Neural interfaces create a connection between neural structures in the body and external electronic devices. Brain-machine interfaces and bioelectric medicine therapies rely on the seamless integration of neural interfaces with the brain, nerves, or spinal cord. However, conventional neural interfaces cannot meet the demands of high channel count, signal fidelity, and signal longevity that these applications require.



In this thesis we characterized the damage resulting from conventional Utah arrays after multiple years of implantation in the cortex of a non-human primate. The neuron density around the electrode shanks was compared to the neuron density of nearby healthy tissue, finding a 73% loss in density around the electrodes. The explanted arrays were imaged and characterized for forms of electrode surface inconsistency. Coating cracks, tip breakage, and parylene cracks were the most common inconsistency. A significantly higher number of tip breakage and coating crack occurrences were found on the edges of the arrays as compared to the middle. In this work, we made clear the need for a minimally damaging alternative to the Utah electrode array.



Neural interfaces composed of carbon fiber electrodes, with a diameter of 6.8 microns, could enable a more seamless integration with the body. Previous work resulted in an array of individuated carbon fiber electrodes that could record reliably high signal-to-noise ratio neural signals from the brain for several months. However, the carbon fiber arrays were limited by only 30% of the electrodes recording neural signals, despite inducing very minimal inflammation. Additionally, it was relatively unknown if carbon fibers would make suitable long-term peripheral neural interfaces. Here, we illustrate the potential of carbon fiber electrodes to meet the needs of a variety of neural applications.



First, we optimized state-of-the-art carbon fiber electrodes to reliably record single unit electrophysiology from the brain. By analyzing the previous manufacturing process, the cause of the low recording yield of the carbon fiber arrays was identified as the consistency of the electrode tip. A novel laser cutting technique was developed to produce a consistent carbon fiber tip geometry, resulting in a near tripling of recording yield of high amplitude chronic neural signals. The longevity of the carbon fiber arrays was also addressed. The conventional polymer coating was compared against platinum iridium coating and an oxygen plasma treatment, both of which outperformed the polymer coating. In this work, we customized carbon fiber electrodes for reliable, long-term neural recording.



Secondly, we translated the carbon fiber technology from the brain to the periphery in an architecture appropriate for chronic implantation. The insertion of carbon fibers into the stiffer structures in the periphery is enabled by sharpening the carbon fibers. The sharpening process combines a butane flame to sharpen the fibers with a water bath to protect the base of the array. Sharpened carbon fibers recorded electrophysiology from the rat vagus nerve and feline dorsal root ganglia, both structures being important targets for bioelectric medicine therapies. The durability of carbon fibers was also displayed when partially embedded carbon fibers in medical-grade silicone withstood thousands of repeated bends without fracture. This work showed that carbon fibers have the electrical and structural properties necessary for chronic application.



Overall, this work highlights the vast potential of carbon fiber electrodes. Through this thesis, future brain-machine interfaces and bioelectric medicine therapies may utilize sub-cellular electrodes such as carbon fibers in medical applications.



Date: Friday, May 7, 2021

Time: 10:00 AM

Zoom: https://umich.zoom.us/j/95839545566 (Zoom link requires prior registration)

Chair: Dr. Cynthia Chestek

For Assistance or Questions
um-bme@umich.edu