Effect of hydrogen peroxide on titanium surfaces: In situimaging and step-polarization impedance spectroscopy ofcommercially pure titanium and titanium, 6-aluminum,4-vanadium

Jane P. Bearinger,1,2,3 Christine A. Orme,2 Jeremy L. Gilbert4

1Department of Biomedical Engineering, Northwestern University, Evanston, Illinois2Chemistry and Material Science, Lawrence Livermore National Laboratory, 7000 East Avenue, Livermore, California3Medical Technology Program, Lawrence Livermore National Laboratory, 7000 East Avenue, Livermore, California4Department of Bioengineering and Neuroscience, Syracuse, 365 Link Hall, Syracuse, New York 13244-5290

Received 6 January 2003; revised 10 March 2003; accepted 20 March 2003

Abstract: To analyze titanium’s response to representativesurgical wound environments, a study was conducted oncommercially pure titanium (CPTi) and titanium, 6-alumi-num, 4-vanadium (Ti-6Al-4V) exposed to phosphate-buff-ered saline (PBS) with 30 mM of hydrogen peroxide (H2O2)added. The study was characterized by simultaneous elec-trochemical atomic force microscopy (EC AFM) and step-polarization impedance spectroscopy (SPIS). Surfaces werecovered with protective oxide domes that indicated topog-raphy changes with potential and time of immersion. Lessoxide dome coarsening was noted on surfaces treated withPBS containing H2O2 than on surfaces exposed to pure PBS.Electrical data deduced from current transients collectedwhile stepping voltage between 0 V and 1 V indicated thatcharge transfer in hydrogen peroxide solutions was an orderof magnitude larger than it was in pure PBS. Oxide (early)resistances of CPTi samples were higher than were Ti-6Al-4V oxide resistances in both types of solutions, but CPTi

oxide resistance was lower in the hydrogen peroxide solu-tion compared to pure PBS. Capacitance data suggest thatCPTi oxide films thicken in hydrogen peroxide solutionmore than they do in pure PBS. Differences in electricalproperties between CPTi and Ti-6Al-4V surfaces suggestthat CPTi, but not Ti-6Al-4V, has catalytic activity on H2O2and that the catalytic activity of CPTi oxide affects its abilityto grow TiO2. Differences in electrical properties are relatedto catalytic and oxidative mechanisms that take place di-rectly on the titanium oxide surface and in wound environ-ments. The study provides a foundation and theoretic basisfor the porous oxide model on commercially pure titaniumexposed to hydrogen peroxide. © 2003 Wiley Periodicals,Inc. J Biomed Mater Res 67A: 702–712, 2003

Key words: titanium; titanium oxide; atomic force micros-copy; electrochemical methods; corrosion; oxidation; cataly-sis; surface structure; interfaces

INTRODUCTION

Hydrogen peroxide (H2O2) is present duringwound healing and may result in increased titaniumdissolution from biomedical implant materials follow-ing implantation. Examination of titanium’s admit-tance (impedance�1) response over a potential rangeshould lead to an increased understanding of hydro-gen peroxide’s effects on titanium in these environ-

ments. This work sought to investigate structural andelectrical changes in titanium and its oxide filmbrought about by addition of hydrogen peroxide to aphosphate-buffered saline (PBS) solution. Changeswere characterized via simultaneous electrochemicalatomic force microscopy (EC AFM) and step-polariza-tion impedance spectroscopy (SPIS). Hydrogen perox-ide’s effects on commercially pure titanium (CPTi)and titanium, 6-aluminum, 4-vanadium (Ti-6Al-4V)were compared.

Hydrogen peroxide is an important oxidizing com-ponent of wound healing with multiple formationpathways in the body. Molecular oxygen may be re-duced to a superoxide anion radical, a highly reactiveand unstable species that is damaging to microbes andcells. Superoxide anion radicals may be doubly pro-tonated in the body to form hydrogen peroxide. Su-

Correspondence to: J. Gilbert; e-mail: gilbert@ecs.syr.eduContract grant sponsor: Department of Energy; contract

grant number: Lawrence Livermore National Laboratory(LLNL) Contract W-7405-Eng.48

Contract grant sponsor: The Medical Technologies Pro-gram and the Materials Research Institute at LLNL

© 2003 Wiley Periodicals, Inc.

peroxide dismutase (SOD), produced by inflamma-tory cells, also may catalyze intermediate oxygenradicals to form hydrogen peroxide. Because of theinability of hydrogen peroxide to distinguish hostfrom invader, the potentially corrosive effects of hy-drogen peroxide on implant materials are a factor toconsider.

Structural studies have been performed on titanium/titanium oxide to characterize its surface topographyand chemical composition. Using scanning tunnelingmicroscopy (STM), various groups have observed theeffect of H2O2 in PBS on titanium. STM images haverevealed much rougher titanium oxide surface topog-raphies with the addition of H2O2 to PBS. Granularstructures in the passive film suggest that oxide dis-solution occurs at localized defects. AFM/STM stud-ies have revealed changes in oxide morphology overtime in acid solution as well. Ex situ work performedwith X-ray photoelectron spectroscopy (XPS) and Au-ger electron spectroscopy (AES) were conducted ontitanium immersed in solution to explore Ti-adsorbatebonding environments. Over time, calcium and phos-phate become incorporated into titanium oxides, andoxidation may continue for years; crystallization mayoccur on thickened oxides and ion incorporation alsomay take place.5–7 While surface chemical and func-tional group characterization is informative, it typi-cally cannot explain morphologic reaction kinetics ordynamic oxide changes.

Electrochemical investigation therefore is an invalu-able tool for providing insight into chemical reactions,potential activity ranges, kinetics, and parameters(e.g., resistance and thickness) of oxide films. Pan et al.performed both electrochemical and structural studieson the effects of hydrogen peroxide on titanium. Itwas determined that hydrogen peroxide pretreatmenton titanium in phosphate-buffered saline may result ina tenfold thickened oxide film and that large amountsof surface hydroxyl groups and phosphate ions areincorporated into films.2,8,9 Oxides were modeled astwo-layered films comprising an inner dense, high-resistance, low-capacitance TiO2 region, and an outerporous, reduced-capacitance and resistance, nonstoi-chiometric oxide containing inorganic phosphate ions.Re-examination of films treated with hydrogen perox-ide at 30 days revealed increased resistance, presum-ably from oxide pores filling with inorganic materi-als.8

In this study, the admittance response of titaniumover a potential range was analyzed in order to exam-ine corrosion resistance more rigorously. Impedanceand structural data simultaneously were collectedwhile testing titanium oxide films of CPTi and Ti-6Al-4V in PBS and PBS with 30 mM of H2O2. Voltagewas stepped and current data were recorded whileconcurrent imaging captured substrate topographicchanges related to potential and solution environ-

ment. The results suggest that CPTi oxide has a cata-lytic effect on hydrogen peroxide conversion to waterand oxygen that is absent in Ti-6Al-4V. It is possiblethat the catalytic activity impedes the expected forma-tion of TiO2 in aqueous solutions. The study providesa theoretic basis and underpinning for the porousoxide model.

MATERIALS AND METHODS

An experimental method was developed for characteriza-tion of titanium and its oxide film in saline solution. Grade4 CPTi and Ti-6Al-4V discs, 16 mm in diameter and 4 mmthick (DePuy Orthopedics, Warsaw, IN), were attached to15-mm-in-diameter magnetic pucks with Master BondEP76M conductive epoxy (Master Bond Inc., Hackensack,NJ) to permit electropolishing and subsequent transfer to anAFM. The titanium discs were individually electropolished(2.25M of H2SO4 in anhydrous CH3OH, 10 V, �30 s, 25°C)using a rotating disc working-electrode (titanium)-typesetup. Samples were rinsed in ultrapure water (18 M�cm),blown dry with nitrogen, and dipped in Kroll’s solution (3%HNO3, 1% HF in H2O) to bring out the metallic grain struc-ture of the samples before use. All chemicals were pur-chased from Sigma Aldrich (Milwaukee, WI).

Titanium samples were placed in the electrochemical fluidcell in the AFM, as described previously.10 CPTi and Ti-6Al-4V substrates were experimentally tested in situ in PBS(control) and PBS with 30 mM of H2O2 (Brite-Life, Orange,CA) solutions. PBS, a physiologically representative solu-tion, was made by dissolving 9 g of crystalline NaCl in 100mL of 0.4M of PO4 and 900 mL of ultrapure H2O to a pH of7.4. As seen in Figure 1 (a,b), the experimental setup in-cluded a three-electrode electrochemical-testing configura-tion employing titanium as the working electrode (WE), Agwire as the quasi-reference electrode (RE), and Pt wire as thecounter electrode (CE). The CE and RE wires were 0.25 mmin diameter and were 99.99% pure materials. All voltagesreported herein are referenced to Ag wire.

Electrochemical measurements were made using a com-puter-controlled potentiostat (Model 362, EG&G, PrincetonApplied Research, Princeton, NJ) external to the AFM. flex-ible wires (NEF32-2546, 0.7 mm, Cooner Wire, Chatsworth,CA) from all electrodes traveling from the titanium andAFM cell head to the differential electrometer on the poten-tiostat. Thus an internal chamber between the titanium andthe AFM cell head served as the electrochemical and imag-ing cell in all experiments even though the electrochemicaland imaging systems were isolated.

Titanium surfaces (CPTi and Ti-6Al-4V) were first imageddry in contact mode with Digital Instruments software(v.4.24 r4). The images were centered on a distinct surfacefeature, such as a triple point of grain boundaries. ThirtymM of H2O2 in PBS were injected into a fluid cell input portvia a syringe coupled to the port by a luer lock, therebyimmersing 0.5 cm2 of titanium, the imaging tip, RE, and CEin solution. The potentiostat applied 0 V to the titaniumsurface from the instant fluid was injected. The syringe was

EFFECT OF H2O2 ON Ti SURFACES 703

left in place during experimentation, and exit port tubingwas clamped off to provide backpressure.

The AFM tip was re-engaged on the surface, and the exactlocation of dry imaging was located. Concurrent imagingand step-polarization impedance spectroscopy (SPIS) was

performed. Images were acquired (512 � 512 lines) everyfifth step as the voltage was ramped up and down between0 V to �1 V in 50-mV increments. Image acquisition wasperformed in contact mode, typically at 6 Hz. Both heightand deflection modes were used. Experimentation lastedapproximately 1 h per sample.

The SPIS technique has been described previously.11

Briefly, small steps in potential (e.g., 50 mV) were applied toan electrochemical interface, and the resulting current tran-sients were collected using A-D methods. All faradic andnonfaradic processes were monitored. Impedance-based in-formation was deduced from Laplace transform analysis ofthe logarithmically monitored current transients at the elec-trochemical interface in the frequency domain. Data pro-vided minimum and maximum currents per step, high andlow frequency (or early and late) resistances, capacitance,and admittance. Experimental and calculated charge trans-fers per each 50-mV step also were determined for bothtypes of titanium/titanium oxide surfaces tested. All statis-tical calculations were performed using Statistica (Statsoft,Tulsa, OK).

RESULTS

Figures 2 and 3 show sets of 20 �m by 20 �mdeflection and height CPTi and Ti-6Al-4V contactmode AFM images. While deflection images are usefulin illustrating edges, height images are best at convey-ing elevation differences (light is high, dark is low).Thus one of each type and surface is shown for clarity.The Figure 2 CPTi images comprise three hexagonallyclose-packed (HCP) �-grains separated by grainboundaries that form a “Y” and come to a triple pointin the middle of the images. This geometry represents

Figure 1. Schematic illustration of (a) side view of EC AFMsystem design, and (b) underside of electrochemical fluidcell. In this view is a vertically flipped view from (a).

Figure 2. AFM deflection (left) and height image (right) of an electropolished and Kroll-etched commercially pure titanium(CPTi) surface (20 �m � 20 �m) in air. A triple point of grain boundaries and three oxide dome-covered � grains are shown.Note the raised oxide ridges at the grain boundaries.

704 BEARINGER, ORME, AND GILBERT

the lowest energy configuration for grains to join, atapproximately 120° angles with respect to each other.The tiny specks that cover the visual field are protec-tive oxide domes that grant corrosion resistance to thesurface. The lower right grain probably is a differentcrystal face than the other two grains present, as evi-denced by a mix of hemispheric and long spine-shaped oxide domes (seen more clearly in the heightimage). Etching was used to emphasize grain struc-ture, and raised ridges of oxide were seen at the grainboundaries.

Distinct transformed �-grains interrupt the globular� phase across the Ti-6Al-4V surface of Figure 3. The

Ti-6Al-4V surface is bimodal; the transformed � iscomprised of retained � (body centered cubic, BCC)and needle-like acicular �. Thermo-mechanical pro-cessing disrupts acicular �, resulting in the high fa-tigue strength, bi-modal microstructure structure seenin Figure 3. Acicular � is difficult to detect in thisimage, but it is easier to detect in subsequent images athigher magnification. AFM images complement pre-viously acquired SEM images of similar titanium alloysurfaces with added height resolution.12

Figures 4 and 5 show 5 �m by 5-�m images of oxidedome-covered CPTi and Ti-6Al-4V surfaces in 30 mMof H2O2 in PBS (deflection images). CPTi images are

Figure 3. AFM deflection (left) and height image (right) of an electropolished and Kroll-etched titanium, 6-aluminum,4-vanadium (Ti-6Al-4V) surface (20 �m � 20 �m) in air. The Ti-6Al-4V contains the � phase covered with oxide domes,occasionally interrupted by the � phase sitting higher on the surface. Also note the differences (especially in the deflectionimage) in oxide morphology on different � grains, presumably due to substrate crystal orientation differences.

Figure 4. AFM deflection images (5�m � 5�m) of electropolished and Kroll-etched commercially pure titanium (CP Ti)hydrated in PBS/H2O2. Images include the same triple point of grain boundaries: (a) CPTi surface at 0 V upon hydration; and(b) CPTi surface after voltage had been ramped linearly up to 1 V and then back down to 0 V. Arrows point to the same oxidedomes (a) before and (b) after potential sweep.

EFFECT OF H2O2 ON Ti SURFACES 705

centered on the same triple point of grain boundariesas in Figure 2, and Ti-6Al-4V images are centered on afew � grains surrounded by predominant � grains.The largest � grain also shows some acicular � ridingdown its left and lower regions. The first images in theseries are from initially hydrated conditions at 0 V; thesecond images are from �30 min after the potential hadbeen ramped up to 1 V and then back down to 0 V.

As discussed previously for titanium, lateral oxidedome growth increases with potential but does notdecrease once potential is ramped or stepped cathod-ically.13 Starting roughness (Ra) was 8.9 nm and de-creased to 7.4 nm at 1 V for CPTi. Upon ramping thepotential back down to 0 V, the roughness was 7.5 nm.It is believed that changes in absolute height take placeunder cathodic potentials, but this cannot be con-firmed at present. The way experiments were con-ducted allowed only for relative height comparisons,not absolute height comparisons. Some tip convolu-tion is assumed in these images, making some featuresappear slightly more rectangular than they probablyare. This cannot be fully eliminated and is a functionof tip size and geometry.

The Ti-6Al-4V surface initially hydrated in 30 mMof H2O2 in PBS at 0 V appears on the left in Figure 5,and the same surface after �30 min, when the poten-tial had been ramped up to 1 V and then back down to0 V, appears on the right. The Ti-6Al-4V images arecentered on a few � grains surrounded by predomi-nant � grains. Impingement on and coarsening ofoxide domes is noted in the � phase but seems to beabsent in the � phase. Roughness measurements (Ra)at starting conditions (0 V), 1 V, and upon returning to0 V were 15.0 nm, 13.1 nm, and 13.9 nm, respectively.As with CPTi, a slight increase in roughness uponcathodic ramping between 1 V and 0 V was noted.

The relative height and width of the main � grain inthe approximate center of the above images weretracked with potential. Initially, the � feature was 87.8nm tall and 1.436 �m wide. Upon anodic ramping to 1V, the feature was 87.5 nm tall and 1.553 �m wide.Cathodic ramping back to 0 V resulted in a � grainheight of 87.6 nm and a width of 1.543 �m. Theseresults indicate a nominal change in the dimensionsfor the � grain examined in H2O2/PBS.

Oxide dome diameter as a function of potential forCPTi is plotted in Figure 6, and dome diameter as afunction of both potential and time is plotted in Figure7. Dome diameters (n 20) were measured usingDigital Instruments software. Figure 6 was fit usinglinear regression methods and resulted in a line char-acterized by Y 38.39X � 183 (R2 0.90). The resultsshown in Figure 7 indicate that dome diameter in-

Figure 5. AFM deflection images (5 �m � 5 �m) of electropolished and Kroll-etched titanium, 6-aluminum, 4-vanadium(Ti-6Al-4V) hydrated in PBS/H2O2. Images include the same central � grains surrounded by a predominant � phase: (a)Ti-6Al-4V surface at 0 V upon hydration; and (b) Ti-6Al-4V surface after voltage had been linearly ramped up to 1 V and thenback down to 0 V.

Figure 6. Oxide dome diameter for CPTi as a function ofapplied potential in 30 mM of H2O2 in PBS.

706 BEARINGER, ORME, AND GILBERT

creased with both potential and time. Linear regressionalso was used to fit time versus dome diameter, accord-ing to the equation Y 3.061X � 174.6 (R2 0.92). Ifmeasured diameters are converted into areas and linearfits again are determined, the associated slopes are12,084 nm2/V and 951 nm2/min. It is interesting to notethat the rate of oxide dome growth with respect to timeis slightly less than that of the rate of oxide dome growthin straight PBS at OCP conditions (1084 nm2/min10). Thefact that a rate of 951 nm2/min was obtained whilepotential was increased one volt suggests that this ratewould be even slower at OCP conditions.

Polarization and impedance data are plotted in Fig-ure 8 for CPTi and in Figure 9 for Ti-6Al-4V. All plotsshow data for experiments conducted in both PBS(control state) and PBS with H2O2. Figure 8 and Figure9(a) indicate maximum (50 �s) and minimum (10 s)currents versus applied potential. Solutions with hy-drogen peroxide show considerable minimum currenthysteresis for both surfaces. Currents from anodicramping of potential in H2O2-containing solutionswere an order of magnitude larger than currents gen-erated under control (PBS) conditions. Maximum cur-rents were on the same order of magnitude betweenthe two solutions, but these currents were still slightlyelevated in hydrogen peroxide. A bubble formed nearthe tubing during potential step 15 of 40 (�0.75 V) onthe Ti-6Al-4V surface, throwing off current data. Thedata from this step were discarded.

Plots in Figures 8(b) and 9(b) contain early resis-tance (Re) versus potential data, and associated plots[Fig. 8(c) and Fig. 9(c)] contain polarization resistance(Rp) versus potential data for CPTi and Ti-6Al-4V sur-faces, respectively. Early resistance also is known assolution resistance. However, we have shown that

early time resistance, as measured by SPIS, reflectsoxide passivity characteristics as well, so the termearly resistance is substituted. On the CPTi surface[Fig. 8(b)] resistances rose higher in pure PBS solutionand remained high as potential was decreased to 0 V.The solution with H2O2 yielded a relatively flat Re

curve on the CPTi surface.The Ti-6Al-4V surface [Fig. 9(b)] presented consid-

erably shifted data; data from both solutions mim-icked each other, and ramped up to 1 V, remainedhigh from 1 V to about 0.4 V and then fell to 0 V again.The Ti-6Al-4V surface exposed to H2O2 yielded datathat lies inside the data from the surface exposed topure PBS. These data suggest that hydrogen peroxideaffects the oxide of pure titanium differently than itaffects the alloy surface oxide, even though the Ti-6Al-4V oxide is thought to be predominantly TiO2. Itis notable that although the maximum currents aresimilar, Re versus potential indicates differentlyshaped curves and values for the two surfaces andsolutions. Figures 8(c) and 9(c) plotted Rp (low fre-quency or late resistance) versus potential for CPTiand Ti-6Al-4V, respectively, as relatively flat, so theywere compared statistically. No significant differencesbetween PBS and PBS/H2O2-treated surfaces data(ANOVA, post hoc Newman-Keuls) were indicated.

Capacitance data were noisier for the CPTi surfacebut indicated higher capacitance values for surfacesexposed to hydrogen peroxide. This elevated capaci-tance was significant for the CPTi surface within theCPTi data sets and among the Ti-6Al-4V surfaces (p 0.04, ANOVA, post hoc Newman-Keuls). Greater ca-pacitance probably results from PBS treatment grant-ing a thicker resultant oxide. This is consistent withthe Re versus the potential data in Figure 8(b), whichalso suggests a thicker resultant oxide treated in PBS.The two sets of Ti-6Al-4V surface capacitances werenot significantly different.

Figures 10 and 11 contain 3D mesh plots of (a) in-phase admittance and (b) out-of-phase admittance as afunction of frequency and potential. CPTi admittance inPBS is plotted in Figure 10 and CPTi admittance inPBS/H2O2 is plotted in Figure 11. Associated Ti-6Al-4Vadmittance curves exhibited similar behavior (data notshown). In-phase admittance (A�) was greatest at lowvoltage and high frequency. Curves dropped off withslightly increasing voltage and reduced to 0 with de-creasing frequency. Hydrogen peroxide appears to leveloff A� values with respect to voltage, as seen in a com-parison of CPTi in Figures 10 and 11. The hydrogenperoxide-treated surfaces had a tighter admittance rangeacross voltage than did the surfaces exposed to purePBS, but they indicated a more linear response withfrequency and rolled off roughly half as fast.

Out-of-phase (or loss) admittance (A�) responseformed characteristic 3D bell-shaped curves for CPTisamples subjected to both solutions. As seen in the

Figure 7. Oxide dome diameter for CPTi as a function ofpotential and time in 30 mM of H2O2 in PBS.

EFFECT OF H2O2 ON Ti SURFACES 707

in-phase admittance curves, the peak values werehigher in pure PBS solution while the response wasmore level across voltage in H2O2-containing solution.The PBS/H2O2 solution yielded lower A� values thanwere yielded in pure PBS solution. The data for purePBS also were noisier with respect to frequency.

The degree of similarity between the surface admit-tance curves for Ti-6Al-4V (not shown) was higherthan that of the CPTi curves for samples tested in thetwo types of solutions. Peak values were much higherfor Ti-6Al-4V than for CPTi, reflecting the lower oxideresistances noted earlier. Within the two Ti-6Al-4Vin-phase admittance plots, A� values were slightlymore elevated in pure PBS than in PBS/H2O2 solution.

Measured charge (i.e., integral of current transients)versus a 50-mV step number and applied potential isplotted in Figure 12 for Ti-6Al-4V exposed to PBS andPBS/H2O2. The results were similar for CPTi surfacesand Ti-6Al-4V surfaces. A representative plot is shown.

Measured charge was orders of magnitude higher forthe Ti-6Al-4V sample in the PBS solution containingH2O2 in positive steps between 0 V and 1 V. Charge thendiminished for the cathodic voltage steps back down to0 V. By �0.5 V, charge had reduced to the level of thePBS-treated sample. Calculated charge per step was 106�C/cm2, so it is notable that the high anodic region ofthe PBS/H2O2 curve contained values above the calcu-lated value (see caption in Fig. 12). Statistical analysis(ANOVA, post hoc Newman-Keuls) revealed that chargefrom both CPTi and Ti-6Al-4V surfaces exposed to hy-drogen peroxide were statistically different from the twosurfaces exposed only to PBS (p 0.0006)

DISCUSSION

Simultaneously probing structural and electricalcharacteristics of titanium oxide films from pure and

Figure 8. Polarization plot and impedance characteristics for CPTi in PBS and PBS/H2O2. Polarization plot (a) indicatesmaximum currents (50 ms after potential steps) and minimum currents (10 s after potential steps) versus voltage. Highfrequency (early time) resistance Re � 1/�A��3�� versus voltage is plotted in (b); polarization resistance (high-frequencyimpedance–low-frequency impedance) Rp � 1/�A��30 � 1/�A��3�� versus voltage is plotted in (c); interfacial capacitance C � �Rs � Rp�/RsRp�A, where �A is the frequency where there is a peak loss admittance, A�] versus voltage is plotted in (d).

708 BEARINGER, ORME, AND GILBERT

alloyed titanium exposed to PBS and PBS containinghydrogen peroxide allowed direct observation of lateraldome spreading, the voltage and time dependence ofoxide processes, and notable surface roughness changes.Polarization, impedance, and surface structure behaviorwere concomitantly reported after simultaneously mon-itoring the pure and alloyed titanium electrode samples.The impedance measurements were collected using SPISand provided concurrent determination of polarizationand electrochemical impedance behavior at the electro-chemical interfaces as a function of potential; surfacestructure behavior was collected using AFM in dry andimmersed environments.

EC AFM images from titanium surfaces exposed toH2O2 in PBS indicate that anodic potential rampinginfluences oxide dome growth, but subsequent ca-thodic ramping does not reduce dome size. This phe-nomenon was noted earlier in pure PBS solutions andrelates to the classification of TiO2 as an n-type semi-

conductor.10,13 N-type semiconductors are thought togrow predominantly at the metal/oxide interface, asopposed to the oxide/solution interface.

In comparing the PBS/H2O2- and pure PBS-im-mersed surfaces, less dome coalescence was noted onthe surfaces exposed to hydrogen peroxide. Lesscoarsening suggests more active sites and more corro-sion susceptibility. Reduced coarsening occurred onboth CPTi and Ti-6Al-4V surfaces and is thought to bedirectly due to the presence of hydrogen peroxide.

These studies also indicated slight decreases in sur-face roughness upon ramping potential anodicallyand slight increases upon ramping potential cathodi-cally. In contrast, STM images previously acquired byPan et al. and Ejov et al. showed more substantialsurface roughening upon adding hydrogen peroxideto PBS.1,8 It is thought that this is because their exper-iments took place over a longer time interval than didthe experiments in this study.

Figure 9. Polarization plot and impedance characteristics for Ti-6Al-4V in PBS and PBS/H2O2. Polarization plot (a) indicatesmaximum currents (50 ms after potential steps) and minimum currents (10 s after potential steps) versus voltage. Highfrequency (early time) resistance versus voltage is plotted in (b); polarization resistance (high-frequency impedance–low-frequency impedance) versus voltage is plotted in (c); interfacial capacitance, C, versus voltage is plotted in (d).

EFFECT OF H2O2 ON Ti SURFACES 709

With regard to electrical effects, the roles of currentand charge transfer on the titanium surfaces have beenconsidered. Corrosion currents from both CPTi andTi-6Al-4V surfaces immersed in PBS/H2O2, taken 10 safter the potential was stepped, show hysteresis be-tween anodic and cathodic scans for late currents of anorder of magnitude (see Figs. 8 and 9). Charge transferincreased concurrently for anodic scans (see Fig. 12),suggesting oxide growth. Oxide growth is thought toincrease oxide resistance, which should have beenreflected in Re versus V plots. However, neither of thePBS/H2O2-immersed surfaces showed increased Re ascompared to the pure PBS-bathed surfaces.

Resistance values from the Ti-6Al-4V surfaces al-most overlay one another although admittance resultssuggest a slightly lower resistance. The CPTi Re val-ues, in contrast, are almost flat, with voltage in thePBS/H2O2 data set. CPTi samples in both environ-ments possessed higher resistances than did the Ti-6Al-4V samples. Both titanium metal compositionsshowed higher charge transfer in H2O2 but possessed

varying electrical properties, which suggests that thehigher residual currents originate from increased ox-ygen formation in solution: oxidation

H2O2 3 O2 � 2H� � 2e� (1)

or reduction at neutral pH,

H2O2 � 2e�3 2OH� (2)

or reduction followed by oxidation at acidic pH.

H2O2 � 2H� � 2e�3 2H2O3 O2 � 4H� � 4e� (3)

A charge exceeding calculated values (see Fig. 12)implies oxidation of hydrogen peroxide to molecularoxygen. A comparison of other electrical properties ofthe two titanium alloys is expanded below.

The very different CPTi and Ti-6Al-4V peroxide-influenced Re curves suggest that different mecha-nisms control the two oxide responses. Peroxide af-fects the resistive behavior of CPTi more than it does

Figure 10. Three-dimensional plot of (a) in-phase (A�)and(b) out-of-phase (A�) admittance of CPTi in PBS versus fre-quency (rad/s) and potential (V).

Figure 11. Three-dimensional plot of (a) in-phase (A�)and(b) out-of-phase (A�) admittance of CPTi in PBS/H2O2 ver-sus frequency (rad/s) and potential (V).

710 BEARINGER, ORME, AND GILBERT

the alloy. A reduction in the Re of the oxide on CPTi isconsistent with the porous oxide model explained byPan, who suggested that when left for 30 h in PBS/H2O2, the titanium “self-heals” and resistance increas-es.9 The rationale of the Pan study was that the chem-ical potential of H2O2 drives Ti4� ion diffusionthrough the oxide, leaving OH� and PO4� ions on thesurface. Cations such as Ca2� then may migrate intothe oxide. Theoretically, oxide pores seal with biologicmolecules and thereby bond more strongly to bone.This supports titanium’s reputation as biocompatiblewithout contradicting the electrical data. The SPIStechniques described herein enabled resistance deter-mination over a range of voltages, leading to a morein-depth theory consistent with the porous oxidemodel.

This theory is introduced with the consideration ofH2O2 activity on titanium in the body. In the body,electrons may reduce oxygen and generate harmfulspecies, such as superoxide anion radical and hydro-gen peroxide:

O2 � e�3O2•� (4)

H� � O2•� 3 HO2

• (5)

HO2• � H� 3 H2O2 (6)

These molecules are necessary to kill off microbeswith their oxidative powers. As they oxidize foreignmolecules, they must be reduced. The superoxide an-ion radical may be reduced to hydroxyl radicals, and

hydrogen peroxide may be reduced to water or hy-droxyl radicals, as seen in Equations (2) and (3). Be-cause wound and infected environments typically areacidic, it is thought that the water-generating reactionwould dominate in the body.

While oxidizing molecules protect against foreignspecies, they also may damage host cells and or-ganelles. The body therefore has defensive mecha-nisms for removing or reducing activity of reactivespecies, including biologic enzymes, such as superox-ide dismutase (SOD, a primary oxygen product ofpolymorphonuclear leukocytes involved in inflamma-tory response) and catalase.14

oxygen radicals � SOD3H2O2 (7)

2H2O2O¡catalase

2H2O � O2 (8)

Titanium shares a common link with these en-zymes. Commercially pure titanium’s photoconduc-tivity converts H2O2 to less harmful species. UV illu-mination of titanium above the bandgap energy (3.2eV), at potentials above the flatband potential, cata-lyzes decomposition of water into molecular oxygenand hydrogen.15 Small amounts of hydrogen peroxidemay be generated in an intermediate step:

2H2OO¡TiO2

2OHS• � 2H� � 2e� (9)

2OHS•O¡

TiO2

(H2O2) (10)

and converted in a final step to oxygen, hydrogen, andelectrons, as in Equation (1). Tafalla stated that thesereactions will proceed without illumination, but withlower efficiency.15

Tengvall, however, found that titanium catalyticallydecomposes hydrogen peroxide and titanium perox-ide at appreciable rates without illumination andwithout giving rise to detectable amounts of hydroxylradicals.16 He reported superoxide oxidation from agel made with metallic titanium and hydrogen perox-ide. Relatively stable and nontoxic intermediate tita-nium–peroxy compounds, such as Ti4�O2

2�, also werefound at the surface.

We believe that TiO2, in concert with the body,catalyzes removal of reactive oxygen species from itslocal environment. As discussed previously, SPIS elec-trical resistance results suggest that the CPTi oxideand Ti-6Al-4V oxide function differently: admittancedata on the Ti-6Al-4V oxide appears relatively unaf-fected by hydrogen peroxide. The early resistance ver-sus potential plot for the CPTi surfaces [Fig. 8(b)]shows that resistances measured in both the PBS and

Figure 12. Charge passed through Ti-6Al-4V oxide films inPBS and PBS/H2O2 as a function of stepped voltage. Calcu-lated charge is: Q � �X�ZFA/MW, where �X � * (V �Vi) � is the anodization rate for titanium (2 nm/V), and V �Vi is 50 mV; � is density of the oxide (4.4 g/cm3); Z is valencestate of the oxide (�4); F is Faraday’s constant (96,480c/mole); A is electrode area (0.5 cm2); and MW is the mo-lecular weight of TiO2 (79.9 g/mole). The area under thecurrent transient curves equals experimental charge and wasdetermined by trapezoid approximation.

EFFECT OF H2O2 ON Ti SURFACES 711

PBS/H2O2 solutions start near the same value. PBSresults show an expected increase in resistance withpotential, which most likely corresponds to TiO2growth. PBS/H2O2 results indicate that film resistanceis substantially independent of potential; however,capacitance data still indicate film growth. This verysurprising result suggests that the film grown uponexposure to PBS/H2O2 solution is not, in fact, TiO2,but rather a hydroxy-titanium- and titanium-peroxy-rich layer that is porous and therefore incapable ofproviding passivity until other biolayers stabilize thesurface. Thus the only material on the sample surfacewith a stoichiometry similar to TiO2 results from thenative oxide present before immersion.

Increased oxide growth on CPTi, compared to Ti-6Al-4V is coupled to microstructure. Ti-6Al-4V’s oxideis thought to be almost the same as CPTi’s in compo-sition.17–19 The underlying microstructure and poros-ity, however, differ. Our results show little change insurface oxide dimensions on � grains of Ti-6Al-4V.Oxygen is more soluble in � titanium than in � tita-nium because of larger internal spaces in the �phase.20 In addition to intermediates formed on CPTi,H2O2 catalysis on its TiO2 film yields a predominanceof oxygen in the local environment. This increases theelectrochemical oxygen gradient through the oxideinto the substrate. While oxygen content also mayincrease slightly in the environment surrounding Ti-6Al-4V surfaces exposed to H2O2, decreased catalyticactivity will result in a decreased electrochemical gra-dient.

CONCLUSIONS

EC AFM and SPIS were simultaneously performedon CPTi and Ti-6Al-4V oxide surfaces exposed to PBSand PBS with 30 mM of H2O2. EC AFM results re-vealed less oxide dome coarsening on oxides exposedto hydrogen peroxide. Charge transfer was an order ofmagnitude higher for H2O2-containing solutions, par-tially because of oxygen evolution. While hydrogenperoxide did not affect Ti-6Al-4V surface electricalproperties, it did greatly affect CPTi oxide resistanceand capacitance between 0 V and 1 V. Intermediates,such as hydroxy and peroxy compounds, formed inthe catalytic conversion of H2O2, compose oxidesgrown in solution containing hydrogen peroxide onthe CPTi surface, while oxides closer to TiO2 form inperoxide-free aqueous solutions.

The authors thank Robert Haskins at DePuy for the com-mercially pure titanium and the titanium, 6-aluminum, 4-va-nadium.

References

1. Ejov AA, Savinov SV, Yaminsky IV. Ex situ scanning tunnelingmicroscopy investigations of the modification of titanium sur-face due to corrosion processes. J Vac Sci Technol B 1994;12:1547–1550.

2. Pan J, Thierry D, Leygraf C. Electrochemical and XPS studies oftitanium for biomaterial applications with respect to the effectof hydrogen peroxide. J Biomed Mater Res 1994;28:113–122.

3. Brown GM, Thundat T, Allison DP, Warmack RJ. Electrochem-ical and in situ atomic force microscopy and scanning tunnel-ing microscopy investigations of titanium in oxalic acid solu-tion. J Vac Sci Technol A 1992;10:3001–3006.

4. Bullock EL, Patthey L, Steinemann SG. Clean and hydroxy-lated rutile TiO2 (110) surfaces studied by X-ray photoelectronspectroscopy. Surface Sci 1996;352–354:504–510.

5. Sundgren J-E, Bodo P, Lundstrom I. Auger electron spectro-scopic studies of the interface between human tissue and im-plants of titanium and stainless steel. J Colloid Interface Sci1986;110:9–20.

6. Effah EAB, Bianco PD, Ducheyne P. Crystal structure of thesurface oxide layer on titanium and its changes arising fromimmersion. J Biomed Mater Res 1995;29:73–80.

7. Healy KE, Ducheyne P. Oxidation kinetics of titanium thinfilms in model physiologic environments. J Colloid InterfaceSci 1991;150:404–417.

8. Pan J, Thierry D, Leygraf C. Hydrogen peroxide toward en-hanced oxide growth on titanium in PBS solution: Blue colora-tion and clinical relevance. J Biomed Mater Res 1996;30:393–402.

9. Pan J, Liao H, Leygraf C, Thierry D, Li J. Variation of oxidefilms on titanium induced by osteoblast-like cell culture andthe influence of an H2O2 pretreatment. J Biomed Mater Res1998;40:244–256.

10. Bearinger JP, Orme CA, Gilbert JL. Direct observation of hy-dration of TiO2 on Ti using AFM: Freely corroding versuspotentiostatically held. Surf Sci 2001;491:370–387.

11. Gilbert JL. Step-polarization impedance spectroscopy of im-plant alloys in physiologic solutions. J Biomed Mater Res 1998;40:233–243.

12. Gilbert JL. Basic sciences: Metals. In: Callaghan JJ, editor. Theadult hip, Vol. 1. Philadelphia: Lippincott-Raven; 1998. p 123–133.

13. Bearinger JP, Orme CA, Gilbert JL. In-situ imaging and imped-ance measurements of titanium surfaces using AFM and SPIS.Biomaterials 2003;24(11):1837–1852.

14. Stryer L. Biochemistry. New York: W. H. Freeman; 1988.15. Tafalla D, Salvador A. Photosynthetic production of H2O2 from

water at n-TiO2 electrodes in a photoelectrochemical cell: In-fluence of electrolyte pH and Pt electrodeposition. J ElectroanalChem 1987;237:225–236.

16. Tengvall P, Elwing H, Lundstrom I. Titanium gel made frommetallic titanium and hydrogen peroxide. J Colloid InterfaceSci 1988;130:405–413.

17. Ask M, Lausmaa J, Kasemo B. Preparation and surface spec-troscopic characterization of oxide films on Ti6Al4V. Appl SurfSci 1988;35:283–301.

18. Lausmaa J, Kasemo B, Mattsson J, Odelius H. Multi-techniquesurface characterization of oxide films on electropolished andanodically oxidized titanium. Appl Surf Sci 1990;45:189–200.

19. Ask M. Rolander J, Lausmaa J, Kasemo B. Microstructure andmorphology of surface oxide films on Ti-6Al-4V. J Mater Res1990;5:1662–1667.

20. Evans UR. The corrosion and oxidation of metals. London:Arnold; 1960.

712 BEARINGER, ORME, AND GILBERT