Scalable batch fabrication of ultrathin flexible neural probes using a bioresorbable silk layer

Rigid neural implants suffer from large mechanical mismatch with brain tissue, leading to tissue damage, scarring, and degradation of recording quality. Flexible probes improve compliance but are difficult to insert because ultrathin devices buckle against meningeal barriers. Existing strategies use removable rigid shuttles or temporary bioresorbable coatings; however, many reported approaches rely on manual assembly, limiting downscaling, throughput, and may increase surgical footprint and tissue trauma. This study addresses the question of how to batch-fabricate ultrathin, flexible neural probes that can be reliably implanted without buckling while retaining flexibility afterward. The authors propose a silk-parylene bilayer in which a biodegradable silk fibroin layer temporarily stiffens an ultrathin parylene probe, with programmable degradation to maintain stiffness long enough for placement and then resorb to restore flexibility.

Prior work established that silicon and metal microelectrodes (Utah, Michigan, microwire arrays) provide high-resolution recordings but suffer from chronic instability due to stiffness mismatch with soft brain tissue. Flexible substrates (parylene, polyimide, PDMS, hydrogels) or thinner stiff materials improve mechanical compliance but face insertion challenges. Two main insertion strategies emerged: external stiff shuttles removed after implantation, and bioresorbable coatings (e.g., PEG, PLA, chitosan, silk fibroin) that dissolve over minutes to days, reducing trauma compared to shuttle withdrawal. Nonetheless, many biodegradable-coating implementations require manual handling, limiting device miniaturization (<10 µm), parallel fabrication, and risking larger surgical footprints that can exacerbate inflammation. Silk fibroin is highlighted in the literature for biocompatibility, tunable biodegradability via β-sheet content, and favorable mechanics, but its integration into a scalable microfabrication process for ultrathin neural probes had not been reported.

• Device concept and materials: A bilayer probe comprising a degradable silk fibroin stiffening layer beneath an ultrathin parylene-C microelectrode array. Parylene is selected for biocompatibility and barrier properties; silk for temporary stiffness (E ≈ 3 GPa), biocompatibility, and programmable biodegradation.
• Batch microfabrication workflow (three stages):
  1. Substrate preparation: A glass wafer is coated with a sacrificial cellulose acetate (CA) layer. A cast silk fibroin layer is deposited atop CA. Film thickness is tuned by silk solution concentration and cast volume; e.g., 0.1 mL/cm² of 7 wt% silk yields ~30 ± 5 µm thickness.
  2. Parylene probe build-up: On the silk layer, a bottom parylene-C layer (3 µm) is deposited, followed by patterning of Au/Ti interconnects and 40 µm-diameter recording sites (Au thickness 200 nm). A top parylene-C encapsulation layer (1 µm) completes the 4 µm-thick parylene stack.
  3. Shaping and release: Reactive ion etching (RIE) using O2/CF4 plasma (optimal 75/25) defines the probe geometry through the silk-parylene bilayer. The CA sacrificial layer is dissolved in acetone to release devices. Probes are bonded to flexible flat cables (FFC) for testing.
• Silk degradation programming: Silk films are immersed in methanol to increase β-sheet crystallinity. In vitro degradation is tested in proteolytic medium (protease XIV, 1 U/mL in PBS). Methanol treatment durations (3, 6, 12, 24 h) modulate degradation times from hours (3–6 h treatments) to up to a week (12–24 h treatments).
• Mechanical insertion tests: Probes are inserted into 1 wt% agarose brain phantom (E ≈ 40 kPa). Force-displacement profiles are recorded, documenting the minimum penetration force and detecting buckling. Multiple insertion-extraction cycles are performed.
• In vivo feasibility: In an anesthetized mouse, dura is resected; the silk-stiffened probe is advanced through pia into somatosensory cortex and left in place for ~1 h to allow silk dissolution; insertion behavior and electrode integrity are observed.
• Compression and durability tests: Axial compression against a hard substrate determines buckling force (Fbuckling). A theoretical model (Euler buckling; clamped–pinned, K=0.7; rectangular cross-section; Eparylene = Esilk = 3 GPa) estimates Fbuckling as a function of length and thickness for comparison with experiments. Cyclic bending (radius 0.3 mm) over 1000 cycles monitors changes in buckling force and checks for delamination.
• Electrochemical/electrical characterization: Impedances are reduced via PEDOT:PSS coatings on Au electrodes; assessment ensures no degradation from fabrication or insertion (details referenced but not fully included in the excerpt).

• Scalable batch fabrication: Up to 80 ultrathin devices per 4-inch glass substrate with precise microfabricated geometries; each probe features four 40 µm-diameter Au recording sites on a 3 mm-long, 250 µm-wide, 4 µm-thick parylene shank atop ~30 µm silk.
• Programmable silk degradation: Methanol treatment tunes β-sheet content and in vitro proteolytic degradation: 3–6 h treatments yield degradation within hours; 12–24 h treatments extend lifetime up to ~1 week in PXIV (1 U/mL) in PBS.
• Insertion performance: In 1 wt% agarose brain phantom (E ≈ 40 kPa), minimum penetration force is ~0.7 mN with no buckling; probes withstand repeated insertion/extraction without damage. Bare 4 µm parylene shanks curl and are unhandleable without the silk stiffener, underscoring the necessity of the silk layer.
• In vivo feasibility: With dura resected, probes penetrate pia and enter mouse somatosensory cortex without buckling; after ~1 h, silk dissolves; electrodes show no visible damage upon retraction.
• Mechanical robustness: Average buckling force against a hard substrate is 10.9 ± 1.3 mN (N=20), ~15× the gel penetration force. Theoretical Euler buckling predicts 7.2 ± 2.3 mN, matching order of magnitude and trends. Under 1000 bending cycles (0.3 mm radius), buckling force decreases only ~10%, with no delamination or peeling observed.
• Electrical interface: Au electrodes coated with PEDOT:PSS for low impedance and high SNR; no adverse effects of the silk stiffener or RIE shaping on electrode integrity reported in the presented tests.

The silk-parylene bilayer strategy directly addresses the key challenge of inserting ultrathin flexible probes without buckling while restoring flexibility post-implantation. By integrating silk fibroin within a standard microfabrication flow, the approach eliminates manual assembly steps that previously limited scaling and miniaturization, enabling high-throughput production of sub-10 µm-thick parylene devices. Methanol-tuned β-sheet crystallinity confers programmable stiffness duration, allowing sufficient time for probe positioning and then dissolution to minimize chronic mechanical mismatch. Insertion tests in brain phantoms and feasibility in vivo (pia penetration) confirm that the temporary stiffener provides adequate rigidity, while compression and fatigue tests demonstrate robustness far exceeding the forces required for tissue entry. Collectively, these results suggest improved implantation with reduced acute trauma and potential for enhanced long-term tissue integration compared with shuttle-based methods or bulkier coatings.

The study introduces a scalable, batch-compatible fabrication method for ultrathin flexible neural probes leveraging a degradable silk fibroin layer as a temporary stiffener for a 4 µm parylene microelectrode array. The process enables precise geometries, robust handling, reliable insertion without buckling, programmable degradation times, and mechanical durability, with successful ex vivo recordings and in vivo penetration through pia. This platform opens avenues for minimally invasive neural interfaces in research and clinical applications. Future work should validate chronic in vivo performance and biocompatibility over extended periods, quantify in vivo degradation kinetics across brain regions and species, optimize coating thickness/treatment for different targets (including dura penetration), scale electrode counts and shank geometries, and further integrate stimulation/sensing modalities while maintaining minimal footprint.

• In vivo data are limited to short-term feasibility with dura resected; chronic recording stability and tissue response over weeks to months are not reported.
• Brain phantom tests do not replicate full meningeal structures or vascularization; insertion forces and behavior may differ across species and targets, especially with intact dura.
• Silk degradation is characterized in vitro (protease XIV in PBS) and by approximate in vivo dissolution time; precise in vivo degradation kinetics and byproducts were not quantified.
• Electrical/electrochemical performance details (impedance spectra, noise levels, long-term stability) are not fully presented in the excerpt.
• Demonstrated devices have four electrodes; scalability to higher channel counts and denser arrays, while suggested, is not experimentally detailed here.