A lithographically-patterned, elastic multi-electrode array for surface stimulation of the spinal cord

Abstract

A new, scalable process for microfabrication of a silicone-based, elastic multi-electrode array (MEA) is presented. The device is constructed by spinning poly(dimethylsiloxane) (PDMS) silicone elastomer onto a glass slide, depositing and patterning gold to construct wires and electrodes, spinning on a second PDMS layer, and then micropatterning the second PDMS layer to expose electrode contacts. The micropatterning of PDMS involves a custom reactive ion etch (RIE) process that preserves the underlying gold thin film. Once completed, the device can be removed from the glass slide for conformal interfacing with neural tissue. Prototype MEAs feature electrodes smaller than those known to be reported on silicone substrate (60 microm diameter exposed electrode area) and were capable of selectively stimulating the surface of the in vitro isolated spinal cord of the juvenile rat. Stretchable serpentine traces were also incorporated into the functional PDMS-based MEA, and their implementation and testing is described.

Figure 1

Fabrication Steps for a PDMS-Substrate Multi-Electrode Array (MEA): (a) Deposition of gold anti-adhesion layer onto glass slide. (b) Spin coating of PDMS. (c) Gold conductor layer deposition and positive photoresist defining conductor pattern. (d) Etching of gold and photoresist removal. (e) Spin coating of insulating PDMS. (f) Aluminum deposition for RIE mask and photoresist defining the mask pattern. (g) Etching of Aluminum and reactive ion etching to remove PDMS. (h) Stripping of aluminum. (i) Removal from glass slide.

Figure 2

Scanning electron microscope image of etched electrode contacts. Three PDMS-substrate MEA electrodes are shown following the reactive ion etch step. The upper PDMS layer has been etched to expose the underlying gold electrode contacts (circular areas above) without incurring any damage to the gold. The lighter area indicates the prior presence of an aluminum mask, which was used to prevent etching of the PDMS insulation over electrode wires and was subsequently stripped before imaging.

Figure 3

Fabricated PDMS-substrate MEA. A final product of our fabrication process is shown wrapped around a wire of similar diameter (2mm) to that of the neonatal intact or hemisected juvenile in vitro rat spinal cord.

Figure 4

Experimental apparatus for electrical evaluation of MEA. To evaluate the electrical properties of the MEA, we measured impedances using a spectrum analyzer at frequencies from 100Hz to 100KHz. A drop of Hanks’ Balanced Salt Solution was used to interface between the MEA and the spectrum analyzer lead (0.015” silver wire). (a) Configurations for measuring impedance of an electrode contact (a) and the insulating PDMS film (b) are shown.

Figure 5

Results for electrical impedance testing of 19 MEA electrodes on 4 arrays. The MEA electrodes consistently exhibited impedance values comparable to those of rigid MEAs (Heuschkel et al., 2002; Oka et al., 1999) (a) and the PDMS-covered traces demonstrated impedance values indicative of insulation (b).

Figure 6

Uniaxial strain device for electro-mechanical evaluation of the PDMS-based MEA. The custom-made apparatus shown was used to clamp the MEAs and apply uniaxial strain in 1% increments.

Figure 7

Serpentine Trace Design for MEA Electrode Traces. (a) An intersecting, serpentine electrode trace pattern confers a greater elasticity to the PDMS-based MEA (electrode trace conductivity failure at > 8% strain for one tested trace, recoverable upon relaxation) than does a (b) non-serpentine pattern of electrode traces (failure at > 3% strain, no recovery, one tested trace). The parallel, serpentine electrode trace pattern shown in (a) conferred no functional elasticity advantages when compared to the non-serpentine pattern (b).

Figure 8

Bending of polyimide and PDMS arrays. We visually compared the ability of PDMS and polyimide array substrates to conform to a bending tube of the same approximate diameter as our in vitro spinal cords. (A) Polyimide film wrapped around a 2mm-diameter tube shows buckling of the polyimide film along the tube. (B) PDMS film wrapped around a 2mm tube conforms more uniformly to the bending tube.

Figure 9

Experimental setup for surface stimulation of the in vitro isolated rat spinal cord (hemisected). The MEA was wrapped around the isolated in vitro spinal cord of a postnatal day 11 rat and two adjacent MEA electrodes (bipolar configuration) were used to stimulate the 5^th thoracic segment's ventrolateral funiculus. The degree of stimulus spread was assessed by recording surface compound action potentials (CAPs) at multiple circumferential sites distant to the site of activation. For recording, a glass suction electrode (40−50 μm internal diameter) was placed 11.5 mm caudal to the stimulation site to record CAP responses in 50 μm lateral increments from the site of the peak response.

Figure 10

Spinal cord white matter response to MEA surface stimulation. MEA (vs. tungsten control) stimulus-evoked compound action potentials (CAPs) on white matter tracts are plotted over lateral recording distance from the site of peak CAP response, for one array wrapped around the thoracic region of one in vitro spinal cord. Recordings were made 11.5 mm caudal to the site of ventrolateral funiculus stimulation, in 50 μm lateral increments. Strength of CAP response was quantified by finding the area under the baseline-subtracted, rectified traces of the CAP recording. Shown are responses to MEA stimuli at threshold value (700 μA/500 μs single current pulse) as well as at 800 μA/500 μs, demonstrating an increase in white matter tract recruitment at greater stimulus values. To compare stimulus selectivity, also shown is the logarithmic trendline for threshold stimulation (300μA/100μs) using a conventional tungsten bipolar electrode.. The rate at which evoked CAP response strength decreased over recording distance compared well to the logarithmic trendline for tungsten bipolar stimulation (R² = 0.8201), indicating a similar stimulus selectivity between the two methods.