Carnegie Mellon University
Browse
Forssell_cmu_0041E_10349.pdf (96.15 MB)

Long-term Insulation of Active Electronics Embedded in Compliant Neural Probes

Download (96.15 MB)
thesis
posted on 2019-01-29, 00:00 authored by Mats ForssellMats Forssell
Compliant neural probes suffer from poor long-term reliability due to the limited insulation capabilitiesof thin polymer films. Reliability is particularly critical for clinical applications wherelifetimes on the order of decades will be required. In addition, large scale recording and stimulationprobes to be used clinically require probes augmented with active circuits, which haveeven stronger requirements for insulation. Parylene has been considered a promising material forneural probes because of its biocompatibility and high barrier capabilities. However, neural probes
insulated with parylene are prone to failure over timescales of days to months. The multilayer insulation consisting of parylene and ALD deposited alumina/titania nanolaminate proposed in this thesis can provide the necessary long term insulation performance and enable active circuit
embedding. This thesis explores the long-term insulation capability of parylene thin films with embedded
chips for the purpose of neural recording and stimulation. The different failure modes of parylene films in phosphate buffered saline (PBS) are investigated through simulation, experiments and modeling. Test structures representative of neural probe insulation are fabricated and their
impedance measured over six months. Experiments designed to mimic the thermal (body temperature
of 37 C) and electrical conditions experienced by probes in recording (small signal voltage applied) and stimulation (1 Vpp applied) cases allow to assess the performance in more realistic conditions. In order to improve on the limited lifetime of parylene probes, additional thin film barrier
layers are evaluated: atomic layer deposited ceramics surrounding the metal and enclosed in the
parylene can limit the rate of degradation of the insulation, without affecting the biocompatibility or mechanical properties of the probes. However because the adhesion of parylene to many materials is poor, the overall insulation might be degraded due to delamination between the
layers. Therefore additional adhesion layers also have to be used to enable the full benefits of
the multilayer insulation. These adhesion layers can be additional ceramic layers, or conformally
deposited polymers. In practice, increasing the electrode count in compliant probes is limited by the wiring area
required to connect all the electrodes to the external circuitry. In order to allow very large electrode
counts (>100 electrodes per probe), multiplexing circuitry is therefore required in the probe. For polymer probes, this requires embedding chips in the probe. A CMOS chip was designed that would allow multiplexing of 8 electrode channels. For recording electrodes, in order to limit the
additional noise generated by multiplexing, amplifiers are present at each electrode. A simple technique for integrating thin chips, explored in this thesis, uses anisotropic conductive film adhesives (ACF) to ensure mechanical and electrical connection to thin silicon chips
that are be flip-chip bonded to the probe before the final parylene layer is deposited. A prototype
CMOS chip allows multiplexing of 8 electrode channels and cascading of multiplexers. For recording electrodes, in order to limit the additional noise generated by multiplexing, amplifiers are present at each electrode. In addition to higher electrode counts, the addition of integrated
circuitry opens the door for many possibilities for neural recordings, including better signal to noise
ratio (amplification), higher throughput (spike sorting/digitization), and wireless capabilities.

History

Date

2019-01-29

Degree Type

  • Dissertation

Department

  • Electrical and Computer Engineering

Degree Name

  • Doctor of Philosophy (PhD)

Advisor(s)

Gary Fedder

Usage metrics

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC