Towards Implantable Bio-Supercapacitors: Pseudocapacitance of Ruthenium Oxide Nanoparticles and Nanosheets in Acids, Buffered Solutions, and Bioelectrolytes

Metal oxides, in particular ruthenium-based oxides, are promising electrode materials for aqueous pseudocapacitors. Strong acids or bases are favored over neutral electrolytes owing to the higher capacitance. Here we explore the pseudocapacitive behavior of ruthenium oxide nanoparticles and nanosheets in near neutral pH as an environmentally benign electrolyte. The pseudocapacitive charge storage in poorly-crystalline hydrous RuO 2 nanoparticles, and highly-crystalline RuO 2 nanosheets were investigated in acetic acid-lithium acetate (AcOH-AcOLi) buffered solutions. It is shown that capacitance values as high as 1,038 F g − 1 can be achieved in AcOH-AcOLi buffered solutions with RuO 2 nanosheets, which is 44% higher than the benchmark RuO 2 · n H 2 O in H 2 SO 4 electrolyte (720 F g − 1 ). Furthermore, comparable performance was obtained in phosphate buffered saline and fetal bovine serum. The mechanism of the pseudocapacitive properties is discussed based on the difference in the surface redox behavior of different RuO 2 nanomaterials in acid, neutral, buffered solutions, and in weak acid. ©

Electrochemical capacitors (also known as supercapacitors) are energy harvesting devices capable of charge and discharging within a few seconds, and cycle life in the order of thousands of cycles. Some metal-oxides are known to provide high capacitance in aqueous electrolytes, owing to the combination of the non-faradaic electrical double layer charging and the faradaic surface or near surface confined redox capacitance (psuedocapacitance). 1 Pseudocapacitance is a phenomenon generally observed in aqueous electrolytes. 2 Acidic or basic electrolytes such as H 2 SO 4 and KOH are favorable in terms of power density owing to the high conductivity. RuO 2 is one of the rare oxides that is stable in both acidic and basic conditions. Hydrous RuO 2 nanoparticles (RuO 2 · nH 2 O; where n is typically 0.5) offers capacitance of ∼700 F g −1 in H 2 SO 4 electrolyte and can be cycled for thousands of cycles with practically no decay, thus is often used for performance benchmarking of new electrode materials. Although studies on the asymmetric systems and applicability of non-aqueous electrolytes to oxide electrodes in order to widen the operating voltage window have recently been initiated, non-aqueous electrolytes have yet to surpass aqueous electrolytes in terms of specific capacitance. [3][4][5][6][7][8] Electrolytes near neutral pH are selected for materials that are not as corrosion-resistant in acids and base, for example manganese oxide. [9][10][11][12] Neutral electrolytes are more environmentally benign and its low corrosiveness allows a wider range in choice for periphery material, such as current collectors and packaging. 13 Despite the RuO 2based material being the model pseudocapacitive material, studies on the electrochemical capacitor behavior in neutral electrolytes are scarce compared to the more popular acidic or basic electrolytes. One of the reasons is that the capacitance of RuO 2 in neutral electrolytes is generally 1/2 of that in sulfuric acid or potassium hydroxide. [14][15][16][17] Nonetheless, the use of neutral pH electrolytes has advantages when used for asymmetric (hybrid) supercapacitors, which are devices that utilize different materials (e.g. metal oxides, carbon, etc) for the positive and negative electrodes. 12 In such a case, the operating voltage window can be extended beyond the thermodynamic 1.2 V limit if the kinetics of gas evolution is extremely slow. Up to now, alkali metal sulfates, nitrates, and chlorides have been used as neutral electrolytes.
We have also used Li 2 SO 4 as the electrolyte for our hybrid supercapacitors based on protected Li anode technology. [18][19][20] This hybrid supercapacitor utilizes a Li ion conducting glass ceramic membrane, which is stable within a limited pH range. We recently communicated that unprecedented capacitance values exceeding that of H 2 SO 4 can be achieved by using an acetic acid-lithium acetate (AcOH-AcOLi) buffered solution with near neutral pH, 20 suggesting the possibility of other new electrolytes.
In this study, emphasis was placed on elucidating the origin and mechanism of the pseudocapacitance of ruthenium based oxides in buffered solutions. Three different nanostructured RuO 2 materials were studied; namely poorly crystalline hydrous RuO 2 nanoparticles, well-crystalline anhydrous RuO 2 nanoparticles, and crystalline RuO 2 nanosheets. The capacitive behavior of these materials were studied in H 2 SO 4 as the acidic electrolyte, or Li 2 SO 4 or AcOLi as neutral electrolyte. Various AcOH-AcOLi buffer solutions with different ionic strength (constant pH) were used, and the ratio of weak acid/conjugated base ratio was also varied. In addition, the role of weak acid was investigated by adding a small amount of AcOH to a supporting electrolyte (Li 2 SO 4 ). Furthermore, phosphate buffered saline and fetal bovine serum were studied as bioelectrolytes for application toward implantable bio-supercapacitors.

Experimental
Ultrapure water (Milli-Q, >18 M cm) was used for all synthesis and characterization. RuO 2 · nH 2 O was prepared by a modified solgel process following literature. 21,22 In a typical synthesis, a 0.3 M NaOH solution was slowly added to a 0.1 M aqueous RuCl 3 solution while maintaining a constant pH value of 7. The precipitate was collected and washed thoroughly to remove by-products. The product was suspended in H 2 O and aged for 72 h at 40 • C. The powder samples were collected and heat treated in air at 150 • C for 17 h to obtain RuO 2 · nH 2 O (n = 0.5). Anhydrous RuO 2 was prepared by calcination of RuO 2 · xH 2 O (Johnson Matthey) at 450 • C for 2 h.
Ruthenium oxide nanosheets (RuO 2 ns) were synthesized following a previously reported method. 23 Briefly, α-NaFeO 2 type NaRuO 2 was synthesized by solid state reaction of Na 2 CO 3 , Ru and RuO 2 (2: 1: 3 molar ratio) at 900 • C for 12 h under Ar atmosphere. Oxidative Electrochemical measurements were carried out using a beakertype electrochemical cell composed of a Pt mesh counter electrode and a Ag/AgCl/KCl (sat.) reference electrode connected with a salt bridge. A Luggin capillary faced the working electrode at a distance of 2 mm. Electrode potentials will be referred to the reversible hydrogen electrode (RHE) potential scale. The working electrodes for RuO 2 · nH 2 O and anhydrous RuO 2 was prepared by coating the active material on a glassy carbon surface (∼200 μg cm −2 ). A thin layer of Nafion ionomer was cast on the electrode as a proton conductive binder. Re-stacked RuO 2 nanosheet electrodes were prepared by dropping a colloidal suspension onto a mirror-polished glassy carbon rod (∼20 μg cm −2 ). The capacitance was calculated by averaging the anodic and cathodic charge. Cyclic voltammetry was conducted in H 2 SO 4 , Li 2 SO 4 , AcOLi, and AcOH-AcOLi at 60 • C unless otherwise noted with the scan rate 2, 5, 20, 50, 200, 500 mV s −1 . Phosphate buffered saline and fetal bovine serum (Biowest, France) was used as-received. The electrolytes used in this study are summarized in Table I.

Results and Discussion
Impact of electrolyte on the pseudocapacitive properties of RuO 2 · nH 2 O.-Cyclic voltammograms of RuO 2 · nH 2 O nanoparticles in 0.5 M H 2 SO 4 , 1.0 M Li 2 SO 4 , 2.0 M AcOLi, and 2.0 M AcOH-AcOLi are shown in Fig. 1A. The voltammograms in 0.5 M H 2 SO 4 are typical of sol-gel derived RuO 2 · nH 2 O nanoparticles with a maximum capacitance of 720 F g −1 at 2 mV s −1 , decreasing by 18% to 589 F g −1 at 500 mV s −1 . A broad redox peak at E 1/2 = 0.60 V vs RHE can be clearly distinguished.
The cyclic voltammograms in 1.0 M Li 2 SO 4 are characterized by a rectangular background current (shown as shaded region in the figure), and a slow irreversible redox process above 0.8 V and below 0.6 V vs RHE on the anodic and cathode scans, respectively. The rectangular background current is superimposed for the other electrolytes assuming that the C dl is the same regardless of electrolyte. Here we are disregarding the size of the (solvated) ions for sake of simplicity. The specific capacitance due to electrical double layer charging (C dl ) in 1.0 M Li 2 SO 4 is estimated as ∼200 F g −1 from the scan rate independent region. This capacitance translates to an estimated surface area of 1,000 m 2 g −1 or 1.0 nm particle size, taking the value of 20 μF cm −2 as a probe value for area specific capacitance. This particle size is in good agreement with the local structure derived by EXFAS 24 and SAXS. 25 The pseudocapacitance due to surface redox processes (C redox ) is calculated by subtracting C dl from the overall capacitance C at the respective scan rates and is shown in Fig. 1B. The slow irreversible redox process above 0.8 V and below 0.6 V can be interpreted as the dissociative adsorption of water according to reaction 1. 15,16 The behavior in 2.0 M AcOLi is similar to Li 2 SO 4 in many aspects. The C dl values in AcOLi and Li 2 SO 4 are both ∼200 F g −1 . The redox peaks due to reaction 1 is observed at E 1/2 = 0.70 V, which is close to the Ru 4+ /Ru 3+ potential according to the Pourbaix diagram 26 for this reaction. An obvious difference is the charge related to reaction 1; The charge is higher in 2.0 M AcOLi. This difference may be due to specific adsorption of SO 4 2− on the oxide surface, hindering the adsorption of water, and thus delaying reaction 1.
The behavior in 2.0 M AcOH-AcOLi seems to be a combination of the behavior in H 2 SO 4 and AcOLi depending on the scan rate. At slow scan rates the voltammograms are similar to that in H 2 SO 4 and a capacitance of 687 F g −1 at 2 mV s −1 is obtained. On the other hand, only electrical double layer charging occurs at fast scan rates in AcOLi, reducing the capacitance to 210 F g −1 at 500 mV s −1 . This phenomenon (lack of pseudocapacitance at high scan rates) is due to the low proton concentration in 2.0 M AcOH-AcOLi. Surface redox process related with hydrated protons will become diffusion limited at high scan rates. If we take the C dl values of ∼200 F g −1 from AcOLi, C redox in 2.0 M AcOH-AcOLi becomes extremely large (487 F g −1 or 64.8 kC mol −1 ). Dividing 64.8 kC mol −1 by the Faraday constant 96.5 kC mol −1 gives a 0.6 electron reaction for C redox . Since not all of the Ru ions will be in the outer shell of the nanoparticle, this value seems to be a reasonable value for pseudocapacitance.
Using the C dl values obtained in neutral electrolytes, we can calculate the C redox in 0.5 M H 2 SO 4 as 400-500 F g −1 . The E 1/2 = 0.60 V vs RHE in H 2 SO 4 is also attributed to reaction 1. 16 Note that if we take differential capacitance that is mostly scan-rate independent in H 2 SO 4 as the C dl (which is often practiced in literature (see for example ref [27,28])), C dl can be estimated as 500 F g −1 . The probe value of 80 μF cm −2 is often used as a measure of the specific capacitance in H 2 SO 4 , which originates from the above mentioned treatment. The results and discussion shown here using various electrolytes suggests that this probe value contains both C dl and C redox charge.
The pseudocapacitive behavior of anhydrous RuO 2 nanoparticles in 0.5 M H 2 SO 4 , 1.0 M Li 2 SO 4 , 2.0 M AcOLi, and 2.0 M AcOH-AcOLi are qualitatively similar to RuO 2 · nH 2 O (Fig. S1). The change in molarity of AcOH-AcOLi between 0.5 and 5.0 M does not affect the voltammograms significantly (Fig. S2). At the lowest concentration of 0.5 M, a peak at E 1/2 = 0.41 V evolves at slow scan rate, which is attributable to the adsorption of AcOH (discussed in detail later).  Fig. 2A. The voltammograms in 0.5 M H 2 SO 4 are quite different from sol-gel derived RuO 2 · nH 2 O nanoparticles in that there is a distinctive large redox pair at E 1/2 = 0.64 V vs. RHE. The capacitance at 2 mV s −1 is 831 F g −1 and decreases by 15% to 703 F g −1 at 500 mV s −1 . Following the case for RuO 2 · nH 2 O nanoparticles, the C dl value can be deduced from the constant dQ/dE background current taken in Li 2 SO 4 , which is C dl ∼300 F g −1 . The deconvoluted C dl and C redox contribution at the respective scan rates are given in Fig. 2B. Using the C dl value and 20 μF cm −2 , the electrochemically accessible surface area is estimated as 1,500 m 2 g −1 . This is much higher than the theoretical surface area of ∼400 m 2 g −1 for a RuO 2 nanosheet crystallite with thickness of 0.7 nm. The estimated C dl ∼300 F g −1 most likely includes pseudocapacitance from fast surface redox processes. C redox due to reaction 1 is observed also for RuO 2 nanosheets in Li 2 SO 4 and AcOLi, although the contribution is much smaller than for RuO 2 · nH 2 O nanoparticles. The cathodic current below 0.6 V and corresponding oxidation current at E = 0.75 V in Li 2 SO 4 and AcOLi ( Fig. 2A(b) and (c)) may be due to hydrogen adsorption. 28 The behavior of RuO 2 nanosheets in 2.0 M AcOH-AcOLi is characterized by two pairs of redox peaks at E 1/2 = 0.41 and 0.60 V vs RHE. The former pair is more scan rate dependent than the latter. The overall capacitance is 958 F g −1 at 2 mV s −1 and 752 F g −1 at 500 mV s −1 . The C redox in H 2 SO 4 and AcOH-AcOLi represents 50 to 70% of the overall capacitance.

Pseudocapacitive properties of RuO 2 nanosheets in various
The distinctive redox peaks observed for RuO 2 nanosheets changes drastically when the ionic strength of the AcOH-AcOLi buffer solution is changed while keeping a constant pH (Fig. S3). 20 The redox pair at E 1/2 = 0.41 V is strongly dependent on the ionic strength, broadening and decreasing in charge with decreasing AcOH-AcOLi concentration. The E 1/2 = 0.60 V peak is less dependent on the AcOH-AcOLi concentration. These observations can be attributed to decreasing concentration of the active species. It is noted that at the highest concentration of 5.0 M AcOH-AcOLi, a remarkable capacitance of 1,038 F g −1 is obtained. 20 In order to gain further insight into the origin of the redox peaks, the pH of the buffer solution was controlled by changing the salt/supporting electrolyte volumetric ratio. As shown in Fig. 3, as the pH is lowered, the cyclic voltammograms become strongly scan rate dependent, which is due to the decrease in the conductivity of the buffer solutions. Figure 3d compares the voltammograms at 2 mV s −1 . The E 1/2 = 0.60 V vs RHE peak is independent of pH (note that in H 2 SO 4 , the E 1/2 = 0.41 and 0.60 V vs RHE peaks overlap). The E 1/2 = 0.60 V vs RHE peak is attributed to reaction 1, similar to the case for RuO 2 · nH 2 O nanoparticles. The E 1/2 = 0.41 V vs RHE (at pH = 5.43) shifts to positive potentials as the pH is decreased. This peak shows a linear relation with pH (Fig. 4), suggesting that it is related to adsorption of protons or hydrated protons. Analysis of the slope shows that this is a 1.5 electron reaction, or 3 electrons per 2 reaction sites. This irregular behavior is consistent with previous studies on hydrous metal oxide films of Ir and Ru. 29-31 Figure 5 shows comparative data of the change in pH for RuO 2 · nH 2 O nanoparticles. The cyclic voltammograms for H 2 SO 4 and 2.0 M AcOH-AcOLi (pH = 5.36) completely overlap, suggesting an analogous charge storage mechanism. On the other hand, the electrochemical behavior in 2.0 M AcOH-AcOLi with pH = 4.44 and 3.42 are quite different. For pH = 3.42, the electrical double layer capacitance is completely lost. We attribute this peculiarity to adsorption of molecular AcOH (not AcO − ) on the surface of RuO 2 , blocking the electrical double layer formation. As the pH is lowered, the relative content of AcOH increases (Table I). In the case of 2.0 M AcOH-AcOLi (pH = 5.36), the adsorption of AcOH is not evident since the concentration of AcOH is much smaller than at pH = 4.44 and 3.42. A similar phenomena was observed for adsorption of CH 3 OH in H 2 SO 4 , where CH 3 OH is adsorbed on RuO 2 · nH 2 O nanoparticles. 23 The reason for the absence of AcOH adsorption on RuO 2 nanosheets cannot be identified at this point, but it should be noted that CH 3 OH also does not adsorb on RuO 2 nanosheets. 23 A weak acid in a supporting electrolyte.- Figure 6 shows cyclic voltammograms when a small amount of AcOH was added to Li 2 SO 4 (0.1 M AcOH + 1.0 M Li 2 SO 4 , pH = 4.16). At slow scan rates    the cyclic voltammograms for RuO 2 · nH 2 O nanoparticles and RuO 2 nanosheets both resemble the behavior in AcOH-AcOLi buffered solutions. The capacitance at 2 mV s −1 is 639 and 962 F g −1 for RuO 2 · nH 2 O nanoparticles and RuO 2 nanosheets, respectively. On the other hand, at fast scan rate, the voltammograms are similar to those in Li 2 SO 4 , since the concentration of AcOH is very small. This clearly shows that protons contribute to the pseudocapacitive behavior of RuO 2 -based nanostructures. Thus, an electrolyte composed of a weak acid and a supporting electrolyte can also be used as an electrolyte for pseudocapacitors. Buffered solutions have the advantage of pH control, ionic strength, and biocompatibility, compared to a weak acid in supporting electrolyte. However, the finding that such a simple mixture can be used as an electrolyte for pseudocapacitors paves the way to a massive combination of new electrolytes to explore.
Bio-supercapacitor based on phosphate buffered saline and fetal bovine serum.-Besides the benefit of control in pH and ionic strength, buffered solutions have the advantage of biocompatibility, as many buffers exist in nature (sea water, blood, internal cell fluids, cells and tissues). Here we demonstrate the use of phosphate buffered saline and fetal bovine serum as bioelectrolytes for supercapacitors applicable to safe and bio-compatible implantable power sources. Figure 7 shows the cyclic voltammograms of RuO 2 nanosheets in phosphate buffered saline and fetal bovine serum at 25 • C. The redox behavior and capacitance are similar to the AcOH-AcOLi system. The specific capacitance at 2 mV s −1 in phosphate buffered saline was 837 F g −1 , and 772 F g −1 in fetal bovine serum. These values are at least twice as large as those of MWCNTs/PANI composite in physiological electrolyte and human serum. 32 The results potentially show that the combination of pseudocapacitive oxide electrodes and bioelectrolytes can afford exceptionally high energy density.

Conclusions
The electrochemical charge storage in buffered solutions using poorly-crystalline hydrous RuO 2 nanoparticles and well-crystalline RuO 2 nanosheets as electrode materials was studied. Capacitance comparable to or higher than H 2 SO 4 were obtained in acetic acidlithium acetate (AcOH-AcOLi) buffered solutions, depending on the ionic strength and pH. At constant pH, AcOH-AcOLi with higher ionic strength (molarity) lead to higher capacitance, owing to the presence of higher concentration of the adsorbant in the electrolyte. When the pH is varied by changing the volume ratio of AcOH-AcOLi, the pseudocapacitive behavior deteriorated with decreasing pH. This phenomenon is discussed based on the decrease in the conductivity of the electrolyte. The highest capacitance of 1,038 F g −1 was obtained in a 5 M AcOH-AcOLi (pH = 5.4) with RuO 2 nanosheet electrodes. RuO 2 nanosheets afforded 20 to 50% higher capacitance than RuO 2 · nH 2 O in all of the electrolytes studied (H 2 SO 4 , Li 2 SO 4 , AcOLi, AcOH-AcOLi, AcOH-Li 2 SO 4 ). The lower capacitance of RuO 2 · nH 2 O in AcOH-AcOLi is due to the adsorption of AcOH molecules, which hinders the formation of the electrical double layer. Moreover, outstanding performance was obtained in phosphate buffered saline (837 F g −1 ) and fetal bovine serum (772 F g −1 ) with RuO 2 nanosheet electrodes. The results presented here demonstrate the effective use of bioelectrolytes for pseudocapacitors applicable to environmentally begin, safe and bio-compatible implantable power sources.