Developing Chronically Reliable, High-Resolution, Flexible Neural Interfaces for Human and Animal Research

Thumbnail Image



Journal Title

Journal ISSN

Volume Title

Repository Usage Stats



The human brain has roughly 86 billion neurons. Deciphering how the brain functions remains difficult because of our limited tools to interface with this massive number of neurons. Many important insights have been discovered by researchers using electrodes that are capable of interfacing with only a few dozen neurons at a time. Clinical diagnoses and surgical treatment plans are currently made with only a few dozen electrode contacts on the brain surface. However, recent research with limited high-resolution sampling has revealed hidden rich neural dynamics that were missed by the low-resolution clinical electrodes. All current neural interface technologies are restricted to record from either a small population of neurons in a small area, or they record greatly averaged and coarsely sampled signals from a large area. To be able to record at high resolution from large areas of the brain requires more electrodes and more wires. Routing more wires out of the head is challenging, due to the space constraints inside the skull. One way to solve this challenge in scalability is to integrate powered electronics directly at each electrode site. However, this incorporating actively-powered electronics in a thin and flexible form factor that must survive for many years implanted in the brain presents a challenge that has never before been solved. To effectively capture micro-scale neural activity simultaneously across large brain regions presents three major technological challenges: accessibility, reliability, and scalability. This dissertation describes my efforts to solve these three challenges. Chapters 2 and 4, describe utilizing commercial flexible printed circuit (FPC) technology to achieve large scale manufacturing of high-resolution neural interfaces. Rather than using conventional polyimide for these neural interfaces, I investigated using liquid crystal polymer (LCP) as the substrate to improve the reliability of the devices. The LCP electrodes were tested in rats for over a year (Chapter 2). I also scaled up the designs and performed pilot intra-operative research studies in humans (Chapter 4). To increase the coverage of these LCP electrodes, I developed a silicone molding technique, which was tested in non-human primates (NHP) for over a year (Chapter 3). This work further demonstrated the utility of high-resolution electrodes in decoding studies. We were able to demonstrate improved decoding using optimal electrode selection. This result will enable future studies to improve decoding accuracy without significantly increasing power consumption by requiring all the channels to be wirelessly transmitted. Next, I solved the challenge of reliably integrating actively-powered electronics into a flexible neural interface, leading to a next-generation neural interface, the Neural Matrix. This technology demonstrated stable recording performance over a year of chronic implantation in rats (Chapter 5). The use of thermally-grown silicon dioxide (SiO2) and capacitive sensing provided key innovations to extend the reliability of the encapsulation for the active array. Finally, the incorporation of hermetic vertical interconnect access (VIA) structures made from highly doped silicon enabled direct faradic sensing and the possibility for electrical stimulation, without compromising device longevity (Chapter 6). These set of results from in vitro, acute, intra-operative, and chronic settings provide a robust demonstration of the novel neural interface technologies I have developed. The methodology and technology developed in this dissertation will enable broad improvements in clinical care and neuroscience research.





Chiang, Chia-Han (2021). Developing Chronically Reliable, High-Resolution, Flexible Neural Interfaces for Human and Animal Research. Dissertation, Duke University. Retrieved from


Dukes student scholarship is made available to the public using a Creative Commons Attribution / Non-commercial / No derivative (CC-BY-NC-ND) license.