Implantable neural devices, fabricated using microfabrication techniques, offer potential for treating neurological disorders. Significant advancements have been made in device design, including the Michigan probe and subsequent iterations featuring SOI shank shaping, increased channel counts, and CMOS-based microelectrodes. Despite successful long-term animal studies, clinical approval for multisite silicon-based microelectrodes remains elusive, unlike single-site devices like the Utah Electrode Array. A key barrier is demonstrating long-term reliability and biostability. While top surfaces are typically protected by layers of biocompatible materials like SiO2, SiN, or polymers, the sidewalls of silicon shanks are usually left unprotected, despite the fact that silicon corrodes in saline environments. Previous attempts to protect sidewalls, such as Parylene C coating, have limitations due to complex serial processes. This study aimed to explore batch-process compatible, low-temperature deposition methods, specifically ALD and ICPCVD of SiO2, to protect sidewalls while maintaining compatibility with standard microfabrication processes and materials.

The literature highlights the challenges in achieving long-term reliability and biostability of implantable silicon-based neural devices. While various protective coatings have been explored, the issue of unprotected sidewalls has received less attention. The corrosion of silicon in physiological environments, even at a low rate, can contribute to device failure over time. Studies have shown the effect of silicon doping on corrosion rates. Existing methods for sidewall protection, such as Parylene C coating, are complex and require serial processing, making them unsuitable for high-throughput fabrication. The need for a batch-process compatible, low-temperature method that does not interfere with standard microfabrication is apparent.

The study utilized test devices with a similar shape and material composition to previously validated neural microelectrodes. Four groups of devices were fabricated: a control group with no sidewall protection and three groups coated with SiO2 using ALD, ICPCVD, and a combination of both. A rapid KOH etch test was used to evaluate the quality of sidewall coverage; KOH etches silicon rapidly but not SiO2. Focused ion beam (FIB) cross-sectioning, scanning electron microscopy (SEM), and 3D extrapolation of etched silicon volumes were used to characterize and quantify the effectiveness of each protection strategy. ALD SiO2 deposition involved approximately 50nm deposition using trismethylaminosilane and ozone as precursors at 130°C. ICPCVD SiO2 deposition resulted in approximately 1323 nm thick films. The KOH etch test involved immersing the devices in 45% KOH at 60°C for 0, 5, and 30 minutes. SEM imaging and 3D modeling were used to quantify etched silicon volumes. FIB-SEM provided high-resolution imaging for analyzing the failure modes of the coatings.

Qualitative SEM observations showed that bare silicon sidewalls suffered the most dissolution during the KOH test, while ALD SiO2 provided the best protection. ICPCVD showed moderate protection initially, but significant etching occurred after 30 minutes. The combination of ALD and ICPCVD showed improved protection compared to ICPCVD alone, but still performed worse than ALD alone. Quantitative analysis of etched silicon volumes using 3D extrapolation from SEM images confirmed these observations. Control devices had an average etched volume of 952,381 µm³, while ALD-protected devices had significantly less etching (33,682 µm³). ICPCVD and ALD/ICPCVD groups exhibited intermediate levels of etching (477,898 µm³ and 322,052 µm³, respectively). One-way ANOVA showed significant differences among the groups (p<0.01). High-resolution SEM analysis revealed that the ALD/ICPCVD film exhibited a scalloped morphology, leading to strip-like detachment of the SiO2 film from the silicon surface, starting near the top edge. FIB-SEM analysis revealed that ALD provided conformal coverage of the sidewalls, while ICPCVD had gaps in the coating near the top, explaining its inferior performance compared to ALD. The residual stress in the ICPCVD film was measured as 32.88 ± 1.58 MPa (compressive).

The findings address the crucial need for improved biostability of implantable silicon-based neural probes. The study demonstrates that ALD SiO2 offers a superior method for protecting the exposed silicon sidewalls compared to ICPCVD or a combination of the two methods. The rapid KOH etch test provides a useful tool for screening and evaluating the quality of sidewall protection, offering a faster alternative to standard accelerated aging tests. The superior performance of ALD is attributed to its highly conformal deposition properties resulting in pinhole-free coverage, particularly at challenging areas like the top edges. The inferior performance of ALD/ICPCVD compared to ALD alone may be attributed to delamination of the coating due to combined compressive stress or stronger adhesion of ALD to ICPCVD than to silicon. Future work could investigate alternative deposition methods or film combinations to further enhance sidewall protection. The study’s findings are relevant to a wide range of silicon-based implantable devices.

This study systematically investigated sidewall protection strategies for implantable silicon-based neural probes. ALD SiO2 demonstrated significantly superior protection against silicon dissolution compared to ICPCVD or a combined approach. The rapid KOH etch test proved effective for evaluating coating quality. These findings highlight ALD as a promising technique for improving the long-term reliability and biocompatibility of implantable silicon devices. Future research should focus on in vivo testing to validate the long-term efficacy of ALD-based sidewall protection and explore other materials and deposition techniques.

The study employed a KOH etch test as an accelerated failure model. While this test is useful for comparing the relative performance of different coatings, it does not perfectly replicate the in vivo environment. The 3D extrapolation of etched silicon volume was based on the assumption of consistent etching along the shank sidewalls. More sophisticated 3D modeling techniques could be used in future studies for more accurate volume quantification. The in vivo validation of the ALD protection technique is yet to be conducted.