ECS Journal of Solid State Science and Technology, 3 (2) N15-N21 (2014) N152162-8769/2014/3(2)/N15/7/$31.00 © The Electrochemical Society

3D Capacitive Sensor Array for Detection of Neural ResponsesBrittany Branch,a,b Jennifer L. Schei,c Kateryna Artyushkova,a,∗ Gautam Gupta,bAndrew M. Dattelbaum,b John S. George,c,z and Dimiter N. Petseva,z

aNanoscience and Microsystems Engineering and The Department of Chemical and Nuclear Engineering, Universityof New Mexico, Albuquerque, New Mexico 87131, USAbCenter for Integrated Nanotechnologies, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USAcPhysics Division, Los Alamos National Laboratory, Los Alamos, New Mexico 8754, USA

Neural degenerative diseases and traumatic injuries affect millions of people worldwide, motivating the development of neuralprosthetic interfaces to restore sensory or motor function in affected individuals. Advances in neural sensing and stimulationinterface technology will allow a more comprehensive understanding of neural function while leading to the development of hybridbiological-electronic sensor devices for robust, functioning neural prosthetic systems. Current techniques of neural activity sensingand stimulation employ multi-electrode arrays (MEAs) that typically incorporate metal electrodes and measure or apply currentsvia an electrochemical junction, leading to corrosion and charge transfer across the electrode-tissue interface. High-density neuralinterface technology will require active circuitry within the implant; the device must withstand corrosion and induce minimal damageat the electrode/tissue interface. The work shown here demonstrates a prototype neural interface device based on capacitive couplingthrough hafnium oxide encapsulation of a novel 3D device architecture, advancing neural sensing technology toward long-termimplantable neural interfaces.© 2013 The Electrochemical Society. [DOI: 10.1149/2.016402jss] All rights reserved.

Manuscript submitted September 30, 2013; revised manuscript received November 22, 2013. Published December 10, 2013.

Neural degenerative diseases and traumatic neurological injuriesaffect millions of people worldwide, motivating the development ofneural prosthetic interfaces to restore sensory or motor function inaffected individuals. Most existing systems are based on a relativelysmall number of relatively simple electrodes, but many applications re-quire higher density arrays, able to selectively record from or stimulateindividual neurons. Such systems will ultimately require incorporationof active circuitry into packages that can withstand chronic implanta-tion. Advances in neural sensing and stimulation interface technologywill lead to the development of hybrid biological-electronic sensordevices for robust, functioning neural prosthetic systems, and alsoprovide a more comprehensive understanding of neural function.

As a specific example, degenerative disease such as retini-tis pigmentosa or macular degeneration affect millions of peopleworldwide.1,2 The leading cause of vision loss is age-related mac-ular degeneration, which affects the photoreceptors of the retina,3

while to some extent sparing the neuronal circuitry, which processes,encodes and transmits visual information to the brain. Recently theFDA has approved a prosthetic retinal implant that can restore lim-ited vision in such patients. However, the underlying technology ofpresent implants will support at most a few hundred electrodes. Forhigh-resolution vision, thousands or even hundreds of thousands ofelectrodes are probably desirable. Further, our increasing ability tocontrol the spatio-temporal patterns of activity of extended popula-tions of neurons, underscores the limitations of our present knowledgeof retinal encoding.

In the past several decades, neural activity imaging and sensingtechniques have been developed to record the activity of individualneurons and small networks.4–10 An emerging goal is to simultane-ously measure the activity of all neurons in the retina or a limitedregion of brain tissue. Advancements in sensing technology will al-low a more comprehensive understanding of retinal processing andencoding leading to the development of hybrid biological-electronicsensor systems for robust, functioning retinal prosthetics.11–15 We an-ticipate that such knowledge will also serve as the foundation formore sophisticated electronic systems that will ultimately emulate theperceptual function of biological vision.

Both basic science and biomedical applications require electro-neural interface devices to withstand corrosion and induce mini-mal damage at the electrode/tissue interface. Current techniques ofneural activity sensing employ multielectrode arrays (MEAs) thatare typically made of metal electrodes and utilize direct current

∗Electrochemical Society Active Member.zE-mail: dimiter@unm.edu; jsg@lanl.gov

measurements.16–19 Encapsulation of MEAs in a protective coatingcan eliminate corrosion and minimize electrochemical charge transferacross the electrode-tissue interface, which induce circuitry failureand neural damage through irreversible Faradic reactions.20–23

To overcome the direct current limitations in existing MEA tech-nology, capacitive charge transport between the electrodes and neuraltissue can be employed. Capacitive sensing could alleviate most ofthe drawbacks of the Faradic current stimulation, including corrosionand evolution of chemically reactive species. Recent work has ad-vanced the technology of capacitive stimulation of neurons25 and thismethod will likely prove important for designing future electroneuralprosthetic systems. Capacitive coupling has been used for noninvasivemeasurements in a neuronal culture environment on a semiconductorchip, but the development of this technology has been limited,24 andpresent systems do not support implantation and chronic recording.Advances in the design and fabrication of systems for high-density ca-pacitive coupling to neural tissue will advance prosthetic applicationsas well as our knowledge of systems neuroscience.

We have developed a three-dimensional micro pillar sensor arraythat utilizes capacitive coupling in order to sense electrical activity inexcised retinal tissue. The device incorporates over 3600 capacitiveprobes organized into 60 independent sensor sites (for compatibilitywith existing electronics) spread over an area of 750 μm2. Each sensorsite consists of an 8 × 8 array of Pt micropillars in order to maximizethe capacitive response, interconnected by leads on the device. Thedevice is fully insulated by atomic layer deposition of hafnium oxide.Electrochemical impedance spectroscopy was used to study the oxidedeposition on the 3D micro pillar sensor array to ensure a pinhole-freedielectric coating. The characteristic impedance magnitudes increaseup to 3 orders of magnitude upon oxide deposition and phase measuresindicate fully capacitive sensor sites.

The capacitive sensor was interfaced with retinal tissue and uponlight stimulation electro-retinograms (ERG) were recorded. Thesefabrication processes and electrochemical impedance measurementsdemonstrate the usefulness of such techniques for building high-density 3D arrays that can be fully encapsulated with a protectivedielectric coating. The work shown here advances the technology to-ward capacitive sensing of neurons with a robust, non-invasive sensingdevice, providing a base for studying neural encoding and process-ing by networks of neurons and for the future development of neuralprosthetics.

Materials and Methods

Fabrication of 3D micro pillar array.— To fabricate the 3D MEAand integrate signal traces and bond pads on a single device a

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 129.24.4.212Downloaded on 2014-01-24 to IP

N16 ECS Journal of Solid State Science and Technology, 3 (2) N15-N21 (2014)

three-mask lithography process using a silicon wafer substrate wasemployed. Electron beam lithography (JEOL JBX-6300ES) was usedto expose an array of dots in ZEP520A photoresist (Zeon Chemicals,Louisville, KY). After E-beam exposure, the pattern was developedand a metal mask was produced by E-beam evaporation of nickelusing the developed pattern as a template. Finally, a lift-off processresulted in arrays of nickel dots 3 μm in diameter with a pitch of4.5 μm. The nickel dots acted as a mask during a deep reactive-ionetching (DRIE, Plasma-Therm SLR Series ICP Bosch), resulting inthree-dimensional (3D) pillar arrays. The nickel mask was removedand the pillar sidewalls were smoothed by dry oxygen annealing anda hydrofluoric acid etch.

Low-pressure chemical vapor deposition (LPCVD, Trion Technol-ogy, Orion III) was used to coat the pillar arrays with a 300 nm layerof SiO2 to electrically insulate the silicon platform. A conformal layerof platinum (Pt) was deposited onto the insulated 3D platform byphysical vapor deposition (PVD) sputtering (Kurt J. Lesker PVD 75).AZ4330 photoresist (AZ Electronic Materials USA Corp.) was usedto mask the Pt coated 3D pillar arrays and the electrical leads wereintegrated by wet etching the conductive layer to insulate each pillararray. Finally a negative resist (AZ nLOF 2070) was used to maskbond pads at the end of the Pt leads. The bond pads were made usingE-beam evaporation of a 20 nm titanium and 100 nm gold layer usingthe developed pattern as a template. A lift-off process resulted in bondpads terminating all Pt traces from each array of pillars. The Pt traceswere insulated with a photoresist (AZ4330) layer, which exposed onlythe center of the 3D micro pillar array for further experimentation.

The overall view of the 3D micro pillar array is shown in Figure 1.Each Pt electrode (60 total) consists of an 8 × 8 array (Inset A) ofthree-dimensional pillars 8 μm high with a 3 μm diameter evenlyspaced with a 4.5 μm pitch (Inset B). Platinum traces patterned fromeach Pt electrode terminate at Ti/Au bond pads on the outside edge ofthe device are shown in Inset C.

Hafnium oxide deposition.— Atomic layer hafnium oxide coat-ings were deposited onto the fabricated Pt pillar electrodes using aPicosun SUNALE R-150B ALD system (Detroit, MI). Pretreatmentof the electrode surface with oxygen plasma in the ALD chamberwas done prior to HfO2 deposition. Tetrakis-dimethylamido hafnium(IV) (TDMAH, SAFC Hitech (USA)) was used as the hafnium metalprecursor with deionized water as the oxygen source for the ALDdeposition of a HfO2 film. TDMAH preheated to 75◦C was delivered

Figure 1. SEM images of the 60-electrode device with three-dimensionalpillar electrodes (Inset A & B) connected by Pt metal traces terminating atbond pads on the outside edge of the device (Inset C).

to the reaction chamber at 100 standard cubic centimeters per minute(sccm) with a pulse time of 1.9 seconds. The water precursor preheatedto 25◦C followed at 200 sccm for 0.1 seconds with a purge of nitrogenat 150 sccm for 6.0 s between each pulse of precursor. The chambertemperature was maintained at 200◦C during the completion of theALD cycles, where one cycle consisted of TDMAH and water pulseswith nitrogen purges in between. The coverage versus the number ofcycles was characterized by X-ray photoelectron spectroscopy (XPS),scanning electron microscopy (SEM) and impedance spectroscopy.

Impedance spectroscopy characterization.— The electrical prop-erties of the HfO2 capacitive layer were characterized by electrochem-ical impedance spectroscopy (EIS) using a CH Instruments (Austin,TX) impedance analyzer. EIS was performed by applying an A.C.potential of 5 mV with a frequency range of 0.1 Hz to 10 kHz. TheHfO2 coated electrodes were individually connected as the workingelectrode and tested in amphibian Ringer solution (107 mM NaCl,2.5 mM KCl, 1.6 mM MgCl2, 1 mM CaCl2, 22.0 mM NaHCO3,10 mM D-glucose) serving as the electrolyte at room temperature,consistent with the experimental conditions of the electrophysiologymeasurements (described below). An Ag/AgCl electrode was used asthe reference electrode and a Pt wire as the counter electrode.

Electrophysiology device testing.— To evaluate the response of the3D micro pillar array in capacitive sensing of the retinal response tolight stimulation, electrophysiology measurements were performed.Leopard frogs (Rana pipiens) were obtained from Kons Scientific(Germantown, WI) and housed in a terrarium under a 12 hour dark/12 hour light cycle. Prior to an electrophysiology experiment, thefrogs were dark-adapted for at least 30 minutes before being rapidlydecapitated, double pithed, and the eyes removed in accordance withstandards approved by the Los Alamos National Laboratory Insti-tutional Animal Care and Use Committee. The retina was carefullyremoved in a bath of amphibian Ringer under dim red light illumina-tion. During device testing the retina was placed ganglion side downon the 3D micro pillar array and a ring of Lucite covered in nylonmesh held the retina in contact with the device. Initially, an 8 secondprestimulus recorded baseline neural activity before a 4 second stim-ulus pulse consisting of a white LED light source shining through anopaque material with a 50 μm spot size. All recordings were madeinside a Faraday cage at room temperature. Data were amplified (105

Ohms input impedence, PZ2, Tucker-Davis Technologies, Alachua,FL), filtered 2–3200 Hz, and digitized at 24.4 kHz using a multi-channel neurophysiology workstation (RZ2, Tucker-Davis Technolo-gies, Alachua, FL).

Results and Discussion

Impedance is a method to assess the charge flow between twosurfaces as a function of excitation frequency. The characterizationof the impedance and phase angle through equivalent circuit (EC)modeling can provide structural information about the material orinterface of the system. It is important to consider the behavior ofboth the Bode magnitude plot and the Bode phase plot when fitting anEC model to accurately describe a physical interface. EIS was usedto study the electrical properties of hafnium oxide growth over 3Dmetal-coated structures. A similar pillar array described in Materialsand Methods, completely coated in Pt, was used for investigation ofatomic layer deposition of hafnium oxide on Pt. Figure 2 shows theBode plot of a 3D Pt electrode before and after HfO2 deposition.A sequential increase in impedance is observed at frequencies lowerthan 100 Hz, with increasing number of HfO2 cycles (Figure 2A).This impedance behavior is characteristic of ALD growth. The layer-by-layer deposition method exposes the growth surface to two half-reactions keeping the precursors separate during the coating processresulting in atomic scale deposition. Sequential reaction cycles resultin larger oxide coverage leading to a more resistive interface. Thisgrowth mechanism is further shown by the phase behavior beingdominated by a capacitive element with increasing oxide coverage, as

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 129.24.4.212Downloaded on 2014-01-24 to IP

ECS Journal of Solid State Science and Technology, 3 (2) N15-N21 (2014) N17

Figure 2. Interface impedance (A) and phase shift (B) vs. frequency of a 3DPt electrode coated with 100–150 cycles of HfO2.

indicated by the shift in phase toward 90 degrees for the frequencyrange between 0.1 and 100 Hz (Figure 2B).

To parameterize and quantify the electrochemical behavior ob-tained by EIS, we used a modified R (CR) circuit. The equivalentcircuit (EC) used to model the impedance data for each deposition isshown in Figure 3. The 3D Pt electrode, shown in Figure 3A, con-sists of the resistance of both the solution and the electrical leads(Rs), which remained constant for each measurement, in series withan interfacial constant phase element (CPE) and polarization resis-tance (RPt) circuit in parallel. This EIS building block is known as theDebye response involving a single time constant accounting for thedelay in polarization at the interface. The Debye response accountsfor a metal interface in water where the separation of charge at thesolid-liquid interface creates a space charge capacitance. The spacecharge capacitance of non-homogenous surfaces with variable currentdensity at the interface is expressed by the Helmholtz model and canbe accounted for by the CPE circuit element. This element accuratelydescribes the electrode surface due to the nature of sputtering Pt ontohigh aspect ratio features. From the circuit analysis (performed usingZ-view software) of the deposited metal film, the values of Rs, RPt

and CPEPt that best fit the data were 281.4 �, 9.128 × 10 + 4 �, and1.349 × 10-5 F, respectively. There is a significant deviation from thepure capacitive behavior of the electrode due to the high surface areamorphology (n value 0.8488). The phase behavior of the electrode,depicted by the right axis of Figure 3A, clearly indicates that the cir-cuit is dominated by a resistive element as the phase is predominantly0 degrees at lower frequencies. The phase behavior along with theoverall impedance (Rs + RPt) indicates the 3D metal surface is highlyconductive.

With sequential deposition of hafnium oxide we would expect theimpedance of the resulting surface to increase. Figure 3B shows thatupon deposition of 100 HfO2 cycles the Pt surface is coated by an insu-lating layer as indicated by the higher impedance. The phase behaviorindicates the Debye response remains and an additional capacitiveresponse is present before the phase shift toward 0 degrees as seen

Figure 3. Interface impedance (left axis) and phase shift (right axis) vs. frequency of a 3D Pt electrode (A) coated with ALD of 100 (B) 125 (C) and 150 (D)cycles of HfO2, along with the respective equivalent circuit models.

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 129.24.4.212Downloaded on 2014-01-24 to IP

N18 ECS Journal of Solid State Science and Technology, 3 (2) N15-N21 (2014)

previously for a conductive surface. This complex phase behavior isa result of the metal surface coated with an oxide layer containing de-fects. The EC model accounting for defects in the oxide layer is shownin the inset of Figure 3B. From the circuit analysis of the depositedoxide film, the values of ROxide, CPEOxide, RDefect, and CPEDefect thatbest fit the data were 5.975 × 10 + 4 �, 1.549 × 10-6 F, 4.146 ×10 + 5 �, and 9.024E-6 F, respectively. The true capacitance can becalculated using the relationship:

C = C P E1n · R

1−Nn [1]

Therefore the capacitance of the oxide is 4.841 × 10-6 F/cm2 com-pared to 7.581 × 10-5 F/cm2 for the electric double layer within thedefect. These values reflect what is expected of an oxide film (0.1–20 μF/cm2) and electric double layer (80–200 μF/cm2) on a roughsurface, validating the EC model proposed here.26

From the EIS analysis of 100 HfO2 cycles it can be concluded thatnucleation and growth of the dielectric layer occurred over the metalmicrostructures, but increased cycles are required to produce a defect-free film encapsulating the 3D surface. Increasing the deposition to125 cycles it can be seen in Figure 3C the phase begins to shift(∼80◦) toward capacitive like behavior, with a single time constantsignifying the relaxation time of the dielectric interface. The EISbehavior is evidence for the 3D Pt electrode fully encapsulated inHfO2. The EC model includes a circuit consisting of an interfacialconstant phase element (CPE) and oxide resistance (Roxide) in parallel(Figure 3C inset). From the circuit analysis of the deposited oxide

film, the ROxide, CPEOxide and the n value was used to calculate thecapacitance of the oxide film (3.844 × 10-6 F/cm2). Although thefilm is uniform and capacitive in nature, the ALD deposition wasincreased to 150 cycles to demonstrate an ideal capacitive layer (90◦

phase). Figure 3D shows a slight increase in impedance as a functionof frequency and a perfect capacitive phase response. The overallcapacitance of the oxide layer was found to be 1.265 × 10-6 F/cm2,decreasing slightly due to the thicker oxide produced at higher cycles.

To further validate the proposed EC models described above, scan-ning electron microscopy (SEM) was used to characterize the HfO2

films at varying degrees of deposition time (100–150 cycles). Figure 4shows SEM micrographs of the bare Pt surface (Figure 4A) and aftersubsequent deposition of hafnium oxide (Figure 4B–4D). After 100cycles of HfO2 ALD, it can be seen that defects occur on the surfaceas illustrated in Figure 4B. Increasing the deposition time by 25 or50 cycles, the defects are eliminated and oxide uniformly covers thesurface as seen for both cases Figure 4C and 4D.

XPS analysis was performed to confirm hafnium oxide growth andquantify the thickness of the resulting films. XPS spectrum shows anincrease in Hf to Pt ratio as the number of cycles increases. For 100cycles, 0.76 at% of Pt is detected, while for 125 cycles this amountdrops to 0.02 at% and no Pt is detected for samples with HfO2 coatedusing 150 cycles. Although it is inherent during these measurementsto propagate into the distal Pt layer, the larger detection of Pt after 100cycles of HfO2 can be attributed to a combination of this propagationand the presence of defects within the layer, this amount drops after125 cyles of HfO2 due to the lack of defects on the surface and

Figure 4. SEM images of the Pt electrode surface (A) coated by 100 cycles of HfO2 (B) resulting in an oxide coating with defects, 125 cycles of HfO2(C) completely covering the Pt surface and 150 cycles (D) resulting in a thicker layer of oxide.

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 129.24.4.212Downloaded on 2014-01-24 to IP

ECS Journal of Solid State Science and Technology, 3 (2) N15-N21 (2014) N19

Figure 5. Bode plot and equivalent circuit modelof a oxide film varying in thickness as a result of105 ALD cycles of HfO2 and the correspondingSEM micrograph (inset) showing the thicker andthinner regions of the film (red arrows).

increasing thickness of the oxide film preventing propagation of thebeam into the underlying Pt layer. This trend is in good correspondencewith SEM data shown previously. Thickness of HfO2 calculated fromangle-resolved X-ray photoelectron spectra was between 5.5–5.9 nmfor 100 cycles and 10.6–10.9 nm for 125 cycles. These results alongwith the EIS analysis provide evidence that the growth mechanismfor atomic layer deposited HfO2 on Pt microstrucutres occurs throughinitial nucleation sites that with subsequent deposition cycles growtogether to form a continuous film. As evident from the XPS results,the film thickness rapidly increases upon film formation, but the initialnucleation and spread of the film takes significant time before thisoccurs. For 150 cycles, the thickness of hafnium oxide becomes largerthan sampling depth of underlying Pt; therefore no calculations couldbe made.

For the application of capacitive sensing, it was imperative that notonly the device was uniformly coated with oxide, but also the oxidelayer was minimized to increase the sensitivity of the device. Afterobserving the film properties for 100 cycles it was concluded thatincreasing the number of cycles slightly will allow enough reactiontime at the surface to fill the defects and ensure complete cover-age, while minimizing the thickness of the film as shown by XPS.Figure 5 shows the Bode plot of a 3D Pt electrode coated with 105ALD cycles of HfO2. It can be seen that two time constants existin the phase behavior, representing two interfaces within the system.The shift in phase toward two capacitive elements indicates the ab-sence of the Debye response and rather two dielectric interfaces onthe solid surface. This type of phase behavior was accounted for in theEC model consisting of two circuits in series with a constant phaseelements in parallel with a resistor (Figure 5). The model representsa HfO2 film with thicker (8.832 × 10-6 F/cm2) and thinner (9.257× 10-5 F/cm2) regions resulting in capacitive values calculated fromthe respective ROxide1, CPEOxide1, ROxide2, CPEOxide2, and n values asshown previously. The EC model was confirmed with characteriza-tion of the oxide film using scanning electron microscopy. Figure 5inset shows the oxide film with varying thickness as indicated by thearrows.

Having established the upper and lower limits of deposition cyclesthat yield complete coverage of high aspect ratio Pt pillar structures,we deposited HfO2 onto a working device (illustrated in Materialsand Methods Figure 1) consisting of 60 individual Pt electrodes withan 8 × 8 array of three-dimensional pillars 8 μm high with a 3 μmdiameter evenly spaced with a 4.5 μm pitch. The device was limitedto 60 individual electrodes to maintain active readout using a system

already in place, but the integration of over 3,800 pillars on a singleplatform proved the fabrication method developed here would allowbuilding high-density MEAs with thousands of active electrodes. Inorder to maximize the capacitance, each sensor site was designed as

Figure 6. Interface impedance (A) and phase shift (B) vs. frequency of a 3DPt pillar electrode measured from the working device before (black diamonds)and after HfO2 deposition (blue circles), along with the respective equivalentcircuit models. The error bars represent the double standard deviation of theaverage value from three independent capacitive electrode measurements foreach data point shown.

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 129.24.4.212Downloaded on 2014-01-24 to IP

N20 ECS Journal of Solid State Science and Technology, 3 (2) N15-N21 (2014)

Figure 7. Responses (left axis) of isolated frog retina to light stimulation from 6 individual capacitive sensors with their respective location within the array(A) and a single capacitive sensor response (B). Stimulation (right axis) starts at 8 seconds and remains on for 4 seconds. Note that the retina characteristicallyresponds to both the onset and offset of light stimulus.

a cluster of pillars and upon oxide deposition; the slightest dielectricfilm was deposited, while ensuring a defectless film. Figure 6 showsthe Bode plot and the respective EC models of a single 3D Pt pillarelectrode before and after 140 HfO2 growth cycles.

As seen previously, there was up to three orders of magnitudeincrease of the impedance upon oxide deposition and the phase indi-cates a capacitive element. The two time constants remain, with theabsence of the Debye response seen previously for an oxide film withdefects, thus confirming an oxide layer with thinner and thicker re-gions, maximizing the capacitance of the device. The EC model againincludes two circuits in series consisting of an interfacial constantphase element (CPE) and oxide resistance (Roxide) in parallel (Figure6). At larger frequencies it can be seen that the experimental datadiverges from the model, resulting in higher than expected impedanceand phase behavior. This type of response is usually an indicationof increased cell resistance and may be accounted by the workingdistance between the reference and working electrodes.

From the circuit analysis of the deposited oxide film the averagecapacitance of the 3D pillar device was calculated to be 4.369 × 10-3F/cm2. The capacitance is significantly larger per unit area comparedto bulk studies shown previously due to the micron size electrodes im-plemented in the prototype capacitive device. The EIS data presentedthus far establishes the method of fabrication developed here resultsin a 3D microstructured device that behaves as a capacitor.

Having achieved the desired electrical properties of the device,we employed it to measure responses of the frog retina to visiblelight stimulation. The light-induced electrical activity of the entireretina is known as the electro-retinogram (ERG). The ERG consistsof specific components (a-, b-, and c-waves) that are located in specificretinal layers. Here we illustrate the detection of these componentsthrough capacitive coupling of the 3D micro pillar electrode arrayand excised retinal tissue. Figure 7A shows simultaneously recordedelectro-retinograms (ERGs) from 6 of the sensing sites with theirrespective location within the array. A single ERG measurement isshown in Figure 7B illustrating the characteristic waves of an ERGresponse.27 No response was recorded in the absence of retina or theabsence of light.

Fourier analysis of the capacitive response recorded previously,results in the frequency distribution for the ERG response. Analysisplaces both the onset and offset retinal responses in the frequencyrange of ∼3 Hz. The physiological response frequency is well in therange of the capacitive regime of the sensor developed here. The resultsshown here of sensing retinal activity through capacitive coupling ofa novel device platform through encapsulation with HfO2 is the firstreport integrating 3D electrodes and capacitive coupling. This workmay lead to high density, minimally invasive, chronic recordings ofneuronal populations.

Conclusions

We have developed a novel three-dimensional micro pillar sen-sor array that utilizes capacitive coupling for the detection of neuralactivity in the retina. The cluster configuration shown here demon-strates the utility of a capacitive pillar device, but optimization andfine tuning to increase resolution can be done through cluster size andspacing, in which inter-pillar cross-talk must be considered at decreas-ing length scales. A systematic study of atomic layer deposited HfO2

on 3D Pt microstructures was conducted with EIS, SEM, and XPS.The analysis of this data characterized the capacitive properties andconfirmed the uniformity as a function of ALD deposition cycles ofthe oxide film. Integration of a highly dense 3D micropillar array thatis fully encapsulated with HfO2 is the first step toward hermeticallysealed devices that incorporate metal electrodes and active electronicsonto a single platform. When interfaced with retina, resolution of thecharacteristic waves of the ERG was achieved. This illustrates the po-tential use of this capacitive technology for interrogation of specificcell types or relatively small collections of neurons and improvingthe selectivity of neural detection, while facilitating mapping of thehigh spatial detail in visual images. We are currently working towardthis goal with the measurement of single neuron action potentials.The incorporation of such a high k dielectric material onto metal mi-crostructures alleviates most of the drawbacks of Faradic current basedmeasurements, advancing neural sensing technology toward less-invasive, higher resolution detection of activity in large populations ofneurons.

Acknowledgments

The LANL LDRD program and the NSF CAREER (CBET0844645) provided funding for this research. This work was per-formed, in part, at the Center for Integrated Nanotechnologies a DOE-BES funded user facility and in the Physics Division at Los AlamosNational Laboratory. The authors thank Dr. S.A. Dayeh for his supportduring the microfabrication effort.

References

1. A. H. Milam, Z. Y. Li, and R. N. Fariss, Prog. Retin. Eye Res., 17, 175 (1998).2. D. T. Hartong, E. L. Berson, and T. P. Dryja, Lancet, 368, 1795 (2006).3. J. Ambati, B. K. Ambati, S. H. Yoo, S. Ianchulev, and A. P. Adamis, Surv. Ophthal-

mol., 48, 257 (2003).4. J. E. Dowling and B. B. Boycott, Proceedings of the Royal Society Series B-Biological

Sciences, 166, 80 (1966).5. H. Kolb, Philos. Trans. R. Soc. Lond. Ser. B-Biol. Sci., 258, 261 (1970).6. T. Yanagida, M. Koshimizu, K. Kawasaki, and D. Yonemura, Documenta Ophthal-

mologica, 67, 355 (1988).

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 129.24.4.212Downloaded on 2014-01-24 to IP

ECS Journal of Solid State Science and Technology, 3 (2) N15-N21 (2014) N21

7. M. S. Humayun, E. de Juan, J. D. Weiland, G. Dagnelie, S. Katona, R. Greenberg,and S. Suzuki, Vision Res., 39, 2569 (1999).

8. F. S. Werblin, J. Physiol.-London, 589, 3691 (2011).9. R. H. Masland, Nat. Neurosci., 4, 877 (2001).

10. P. K. Campbell, K. E. Jones, R. J. Huber, K. W. Horch, and R. A. Normann, IeeeTransactions on Biomedical Engineering, 38, 758 (1991).

11. J. O. Winter, S. F. Cogan, and J. F. Rizzo, J. Biomater. Sci.-Polym. Ed., 18, 1031(2007).

12. M. S. Humayun, J. D. Weiland, G. Y. Fujii, R. Greenberg, R. Williamson, J. Little,B. Mech, V. Cimmarusti, G. Van Boemel, G. Dagnelie, and E. de Juan, Vision Res.,43, 2573 (2003).

13. J. D. Weiland, W. T. Liu, and M. S. Humayun, in Annual Review of BiomedicalEngineering, p. 361, Annual Reviews, Palo Alto (2005).

14. J. Viventi, D. H. Kim, L. Vigeland, E. S. Frechette, J. A. Blanco, Y. S. Kim,A. E. Avrin, V. R. Tiruvadi, S. W. Hwang, A. C. Vanleer, D. F. Wulsin, K. Davis,C. E. Gelber, L. Palmer, J. Van der Spiegel, J. Wu, J. L. Xiao, Y. G. Huang,D. Contreras, J. A. Rogers, and B. Litt, Nat. Neurosci., 14, 1599 (2011).

15. N. G. Hatsopoulos and J. P. Donoghue, Annual Review of Neuroscience, 32, 249(2009).

16. S. F. Cogan, in Annual Review of Biomedical Engineering, p. 275, Annual Reviews,Palo Alto (2008).

17. A. Stett, U. Egert, E. Guenther, F. Hofmann, T. Meyer, W. Nisch, and H. Haemmerle,Anal. Bioanal. Chem., 377, 486 (2003).

18. A. F. M. Johnstone, G. W. Gross, D. G. Weiss, O. H. U. Schroeder, A. Gramowski,and T. J. Shafer, Neurotoxicology, 31, 331 (2010).

19. G. Buzsaki, Nature Neuroscience, 7, 446 (2004).20. S. B. Brummer and M. J. Turner, IEEE Trans. Biomed. Eng., 24, 59 (1977).21. G. Herlem, C. Goux, B. Fahys, F. Dominati, A. M. Goncalves, C. Mathieu, E. Sutter,

A. Trokourey, and J. F. Penneau, Journal of Electroanalytical Chemistry, 435, 259(1997).

22. F. Fassbender, G. Schmitt, M. J. Schoning, H. Luth, G. Buss, and J. W. Schultze,Sensors and Actuators B-Chemical, 68, 128 (2000).

23. Y. Temiz, A. Ferretti, Y. Leblebici, and C. Guiducci, Lab on a Chip, 12, 4920 (2012).24. P. Fromherz and A. Stett, Phys. Rev. Lett., 75, 1670 (1995).25. J. T. Robinson, M. Jorgolli, A. K. Shalek, M. H. Yoon, R. S. Gertner, and H. Park,

Nat. Nanotechnol., 7, 180 (2012).26. Encyclopedia of Electrochemistry, in, S. Bard and S. Frankel Editor, p. 227 (2002).27. K. T. Brown, Vision Res., 8, 633 (1968).

  ecsdl.org/site/terms_use address. Redistribution subject to ECS license or copyright; see 129.24.4.212Downloaded on 2014-01-24 to IP