Electron transfer processes occurring on platinum neural stimulating electrodes: calculated charge-storage capacities are inaccessible during applied stimulation

A steady-state i-E curve for a platinum electrode in N₂-saturated 0.15 M H₂SO₄ over the entire water window (−0.25 V to  +1.25 V versus Ag|AgCl) is presented in figure 2(a). The potential was swept at a constant rate of 100 mV s⁻¹ between the positive and negative limits while the current was being recorded. For a comprehensive explanation of the theory behind i-E curves (or i-V_e profiles, cyclic voltammograms) and how to record and interpret them, see our previous publication [34]. Typical platinum behavior was indicated by the observed current peaks: oxide formation and reduction, atomic hydrogen (H) adsorption and desorption, and molecular hydrogen (H₂) evolution and desorption. For the i-E curve in figure 2(a), the cathodic charge transfer associated with double-layer charging and the hydrogen adsorption reactions in the region between  −0.22 V and  +0.05 V (blue shaded area) was calculated to be 698 µC cm⁻². This indicates a roughness factor of about 3, assuming a monolayer of adsorbed atomic hydrogen corresponds to 210 µC cm⁻² [26, 27].

In figure 2(b), the steady-state i-E curve for a platinum electrode in O₂-saturated 0.15 M H₂SO₄ over the entire water window is shown. In the O₂-saturated environment, at potentials negative of the OCP (+0.76 V), platinum reduces dissolved O₂ molecules to form the superoxide radical ($\text{O}_{2}^{\bullet -}$) [35–37] or other species like H₂O₂. This reduction reaction provides a source of negative current that shifts the i-E curve downward in this potential range. The upper limit of the i-E curves in figure 2 was chosen to be  +1.25 V to prevent the formation of molecular oxygen (O₂), which occurs at potentials above this limit.

Shown in figure 3(a) is an i-E curve of a platinum electrode scanned from the mean OCP to the negative limit of the water window (approximately  +0.55 V to  −0.25 V) for 100 cycles in N₂-saturated H₂SO₄. Cycling through this potential range more accurately reflects the primary potential window platinum electrodes encounter during neural stimulation pulsing. Prior to initiating this sequence of i-E curves, the as-manufactured electrode had been stored in air, rinsed with deionized distilled water, then submerged in the N₂-saturated electrolyte. In the first cycle (bold gray trace), hydrogen (H) adsorption–desorption peaks were relatively undefined compared to the same region (approximately  −0.22 V to  +0.05 V) in figure 2(a) when scanning over the wider window (−0.25 V to  +1.25 V). At the negative end of the i-E curve, molecular hydrogen (H₂) is formed from the reduction of water molecules. In cycle 2 (first of the thin black traces), the cathodic charge density associated with double-layer charging and hydrogen adsorption was 438 µC cm^−2 6. As scanning progressed, the current associated with hydrogen adsorption steadily decreased, and by the 100th cycle (bold black trace; deemed as steady-state), cathodic charge density had decreased to 326 µC cm⁻². A similar decrease in hydrogen adsorption current was observed in O₂-saturated sulfuric acid (figure 3(b)).

After the 100th cycle of figure 3(a), the electrode was then scanned over the entire water window (−0.25 V and  +1.25 V), and the resulting i-E curves are shown in figure 4(a). By polarizing the electrode through the oxide formation region (> +0.55 V) repeatedly, charge transfer in the hydrogen (H) adsorption/desorption region gradually increases over time. By the 50th cycle (bold black trace), the i-E curve has reached a steady state, and the hydrogen adsorption–desorption peaks became well defined. The cathodic charge in the hydrogen adsorption region (−0.22 V to  +0.05 V) in the last cycle, deemed the 'cleaned state', was calculated to be 698 µC cm⁻². This last cycle is the i-E curve presented in figure 2(a).

The i-E curve of platinum in O₂-saturated H₂SO₄ between  −0.25 V and  +1.25 V is shown in figure 4(b). Similar to the N₂-saturated case, as cycling over the entire water window progressed, the H adsorption current increased to a steady-state value within 50 cycles. This last cycle is the i-E curve presented in figure 2(b).

In figure 5(a), i-E curves for the same platinum electrode in N₂-saturated H₂SO₄ immediately after the electrode had been cycled through the oxide formation region to a steady state (after 50th cycle in figure 4(a)) are shown. In the first cycle (bold gray trace) over the narrower potential window, the charge density for H adsorption was 690 µC cm⁻². As scanning progressed, however, the current associated with H adsorption steadily decreased. Charge transfer in the H adsorption region decreased from 690 µC cm⁻² to 435 µC cm⁻² by the 100th cycle (bold black trace). Comparing this to the fully developed electrode represented in figure 4(a), there was about a 38% decrease in H-adsorption charge transfer when operating platinum in the narrower window.

i-E curves of the platinum electrode in O₂-saturated H₂SO₄ over a sequence of 100 cycles in the narrow potential window (immediately after the 50th cycle in figure 4(b) are presented in figure 5(b). Over the progression from the first cycle (gray) to steady state (black), the current associated with H adsorption decreased similarly to the N₂-saturated case.

Beyond the protons, hydroxide ions, water molecules, and dissolved O₂ studied in diluted sulfuric acid, two important components of living tissue should be considered when analyzing electrodes for neural stimulation: proteins and chloride ions (Cl⁻). Both can have significant effects on the electrochemical activity of the electrode and may alter the amount of charge that can be injected through platinum.

Previously, we have shown that phosphate-buffered saline electrolytes may provide inaccurate electrochemical results since the phosphate groups are electrochemically active in certain potential regions [28]. So, we attempted to evaluate the electrochemical response of platinum in an unbuffered saline solution as a vehicle for examining the effect of Cl⁻. i-E curves of a platinum disk electrode in 0.15 M H₂SO₄ (black trace) and in unbuffered 0.15 M NaCl (blue traces) are overlaid in figure 6. Several visual differences between i-E curves in the two electrolytes are apparent. One major difference is that the sulfuric acid i-E curve achieved a stable steady state, whereas the unbuffered NaCl i-E curves continually changed over time. The oxide reduction current decreased with time and an additional small reduction peak began to appear near  +0.50 V (see red arrows). There was also a difference in the apparent water window of platinum in these electrolytes. The negative limit of the i-E curves in unbuffered NaCl is shifted 0.55 V negative of the i-E curve in sulfuric acid. This shift is due to the difference in pH between sulfuric acid (pH  =  0.9) and unbuffered NaCl (pH ~7, unstable) [38]. The overall width of the water window was larger in unbuffered NaCl (1.8 V) than in sulfuric acid (1.5 V). Also, the shapes and amplitudes of the hydrogen adsorption–desorption peaks differ. The differences observed between dilute sulfuric acid and unbuffered NaCl may be explained by variations in pH [38, 39] and ion adsorption effects, which are further studied below. Because of the instability of the i-E curve in unbuffered NaCl, it is not normally used as a test electrolyte for careful analyses of platinum (or other metal) electrochemistry. As discovered previously, even phosphate-buffered NaCl electrolytes exhibit electrochemical reactions that may not occur in vivo [28].

Zoom In Zoom Out Reset image size

In a related experiment, a platinum electrode in sulfuric acid was first scanned through the water window until steady state was reached (figure 7(a), black trace). Then, aliquots of 0.15 M NaCl were added to determine how the chloride ion (Cl⁻) affects the electrochemical activity. Upon addition of 1 mL of the NaCl electrolyte (final NaCl concentration of approximately 3 mM) and cycling to steady state, the current response for oxide formation decreases in the potential range from  +0.55 V to  +0.90 V, but appears relatively unchanged above  +0.90 V. Also, the relative amplitudes of the hydrogen adsorption peaks change. Addition of 5 ml NaCl (final concentration of approximately 15 mM) resulted in an additional reduction in oxide formation current, and slight changes to the hydrogen adsorption peaks.

Zoom In Zoom Out Reset image size

Further insight on the behavior of platinum electrodes can be gained from the work by Robblee et al [40]. Robblee showed that under pulsed conditions, less platinum was detected in solution when human serum albumin (a protein) was added to a phosphate buffered saline electrolyte (i.e. less platinum dissolution occurred). To gain a better understanding of this observation, bovine serum albumin (BSA) solution was added incrementally to sulfuric acid electrolyte and its effect was observed (figure 7(b)). Serum albumins are globular proteins and human serum albumin is a prominent constituent of cerebrospinal fluid. Shown in figure 7(b), a 1 ml addition of BSA solution results in current suppression over most of the oxide formation region (+0.55 V to  +1.20 V) Additional amounts of BSA (up to 9 mL) did not result in a significant further change in the i-E curve. The approximate final concentrations of BSA in solution for the 1, 2, 5, and 9 ml additions were 4, 8, 20 and 36 µg ml⁻¹, respectively. With BSA and other organic molecules, the suppression of oxide formation and H adsorption has been reported as an overall surface blocking effect [41, 42], where the organic molecules physically block platinum surface sites from participating in redox reactions.

In this study, dilute sulfuric acid was used as the electrolyte for platinum because it is a relatively well-understood electrochemical electrode-electrolyte system. Also, it contains protons, hydroxide ions, and water molecules, all of which are present in any living system. Protons and hydroxide ions are present in both sulfuric acid and in vivo, although in different proportions, giving rise to a pH difference and the resulting shift in potential. The sulfate ($\text{SO}_{4}^{\text{2}-}$) and bisulfate ($\text{HSO}_{4}^{-}$) anions in sulfuric acid adsorb to the platinum surface, but do not have a significant inhibiting effect on oxide formation or hydrogen (H) adsorption [41, 43–45]. Understanding the electron transfer processes involving protons and water provides the basis for adding other candidate reactants to the test electrolyte. In our experiments in unbuffered NaCl, it was not possible to obtain a well-developed and stable i-E curve that can be used as a standard (see figure 6). In our previous study in phosphate-buffered saline, it was clear that the phosphate components can be electrochemically active [28]. However, in a sulfuric acid electrolyte, a stable i-E curve could be achieved relatively easily and its characteristics matched very well to platinum/sulfuric-acid i-E curves in literature [42, 46–50]. This stable curve provides a baseline electrochemical response of platinum to which other test components can be added to solution for analysis.

When an electrode is submerged in an electrolyte, the electrode potential stabilizes at the point (known as a mixed potential) where current for cathodic processes is balanced with current for anodic processes, resulting in zero net current across the interface. This stable point is called the open-circuit potential (OCP). For an electrolyte containing O₂, current for O₂ reduction (a cathodic process) is balanced with current from anodic processes. Hoare et al deduced that the OCP is a mixed potential arising from oxygen reduction and platinum oxide formation reactions [51]. An analysis by Rand and Woods showed that the likely anodic processes were platinum dissolution and oxidation of organic contaminants ('impurities') in solution [46]. Without O₂ present, there is no cathodic current from oxygen reduction, so the electrode rests at a more negative potential because the electrode must find a cathodic process with which to balance the anodic current from dissolution and organic oxidation. This shift was observed in our experiments (see figure 1 and table 1), where the OCP in O₂-saturated electrolyte was positive (anodic) to the OCP in deoxygenated electrolyte. In this study, O₂ reduction was observed at all potentials negative to the OCP when O₂ was present. With cathodic-first neural stimulation starting from OCP, the electrode will travel through this potential region where O₂ reduction is thermodynamically favorable. Therefore, future electrochemical studies relating to neural stimulation should consider an O₂ concentration near that found in neural environments (~5%) [52]. The consequences of oxygen reduction can be the generation of reactive oxygen species, such as superoxide and hydrogen peroxide, which may not be reversible upon the application of the positive phase of the biphasic pulse. Superoxide radicals are known to react with nitric oxide (NO) to form peroxynitrite, which can be toxic to cells in the vicinity of the stimulating electrode [53–55]. Because NO is a prominent vasodilator, depletion of this molecule through the reaction with superoxide may result in vessel constriction near neural tissue. However, reactive oxygen species are also a consequence of normal metabolic processes and scavengers are naturally present to mitigate them.

When a platinum i-E curve appears in a publication concerning neural stimulation, it is almost always a steady-state i-E curve in an oxygen-free electrolyte, as shown in figure 2(a). It is important to realize that, prior to recording this i-E curve, the electrode potential had been swept many times over the entire potential range shown on the potential scale, as shown in figure 4(a). With each sweep cycle, more and more charge can be reversibly transferred across the electrode interface until little or no increase is observed with additional cycling. The i-E curve at the end of the cycling process is referred to as a steady-state i-E curve or steady-state cyclic voltammogram (CV). Reaching this steady state takes several minutes of cycling, whereas a neural stimulation pulse may only last microseconds. In the case of figure 4, the electrode was cycled over the  −0.25 V to  +1.25 V range fifty times. Such a sweeping or cycling procedure, sometimes referred to as cleaning the electrode, is not often discussed in the neural prosthesis literature, and that absence can be misleading to a novice. After sweeping the potential over the  −0.25 V to  +1.25 V range fifty times, the charge stored in the hydrogen adsorption region increased from ~400 µC cm⁻² to ~700 µC cm⁻². Sweeping the potential over the  −0.25 V to  +0.55 V range did not improve the CSC of the hydrogen adsorption reaction (figure 3(a)). Comparing the results obtained from figure 3 with those of figure 4, it is apparent that to achieve the enhanced charge storage of hydrogen adsorption, the electrode potential needs to be swept through the platinum oxide formation region. Further, as shown in figure 5, the cleaned or enhanced hydrogen adsorption region cannot be maintained if the electrode potential is not swept into the range where oxide is formed.

One might think of applying an anodic bias to the electrode prior to the application of a cathodic neural stimulation pulse to develop surface sites for hydrogen adsorption. The results of these experiments suggest that this may not enhance charge transfer unless the positive bias was applied repeatedly (or for long periods of time) and that some of the exposed sites may be blocked during the inter-pulse interval. Applying an anodic bias on platinum may have other consequences in vivo such as enhanced corrosion through the formation of platinum-chloride complexes [20, 56–58].

In the electrochemistry literature of the 1950s and 60s, several investigators reported altered electrochemical activity on platinum depending on the potential range accessed by the electrode. Out of these observations, two theories on the cause of the activity changes arose: (1) scanning through more positive potentials oxidizes contaminants or organic materials in the electrolyte that had adsorbed to the electrode surface, thus freeing surface sites for charge transfer [59–62] and (2) scanning into the oxide formation potential region elicits 'activation' of the platinum surface, producing a layer of more active platinum sites [63–67]. French and Kuwana estimated that the activated state would revert to the inactivated state with a half-life of 46 minutes [67]. More recently, Hibbert et al demonstrated the inhibition of hydrogen adsorption by cycling a platinum electrode in a phosphate buffered saline electrolyte containing the protein human serum albumin (HSA) and/or amino acids [31]. From our review of literature, a consensus on the true cause of hydrogen adsorption inhibition never developed. Regardless of the cause, the implications for this activity change are clear. During neural stimulation, if the platinum electrode is mostly exposed to potentials negative of OCP, then the electrode will not be able to transfer as much charge as it would if the surface had been fully 'activated' or 'cleaned of contaminants' by cycling through the entire oxide formation region.

The protons, hydroxide ions, and water molecules in sulfuric acid are not the only components to consider when evaluating the electrochemistry of a stimulating electrode. One significant ion present in the body is the chloride ion (Cl⁻). Adding chloride (0.15 M NaCl) to the sulfuric acid electrolyte (figure 7(a)) demonstrated that the presence of Cl⁻ decreases the overall amount of oxide formation and alters the relative peak heights for H adsorption-desorption. Several researchers have investigated ion adsorption on the platinum surface [41, 47, 68, 69]. Bagotzky et al demonstrated that Cl⁻ adsorbs to the surface and changes the energy distribution of H adsorption, but does not decrease the amount that occurs (a full monolayer of H can still be adsorbed). Novak and Conway demonstrated that the adsorption of Cl⁻ acts to suppress oxide film formation [47]. When Cl⁻ is adsorbed, it can be oxidized through the corrosion reaction in (4.4.1).

$$\begin{eqnarray}&&\text{P}{{\text{t}}_{(\text{s})}}~+~4\text{Cl}_{(\text{ad})}^{-}~\leftrightharpoons ~[\text{PtC}{{\text{l}}_{\text{4}}}]_{(\text{aq)} ~ }^{2-}+~2{{e}^{-}}\end{eqnarray} \tag{ 4.4.1 }$$

$$\begin{eqnarray}&&E=+0.\text{49}\,\text{V}\,\text{versus}\,\text{Ag}|\text{AgCl} ~ \left(\text{at}\,\text{25}\,{}^\circ \text{C}, ~\text{1}\,\text{atm},\,\text{pH}=0.\text{9}\right)\end{eqnarray}$$

Another aspect of living tissue that will affect the electrochemical performance of an electrode is the presence of organic materials like proteins. The experiment in figure 7(b) demonstrated that BSA suppresses oxide formation and alters H adsorption. Evidence in literature demonstrates that BSA readily adsorbs to the platinum electrode surface [70]. Bagotzky et al found that when organic molecules adsorb to the surface of platinum, both oxide formation and the amount of H adsorption decrease [41]. The suppression of oxide formation current in figure 7(b) indicates that BSA is blocking the platinum surface from being fully oxidized [42]. By examining the H desorption (positive) peaks in figure 7(b), it is apparent that less H adsorption had happened when BSA was present. It has also been shown that organic molecules and anions compete for surface sites when adsorbing on platinum [41]. When both BSA and chloride are present, BSA may occupy surface sites to which chloride ions would otherwise adsorb. With less chloride adsorbed, the corrosion reaction above would be suppressed. Robblee measured decreased platinum corrosion products with the addition of human serum albumin to phosphate buffered saline [40].

Even with high-quality reagents and deionized distilled water, contaminants may be present in solution that can affect electrochemical measurements. Most would consider the sulfuric-acid-only i-E curves in figures 6 and 7 as 'textbook' i-E curves of platinum in sulfuric acid. Specifically, the pronounced hump at the beginning of the oxide formation region (+0.55 V to  +0.9 V), the well-centered double-layer region (near  +0.1 V), and the sharp H adsorption–desorption peaks (−0.22 V to  +0.05 V) are all indicative of a high-purity de-oxygenated electrolyte and a pure platinum electrode that has been scanned to steady state. Its characteristics are very similar to i-E curves of platinum in sulfuric acid presented in other literature [42, 46, 47, 71]. The slow rise of oxide formation current in the sulfuric acid i-E curve in figure 2(a), however, may indicate that organic contaminants were present in the sulfuric acid electrolyte used in figures 2–5. In those experiments, ultrapure water was not used to prepare the electrolyte and paraffin film was used to cover the electrochemical cell, both of which are possible sources of organic contamination. Note the similarity in the shape of the oxide formation shoulder (+0.55 V to  +1.2 V) in figures 2(a) (possible organic contaminants) and 7(b), where organics (BSA) were intentionally added to solution. That observation supports the theory that scanning through more positive potentials (as demonstrated in figure 4) oxidizes organic molecules on the electrode surface, thus freeing surface sites for charge transfer via the hydrogen (H) adsorption reaction.