Corrosion of tungsten microelectrodes used in neural recording applications

download Corrosion of tungsten microelectrodes used in neural recording applications

of 14

description

Corrosion of tungsten microelectrodes used in neural recording applications

Transcript of Corrosion of tungsten microelectrodes used in neural recording applications

  • Journal of Neuroscience Methods 198 (2011) 158171

    Contents lists available at ScienceDirect

    Journal of Neuroscience Methods

    journa l homepage: www.e lsev ier .com

    Corrosi ne

    Erin Patr azua Department ob Department oc Department o

    a r t i c l

    Article history:Received 17 SeReceived in reAccepted 15 M

    Keywords:Tungsten corroTungsten micrNeural recordi

    -termor ged ford to tr the30mfoundtungstima

    from the electrochemical impedance spectroscopy data. The results show that tungsten microwires inan electrolyte of PBS have a corrosion rate of 300700m/yr. The corrosion rate for tungsten microwiresin an electrolyte containing PBS and 30mM H2O2 is accelerated to 10,00020,000m/yr. The corrosionrate was found to be controlled by the concentration of the reacting species in the cathodic reaction(e.g. O2 and H2O2). The in vivo corrosion rate, averaged over the duration of implantation, was estimated

    1. Introdu

    Electrocthe suitabiltal, and neucause failurmetal ionsimplants foand is stillliterature ismetal oxideTurner, 197informationfor neural r

    Tungstencal microelwell as mamicrowireset al., 1999;

    CorresponE-mail add

    0165-0270/$ doi:10.1016/j.to be 100m/yr. The reduced in vivo corrosion rate as compared to the bench-top rate is attributed todecreased rate of oxygendiffusion causedby the presence of a biological lmand a reduced concentrationof available oxygen in the brain.

    2011 Elsevier B.V. All rights reserved.

    ction

    hemical reactionsat themetal/tissue interfaceconstrainity of biological implant materials for orthopedic, den-roprosthetic applications. Corrosion, for example, cane of an implanted device, and the associated release ofmay itself provoke a toxic reaction. Corrosion of metalr orthopedic applications has been extensively studiedan ongoing issue (Jacobs et al., 1998). A large body ofavailable on the electrochemical analysis of metals ands used for neural stimulation purposes (Brummer and7; Norlin et al., 2005, 2002; Merrill et al., 2005), but lessis available on the corrosion properties of metals used

    ecording applications.is commonly used for recording sites on intracorti-

    ectrode arrays. Commercial microelectrode arrays, asny noncommercial arrays, are made with tungsten(Williams et al., 1999; Carmena et al., 2003; ChapinPatrick et al., 2008). The strength, rigidity, and record-

    ding author. Tel.: +1 352 392 6207.ress: [email protected] (M.E. Orazem).

    ing properties of tungsten electrodes formed from 50m diametermicrowires make them desirable for intracortical applications. Themicrowires can be inserted into neural tissue without buckling andare easilymanipulated into arrays. However, tungsten is not imper-vious to corrosion. Corrosion of tungsten via-plugs has been foundto be a failure mechanism for microelectronic integrated circuits(Bothra et al., 1999). Tungsten coils, due to their thrombogenic-ity, had been used clinically to occlude unwanted vasculature andhave since been taken off themarket due to degradation of the coils(Peuster et al., 2003). Sanchez et al. (2006) found apparent modi-cation of tungsten microwires by corrosion after four weeks ofin vivo implantation. While corrosion-related effects are apparentfor tungsten microwires used in neural recording applications, thenature of the electrochemical reactions is not known. For example,the formation of an oxide or the direct dissolution of tungsten areboth possible outcomes of the corrosion process. Development ofan understanding of the electrochemical reactions provides insightinto the stability of tungsten microelectrodes for long-term in vivorecordings.

    In addition to causing a potential for local toxicity due to diffu-sion of corrosion products into the cortex, corrosion of the tungstenelectrodes may impede successful long-term recording. It has beenshown that amicroelectrodemust be placed less than 200mfrom

    see front matter 2011 Elsevier B.V. All rights reserved.jneumeth.2011.03.012on of tungsten microelectrodes used in

    icka, Mark E. Orazemb,, Justin C. Sanchezc, Toshikf Electrical and Computer Engineering, University of Florida, Gainesville, FL 32611, USAf Chemical Engineering, University of Florida, Gainesville, FL 32611, USAf Biomedical Engineering, University of Miami, Coral Gables, FL 33146, USA

    e i n f o

    ptember 2010vised form 24 February 2011arch 2011

    sionowire arrayng electrode

    a b s t r a c t

    In neuroprosthetic applications, longactivity of neural populations used fof tungsten microwire electrodes usetrolled bench-top study and compareTwo electrolytes were investigated fosaline (PBS) and 0.9% PBS containingtions responsible for corrosion wereof Pourbaix diagrams. Dissolution ofmechanism. The corrosion rate was e/ locate / jneumeth

    ural recording applications

    Nishidaa

    electrode viability is necessary for robust recording of thenerating communication and control signals. The corrosionintracortical recording applications was analyzed in a con-

    he corrosion of tungsten microwires used in an in vivo study.bench-top electrochemical analysis: 0.9% phosphate bufferedM of hydrogen peroxide. The oxidation and reduction reac-by measurement of the open circuit potential and analysis

    sten to form the tungstic ion was found to be the corrosionted from the polarization resistance, which was extrapolated

  • E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171 159

    the cell body to record the action potential of a neuron (Henzeet al., 2000). Over time, dissolution of the tungsten increases thedistance between the recording surface and the neuron such thatthe action potential may become unmeasurable. Also, as the tung-sten microwformed. Thhinder theimpedanceinterface imthe measuroxide couldsimilar effe

    Althougdocumenteand Osseo-Aments (KelWu, 1971),corrosion inprehensivedissolutionaspotentialby tungstenet al. (2003tungsten cosolution, viawould dissoa corrosionbe used to e

    Thereforandquantiin intracortmeasuremedynamic inchemical reexperimenttrode and rThese expetop. Convernear an imptory responOxidizing spand the hyduring thethese, onlyMoreover, ttioned specthe inammin vivo cond

    The objemicroelectrical saline aelectrochempared to thewidely usedfor such apet al., 1997corrosion oenvironmen

    2. Experim

    This sectrochemicatechnique e(EIS). Electr

    tal technique in which sinusoidal modulation of an input signal isused to obtain the transfer function for an electrochemical system(Orazem and Tribollet, 2008). In its usual application, the modu-lated input is potential, the measured response is current, and the

    r funinedpectanceort pped

    rateanceandals to

    ateri

    roeleins

    tedremetrolyare

    ed. Aetatid sidea ndes. Theencas

    edstruss. Eire i

    2-3Md-facess oh silthearevivo

    latedire).weresulatfabr

    ny (3Howwithed forepoanceulatd a selecnd pprodby reandPanen pral im0mMas cmaire dissolves, a hollow tube made of the insulation isis tube may be subject to cellular build-up that woulddiffusion of ionic species and effectively increase theat the electrode/electrolyte interface. The increasedpedance would ultimately decrease the magnitude ofed action potential. Moreover, the formation of a thickalso increase the interface impedance and create a

    ct.h corrosion and anodic dissolution of tungsten are welld for both acidic (Lillard et al., 1998; Anik, 2009; Aniksare, 2002; Johnson and Wu, 1971) and basic environ-

    sey, 1977; Anik and Osseo-Asare, 2002; Johnson andthe rate and electrochemical mechanism of tungstenbiological solutions arenotwell documented. In a com-study, Anik and Osseo-Asare (2002) showed that theof tungsten depends on specic system conditions suchandpH.One studyused conditions similar to those seenmicrowires in neural recording applications. Peuster

    ) performed an in vitro assessment of the corrosion ofils in Ringers solution, a type of physiological salineweight lossmeasurements and concluded that one coillve in 6 years. Their analysis however, did not specifyrate in units of mass per area per time and thus cannotstimate corrosion in other systems.e, a study that determines the corrosion mechanismes the corrosion rate for tungstenmicroelectrodes usedical recording applications are needed. However, thent of a corrosion rate characteristic of an electrode invivo conditions and a controlled analysis of the electro-actions taking place cannot be performed by the same. The electrochemical analysis requires a tungsten elec-eference electrode that are not suited for implantation.riments are most effectively preformed on the bench-sely, the dynamic composition of the extracellular uidlanted electrode undergoing a foreign-body inamma-se is challenging to simulate in bench-top experiments.ecies such as hydrogenperoxide, the superoxide anion,

    droxyl radical are produced by the reactive microgliainammatory response (Burke and Lewis, 2002). Ofhydrogen peroxide is easily combined into a solution.he absence of documented concentrations of the men-ies and relative pH changes thatmay be associatedwithatory responsemake it difcult tomodel accurately theitions.ct of this work is to explore the stability of tungstenodes in a controlled bench-top study using physiolog-nd physiological saline with hydrogen peroxide. Theical behavior of tungsten microelectrodes are com-behavior of platinum microelectrodes, which are alsofor neural recording applications and considered inert

    plications (de Haro et al., 2002; Kipke, 2003; Maynard). The goal is to provide physical understanding of thef tungsten microelectrodes in saline-based, bench-topts that can help explain observations made in vivo.

    ental methods

    tion provides a description of the materials and elec-l instrumentation used. The principal electrochemicalmployedwas electrochemical impedance spectroscopyochemical impedance spectroscopy is an experimen-

    transfeis obtaterm simpedtransphow imrosionImpednique,materi

    2.1. M

    Micinnitefabricameasuto-elecsurfaceis needinterprexposeface arelectroin vivowiresperforming thethe glamicrowtek 30one enroughntor witused insurfaceused ingold-pFine Wtrodeswas inease of

    Mastudy.trodesnot usresultsimpedthe insshowe

    The(PBS) aone byduced(Burkemate.hydrogstructufrom 1H2O2 wused to7.4.ction is represented as an impedance. The impedanceat different modulation frequencies, thus invoking theroscopy. Through use of system-specic models, theresponse can be interpreted in terms of kinetic andarameters. Frateur et al. (1999), for example, showance spectroscopy can be used to determine the cor-for cast iron pipes used in France for water distribution.spectroscopy is a powerful and sensitive in situ tech-it was used here to explore the sensitivity of electrodecorrosion reactions.

    als

    ctrodes approximating disc electrodes with a semi-ulation layer and well-dened surface areas wereand used as the working electrode in the EISnts. For electrochemical characterization of the metal-te interface, an electrode surface with a well-deneda and a perfect seal between electrode and insulationrtifacts caused by imperfect insulation seals confoundon of electrochemical phenomena at the interface, andewalls affect the accuracy of the estimation of the sur-eeded for calculation of the corrosion rate. Thus, themade for EIS differ from the electrodes typically usedEIS electrodes were made with tungsten or platinumed in glass or epoxy. The glass insulation process wasby placing thewires in borosilicate glass tubes andheat-cture in a furnace at a temperature (800 C) that softenslectrodes insulated in epoxy were made by placing then a glass tube andlling the tubewith epoxy resin (Epo-). With glass or epoxy sealed tightly around the wire,e was polished with SiC sandpapers resulting in a nalf 1mand the other endwas secured to a gold connec-ver epoxy (Ablebond 84-1LMI). The polishing process,present work to ensure an accurate assessment of thea, is not performed on tungsten microwires typically. Working electrodes were made using 50m diametertungsten, tungsten, or platinum microwires (CaliforniaThe gold-plated tungsten and platinum working elec-insulated with glass, and the tungsten-only electrode

    ed in epoxy. Glass or epoxy insulation was chosen forication. Each has sufcient insulating characteristics.20) replicates of each electrode type were used in thisever, because of the sensitivity of the EIS results to elec-an imperfect insulation seal, many of the data were

    r the calculation of the corrosion rate. The impedancerted in the presentwork are those determined by a newtechnique (Patrick, 2010) to have a good seal betweenor and the electrode. All electrodes of the same typeimilar impedance trend.trolytes chosen were 0.9% phosphate buffered salinehosphate buffered saline with H2O2 added to simulateuct of an inammatory response. Since H2O2 is pro-activemicroglia that surround the implantedmicrowireLewis, 2002), local concentrations are difcult to esti-et al. (1996) and Fonesca and Barbosa (2001) usederoxide and PBS to assess corrosion of titanium forplantmaterials. They used concentrations that rangedto 100mM. A conservative concentration of 30mM

    hosen for this study. A salt mixture (Sigma P-5368) waske the PBS with composition given in Table 1 and pH of

  • 160 E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171

    Table 1Composition of phosphate buffered saline (PBS).

    Chemical compounds Concentration (M)

    NaCl 0.138HCl 0.0027Na2HPO4 0.01KH2PO4 0.00176H2O2a 0.03

    a Concentration of H2O2 for electrolyte including PBS and H2O2.

    2.2. Instrumentation

    Electrochemical impedance spectroscopy measurements wereperformed under potentiostatic control using a Gamry 300 Gseries potentiostat/glavanostat. Cyclic voltammetry was also per-formed with the Gamry system with voltage sweep rates of 50and 100mV/s. A silver/silver chloride reference electrode (Bio-Analytical Systems RE-5B) was used in this study. A large-areaPt counterelectrode was used for measurements in PBS elec-trolyte. A large-area titanium counterelectrode was used for theexperiments in electrolytes containing H2O2 because Pt wasshown to be reactive with hydrogen peroxide. The EIS perturba-tion voltage amplitude was 10mV. The frequency range for allexperiments was 0.1Hz to 20kHz, and all measurements weretaken at the open circuit potential (OCP). The low-frequencyvalue of 0.1Hz was chosen to limit the inuence of nonstation-ary behavior, and the high-frequency value of 20kHz was thelargest frequency found to have minimal inuence of instru-mental artifacts. The resulting frequency range was sufcient tocharacterize the reaction and transport processes associated withthe electrochemical reactions. The microelectrodes were polishedand thorouimmersionthe OCP totaken.

    3. Experimental results

    Impedance spectroscopy was used to identify potential electro-chemical reactions for themicroelectrodes in simulatedbodyuids.The conclusions drawn from the impedance resultswere supportedby optical photographs, thermodynamic analysis, and comparisonto in vivo results.

    3.1. Impedance response in PBS

    The impedance response obtained for platinum and tungstenmicroelectrodes immersed in PBS is presented in Fig. 1. The resultsobtained for the three electrode materials are not easily differen-tiated in plots of real impedance as a function of frequency, asshown in Fig. 1(a). In contrast, signicant differences are apparentin plots of the logarithm of the imaginary impedance as a func-tion of frequency, Fig. 1(b), and Nyquist plots, Fig. 1(c). While theplot of the real impedance as a function of frequency is presentedin Fig. 1 for completeness, similar plots are not presented for otherexperimental systems due to the lack of sensitivity to experimentalcondition.

    The impedance response for the platinum electrode is seen asa straight line in Fig. 1(b) and (c). This response is characteristicof a blocking or nonreactive electrode. The line passing throughplatinum impedance data presented as the logarithmic plot ofimaginary impedance as a function of frequency for the platinum,shown in Fig 1(b), has a slope equal to 0.9. This result is consis-tentwithconstant-phase-element (CPE)distributed-time-constantbehavior. All electrodes showed similar (CPE) behavior as indicatedby the slopes of the imaginary impedance as a function of frequencydiffering from negative one (Orazem et al., 2006) in Fig. 1(b). Thenite value of impedance observed in Fig. 1(c) at low frequencies

    tungsten microelectrodes suggests the presence of faradaicns.blocin se

    10

    lated W

    Fig. 1. Impedaa function of frghly rinsed in deionized water immediately prior toin the electrolyte. An elapsed period of 5min allowedstabilize before the impedance measurements were

    for thereactio

    Thetor, Re,

    101 101 103 105

    100

    102

    104

    f / Hz

    Z r /

    cm

    2

    WAuplated WPt

    100

    102

    104

    Z j

    /

    cm

    2

    4

    6

    8

    10

    Z j

    / k

    cm

    2

    WAupPt

    0.16 Hz

    0.25 Hz0 2 4 6 80

    2

    Zr / k cm2

    0.06 Hz

    nce response in PBS with electrode material as a parameter: (a) real part of the impedanequency; and (c) complex impedance plane (Nyquist) plot.king and reactive systems can be modeled with a resis-ries with a constant phase element, ZCPE, or the parallel

    1 101 103 105f / Hz

    WAuplated WPt

    slope=0.9

    ce as a function of frequency; (b) imaginary part of the impedance as

  • E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171 161

    Re

    Rp

    ZCPE

    ReZCPE

    Reactive SystemBlocking System

    Fig. 2. Equivalent circuits for blocking and reactive electrochemical systems (see,e.g., Orazem and Tribollet, 2008).

    combination of a resistor, Rp, and constant phase element, respec-tively, as shown in Fig. 2 (see, e.g., Orazem and Tribollet, 2008)where Re is the resistance of the electrolyte, Rp is the polarizationresistance, and ZCPE is the constant phase element impedance givenby

    Zcpe = 1(j)Q

    (1)

    where is the frequency, is a number from 0 to 1, and Q is theCPE coefcientwith units s/ cm2 (Brug et al., 1984). As discussedby Orazem et al. (2006), the high-frequency slopes of the lines pre-sented in Fig. 1(b) are equal to . When =1, the parameter Q isthe double-layer or interfacial capacitance.

    Impedance measurements taken after long periods of immer-sion in the eand give introde. Impein Fig. 3 wdays in theor nonreactgold-platedin Fig. 4witization resisthe impedalar to that scomparablelarger than

    Under oand anodicor more cattungsten elthe oxygen

    101 101 103 10510

    2

    101

    100

    101

    102

    103

    Z j

    /

    cm

    2

    f / Hz

    time (days)

    400

    600

    Z j

    /

    cm

    2 time (days)1.0 Hz0.5 Hz

    0.6 Hz

    a

    b

    mpedaith elof frequency; and (b) complex impedance plane (Nyquist) plot.

    of the faradaic reaction to the concentration of oxygen areted in Fig. 5. The concentration of oxygen was decreased byngN2 into the solution. After 1h of O2 displacement by nitro-bbling, the EIS results show that the polarization resistanceedwith decreasedO2 content. These results suggest that theion of oxygen was the rate-limiting cathodic reaction at thede surface.

    4

    6

    8

    10

    12

    14

    day 1day 20

    0.1 Hz

    Fig. 3. Impedaimpedance plalectrolyte illustrate the nonreactive nature of platinumsight into the reactive nature of the tungsten elec-dance spectra for platinum microelectrodes, presentedith elapsed time as a parameter show that, after 20PBS electrolyte, the platinum still exhibits blocking,ive, behavior. In contrast, the impedance results for atungsten microelectrode in PBS electrolyte presented

    h elapsed time as a parameter show a decrease in polar-tance over a period of 15 days. The inuence of time fornce response of the pure tungsten electrode was simi-hown for the gold-clad tungsten in Fig. 4, and, at eachtime, the impedance of the pure tungsten was alwaysthat of the gold-clad tungsten.pen circuit conditions, the net current is equal to zero,faradaic reactions, if present, must be balanced by onehodic reactions. The dominant cathodic reaction on theectrode was found by testing the system reactivity toconcentration. Impedance results showing the depen-

    Fig. 4. Iperiod wfunction

    dencepresenbubbligen buincreasreductelectro

    101

    102

    103

    104

    105

    Z j

    /

    cm

    2

    day 1day 20

    Z j

    / k

    cm

    2

    a b101 101 103 105

    100

    f /Hz

    00

    2

    nce response of a platinum electrode in PBS with elapsed time as a parameter: (a) imaginne (Nyquist) plot.0 200 400 600 800 1000 12000

    200

    Zr / cm2

    nce response of gold-plated tungsten electrodes in PBS over a 15 dayapsed time as a parameter: (a) imaginary part of the impedance as a2 4 6 8 10

    Zr / k cm2

    ary part of the impedance as a function of frequency; and (b) complex

  • 162 E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171

    103

    104

    Z /

    cm

    2

    aerated PBSPBS with reduced O2

    Z

    (

    cm

    2 )a

    b

    Fig. 5. Impedparameter: (a)complex impe

    3.2. Impeda

    The oxidfor charge tused to exption on botimpedancetrodes in anFig. 6. The tindicates thtungsten animpedanceand cannottive systemplatinum elmoredetailet al., 2009systems sho

    3.3. Photog

    Anoptictrode beforpresented iThe electrod

    Optical pgold-platedfrom one toacteristic ofshown in Fiened as com

    2

    Auplated WPt

    Z j

    /

    cm

    2

    a

    b

    mpedaterialcy; an102 100 102 104 106101

    100

    101

    102

    f / Hz

    j

    0 500 1000 1500 20000

    500

    1000

    1500

    Zr ( cm2)

    j

    aerated PBSPBS with reduced O2

    0.5 Hz

    1.6 Hz

    ance response of a tungsten electrode in PBS with O2 content as aimaginary part of the impedance as a function of frequency; and (b)dance plane (Nyquist) plot.

    Fig. 6. Itrode mfrequennce response in PBS with 30mM H2O2

    izing effect of H2O2 allows another cathodic pathwayransfer via the reduction of H2O2. EIS results were againerimentally ascertain the presence of a faradaic reac-h platinum and gold-plated tungsten electrodes. Theresponse for platinum and gold-plated tungsten elec-electrolyte containing PBS and H2O2 is presented in

    endency toward a nite impedance at low frequenciesat faradaic reactions occurred on both the gold-platedd platinum electrodes when H2O2 was present. Theresponse of the platinum electrode is more complexbe modeled by the equivalent circuit for a simple reac-illustrated in Fig. 2; however, the completemodel of theectrochemical reaction is not the focus of this work andon its impedance response canbe foundelsewhere (Yoo). The main result is that both platinum and tungstenw faradaic behavior at low frequencies.

    raphic evidence for tungsten corrosion

    al photograph of the polished surface of a tungsten elec-e immersion in PBS is presented in Fig. 7(a). The imagen Fig. 7(b) was taken after immersion in PBS for 23 days.e is covered by a porous layer and is visibly roughened.hotographs in Fig. 8 reveal the condition of six differenttungsten electrodes before and after immersion in PBSsix days. A photograph of an electrode surface char-the condition of all the electrodes before immersion is

    g. 8(a). After one day in saline, the surface looked rough-pared to theprevious state. For the electrode immersed

    for two daycircular secfrom the suout of focuother electrcircular depwith time.was depreslayer. The imthe entire twith the cesuggest thaeral ratherthe photogr

    Optical pelectrode imfor an elapsrecesses weface and thperimeter ocurrent distdensity is tprogressedtrode and athe order osuggest a msaline solutcases withoan oxide lay

    The comdent from101 101 103 105

    100

    101

    f / Hz

    Z j

    /

    cm

    0 10 20 30 40 500

    10

    20

    30

    Zr / cm2

    Auplated WPt

    20 Hz

    0.06 Hz

    2 Hz40 Hz

    ance response in an electrolyte containing PBS and H2O2 with elec-as a parameter: (a) imaginary part of the impedance as a function of

    d (b) complex impedance plane (Nyquist) plot.s in PBS, manipulation of the depth of eld revealed ation in the center of the electrode that was depressedrface by approximately 0.5m. This circular region iss in the photograph. A similar trend was seen for theodes left in PBS for longer periods. Each imagedisplays aression in the center of the electrode that grows deeperFor the electrode left in saline for ve days, the centersed approximately 4m and was covered by a porousage of the electrode left in PBS for six days shows that

    ungsten surface was recessed from the polished surfacenter being even more depressed. These photographst the gold-plated tungsten electrodes experienced gen-than pitting corrosion. An oxide layer is not apparent inaphs.hotographs of the surface of one gold-plated tungstenmersed in PBS and 30mM H2O2 are shown in Fig. 9

    ed time extending to 24h. After 1h in the electrolyte,re seen around the edges of the goldtungsten inter-

    e tungsten surface was roughened. The recesses on thef the electrode are expected for a primary or secondaryribution, where, due to the disk geometry, the currenthe highest at the perimeter (Newman, 2004). As time, the recesses expanded inward to the center of the elec-fter 24h, the bulk of the tungsten was depressed onf 10m below the original surface. These photographsuch higher corrosion rate for gold-plated tungsten inions containing H2O2 than was seen for the other twout H2O2. These photographs also show no evidence ofer.parative stability of the platinum electrode is evi-

    images taken before immersion and after 14 days of

  • E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171 163

    Fig. 7. Optical photographs of a tungsten electrode (a) before and (b) after immersion in PBS for 23 days. The nominal diameter of the electrode is 50m.

    immersion in PBS, as shown in Fig. 10. No difference is evidentbetween the photographs taken before and after immersion.

    The in vitro results presented here are consistent with scanningelectron microscope (SEM) images presented in Fig. 11 of gold-plated tungsten microelectrodes, insulated with polyimide andused for in vivo neuronal recording in another study (Patrick et al.,2010). All tthe UniversCommitteepositioner i2.5mm latet al., 2010microwiresand after 87trode surfacbeen cut wtypically ha

    the tungsten wires, beneath the polyimide insulation, to facilitatesoldering of thewire to the connecting substrate. Though not easilyseen in the SEM images, a thin ring of gold between the polyimideand tungsten is exposed to the biological environment. Image cshows the changed state of the tungsten wire after 87 days in vivo.The inset in c highlights the area of gold exposed to the tissue.

    e difarealyimielectrgy-d rep EDEDS

    xygebe

    nd i

    Fig. 8. Opticaland (g) six dayhe steps involved with the animal were approved byity of Florida IACUC (Institutional Animal Care and Use). The electrode array was inserted using a stereotaxicnto a craniotomy located (+1mm anterior to bregma,eral) in an adult male SpragueDawley rat (Patrick). The images presented in Fig. 11 shows the tungstenused in the intracortical microelectrode array beforedays implantation. Images labeled a and b show elec-es (cross section of the insulated wires) after they haveith a dicing saw prior to implantation. The microwiresve a sub-micron layer of gold on the outside surface of

    Notabllargerthe poof the

    EneexposeThe tobottombon, obio-lmtrode aphotographs of gold-plated tungsten electrodes (a) before and after immersion in PBS fos. The nominal diameter of each electrode is 50m.ferences between the before and after images include aof gold exposed in the after state as well as a swelling ofde insulation and the presence of a lm on the surfacerode.dispersive X-ray spectroscopy (EDS) results for twogions on the electrode surface are presented in Fig. 12.S result is characteristic of a bare tungsten surface. Theresult is characteristic of a bio-lm that is rich in car-

    n, and nitrogen. Thus, it is assumed that part of thecame dislodged on the top right portion of the elec-s exposing the tungsten surface. The SEM image showsr (b) one day, (c) two days, (d) three days, (e) four days, (f) ve days,

  • 164 E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171

    Fig. 9. Opticaland (e) 23h. T

    Fig. 10

    that the tunture of thesimilar to tbench-top

    Fig. 11. SEM iet al., 2010). Tphotographs of a gold-plated tungsten electrodes (a) before and after immersion in an ehe nominal diameter of the electrode is 50m.

    . Optical photographs of a platinum electrode (a) before and (b) after immersion in PBS

    gsten is recessed within the gold plating and the struc-tungsten surface has changed from smooth to rough,he general trend with the tungsten electrodes in thestudy. To quantify how far the tungsten surface was

    recessed wthe implan(sodium hyA single rin

    mages of tungsten micro-wires: (a) and (b) typical electrodes before implantation; andhe inset in (c) highlights the area of gold exposed to the tissue. The scale bar in (a) and (clectrolyte containing PBS and 30mM H2O2 for (b) 1h, (c) 2h, (d) 3h,

    for 14 days. The nominal diameter of the electrode is 50m.

    ith the gold plating and insulation, three wires fromted array were cleaned in a dilute solution of bleachpochlorite) until the adherent bio-lm was removed.sing operation was found to be sufcient to remove the

    (c) typical electrodes after implantation in vivo for 87 days (Patrick) is 100m. The scale bar in (b) is 50m.

  • E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171 165

    Fig. 12. EDS re lts area bio-lm (bot

    bio-lm. Inrm that thfrom the toinsulation tfocal lengthequippedwof this meathe electrodsurface was248m.

    4. Analysis

    This sectrodes andAnalysis ussible electrpotential aoccur on anof pH (Poucal speciesfrom the digenerated bsion 1.3.33database th(1966).

    4.1. Thermotungsten

    A thermgiven in Figcomputer pSystems Inc

    r differatees labAbole anposlinesults for two sites on one electrode after implantation in vivo for 87 days. The resutom graph). The scale bar in the photograph is 50m.

    spection under an optical microscope was used to con-e bio-lm was removed by this procedure. The distancep surface of the existing gold plating and polyimideo the tungsten was measured by manipulation of theof an optical microscope. The optical microscope wasith digital readout (Quadra-Chek 200) and the precisionsurement was estimated to be within 1m. In three of

    tion foto gennal linwater.possibtion) isdashedes in the array, the average distance that the tungstenrecessed within the gold plating was measured to be

    and discussion

    tion analyzes faradaic reactions occurring on the elec-discusses their implications on in vivo applications.ing Pourbaix diagrams provides information on pos-ochemical reactions. The Pourbaix diagram shows thet which chemical and electrochemical reactions mayelectrode surface in a specic electrolyte as a functionrbaix, 1966). The thermodynamic stability of chemi-at various potential and pH ranges may be ascertainedagram. The Pourbaix diagrams shown in this study arey a computer program (CorrosionAnalyzer 1.3 Revi-by OLI Systems Inc.), which uses a publicly availableat takes intoaccountnonidealitiesnot found inPourbaix

    dynamic analysis of electrochemical reactions on

    odynamic analysis of the stability of tungsten in PBS is. 13 in the form of a Pourbaix diagram, generated by arogram (CorrosionAnalyzer 1.3 Revision 1.3.33 by OLI.). Theoretical calculations based on the Nernst equa-

    regions desand corrositallymeasutungsten elical speciesgiven in Tabstable at thepossible inThe overallion is given

    W + 4H2O

    Possiblegiven by

    O2 + 2H2O

    and

    2H2O + 2e

    The strosuggests threaction, gi

    Anothertungsten oxRaman specWO3. The Pcharacteristic of a bare tungsten and gold surface (top graph) and of

    erent chemical species at room temperature were usedthe lines on the diagram (Pourbaix, 1966). The diago-eled a and b correspond to the limits of the stability of

    ve line a, oxidation of water (i.e., oxygen evolution) isd below line b, reduction of water (i.e., hydrogen evolu-sible. Between those lines, water is stable. The verticalgives thenatural pHof the electrolyte as 7.4. The shaded

    ignate states of passivation (WO3 (s)), immunity (W(s)),on (WO24 ). The box shows the range of the experimen-redvalues of the open-circuit potential for a gold-platedectrode in PBS over a period of 15days. A list of all chem-considered in the formation of the Pourbaix diagram isle 2. The diagram shows that the tungstic ion WO24 isopen-circuit potential and that corrosion reactions are

    this system since the box is in the white (WO24 ) area.anodic electrochemical reaction producing the tungsticby (Kelsey, 1977; Pourbaix, 1966).

    WO24 + 8H+ + 6e (2)

    cathodic reactions are reduction of water and oxygen

    + 4e 4OH (3)

    H2 + 2OH (4)

    ng effect of electrolyte oxygen content seen in Fig. 5at the reduction of oxygen is the dominant cathodicven by reaction (3).possible chemical reaction is the dissolution of aide. Lillard et al. (1998) showed via surface-enhanced-troscopy that tungsten exposed to air has a native oxideourbaix diagram suggests that WO3 dissolves at the cell

  • 166 E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171

    Fig. 13. Pourbaix diagram for tungsten in PBS. The box shows range of open circuit potential measured over a period of 15 days. The diagrams were generated byCorrosionAnalyzer 1.3 Revision 1.3.33 by OLI Systems Inc.

    equilibrium potential (0.4V vs. SHE) and pH as given by reaction(5).

    WO3 + H2O WO24 + 2H+ (5)Therefore, even if a native oxide layer exists initially on the

    tungsten surface that inhibits dissolution, eventually the oxide willdissolve and expose the tungsten surface.

    The addition of hydrogen peroxide to the PBS allows anothercathodic reaction, the reduction of H2O2, to occur on thetungsten electrode surface. A Pourbaix diagram showing theOCP range for a gold-plated tungsten electrode in an elec-trolyte containing PBS and 30mM of H2O2 is given in Fig. 14.The chemical species considered for the Pourbaix diagram arethe same as listed in Table 2 with the addition of H2O2. The

    Table 2Species considered in the calculation of the Pourbaix diagram presented as Fig. 13. The list of species was generated by CorrosionAnalyzer 1.3 Revision 1.3.33 by OLI SystemsInc.

    Aqueous phase Solid phase Vapor phase

    Water Pentasodium triphosphorous decaoxide WaterChloride ion(1) Pentasodium triphosphorous decaoxide hexahydrate HydrogenDihydrogen er)DihydrogenHydrogen ophosHydrogen ch hosphHydrogen io hosphHydrogen or hate(VHydrogen pyHydrogen tu rateHydroxide ioOrthophospOxygenPhosphate ioPotassium chPotassium ioPyrophosphPyrophosphSodium ion(TrihydrogenTungsten(VITungstic(VI)orthophosphate(V) ion(1) Phosphorus pentoxide (dimpyrophosphate(V) ion(2) Potassium chloride

    Potassium dihydrogen orthloride Potassium hydrogen orthopn(+1) Potassium hydrogen orthopthophosphate(V) ion(2) Potassium hydrogen phosprophosphate(V) ion(3) Potassium hydroxidengstate(VI) ion(1) Potassium hydroxide dihyd

    n(1) Potassium hydroxide monohydra

    horic acid Potassium orthophosphate(V)Potassium orthophosphate(V) he

    n(3) Potassium orthophosphate(V) triloride Potassium tungstate(VI)n(+1) Sodium chlorideate ion(4) Sodium dihydrogen orthophosphoric(V) acid Sodium dihydrogen orthophosph+1) Sodium dihydrogen orthoporthohpyrophosphate(V) ion(1) Sodium hydrogen orthophosphat) tetraoxide ion(2) Sodium hydrogen orthophosphatacid Sodium hydrogen orthophosphat

    Sodium hydrogen orthophosphatSodium hydroxideSodium hydroxide monohydrateSodium orthohosphateSodium orthophosphate hexahydSodium orthophosphate hydroxidSodium orthophosphate monohySodium orthophosphate octahydSodium pyrophosphate decahydrSodium tungstate(VI)Sodium tungstate(VI) dihydrateTungstenTungsten(VI) oxideHydrogen chlorideOxygen

    phate(V)ate(V) hexahydrateate(V) trihydrate)te

    ptahydratehydrate

    ate dihydrateate monohydrateosphateee dihydratee dodecahydratee heptahydrate

    ratee dodecahydratedraterateate

  • E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171 167

    Fig. 14. Pourb shows the measured range of the open-circuit potential over a period of 2days. The diag

    regions oflayed on tH2O2 is pose.g.,

    H2O2 + 2H+

    Thus, fortrode in PBby reactionreaction (3experimentmechanismnot the forthe experimpearance otime.

    4.2. Goldt

    The goldple, givingGold is stab(Pourbaix, 1surface forsion rate ofgold that wfurther expincreasing tsequently tsince the nezero.

    This prodiagram shFig. 15. Forand (3)), aroccurs onlyoccurs onlyin responsetion of oxygby the shiftungsten arsten is incre

    log current

    Pote

    nal

    ecorr,1

    Icorr,1

    ecorr, 4

    Icorr,4

    Hypothetical Evansdiagram for the gold-plated tungsten electrode showingt of increased cathode surface area on the galvanic interaction of tungsten.

    alues. This explanation is supported by observations of theircuit potential for the gold-plated tungsten electrodes andee tungsten electrodes in PBS, shown in Fig. 16 as functionssed time. The gold-plated tungsten system shows a nonlin-nd toward more positive potentials; whereas, the gold-freeaix diagram for tungsten in an electrolyte containing PBS and 30mM H2O2. The boxrams were generated by CorrosionAnalyzer 1.3 Revision 1.3.33 by OLI Systems Inc.

    stability of H2O2 given by Pourbaix (1966) are over-he diagram. Below the line labeled 1, reduction ofsible and above line 2, oxidation of H2O2 is possible,

    + 2e 2H2O (6)an electrochemical system containing a tungsten elec-S and H2O2, the possible anodic reaction is given(2), and possible cathodic reactions are given by

    ) and reaction (6). This thermodynamic analysis atally observed potentials suggests that the corrosionfor tungsten is dissolution into the tungstic ion and

    mation of a stable metal oxide. This result supportsental results of Figs. 711 which show the disap-

    f tungsten from the surface of the electrodes over

    ungsten galvanic couple

    -plated-tungsten electrode constitutes a galvanic cou-gold a signicant role in the reactivity of the system.le at the equilibrium potential and pH of the system

    Fig. 15.the effecand gold

    noble vopen-cgold-frof elapear tre966) and thus will not corrode. However, it provides athe cathodic reaction and thereby increases the corro-the tungsten (Jones, 1992). As the tungsten dissolves,as plated on the outer surface of the tungsten becomesosed. The surface area of the gold increases, therebyhe exchange current for the cathodic reaction, and con-he anodic reaction rate must increase to compensate,t current at the open-circuit potential must be equal to

    cess may be modeled using mixed potential theory. Aowing a proposed mixed potential theory is given insimplicity, only the dominant reactions (reactions (2)e included. It is assumed that the oxidation reactionon the tungsten (anode), and the reduction reactionon the gold (cathode). As the cathodic current increasesto the increasing cathode surface area for the reduc-en, a larger corrosion current is established, as shown

    t from Icorr,1 to Icorr,4. Under the assumption that theea is constant, the resulting corrosion rate for tung-ased, and the potential of the system increases to more

    0

    0.

    0

    0.

    0

    OCP

    (v. SH

    E) / V

    Fig. 16. Open-and tungstenIncreasing Au surface area

    I0,4I0,10 5 10 15 20 25.5

    45

    .4

    35

    .3

    day

    W onlyAuplated W

    circuit potential as a functionof elapsed time for gold-plated tungstenelectrodes in PBS electrolyte.

  • 168 E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171

    Table 3Corrosion rates for tungsten estimated from the impedance response.

    System Rp ( cm2) rcorr (m/yr)

    W in PBS 1700 200 rcorr 500W/Au in PBSW/Au in PBS

    tungsten syair-formed

    Two faclogical salilimited byof the tuncouple existions.

    4.3. Rate of

    Since exthat the metungstic ionpolarizationGeary, 1957

    Rp = 2.3icorwhere a arespectivelyGeary, 1957mated fromasymptoticNyquist plosion rate by

    rcorr = icorranF

    where a isFaradays coloss per unito obtain a

    If the Tafbe approxim0.12 V for If the mostonly a factocorrosion ra

    The corrPBS, gold-pand30mMresistance fsion was umodel given

    Z = R0 +k

    k=

    The meaelectrolyte relement isacteristic tiresponse ofwere polishImpedanceof 5min. Thcalculated fimpedance

    5

    10

    1500

    2000

    j

    W only in PBSAuplated W in PBS

    1

    2

    3

    4

    Zr / cm2

    Z j

    /

    cm

    2

    a

    b

    Estimation of polarization resistance, Rp, for three tungsten systems. Theresent the results of regression of a Voigt model, Eq. (9) to the data. (a)impedance plane (Nyquist) plot with electrode material and electrolyte as

    eter; and (b) enlarged section of (a) emphasizing the results for gold-platedin PBS with added H2O2.

    e of 200500m/yr. The galvanic couple of the gold-plateden microwire acts to increase the corrosion rate of the sys-PBS alone to 300700m/yr. However, the corrosion ratested for the gold-plated tungsten systems are for the initialde area; the rates could increase as more of the gold surfacees exposed as describedpreviously. The additionof 30mMofen peroxide to the PBS signicantly increases the corrosiona gold-plated tungsten electrode to 10,00020,000m/yr.

    us EIS results showed that the corrosion rates are limitedcathodic reaction. Thus, the corrosion rates for each systempendent on the concentrations of oxygen or hydrogen per-respectively, and may also be controlled by diffusion of thes.

    bench-top results may be compared to the corrosionetermined for the tungsten microelectrodes that wereted in vivo. The average recession of the tungsten surfacean 87 day implant period of three different micro-des was 248m yielding an estimated annual rate33m/yr.corrosion rate results from the bench-top experimentsond to a worst-case scenario for in vivo situations. Thetop results showed that the corrosion rate is limited byte of the dominant cathodic reaction. Therefore, the con-1200 300 rcorr 700+ H2O2 40 10, 000 rcorr 20, 000

    stem shows a more stable open-circuit potential afteroxides are removed.tors control the corrosion rate of tungsten in bio-ne solutions. The rate of corrosion of tungsten isthe cathodic reaction. Moreover, the corrosion rategsten is increased by the goldtungsten galvanicting in microelectrodes used for intracortical applica-

    tungsten corrosion

    perimental results and thermodynamic analysis showchanism of tungsten corrosion is dissolution to form, the corrosion rate can be calculated easily from theresistance via the SternGeary equation (Stern and

    ; Jones, 1992)

    ac

    r(a + c)(7)

    nd c are the anodic and cathodic Tafel constants,, and icorr is the corrosion current density. (Stern and; Jones, 1992). The polarization resistance may be esti-the impedance data by subtracting the low frequencyvalue from the high frequency asymptotic value on at. The corrosion current density is related to the corro-

    (8)

    the atomic weight, n is the oxidation number, and F isnstant. Eq. (8) gives a corrosion rate in terms of mass

    t area per time and is divided by the density of themetalcorrosion penetration rate, e.g., m/yr.el constants are not known, the corrosion rate may stillated. The Tafel constants range between 0.06 V and

    a and between 0.12 and innity for c (Jones, 1992).extreme values are used, the corrosion rate varies byr of two. These values are used for the calculation of thete in this work.osion rates of the three tungsten systems (tungsten inlated tungsten in PBS, and gold-plated tungsten in PBSH2O2)werequantiedbyextrapolating thepolarizationrom EIS data. Complex nonlinear least squares regres-sed to t the experimental data to the measurement

    as (Orazem, 2004).

    1

    Rk1 + jk

    (9)

    surement model consists of a resistor, R0, modeling theesistance, in serieswithkVoigt elements,where aVoigta capacitor in parallel with a resistor that has a char-me constant k =2RC. Nyquist plots of the impedancethe three systems are shown in Fig. 17(a). All electrodesed immediately prior to submersion in the electrolyte.

    Z

    /

    cm

    2

    Fig. 17.lines repComplexa paramtungsten

    at a rattungsttem inpredicelectrobecomhydrograte ofPrevioby theare deoxide,specie

    Therate dimplanafterelectroof 100

    Thecorrespbench-the rameasurements were performed after a settling periode corresponding corrosion rate for each system wasrom Eq. (8), and the results are given in Table 3. Theresults predict that tungsten in aerated PBSwill corrode

    centrationThe concenbench-topoxygen in0 500 1000 1500 2000 2500 30000

    00

    00

    Zr / cm2

    Auplated W in PBS + H2O20.16 Hz

    1.0 Hz

    0 10 20 30 40 50 600

    0

    0

    0

    0W only in PBSAuplated W in PBSAuplated W in PBS + H2O2

    10 Hzof the reactive species, O2 or H2O2, play a major role.tration of dissolved oxygen in the PBS used in theexperiments is determined by the partial pressure ofthe air (150mmHg). In comparison, the partial pres-

  • E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171 169

    Fig. 18. Pourbaix diagram for platinum in PBS with 30mM hydrogen peroxide. The red box shows range of measured open circuit potential. The diagrams were generatedby CorrosionAnalyzer 1.3 Revision 1.3.33 by OLI Systems Inc. (For interpretation of the references to color in this gure legend, the reader is referred to the web version ofthe article.)

    sure of oxygen in the extracellular uid of the cortex is between20 and 35mmHg (Maas et al., 1993; Sarrafzadeh et al., 1998).Moreover, the elevated temperature in the biological environ-ment decreases the solubility of oxygen, thus, further reducingthe expected corrosion rate (Benson and Krause, 1984). Anotherfactor that would reduce the corrosion rate in vivo is the pres-ence of a bio-lm. The biological lm that forms on the electrode

    surface, seen in the SEM images of the implanted tungstenmicrowire (Fig. 12), may act to impede diffusion of chemicalspecies and effectively control the corrosion rate. It is antic-ipated, therefore, that the in vivo corrosion rate will changewith time due to the dynamic nature of the wound healing andpersistent foreign-body immune responses and growth of bio-lms.

    Table 4Species considered in the calculation of the Pourbaix diagram presented as Fig. 18. The list of species was generated by CorrosionAnalyzer 1.3 Revision 1.3.33 by OLI SystemsInc.

    Aqueous phase Solid phase Vapor phase

    Water Pentasodium triphosphorous decaoxide WaterChloride ion(1) Pentasodium triphosphorous decaoxide hexahydrate HydrogenDihydrogen er)DihydrogenHydrogenHydrogen chHydrogen ioHydrogen orHydrogen pe ophosHydrogen pe hosphHydrogen py hosphHydroxide ioOrthophospOxygenPhosphate ioPlatinum ionPlatinum ionPlatinum(II)Platinum(II)Platinum(II)Platinum(II)Platinum(II)Platinum(II)Platinum(IVPlatinum(IVPlatinum(IVPlatinum(IVPlatinum(IVPlatinum(IVPotassium cPotassium ioPyrophosphPyrophosphSodium ion(Trihydrogenorthophosphate(V) ion(1) Phosphorus pentoxide (dimpyrophosphate(V) ion(2) Platinum

    Platinum(II) chlorideloride Platinum(II) hydroxiden(+1) Platinum(IV) chloridethophosphate(V) ion(2) Potassium chlorideroxide Potassium dihydrogen orthroxide ion(1) Potassium hydrogen orthoprophosphate(V) ion(3) Potassium hydrogen orthop

    n(1) Potassium hydrogen phosphate(V

    horic acid Potassium hydroxidePotassium hydroxide dihydrate

    n(3) Potassium hydroxide monohydra(+2) Potassium orthophosphate(V)(+4) Potassium orthophosphate(V) hepchloride Potassium orthophosphate(V) trihhydroxide Sodium chloridemonochloride ion(+1) Sodium dihydrogen orthophosphmonohydroxide ion(+1) Sodium dihydrogen orthophosphtetrachloride ion(2) Sodium dihydrogen orthoporthohtrichloride ion(1) Sodium hexachloroplatinate(IV) h) chloride Sodium hydrogen orthophosphat) dichloride ion(+2) Sodium hydrogen orthophosphat) hexachloride ion(2) Sodium hydrogen orthophosphat) monochloride ion(+3) Sodium hydrogen orthophosphat) pentachloride ion(1) Sodium hydroxide) trichloride ion(+1) Sodium hydroxide monohydratehloride Sodium orthohosphaten(+1) Sodium orthophosphate hexahydate ion(4) Sodium orthophosphate hydroxidoric(V) acid Sodium orthophosphate monohy+1) Sodium orthophosphate octahydrpyrophosphate(V) ion(1) Sodium pyrophosphate decahydrHydrogen chlorideHydrogen peroxideOxygen

    phate(V)ate(V) hexahydrateate(V) trihydrate

    )

    te

    tahydrateydrate

    ate dihydrateate monohydrateosphateexahydrateee dihydratee dodecahydratee heptahydrate

    ratee dodecahydratedrateateate

  • 170 E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171

    The bench-top results suggest that dissolution of tungsten isthe corrosion mechanism in vivo. The Pourbaix diagrams gener-ated for the bench-top study show that tungsten dissolution intothe tungstic ion persists over a wide range of pH. Thus, even ifthe immunuid, such amechanism

    Tungstenments in thelectrode ctant questiofocus of thinot a proofformance oin vivo studof the corrotant point ton their pestayed in plbeen movinlayer was narrays (witmance fordecreasingbeforenineotic responunderstand

    4.4. Thermoplatinum

    As showelectrodesis importanoccurring odiagram coated for plais shown insuperimpos1 the reducdation of Hthe two reawater on a popen-circuidiagram shtion in this

    The Pourespondinghydrogen aassumed an

    PtH Pt +and reactio

    Cyclic vchemical reof the plat100mV/s inet al., 1999;inferred to bhydrogen a(0.045V v

    Hydroge(10) and (6hence it ismost likelymay even l

    1.1.5

    1

    0.5

    0

    0.5

    1

    1.5

    2

    100 mV/s50 mV/s

    Cyclic30mM

    ted mesulin bi

    clus

    rateghichm mGrapsencresirosiofor eialsed pogsteenvirant eof tminathe

    The-plattedtungmMcorn of implantation, was estimated to be approximately/yr,which is smaller than the in vitro estimation of tungsten

    ion rate of 300700m/yr. The reduced corrosion rate in thesetting as compared to the in vitro setting is attributed tosed rate of diffusion caused by the presence of a biologicald a reduced concentration of available oxygen in the brainpared to electrolyte exposed to air.inum was shown to be a much better choice for intracorticallectrodes. Platinum is effectively unreactive in phosphated saline solutions and although a faradic response wassolution containing hydrogen peroxide, the electrochemi-

    ctions, reductionofH2O2 andhydrogen-atomplating, shouldss adverse effects on long-term recording.long-termviabilityof implantedmicroelectrodes is anongo-ic of research. This study provides new information on thenismof tungsten corrosion in biological saline environmentsows that corrosion should not be neglected in future stud-e response increases the local pH of the extracellularchange in pH is not expected to change the corrosion

    .microelectrodes have been used in chronic experi-

    e past with mixed results. The inuence of tungstenorrosion on recording performance, though an impor-n, is convoluted with many other factors and is not thes study. Successful recording performance is, in itself,that corrosion did not take place. The recording per-

    f the tungsten microelectrode array used in the presenty did not diminish over a period of 42 days, in spitesion evident in Fig. 12 (Patrick et al., 2010). An impor-o note is that the tungsten electrodes had a gold layerrimeter that did not undergo corrosion and in effectace. Thus, even though the tungsten surface could haveg away from a potential neuron, the conducting goldot. In the study by Williams et al. (1999), four tungstenhout gold plating) showed reliable recording perfor-nine weeks or more, while four other arrays showedperformance from week one to their termination dateweeks. Additional studies that couple thebiotic andabi-ses observed here and in vivo are required for unifyinging of the mechanisms of electrode failure.

    dynamic analysis of electrochemical reactions on

    n in Fig. 6, a faradaic reaction occurs on platinumin saline solutions containing hydrogen peroxide. Itt to know what electrochemical reactions could ben the platinum electrode when implanted. A Pourbaixnsidering the chemical species in Table 4 was gener-tinum in an electrolyte of PBS and 30mM H2O2 andFig. 18. Curves representing the stability of H2O2 areed on the Pourbaix diagram. At potentials below curvetion of H2O2 is possible and above curve 2, the oxi-2O2 is possible (Pourbaix, 1966). In the region wherections overlap, hydrogen peroxide may dissociate intolatinum surface. The box shows the range of measuredt potential for three platinum electrodes. The Pourbaixows that reduction of H2O2 is a possible cathodic reac-system.rbaix diagram does not, however, suggest possible cor-anodic reactions. Brummer and Turner (1977) proposedsorption to be possible charge transfer pathways. Theodic and cathodic reactions are given respectively by

    H+ + e (10)n (6).oltammetry was used to further analyze the electro-actions at the platinum surface. Cyclic voltammogramsinum electrodes are shown for scan rates of 50 andFig. 19. By comparing these resultswith literature (RenYoo et al., 2009), the two peaks labeled Hw and Hs aree the weak and strong hydrogen adsorption peaks. Thedsorption peaks occur near the open circuit potentials. SHE) of the cell.n adsorption and hydrogen peroxide reduction, Eqs.), have no adverse effect on the platinum surface,expected to be stable. Effects on the brain tissue arebenign. The reduction of hydrogen peroxide to water

    essen degradation to nearby cells or other parts of the

    curr

    ent /

    A

    Fig. 19.PBS and

    implanthese rchoice

    5. Con

    A sting wplatinulished.the preizationthe corgramspotentprovid

    TunsalinedomindationThe dosten towater.a goldestimaplatedand 30

    Theduratio100mcorrosin vivodecrealm anas com

    Platmicroebuffereseen incal reahave le

    Theing topmechaand sh5 1 0.5 0 0.5 1 1.5voltage (v. SHE) / V

    Hw

    Hs

    voltammogram for a platinum electrode in an electrolyte containingH2O2.

    icroelectrode (e.g., the polymer insulation). Therefore,ts conrm that platinum as a recording site is a goodological solutions.

    ions

    y for ascertaining the nature of charge transfer includ-faradaic reactions are occurring on tungsten andicroelectrodes in saline environments has been estab-hical analysis of impedance data was used to ascertaine of faradaic reactions, and extrapolation of the polar-stance from the impedance data was used to estimaten rate of corroding systems. Analysis of Pourbaix dia-ach system at experimentally measured open-circuit

    corroborated the presence of tungsten corrosion andssible electrochemical reactions for each system.

    n was shown to corrode in bench-top physiologicalonments without the formation of a stable oxide. Thelectrochemical reactions on tungsten in PBS were oxi-ungsten to the tungstic ion and reduction of oxygen.nt reactions in PBS with H2O2 were oxidation of tung-tungstic ion and reduction of oxygen and H2O2 to

    corrosion rates for a bare-tungsten microelectrode anded tungsten microelectrode in physiological PBS wereto be 300700m/yr. The corrosion rate of a gold-sten microelectrode in an electrolyte containing PBSH2O2 was 10,00020,000m/yr.rosion rate in the in vivo setting, averaged over the

  • E. Patrick et al. / Journal of Neuroscience Methods 198 (2011) 158171 171

    ies on the viability of tungsten electrodes. Moreover, the role ofcorrosion byproducts, e.g., the tungstic ion, should be included inthe examination of tissue reactivity or subsequent toxic effects onthe neural tissue. The ndings in this study discourage the use ofthin-lm tungsten electrodes in any aqueous environment whereoxygen or other oxidizing species are present.

    Acknowledgements

    This work was supported by NIH Grant NS053561. The SEMimages were provided by the Major Analytical InstrumentationCenter (MAIC) at the University of Florida. We also thank VincentVivier for the use of the platinum microelectrodes and for the help-ful discussions on their fabrication.

    References

    Anik M. Ph-dependent anodic reaction behavior of tungsten in acidic phosphatesolutions. Electrochim Acta 2009;54(15):394351.

    AnikM,Osseo-AsareK. Effect of phon the anodic behavior of tungsten. J ElectrochemSoc 2002;149(6):B22433.

    Benson BB, Krause Jr D. The concentration and isotopic fractionation of oxygen dis-solved in fOceanogr

    Bothra S, Surtungsten p

    Brug G, den Eimpedanctroanal Ch

    Brummer SB, Ttion of the1977;BME

    Burke B, LewisCarmena JM, L

    Learning tmates. PLo

    Chapin JK, Moarm using1999;2(7)

    Fonesca C, Bing H2O22001;43:5

    Frateur I, Deslwater syst1999;44:4

    deHaro C,Masplatinumarrays. Bio

    Henze D, Borhpredicted2000;84:3

    Jacobs JJ, Gilborthopaed

    Johnson JW,1971;118(

    Jones DA. Prinlishing Co;

    Kelsey GS. Th1977;124(

    Kipke DR. Silicon-substrate intracortical microelectrode arrays for long-termrecording of neuronal spike activity in cerebral cortex. IEEE Trans Rehabil Eng2003;11:1515.

    Lillard RS, Kanner GS, Butt DP. The nature of oxide lms on tungsten in acidic andalkaline solutions. J Electrochem Soc 1998;145(8):271825.

    Maas A, Fleckenstein W, de Jong D, van Santbrink H. Monitoring cerebral oxy-genation: experimental studies and preliminary clinical results of continuousmonitoring of cerebrospinal uid and brain tissue oxygen tension. Acta Neu-rochir Suppl (Wien) 1993;59:507.

    Maynard E, Nordhausen C, Normann RA. The Utah intracortical electrode array: arecording structure for potential brain-computer interfaces. ElectroencephalogrClin Neurophysiol 1997;102:22839.

    Merrill DR, Bikson M, Jefferys JGR. Electrical stimulation of excitable tissue: designof efcacious and safe protocols. J Neurosci Methods 2005;141(2):17198,01650270, doi:10.1016/j.jneumeth.2004.10.020.

    Newman J. Electrochemical systems. Hoboken, NJ: John Wiley and Sons, Inc; 2004.Norlin A, Pan J, Leygral C. Investigation of interfacial capacitance of Pt, Ti, and

    TiN coated electrodes by electrochemical impedance spectroscopy. Biomol Eng2002;19:6771.

    Norlin A, Pan J, Leygral C. Investigation of electrochemical behavior for stimulat-ing/sensing materials for pacemaker electrode applications. J Electrochem Soc2005;152(2):J715.

    Orazem ME. A systematic approach toward error structure identication forimpedance spectroscopy. J Electroanal Chem 2004;572:31727.

    Orazem ME, Pbre N, Tribollet B. A new look at graphical representation ofimpedance data. J Electrochem Soc 2006;153:B12936.

    Orazem ME, Tribollet B. Electrochemical impedance spectroscopy. Hoboken, NJ:Wiley & Sons; 2008.heirry D, Leygraf C. Electrochemical impedance spectroscopy study ofpassiv;41:1E. D

    nmacE, Sanpoweneerinferenc, Sanke andem. InrnatioM, Finitro asateriM. A

    uX, LisitiontrodesJ, Albowireral Sysdeh AenatilicoxGearylariza

    s JC, Rire m;4(3)Jang JHdocapmuirreshwater and seawater in equilibrium with the atmosphere. Limnol1984;29:62032.H, Liang V. A new failure mechanism by corrosion of tungsten in alug process. Microelectron Reliabil 1999;39(1):5968.eden AV, Sluyters-Rehbach M, Sluyters J. The analysis of electrodees complicated by the presence of a constant phase element. Elec-em 1984;176:27595.urner MJ. Electrochemical considerations for safe electrical stimula-nervous system with platinum electrodes. IEEE Trans Bio-Med Eng

    -24(1):5963.C. The macrophage. Oxford: Oxford Medical Publications; 2002.ebedev MA, Crist RE, ODoherty JE, Santucci DM, Dimitrov DF, et al.o control a brain-machine interface for reaching and grasping by pri-S Biol 2003;1(2):e42.xon KA, Markowitz RS, Nicolelis MAL. Real-time control of a robotsimultaneously recorded neurons in the motor cortex. Nat Neurosci:66470.arbosa M. Corrosion behavior of titanium in biouids contain-studied by electrochemical impedance spectroscopy. Corros Sci4759.ouis C, Orazem ME, Tribollet B. Modeling of the cast iron/drinkingem by electrochemical impedance spectroscopy. Electrochim Acta34556.R, Abadal G,Munoz J, Perez-Murano F, Dominguez C. Electrochemical

    coatings for improving performance of implantable microelectrodematerials 2002;23(23):451521.egyi Z, Sciscvari J, Mamiya A, Harris K, Buzaki G. Intracellular featuresby extracellular recordings in thehippocampus in vivo. J Neurophysiol90400.ert JL, Urban RM. Current concepts review corrosion of metalic implants. J Bone Joint Surg 1998;80(2):26882.Wu CL. The anodic dissolution of tungsten. J Electrochem Soc12):190912.ciples and prevention of corrosion. New York, NY: Maxmilliam Pub-1992.

    e anodic oxidation of tungsten in aqueous base. J Electrochem Soc6):8149.

    JohnPan J, T

    the1996

    Patrickbrai

    Patricklow-EngiCon

    Patrick Estratsystinte

    Peusterin vBiom

    PourbaixRenB,X

    tranelec

    SanchezmicrNeu

    Sarrafzaoxygwith

    SternM,of po

    Williamof w1999

    YooHD,pseuLange oxide lm on titanium for implant application. Electrochem Acta14353.esign, fabrication, and characterization of microelectrodes forhine interfaces. Ph.D. thesis; University of Florida; 2010.kar V, Rowe W, Sanchez JC, Nishida T. An implantable integratedr amplier-microelectrode array for brainmachine interfaces. In:g in medicine and biology society (EMBC), 2010 Annual Internationale of the IEEE, 2010; Aug. 31 2010Sept. 4 2010, pp. 18161819.ar V, Rowe W, Yen SF, Sanchez JC, Nishida T. Flexible polymer sub-tungsten microelectrode array for an implantable neural recording: Engineering in medicine and biology society. EMBS 08. 30th annualnal conference of the IEEE; 2008. p. 315861.k C, von Schnakenburg C. Biocompatibility of corroding tungsten coils:sessment of degradation kinetics and cytotoxicity on human cells.als 2003;24:405761.tlas of electrochemical equilibria. Oxford: Pergamon Press Ltd; 1966.XQ,CaiWB, TianZQ. Extending surfaceRamanspectroscopic studies tometals for practical applications: II. Hydrogen adsorption at platinum. Surf Sci 1999;427428:15761.a N, Nishida T, Batich C, Carney P. Structural modications in chronicelectrodes for cortical neuroprosthetics: a case study. IEEE Transt Rehabil Eng 2006:21721., Kiening K, Bardt T, Schneider G, Unterberg A, Lanksch W. Cerebralon in contusioned vs. nonlesioned brain tissue: monitoring of ptio2and paratrend. Acta Neurochir Suppl (Wien) 1998;71:1869.AL. Electrochemical polarization: I. A theoretical analysis of the shapetion curves. J Electrochem Soc 1957;104:5663.ennaker RL, Kipke DR. Long-term neural recording characteristicsicroelectrode arrays implanted in cerebral cortex. Brain Res Protoc:30313., KaBH, RheeCK,OhSM. Impedance analysis for hydrogenadsorptionacitance and electrochemically active surface area of pt electrode.2009;25(19):1194754.

    Corrosion of tungsten microelectrodes used in neural recording applications1 Introduction2 Experimental methods2.1 Materials2.2 Instrumentation

    3 Experimental results3.1 Impedance response in PBS3.2 Impedance response in PBS with 30mM H2O23.3 Photographic evidence for tungsten corrosion

    4 Analysis and discussion4.1 Thermodynamic analysis of electrochemical reactions on tungsten4.2 Goldtungsten galvanic couple4.3 Rate of tungsten corrosion4.4 Thermodynamic analysis of electrochemical reactions on platinum

    5 ConclusionsAcknowledgementsReferences