat cortical neurons were collected and cultured as described by Callizot et al. [39] from fetuses of pregnant female Wistar rats provided by Neuro-Sys, Gardanne.The cells were seeded at a density of 20 000 cells per well in a 96-well plate, precoated with poly-d-lysine (Sigma-Aldrich, M w 70 000-150 000 amu, 0.1 mg mL −1 ), and were cultured in a humidified incubator at 37 °C and 5% CO 2 .Half of the medium was changed twice per week.
Cytotoxicity Test of PVA on Primary Rat Cortical Neurons: To assess the cytocompatibility of the bioresorbable shanks, a resazurin-based assay, PrestoBlue Cell Viability assay (Thermo Fisher Scientific), was used as an indicator of cell viability through their ability to reduce the dye reagent content.The cells were kept in culture for 10 days before performing the cytotoxicity test to ensure the maturation of the neurons.Prior to the conduction of the test, the PVA shanks were sterilized under UV exposure for 1 h.Afterward, the shanks were added to the wells.Cells cultured in wells without the addition of any shanks were considered as a control.The emission fluorescence was measured at 24 h, 48 h, and 7 days in culture (D7).The increase in the emission fluorescence acts as a viability assay.At each time point, the cell culture medium in the wells was replaced with the fresh medium containing PrestoBlue reagent (10% v/v in cell medium), and cells were incubated for a further 2 h minimum.The fluorescence was measured at 590-615 nm by using the microplate reader (Tecan Infinite M1000); experiments were performed in triplicate.Results are concluded in Figure 5b.
Statistical Analysis: Data within the text were expressed as mean ± standard error (SE).Statistical analyses were performed using OriginPro (OriginLab Corporation, Northampton, MA, USA) software assessing the statistical significance using analysis of variance (ANOVA) with pair comparisons according to the Tukey test.
In Vivo Evaluations-Animal surgery: Animal implantation was carried out on a set of 2 C57BL/6JRj male mice of 9 weeks old, following the kainate mesial temporal lobe epilepsy model (mTLE) as described in ref. [40].The experiment respects the European Union directive in use (Dir 2010/63/UE) and is approved by the ethics committee on animal experimentation of Rennes University and received agreement from the French national legal entities (agreement APAFIS #35019-2022012716305337).During the surgery, animals were under anesthesia and analgesia in a stereotaxic frame.The probe was inserted above the DG region in the right hippocampus (AP = −2.0mm, ML = −1.5 mm, DV = −2 mm).Electrode placement is determined according to the atlas of the mouse brain ("Paxinos and Franklin" the Mouse Brain in Stereotaxic Coordinates, Compact -5th Edition, n.d.).Once the planar electrode had been positioned, a drop of surgical glue was applied to fix the electrode to the skull, then the PVA shank was wetted with physiological saline to dissolve it and restore the flexibility of the pa-C.A gold electrode was inserted in the skull above the cerebellum as a reference.The connector was fixed to the mouse's skull via dental acrylic cement.Signal recordings: Fast ripples (FR) with the frequency band of 250-600 Hz were recorded for 2 h with an EEG monitoring system (Deltamed TM) at a sampling frequency of 2048 Hz.Signal processing: FRs are manually classified considering time and frequency criteria.To be labeled as a "True Fast-Ripple," an event of interest must: i) include at least four clear oscillations in the FRs band (200-600 Hz); ii) have the amplitude of an oscillation at least twice the amplitude of the background; iii) evoke a well-defined spot on the spectrogram, that is not a harmonic of lower frequency oscillations like ripples (120-200 Hz).These benchmarks are critical since FRs are easily mistaken for noise and artifacts due to the presence of many sharp events.In addition, the bandpass filtering of FRs is misleading since it can lead to misinterpretation of "false-Ripples." [41]Thus, the spectral decomposition is done using a convolution between the signal and Gabor wavelets (Reproduced from ref. [42]).
Eight Gabor functions were defined to decompose the signal on a filter bank defined by the following frequency bands:  [0.5-3.After the visual pre-selection and classification, the time frames identified as true FR contain not only the high-frequency event itself but also activity before and after.The segmentation of the FRs is ensured by the algorithm detailed in ref. [43].The result was the time index of onset and offset of the FR.The comparison between the two kinds of electrodes was then performed by comparing the energy of the same recorded FR on the FR and background bands, defined by Equations ( 1) and (2).
Where X FR and X Background refer to the same signal on the FR and Background bands, respectively.

Figure 1 .
Figure 1.Microfabrication process of PVA shanks with a physically modulated crystallinity: A) Schematics of the PVA physically cross-linking by freezingthawing: PVA solution at room temperature (1); Frozen PVA in liquid nitrogen bath at room temperature and the creation of ice domains separating the PVA chains from water (2); c-PVA formation during the thawing process (3).B) DSC curves of PVA versus c-PVA scanned in the temperature range between 150 and 250 °C revealing the T m of crystalline domains.C) Schematic illustration of the shank microfabrication process on a PVA substrate.D) SEM image of the etching profile of c-PVA substrate, defined by the yellow box after the micropatterning process.E) Etching rate evaluation of PVA material after 0 to 4 F-T cycles.The mean values are presented over three samples.F) Top-and cross-section views of the final micropatterned c-PVA shank.The thickness of the shank is 80 μm.

Figure 2 .
Figure 2. p-PVA versus c-PVA degradation profile: A) Degradation time of p-PVA at different thicknesses from 25 to 120 μm (left) and c-PVA at 80+ μm thickness at 1 to 4 F-T cycles.Mean values are presented from five samples.B) p-PVA and C) c-PVA degradation (four F-T cycles) profile in the agarose gel (1%) mimicking the brain tissue consistency.The thickness of both types of shanks is 80 μm.

Figure 3 .
Figure 3. Mechanical characteristics of p-PVA vs. c-PVA: A) Visual representation of increased stiffness and planarity by F-T cycles of PVA substrate and cross-linked PVA.Compression test of B) PVA, showing its complete folding by applying 2 N force and C) c-PVA, showing its breakage by applying 15 N force without bending.D) Young's modulus measurement of PVA and c-PVA after different F-T cycles.Mean values are presented over three samples.E) Experimental set-up of the compression test and the insertion force measurement where the PVA shanks with 2, 3, and 4 F-T cycles penetrate a 50 μm thick PDMS membrane, mimicking rat dura matter.Samples with high F-T cycle numbers require lower force.Standard deviations are plotted over three samples.

Figure 4 .
Figure 4. Implantability of flexible MEA using PVA bioresorbable shanks: A) PVA shank assembled with flexible probes from top-view (top), 30°angle (middle) and cross-section (bottom) views.Flexible probe insertion in agarose gel (1%) by B) PVA shank and C) c-PVA with 4 F-T cycles for a duration of 30 min.

Figure 5 .
Figure 5.In vitro and in vivo evaluations of PVA-assisted neural probes: A) Electrochemical impedance spectroscopy of flexible electrodes combined with p-PVA and c-PVA shank at 1 kHz in PBS-formulated agarose gel.Stars indicate the initial electrode's impedance without the PVA presence.B) PrestoBlue cell viability assay was performed on cortical neurons, cultured with bioresorbable shanks; PVA at 24 h, 48 h, and 7 days in culture.The differences are statistically significant for 48 h and D7 comparisons of the PVA shank with the control (***p < 0.001).C) The boxplot of signal energy of fast ripples (250-600 Hz) from two pairs of electrodes implanted in the hippocampus (blue, yellow) and cortex (green, red) of epileptic mice.Whiskers extend to lower extreme.Outliers are plotted as separate single data points in dark dots.In the circle is a picture of craniotomy with implanted PaC flexible MEA.