Af

KEa

b

a

ARRAA

KEEPPSR

1

eeswsotdip

(

0h

Electrochimica Acta 111 (2013) 888– 897

Contents lists available at ScienceDirect

Electrochimica Acta

jou rn al hom ep age: www.elsev ier .com/ locate /e lec tac ta

high performance pseudocapacitive suspension electrodeor the electrochemical flow capacitor

elsey B. Hatzell a, Majid Beidaghia, Jonathan W. Camposa, Christopher R. Dennisona,b,min C. Kumburb,∗, Yury Gogotsi a,∗∗

A.J. Drexel Nanotechnology Institute, Department of Materials Science and Engineering, Drexel University, Philadelphia, PA 19104, USAElectrochemical Energy Systems Laboratory, Department of Mechanical Engineering and Mechanics, Drexel University, Philadelphia, PA 19104, USA

r t i c l e i n f o

rticle history:eceived 1 May 2013eceived in revised form 17 July 2013ccepted 17 August 2013vailable online 30 August 2013

eywords:lectrochemical flow capacitornergy storage-phenylenediamineseudocapacitorupercapacitor

a b s t r a c t

The electrochemical flow capacitor (EFC) is a new technology for grid energy storage that is based onthe fundamental principles of supercapacitors. The EFC benefits from the advantages of both superca-pacitors and flow batteries in that it is capable of rapid charging/discharging, has a long cycle lifetime,and enables energy storage and power to be decoupled and optimized for the desired application. Theunique aspect of the EFC is that it utilizes a flowable carbon-electrolyte suspension (slurry) for capac-itive energy storage. Similar to traditional supercapacitor electrodes, this aqueous slurry is limited interms of energy density, when compared to batteries. To address this limitation, in this study a pseudo-capacitive additive has been explored to increase capacitance. A carbon-electrolyte slurry prepared withp-phenylenediamine (PPD), a redox mediator, shows an increased capacitance on the order of 86% whencompared with KOH electrolytes, and a 130% increase when compared to previously reported neutralelectrolyte based slurries. The redox-mediated slurry also appears to benefit from a decrease in ohmic

edox mediator resistance with increasing concentrations of PPD, most likely a result of an increase in the ionic diffusioncoefficient. Among the tested slurries, a concentration of 0.139 M of PPD in 2 M KOH electrolyte yields thelargest capacitance and rate handling performance in both cyclic voltammetry and galvanostatic cyclingexperiments. The improved performance is attributed to the addition of quick faradaic reactions at theelectrolyte-electrode interface as PPD undergoes a two-proton/two-electron reduction and oxidationreaction during cycling.

© 2013 Elsevier Ltd. All rights reserved.

. Introduction

Currently, the primary technologies employed for grid-levelnergy storage are pumped hydroelectric and compressed airnergy storage [1]. These technologies are geographically con-trained and require large capital investments when comparedith electrochemical energy systems (EES), such as batteries and

upercapacitors [2–4]. Thus, EES systems have emerged as the mostpportunistic type of technology for grid-scale applications [1,5]. Inransportation or portable electronics applications, where energy

ensity is the critical factor, batteries are favored [2]. However,

n large-scale stationary storage applications, the scalability, lowower density, and cost hinder the widespread implementation of

∗ Corresponding author. Tel.: +1 215 895 5871; fax: +1 215 895 1478.∗∗ Corresponding author. Tel.: +1 215 895 6446; fax: +1 215 895 1934.

E-mail addresses: eck32@drexel.edu (E.C. Kumbur), gogotsi@drexel.eduY. Gogotsi).

013-4686/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.electacta.2013.08.095

battery systems [6,7]. The critical performance requirements forgrid-level energy storage are cost, scalability, and lifetime [8]. Fur-thermore, the market for stationary energy storage is vast, requiringcapacities that range from kW to GW systems which most likelycannot be achieved with a single technology [1,9]. Thus, simi-lar to energy generation systems, there is a need for a portfolioapproach of tailored energy storage systems designed for spe-cific applications such as voltage and frequency regulation, loadleveling, and bulk power management [1,9]. In order to addressthe scalability requirements and broader power/energy range, var-ious flow-assisted electrochemical systems have been recentlyproposed. The architecture seen in the redox flow battery [10],semi-solid lithium battery [11], and electrochemical flow capac-itor (EFC) [12,13] offers tremendous design flexibility for use ingrid-level energy storage.

Recently, a new energy storage concept called the electrochem-ical flow capacitor (EFC) has been proposed by our group at DrexelUniversity. The EFC is a rechargeable EES that utilizes a flow batteryarchitecture and is based on the fundamental working principles of

K.B. Hatzell et al. / Electrochimica Acta 111 (2013) 888– 897 889

Fig. 1. (a) Operational schematic of the electrochemical flow capacitor. Uncharged slurry flows through polarized plates and charged. At the pore level, electrode neutralityi rry isd l part

sfleKotsv(cdrscpaacepc

s maintained at the interface between the electrolyte and active material. This sluischarge. (b) Schematic of a carbon|electrolyte interface between charged spherica

upercapacitors [3,12]. The primary difference between traditionalow cells and the EFC is that the EFC utilizes a flowable carbon-lectrolyte electrode, similar to the ‘slurry electrode’ introduced byastening et al. [14], for capacitive energy storage (Fig. 1b). Duringperation the slurry is pumped from a storage reservoir throughwo polarized plates (charging process). Once fully charged, thelurry is pumped out of the cell and stored in external reser-oirs until the process is reversed and the slurry is dischargedFig. 1a). The charged slurry stores charge electrostatically at thearbon/electrolyte interface which allows for rapid charging andischarging (Fig. 1c). Traditional flow batteries suffer from slowesponse rates due to faradaic reactions at the electrode-electrolyteurface and have limited cycle lifetimes (<12,000 cycles) which is aritical factor for grid scale energy storage [8,15]. The EFC in com-arison is capable of rapid charging/discharging, is based on cheapnd abundant materials (e.g., carbon), and enables energy storagend power to be decoupled and optimized. Furthermore, since the

harge storage mechanism is physical, the EFC has the potential toxhibit long cycle life, a characteristic commonly seen in superca-acitors. Power is directly related to stack size, and energy capacityan be scaled according to reservoir volume.

then pumped into external reservoirs for storage. The process is reversed duringicles and (c) SEM image of carbon beads.

Although the EFC has high power density, its energy densityneeds to be improved in order to compete with alternative systemsfor grid-level energy storage applications. The energy released bya supercapacitor, E, is:

E = .5CV2 (1)

where C is the capacitance (in F), and V is the operational voltagewindow (in V). Therefore, in order to increase the energy density,one can either increase the voltage window or the active mate-rial’s capacitance. For aqueous electrolytes, the voltage window isthermodynamically limited to approximately 1.0 V because waterdecomposition occurs above 1.23 V [16]. Nevertheless, it has beenshown that the voltage window of supercapacitors based on neutralelectrolytes may extend beyond 1.23 V because the di-hydrogenoverpotential is higher for neutral electrolytes[16–20]. In a recentstudy on the EFC, flowable electrodes based on sodium sulfatedemonstrated voltage stability up to ∼1.5 V[19]. While an increased

voltage window will yield increased energy density, the currentstudy seeks to examine methods to enhance the capacitance of theflowable electrode. Thus, this study restricts its focus to availablemethods to enhance the capacitance of the flowable electrode.

8 imica

t(dctmmpiahefcbti[ssip[tisiswaiiimtt

dpfipsrhc

2

2

PaflSSissfPa

90 K.B. Hatzell et al. / Electroch

One method for increasing capacitance is through the addi-ion of quick redox reactions at the electrode-electrolyte interfacepseudocapacitive effects). The primary means for achieving pseu-ocapacitance is through the use of transition metal oxides [21–23],onducting polymers [21,24] or the addition of surface func-ionalities such as nitrogen or oxygen to the surface of carbon

aterials [25]. Recently a fourth strategy, the use of redox-ediated electrolytes, has emerged as a new method of achieving

seudocapacitive behavior [26]. The addition of redox-mediatorsnto electrolytes has been shown to increase ionic conductivitynd provide an additional charge storage mechanism, and thusas been explored in supercapacitor applications as a way tonhance specific capacitance [27]. Furthermore, there is no needor surface modification because the redox species react at thearbon|electrolyte interface. Most redox-mediated electrolytes areased on proton-coupled electron transfer (PCET) mechanismshat are pH-dependent and prefer basic or acidic environmentsn order to facilitate proton transfer at the active material surface28–30]. In supercapacitors, electrochemically active compoundsuch as indigo carmine [31] and hydroquinone [33], have shownignificant capacitance increases in acidic mediums. In addition,t has been recently shown that aromatic diamines such as p-henylenediamine (PPD) [21,34–37], and m-phenylenediamine38] (MPD) are possible redox mediators in basic electrolyte sys-ems. A MnO2-based supercapacitor[36] demonstrated a 6-foldncrease in capacitance and an activated carbon supercapacitor[35]howed a 4-fold increase in capacitance after the addition of PPDnto KOH electrolytes[36]. Thus, PPD as a redox mediator has beenhown to dramatically enhance capacitance. Pseudocapacitance,here the source of the redox reaction is in the electrolyte, is favor-

ble for systems such as the EFC because the major componentn the slurry is the electrolyte. Thus, redox-mediators integratednto a capacitive slurry offer opportunities for increased capac-tance and enhanced electrochemical performance. However, in

any cases pseudocapacitors suffer from poor rate capacities dueo irreversible redox reactions and physicochemical changes thatake place during charging and discharging [39].

The objective of this study is to develop a carbon-based pseu-ocapacitive flowable electrode for the EFC that utilizes thehenylenediamine/p-phenylenediimine (PPD/PPI) redox coupleor additional storage of energy. Herein, a redox-mediated slurrys prepared by the addition of PPD to 2 M KOH electrolyte and theerformance of the slurry is characterized electrochemically. Thistudy also examines the effectiveness of PPD in a suspension envi-onment (flowable electrode) and assesses its performance (rateandling and capacitance) across a wide range of rates and con-entrations.

. Materials and methods

.1. Chemicals

All the reagents used in this study were of analytical grade.PD anti-fade reagent supplied by Sigma–Aldrich was employeds a redox mediator, and 2 M potassium hydroxide (KOH) suppliedrom Alfa Aesar, was used as the supporting electrolyte. A base-ine experiment was performed in 1 M sodium sulfate supplied byigma-Aldrich for comparison with KOH electrolyte experiments.ix different concentrations of the redox mediators were exam-ned: 0.046 M, 0.092 M, 0.139 M, 0.185 M, and 0.277 M PPD in theupporting electrolyte solution (2 M KOH). Different concentration

olutions were explored in order to identify the optimal conditionsor improved system performance (capacitance and rate handling).reviously, it has been shown high concentrations of redox medi-tors can lead to aggregation of free ions and thus diminish the

Acta 111 (2013) 888– 897

performance [38]. Thus, this study intends to demonstrate theoptimal concentration for a flowable electrode system. The molarconcentrations correspond to increments of 5 mg/ml solutions.The PPD was sonicated in 2 M KOH electrolyte until completelydissolved. All solutions were prepared immediately prior to thetesting.

2.2. Electrode active materials

Carbon beads derived from phenolic resins were obtained fromMAST Carbon (United Kingdom) and used as the active capacitivematerial in the slurry electrodes (Fig. 1d). In our previous work wecharacterized the active material, MAST carbon beads, extensivelyin terms of the physical attributes and electrochemical perfor-mance [12,19]. Three different sizes (diameter) of MAST beads wereexamined ranging from 9 �m to 500 �m, and it was determinedthat the mid-size beads (particle size: 125–250 �m) produced thebest electrochemical and rheological performance [19], and thuswas selected as the active material in this study.

N2 gas sorption was carried out in a Quadrasorb gas sorptioninstrument (Quantachrome, USA) and the average, volume-weighted pore size was derived from the cumulative porevolume assuming slit-shaped pores and using the quenched-soliddensity function theory (QSDFT) algorithm. The pore size distri-butions of the carbon beads (average particle size: 161 ± 35 �m)had an average volume-weighted pore size of 4.7 nm. TheBrunauer–Emmett–Teller (BET) method was used to calculate aspecific surface area (SSA) of 1576 m2/g. This value is in agreementwith the SSA value calculated from QSDFT with a maximum of 10%deviation. Carbon black (100% compressed; Alfa Aesar, USA) wasemployed as the conductive additive between carbon beads. Forcalculation purposes, the active mass was considered to be the totalmass of the carbon black and activated carbon in the slurry. A ZeissSupra 50VP scanning electron microscope (Carl Zeiss AG, Germany)operating at 3 kV was used for observing any surface changes to thecarbon beads as a result of the redox-mediator.

2.3. Preparation of slurry electrodes

The slurry electrodes were prepared by mixing carbon beadswith carbon black (100% compressed; Alfa Aesar, USA) to achievea 9:1 weight ratio, as this ratio is defined as the baseline condi-tion in our previous study. The carbon black served as a conductiveadditive. The aqueous electrolyte (2 M KOH) was then added toachieve 23 wt% (solid-to-liquid ratio), as it was shown previouslyto exhibit favorable capacitive and rheological properties [12,19].In order to achieve a wetted carbon surface to adsorb the carbonblack, an incremental amount of deionized water (<5 wt% of slurry)was weighed and added to the slurry mixture. The mixture (withthe excess deionized water) was mildly heated and stirred until ahomogenous slurry was obtained, and the excess deionized waterevaporated.

2.4. Electrochemical performance

The slurry performance was studied in a symmetric static cellconfiguration, with equal masses in both electrode supensions.The static cell is comprised of two stainless steel current collec-tors, onto which the slurry was uniformly loaded into a channelregion defined by latex gaskets. The channel region (active area)was 1 cm2. The latex gaskets were 610 �m thick when completelycompressed prior to testing. A polyvinylidene fluoride (PVDF)

membrane separator with a mesh width of 100 nm (Durapore®;Merck Millipore, Germany) was used as the separator between thetwo slurry electrodes. All electrodes tested had a carbon content of32 mg per electrode and a solid fraction of 23 wt%.

K.B. Hatzell et al. / Electrochimica Acta 111 (2013) 888– 897 891

F face, (P

tL2m

C

wriscefit

oetic

C

wta

Fo

ig. 2. SEM image of activated carbon beads. (a) As-received beads with pristine surPD. Insets show magnified images of the bead surfaces.

All electrochemical measurements were performed at roomemperature with a VMP3 or VSP potentiostat/galvanostat (Bio-ogic, France). Cyclic voltammetry (CV) was carried out at 2, 5, 10,0, 50, and 100 mV s−1 sweep rates. From CV, the specific gravi-etric capacitance (Csp) was derived using the following equation.

sp = 2�E

×∫

idV

�m(2)

here �E is the width of the voltage window, i is the discharge cur-ent, V is the voltage, � is the sweep rate, and m is the mass of carbonn one electrode. The factor of two accounts for the two electrodeetup, assuming that the charge is evenly distributed between twoapacitors in series [40]. Three flowable electrodes were tested forach case (concentration), and an average value is reported in thegures. Error bars are plotted for each average value and representhe standard deviation from the average capacitance.

All values for the capacitance were normalized by the weightf the carbon material in one electrode, not the total slurry mass, tonable a direct comparison with conventional supercapacitor elec-rodes (which are also normalized to the content of active materialn a single electrode). Csp was also calculated from galvanostaticycling (GC) using Eq. (3):

sp = 2i(3)

m × dV/dt

here dV/dt is the slope of the entire discharge curve starting fromhe bottom of the IR drop and ending at the time when the volt-ge is equal to zero. The whole slope was taken to best account for

ig. 3. (a) Cyclic voltammograms (CV) for carbon-electrolyte based slurries with 1 M sodiumf base-case slurries from 2–100 mV s−1, showing decreasing capacitance with increasing

b) Carbon beads from slurry after being cycled 1000 times in 2 M KOH and 0.139 M

the redox behavior exhibited in the galvanostatic charge/dischargeresults. The equivalent series resistance (ESR) of the cell was deter-mined by dividing the total change in voltage of the IR drop ina galvanostatic cycle with the total change in current (betweencharge and discharge) using Eq. (4).

ESR = �V

�I= �V∣∣Icharge

∣∣+ ∣∣Idischarge

∣∣ = �V

2I(4)

Potentiodynamic electrochemical impedance spectroscopy(PEIS) was used to assess the change in ohmic resistance due tochanges in concentrations of PPD in the slurries. The ohmic resis-tance of the static cell, R�, was determined from the impedancespectrum intersection with the real axis. This value is representa-tive of the total resistances associated with the current collectors,active material, electrolyte, and separator [41]. With all characteris-tics held constant (except PPD concentration), the ohmic resistancehighlights the kinetic changes in the slurry due to the addition ofPPD. Furthermore, PEIS was used to evaluate the changes in iondiffusion at the surface of the electrode. In order to characterizechanges in the diffusion coefficient, a Randles circuit was employedto analyze the impedance responses. All spectra were run in the fre-quency range of 200 kHz to 10 mHz at a sine wave signal amplitudeof 10 mV.

3. Results and discussion

The morphology of the carbon beads with and without PPD wasanalyzed by a scanning electron microscope (SEM) as seen in Fig. 2.

sulfate and 2 M potassium hydroxide electrolytes at 2 mV s−1. (b) Rate dependence sweep rates.

892 K.B. Hatzell et al. / Electrochimica Acta 111 (2013) 888– 897

F in 2 Mv

PticscbsdwibofibtndetaKiiTstatWgmpte

csKiiscmepc6

ig. 4. (a) Cyclic voltammograms at 2 mV s−1 of redox-active slurries (23 wt% CB2)

arious concentrations of p-phenylenediamine in the slurries based on 2 M KOH.

PD, as a monomer, has been reported to polymerize under elec-rochemical oxidative conditions [42]. Polymerization can resultn electrode “poisoning”, or the coating of a polymeric film on thearbon surface [43]. This can cause a blocking effect on the carbonurface that can result in limited ion mobility and poor electroniconductivity. These limitations can be observed electrochemicallyy a rapid decrease in the peak current of a CV plot during succes-ive cycles. In order to examine whether polymerization occurreduring cycling, a SEM micrograph of pristine carbon beads, Fig. 2a,as compared with a SEM micrograph of cycled slurry (0.139 M PPD

n 2 M KOH electrolyte) after undergoing 1000 cycles at 20 mV s−1

etween 0 V and 1 V. From the SEM micrographs in Fig. 2b, it isbserved that the cycled slurry shows little evidence of polymerlms on the carbon surface. The noticeable particles around the car-on beads in Fig. 2b are aggregates of carbon black, which is addedo the slurry as a conductive additive. Furthermore, it should beoted that no drastic declines in the peak current were observeduring the first 20 CV cycles typically when polymerization isxpected to occur. A gas sorption analysis was taken on a sample ofhe cycled slurry with a concentration of 0.139 M PPD. The surfacerea of the cycled slurry (carbon beads, carbon black, PPD, and 2 MOH) exhibited a BET surface area of ∼1210 m2/g. Carbon black

s approximately 10 wt% of the slurry, and explains this decreasen surface area when compared to dry carbon beads (∼1576 m2/g).hus, the fact that a large decrease in surface area was not observedupports the notion that the PPD is not polymerizing. This is advan-ageous for our system because polymerization during oxidationnd reduction cycles forms polymer chains which could contributeo irreversible reactions and deterioration of PPD over the cycle life.

e assessed only the flowable electrode concentration with thereatest overall performance in order to ascertain whether poly-erization occurred. It is expected that below 0.139 M, there is no

olymerization. In later sections we assess the effect of the concen-ration of PPD on the electrochemical performance of the flowablelectrode.

Cyclic voltammetry (CV) was utilized for characterizing theapacitance of the slurries in a static cell configuration. Fig. 3ahows that for the cases without the redox mediator (i.e., 2 MOH and 1 M Na2SO4 electrolyte), the slurries demonstrate a typ-

cal quasi-rectangular shape without redox peaks, implying andeal electric double layer capacitor [44]. KOH electrolyte basedlurries exhibit a higher capacitance (118 F g−1 at 2 mV s−1) whenompared to sodium sulfate solution (94 F g−1 at 2 mV s−1). Thisay be attributed to the fact that protons as mobile species may

xhibit faster ionic diffusion inside the pores and access a largerore surface causing a greater capacitance [45]. Furthermore, theonductivity of 1 M Na2SO4 electrolyte is observed to be about0 mS cm−1 whereas 2 M KOH electrolyte has a conductivity of

KOH electrolyte with various amounts of PPD added. (b) Rate performance for the

approximately 170 mS cm−1, facilitating transport throughout theslurry medium. The KOH based flowable electrode exhibits highercapacitance, but also exhibits signs of faradaic reactions as the volt-age approaches 1 V in Fig. 3a. The sudden increase in current at 1 V isa sign that electrolysis of water occurs which evolves hydrogen andoxygen as byproducts [46]. In constrast, neutral electrolytes have astability window that reaches around 1.6 V because of a high over-potential for di-hydrogen evolution, whereas KOH electrolytes arelimited to about 1 V [17]. Fig. 3b shows the rate performance for theslurries without a redox-mediated slurry. Across all rates tested,the 2 M KOH slurry outperformed the sodium sulfate based slur-ries as a working electrolyte for the flowable electrode (Fig. 3b).Above 20 mV s−1 both slurries exhibit a decline in rate performance(capacitance decay). At 100 mV s−1 the Na2SO4 based slurry saw a66% decrease from its initial capacitance taken at 2 mV s−1, which iscomparable to the 2 M KOH electrolyte which exhibited 65% capac-itance decay. The observed capacitance decay at high sweep ratesfor both slurries is a result of ion transport limitations, which is typ-ically seen in porous activated carbon [47]. Diffusion limitations inslurry arise as a result of the slurry’s thickness and as a result of thenetwork of particle-particle contacts. In traditional supercapacitorswith film electrodes, it has been shown that increased thicknessyields increased barriers to ion diffusion within the inner regionof the electrode [47]. This results in longer diffusion paths, higherelectrode resistivity, and poor power performance at high rates.This explanation can be transferred to slurry electrodes which areon average three to five times thicker than film electrodes, and canexplain the decreased power performance at high rates. In addi-tion to diffusion length challenges, while loading the cell duringcharging, some losses can arise when beads are not in direct con-tact, and transport paths are broken. Nevertheless, the desired ratehandling of the slurry for a grid scale applications will most likelylie below 20 mV s−1. which is very well in the range of acceptableperformance for the EFC.

Fig. 4 shows the effect on capacitance and rate handling withincreasing concentrations of the redox mediator (PPD) in the slurry.In contrast to the slurries without redox mediators, the redox-active slurries demonstrate an increased capacitance. The areaunder CV curve is proportional to the capacitance, thus in Fig. 4athe increase in capacitance can be observed qualitatively by theexpansion of the CV curve with changing concentration of PPD. Theanodic and cathodic peaks centered around 0.4 V can be attributedto a pseudocapacitive behavior produced by the redox couple p-phenylenediamine and p-phenylenediimine (PPD/PPI), resulting in

a two-electron two-proton transfer reaction (see Fig. 5) [29]. Fol-lowing the redox peak, the CVs of each redox-active slurry (acrossall concentrations of PPD) exhibits a relatively flat region with a cur-rent density of about 0.175 A/g. The similar voltammetric current

K.B. Hatzell et al. / Electrochimica Acta 111 (2013) 888– 897 893

Fig. 5. Standard two-proton/two-electron oxidation and reduction reaction ofpst

daccprTvs

crt2dnc2csidtws

5(tsrIactdfrAe∼b

TOs

Fig. 6. Capacitance as a function of changing concentrations of p-phenylenediamine−1 −1

-phenylenediamine to p-phenylenediimine. (dark gray, dark blue and white corre-pond to carbon, nitrogen and hydrogen atoms). (For interpretation of the referenceso color in this figure legend, the reader is referred to the web version of the article.)

ensities implies that for each concentration studied, the slurriesppear to experience similar double layer storage [48]. The CVurves show that the capacitance increases as a function of con-entration of PPD. This behavior can be attributed to increases inseudocapacitive charge storage by the fully reversible PPD/PPIedox couple, rather than changes in double layer charge storage.his can be observed by the varying redox peak heights on theoltammagram shown in Fig. 4a. The effect of PPD on the chargetorage mechanisms will be discussed in detail later.

As shown in Fig. 4b, between the sweep rates of 2–20 mV s−1,apacitances remain relatively stable for the redox-mediated slur-ies. The PPD concentrated slurries (0.046–0.277 M) demonstrate aypical decreasing capacitance with increasing sweep rates. Above0 mV s−1, all slurries are observed to demonstrate a more rapidrop in capacitance (Fig. 4b). This capacitance drop is more pro-ounced in the redox-mediated electrolyte with 0.277 M PPDoncentration which exhibits a 69% capacitance decay between

and 100 mV s−1. Table 1 shows a comparison of the observedapacitance decay between 2 and 100 mV s−1 for each redox-activelurry studied. The slurry with 0.139 M PPD shows a 57% capac-tance decay, and the slurry without PPD has a 65% capacitanceecay, indicating that rate handling can be improved by the addi-ion of PPD into the slurry. However, the deepest capacitance decayas observed in the highest concentration PPD slurry which under-

cores the concentration limitations of PPD.From the CV experiments (Fig. 4), at low sweep rates below

mV s−1, the capacitance increases for all concentrations studied0.046–0.277 M). Fig. 6 shows the effects of varying PPD concen-ration on the both rate handling and capacitance of the slurry. Atweep rates greater than 5 mV s−1, the optimal concentration forate handling and capacitance is 0.139 M PPD in 2 M KOH solutions.t appears that at high sweep rates, transport properties becomen issue for highly concentrated solutions. In slurries with highoncentrations of the PPD, the PPD may obstruct ion mobility athe pore level, and thus might contribute to the deep capacitanceecay with sweep rate. Thus, in order to optimize for both rate per-ormance and capacitance, 0.139 M of PPD (2 M KOH) yields the bestesults for the rates that an EFC is expected to operate (<20 mV s−1).

t a concentration of 0.139 M, for our two-electrode set-up, it isstimated that the increased capacitance only amounts to utilizing3.26% of the available PPD molecules in the suspension. This num-er suggests that with further optimization of the electrode, we can

able 1bserved decreases in capacitances for different PPD concentrated redox-active

lurries after increasing the sweep rate from 2 mV s−1 to 100 mV s−1.

Concentration ofPPD (M L−1)

0 0.046 0.092 0.139 0.184 0.277

Capacity decay (%) 64 53 52 57 60 69

and sweep rates ranging from 2 mV s to 100 mV s .

achieve an even higher capacitance in the flowable electrode usingPPD as a redox-mediator.

Utilizing a partition method, commonly applied in examiningthe electrochemically active sites at the outer and inner regionsof a metal oxides electrode [48–50], it is possible to distinguishbetween pseudocapacitive and double layer charge storage mech-anisms for the various redox-mediated slurries (0.046–0.277 M).In metal oxides, the inner region is characterized by slower diffu-sion dynamics, due to extended diffusion lengths in comparisonto the surface accessibility. Thus, in this analysis the slower diffu-sion dynamics (i.e. inner region) is analogous to pseudocapacitiveeffects that occur at slower timescales than electric double layercharge storage mechanisms. As demonstrated in Fig. 4b, the capac-itance (charge storage) decreases as the sweep rate (�) increases.Thus, if we assume semi-infinite linear diffusion, the charge storagemechanism is expected to be linearly related to �−1/2 [49].

The results from the partition method are shown in Fig. 7. Thetotal voltammetric charge can be extrapolated from the plot of 1/qvs. �1/2 as � → 0 (Fig. 7b), and the double-layer charge, qdl can beestimated from the plot of q vs. �−1/2 as � → ∞ (Fig. 7a) [48]. Inthis analysis, a linear fit is used for the data at low sweep rates (2,5, 10, and 20 mV s−1), which demonstrated the linear dependencebetween q and �−1/2 and 1/q vs. �1/2. At higher rates polarizationcan occur, and thus are not included in the in the linear fit [51].The fitted regions are represented by the shaded regions on Fig. 7aand b. The total and double layer charge storage values (q) wereobtained from the cathodic portion of the CV curve:

q =∫

Idt (5)

Intuitively, as the sweep rate goes to infinity, the total chargewill be stored electrostatically in the double layer, because faradaicreactions are too slow to occur. Furthermore, as the sweep rateapproaches to zero, or a very slow rate, one can estimate thetotal charge storage possible because both faradaic and doublelayer charge storage mechanism can occur concurrently. Thus,these approximations are valid assumptions for deciphering thedifferent storage mechanisms. Finally, the difference between thetotal charge and the double-layer charge is assumed to equal

the pseudocapacitive charge storage contributions. In Fig. 7c, thepseudocapacitive contributions increase with concentration, whiledouble layer storage appears to remain relatively stable acrossall concentrations studied. The stable double layer charge storage

894 K.B. Hatzell et al. / Electrochimica

Fig. 7. (a) Double layer charge storage contribution for each concentration of p-phenylenediamine (� = sweep rate), (b) the total charge storage mechanism, derivedas the sweep rate approaches very 0 mV s−1, (c) partitioned charge storage quantities(total, double layer, and pseudocapacitive contributions).

Acta 111 (2013) 888– 897

mechanism can also be observed in Fig. 4a as the current den-sity after the redox peak for each concentration is approximatelyequal. Furthermore, this is expected because for all slurries stud-ied, the same active material with the similar surface areas wereused. Overall, double layer charge storage appears to contributea greater proportion of the charge storage over all concentrationsof PPD tested, except at 0.277 M of PPD, where double layer andpseudocapacitive charge storage mechanisms are approximatelyequal. This analysis demonstrates that the highest charge storagewas observed in the slurry that had 0.277 M of PPD. This observationis qualitatively confirmed, by the CV plot shown in Fig. 4a as 0.277 Mdemonstrates the greatest area at 2 mV s−1. However, at increasedrates it is expected that the charge storage at 0.277 M would besub-optimal to 0.139 M because it is performance degrades above2 mV s−1 (Fig. 6).

Fig. 8 shows typical galvanostatic cycling for the redox medi-ated slurries at a charge rate of 200 mA g−1. Each curve shows asimilar charge and discharge time which indicates good coulombicefficiency across all redox slurries examined. Typical supercapcac-itors demonstrate triangular shape galvanostatic curves. However,for these pseudocapacitive slurries, we see reduction and oxidationbumps in the middle region, which indicates that redox reactionsoccur. The fact that charge and discharge curves are symmetricemphasizes the reversibility of the redox-couple in KOH elec-trolytes. Furthermore, the slurry with 0.139 M of PPD charges in∼600 s (∼6 C), which demonstrates its ability charge and dischargequickly, a characteristic that is desired in grid energy storage appli-cations. The equivalent series resistance which reduces the poweroutput of supercapacitors is found to be low, around 1.5 � cm2

(Fig. 8a). Moreover, Fig. 8b shows that the capacitance appears to beoptimized at 0.139 M of PPD in 2 M KOH. In the CV experiments itwas shown that for rates above 2 mV s−1, 0.139 M of PPD surpasses0.277 M PPD in terms of capacitance. This trend is confirmed withthe galvanostatic results shown in Fig. 8.

Fig. 9 shows the Nyquist plot for all redox-mediated slur-ries. The Nyquist plots exhibit a small semi-circle in the highfrequency region, and a linear region at low frequency. The imag-inary impedance is linearly related to the real impedance at lowfrequency which indicates a diffusion controlled system. Withincreasing concentrations of PPD, we see a decrease in the ohmicresistance (R�) from 0.8 to 0.2 � cm2. The maximum specific powerfor a supercapacitor can be found from:

Pmax = V2max

4R˝m(6)

where V is the voltage window, and m is the mass of the activematerial in both electrodes. When all the parameters are held con-stant, except for concentration, the only parameter that changesis the ohmic resistance. Thus, we see with increasing concentra-tions of PPD, R� decreases and thus the maximum specific powerincreases.

A typical Randles circuit was employed to distinguish the effectsof PPD concentration on the faradaic impedance (Fig. 9b inset). ARandles circuit is comprised of a resistor (which represents ohmicor solution resistance) in series with a parallel RC circuit. The par-allel RC circuit is characterized by a branch with a capacitor whichrepresents pure capacitance, and another branch that captures thefaradaic impedance. The faradaic impedance is comprised of chargetransfer resistance and a Warburg impedance which representsmass transfer losses [52]. For the pseudocapacitive slurry, we areinterested in examining the changes in Warburg impedance with

concentration of PPD. The Warburg impedance, for the equivalentcircuit described above can be calculate from [39,52]:

Zw = [�ω−1/2 − j(�ω−1/2)] = ω−1/2�[1 − j] (7)

K.B. Hatzell et al. / Electrochimica Acta 111 (2013) 888– 897 895

Fig. 8. (a) Galvanostatic charging/discharging at a constant current density of 200 mA g−1 for the various concentrations of PPD in 2 M KOH electrolyte. (b) Capacitanced

witWi

�

wcfctttt

ttdmtt

Fo

ependence on concentration of PPD in 2 M KOH electrolyte.

here j is the imaginary number, ω is the angular frequency, and �s Warburg coefficient. Thus the slope of a plot of ω−1/2 versus eitherhe real or imaginary part of the impedance spectra provides the

arburg coefficient (Randles plot, Fig. 9b). The Warburg coefficients defined as [52]:

= RT

nF2A√

2

(1

D1/20 C∗

0

)(8)

here R is the ideal gas constant, T is the temperature, n is theharge transfer number, A is the area of the electrode surface, F isaraday’s constant, D0 is the diffusion coefficient, and C0 is the bulkoncentration. The effects on the diffusion coefficient with concen-ration of PPD can be determined qualitatively by examining howhe magnitude of the Warburg coefficient changes with concentra-ion of PPD because the diffusion coefficient is inversely related tohe Warburg coefficient.

The slope of the Randles plot is �, the Warburg coefficient. Fromhe Randles plot (Fig. 9b) it can be seen that increasing the concen-ration of PPD decreases the ohmic resistance, and increases the

iffusion coefficient (Fig. 9b). The diffusion coefficient reflects theobility of the ionic species across the supercapacitor. Increasing

he diffusion coefficient is advanatageous for obtaining fast reac-ions. Thus, the addition of PPD increases the performance of the

ig. 9. (a) Nyquist plots for the various redox slurries in 2 M KOH electrolyte across the fren KOH electrolyte, with an inset of a Randles circuit.

supercapacitor initially by both decreasing the ohmic resistanceand increasing the diffusion coefficient of the ionic species.

Fig. 10b shows the specific discharge capacitances for a slurrycontaining 0.139 M PPD in 2 M KOH during successive cycling.Over 1000 CV cycles (20 mV s−1), an 11% decrease in capacitancewas observed. In typical pseudocapacitor systems cycle life isoften an issue because volume changes during the redox reaction(metal oxides and conducting polymers) leads to losses of activematerial [53]. This loss is typically alleviated by integrating pseu-docapacitive material into a carbon support. In our carbon-basedredox-mediated slurries the redox peaks appears to compresstoward the inside of the CV curve with cycling, which represents adecrease in capacity (Fig. 10a). It can be assumed from this transi-tion that the system is losing faradaic charge storage over the cyclelifetime as a result of possible loss of active PPD. This is evidentbecause the flat region beyond the redox bump remains relativelystable, indicating that we are not losing double layer charge stor-age with cycling. The loss of faradaic charge storage could be due todegradation of the redox-active PPD during cycling. Furthermore,deterioation might be related to pH-related changes of the elec-

trolyte which would stifle redox reactions. This test was done in astatic cell, which does not fully mimic the conditions that wouldbe encountered during flow cell operations. In an actual flow sys-tem, a simple feedback control system could be implemented in

quency range from 10 mHz to 200 kHz. (b) Randles plot of the redox slurries based

896 K.B. Hatzell et al. / Electrochimica Acta 111 (2013) 888– 897

F ious co

tac

4

ccrcitmousscitsascttiTrt

A

IEofsFw1lpf

[

[

[

[

[

[

[

[

[

[

ig. 10. (a) Cyclic voltammogramof 0.139 M PPD in 2 M KOH electrolyte slurry at varf cycle number.

he storage tanks to ensure that pH remained constant, and thective material weight percentage remains constant, which mayontribute to better electrolyte stability.

. Conclusions

The high power, low cost, and long lifetime characteristics of theapacitive flowable carbon-based electrodes in the electrochemi-al flow capacitor (EFC) offer the opportunity to address a diverseange of applications on the grid level. The limited energy density,ommon in carbon-based electrodes that use physical adsorption ofons as their charge storage mechanism, can be enhanced throughhe addition of pseudocapacitance. This study introduced a redox-

ediated slurry as a facile method for increasing the energy densityf the flowable electrodes for the EFC. A pseudocapacitive slurrytilizing a PPD/PPI redox-couple in 2 M KOH electrolyte demon-trated high capacitance for the EFC. The alkaline redox-mediatedlurry shows increases in capacitance on the order of 86% whenompared alkaline solutions without redox mediators, and a 130%ncrease compared to previously reported neutral slurries [12]. Fur-hermore, the rate handling performance of the redox-mediatedlurry increased from approximately 64% capacitance decay tobout 57% between rates of 2–100 mV s−1. The redox-mediatedlurry is found to decrease the ohmic resistance, with increasingoncentrations of PPD, which can be described by the increase inhe ionic diffusion coefficient. Among tested slurries, a concentra-ion of 0.139 M of PPD in 2 M KOH yielded the highest capacitancesn both CV and GC experiments over a wide range of charging rates.he high performance is attributed to the addition of quick redoxeactions at the electrolyte/electrode interface as PPD undergoes awo-proton/two-electron reduction and oxidation during cycling.

cknowledgements

Y.G. and M.B. would like to acknowledge support from the Fluidnterface Reactions, Structures and Transport (FIRST) Center, annergy Frontier Research Center funded by the U.S. Departmentf Energy, Office of Science, Office of Basic Energy. Partial supportrom the Ben Franklin Technology Partners of Southeastern Penn-ylvania group (Grant# 001389-002) and the National Scienceoundation (NSF) (Grant# 1242519) is also appreciated. K.B.H.as supported by the NSF Graduate Research Fellowship (Grant#

002809). J.C. was supported by the NSF Bridge to the Doctorate Fel-owship (Grant# 1026641). C.R.D. was supported by the NSF IGERTrogram (Grant# 0654313). The authors would also like B. Dyatkinor assistance with gas adsorptions data and helpful conversations.

[

[

ycles over a lifetime test. Static cell cycled at 20 mV s−1. (b) Capacitance as a function

SEM was carried out using instruments in the Centralized ResearchFacility (CRF) of the College of Engineering at Drexel University

References

[1] B. Dunn, H. Kamath, J.-M. Tarascon, Electrical energy storage for the grid: abattery of choices, Science 334 (2011) 928–935.

[2] M. Armand, J.-M. Tarascon, Building better batteries, Nature 451 (2008)652–657.

[3] P. Simon, Y. Gogotsi, Materials for electrochemical capacitors, Nat. Mater. 7(2008) 845–854.

[4] J.-M. Tarascon, M. Armand, Issues and challenges facing rechargeable lithiumbatteries, Nature 414 (2001) 359–367.

[5] Z. Yang, J. Zhang, M.C. Kintner-Meyer, X. Lu, D. Choi, J.P. Lemmon, J. Liu, Elec-trochemical energy storage for green grid, Chem. Rev. 111 (2011) 3577.

[6] Department of Energy, Basic Research Needs for Electrical Energy Storage, 2007.[7] M. Anderman, Status and prospects technology for hybrid electric vehicles,

including plug-in hybrid vehicles, Briefing to the U.S. Senate Committee onEnergy and Natural Resources by the President of Advanced Automotive Bat-teries, 2007.

[8] C.J. Barnhart, S.M. Benson, On the importance of reducing the energetic andmaterial demands of electrical energy storage, Energy Environ. Sci. 6 (2013)1083–1092.

[9] J. Liu, Addressing the grand challenges in energy storage, Adv. Funct. Mater. 23(2013) 924–928.

10] M. Skyllas-Kazacos, M. Rychcik, R.G. Robins, A. Fane, M. Green, New all-vanadium redox flow cell, J. Electrochem. Soc. 133 (1986) 1057–1058.

11] M. Duduta, B. Ho, V.C. Wood, P. Limthongkul, V.E. Brunini, W.C. Carter,Y.M. Chiang, Semi-solid lithium rechargeable flow battery, Adv. Energy Mater.1 (2011) 511–516.

12] V. Presser, C.R. Dennison, J. Campos, K.W. Knehr, E.C. Kumbur, Y. Gogotsi, Elec-trochemical flow cells: the electrochemical flow capacitor: a new concept forrapid energy storage and recovery, Adv. Energy Mater. 2 (2012) 911.

13] Y. Gogotsi, V. Presser, E. Kumbur, Electrochemical Flow Capacitor. WO PatentApplication 2,012,112,481, 2012.

14] B. Kastening, T. Bointowitz, M. Heins, Design of a slurry electrode reactor sys-tem, J. Appl. Electrochem. 27 (1997) 147–152.

15] M. Skyllas-Kazacos, M. Chakrabarti, S. Hajimolana, F. Mjalli, M. Saleem, Progressin flow battery research and development, J. Electrochem. Soc. 158 (2011)R55–R79.

16] K. Fic, G. Lota, M. Meller, E. Frackowiak, Novel insight into neutral mediumas electrolyte for high-voltage supercapacitors, Energy Environ. Sci. 5 (2012)5842–5850.

17] L. Demarconnay, E. Raymundo-Pinero, F. Béguin, A symmetric carbon/carbonsupercapacitor operating at 1.6 V by using a neutral aqueous solution, Elec-trochem. Commun. 12 (2010) 1275–1278.

18] Q. Gao, L. Demarconnay, E. Raymundo-Pinero, F. Béguin, Exploring the largevoltage range of carbon/carbon supercapacitors in aqueous lithium sulfateelectrolyte, Energy Environ. Sci. 5 (2012) 9611–9617.

19] J. Campos, M. Beidaghi, K.B. Hatzell, C. Dennison, V. Presser, E.C. Kumbur, Y.Gogotsi, Investigation of carbon materials for use as a flowable electrode inelectrochemical flow capacitors, Electrochim. Acta 98 (2013) 123–130.

20] X. Lang, A. Hirata, T. Fujita, M. Chen, Nanoporous metal/oxide hybrid electrodesfor electrochemical supercapacitors, Nat. Nanotechnol. 6 (2011) 232–236.

21] H. Yu, J. Wu, J. Lin, L. Fan, M. Huang, Y. Lin, Y. Li, F. Yu, Z. Qiu, A reversibleredox strategy for SWCNT-based supercapacitors using a high-performanceelectrolyte, ChemPhysChem 14 (2013) 394–399.

imica

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

K.B. Hatzell et al. / Electroch

22] B. Conway, V. Birss, J. Wojtowicz, The role and utilization of pseudocapac-itance for energy storage by supercapacitors, J. Power Sources 66 (1997)1–14.

23] T. Brezesinski, J. Wang, S.H. Tolbert, B. Dunn, Ordered mesoporous [alpha]-MoO3 with iso-oriented nanocrystalline walls for thin-film pseudocapacitors,Nat. Mater. 9 (2010) 146–151.

24] V. Khomenko, E. Frackowiak, F. Beguin, Determination of the specific capaci-tance of conducting polymer/nanotubes composite electrodes using differentcell configurations, Electrochim. Acta 50 (2005) 2499–2506.

25] T. Brousse, P.-L. Taberna, O. Crosnier, R. Dugas, P. Guillemet, Y. Scudeller, Y.Zhou, F. Favier, D. Bélanger, P. Simon, Long-term cycling behavior of asymmet-ric activated carbon/MnO2 aqueous electrochemical supercapacitor, J. PowerSources 173 (2007) 633–641.

26] S. Roldán, C. Blanco, M. Granda, R. Menéndez, R. Santamaría, Towards a fur-ther generation of high energy carbon based capacitors by using redox-activeelectrolytes, Angew. Chem. Int. Ed. 50 (2011) 1699–1701.

27] H. Yu, J. Wu, L. Fan, Y. Lin, K. Xu, Z. Tang, C. Cheng, S. Tang, J. Lin, M. Huang, A novelredox-mediated gel polymer electrolyte for high-performance supercapacitor,J. Power Sources 15 (2011) 402–407.

28] S. Horvath, L.E. Fernandez, A.M. Appel, S. Hammes-Schiffer, pH-Dependentreduction potentials and proton-coupled electron transfer mechanisms inhydrogen-producing nickel molecular electrocatalysts, Inorg. Chem. 52 (2013)3643–3652.

29] N.J. Ke, S.-S. Lu, S.-H. Cheng, A strategy for the determination of dopamine ata bare glassy carbon electrode: p-phenylenediamine as a nucleophile, Elec-trochem. Commun. 8 (2006) 1514–1520.

30] L.A. Clare, L.E. Rojas-Sligh, S.M. Maciejewski, K. Kangas, J.E. Woods, L.J. Deiner,A. Cooksy, D.K. Smith, The effect of H-bonding and proton transfer on thevoltammetry of 2,3,5,6-tetramethyl-p-phenylenediamine in acetonitrile. Anunexpectedly complex mechanism for a simple redox couple, J. Phys. Chem.C 114 (2010) 8938–8949.

31] S. Roldán, Z. González, C. Blanco, M. Granda, R. Menéndez, R. Santamaría, Redox-active electrolyte for carbon nanotube-based electric double layer capacitors,Electrochim. Acta 56 (2011) 3401–3405.

33] S. Roldan, M. Granda, R. Menendez, R. Santamaria, C. Blanco, Mechanisms ofenergy storage in carbon-based supercapacitors modified with a quinoid redox-active electrolyte, J. Phys. Chem. C 115 (2011) 17606–17611.

34] M. Hébert, D. Rochefort, Electrode passivation by reaction products of theelectrochemical and enzymatic oxidation of p-phenylenediamine, Electrochim.Acta 53 (2008) 5272–5279.

35] H. Yu, J. Wu, L. Fan, G.G. Luo, J. Lin, M. Huang, A simple and high-effective

electrolyte mediated with p-phenylenediamine for supercapacitor, J. Mater.Chem. 22 (2012) 19025–19030.

36] H. Yu, J. Wu, L. Fan, Y. Lin, S. Chen, Y. Chen, J. Wang, M. Huang, J. Lin, Z. Lan,Application of a novel redox-active electrolyte in MnO2-based supercapacitors,Sci. China Chem. 55 (2012) 1319–1324.

[

[

Acta 111 (2013) 888– 897 897

37] Z. Liu, H. Zhou, Z. Huang, W. Wang, F. Zeng, Y. Kuang, Graphene covalentlyfunctionalized with poly (p-phenylenediamine) as high performance electrodematerial for supercapacitors, J. Mater. Chem. A 1 (10) (2013) 3454–3462.

38] H. Yu, L. Fan, J. Wu, Y. Lin, J. Lin, M. Huang, Z. Lan, Redox-active alkaline elec-trolyte for carbon-based supercapacitor with pseudocapacitive performanceand excellent cyclability, RSC Adv. 2 (17) (2012) 6736–6740.

39] B.G. Choi, M.H. Yang, S.C. Jung, K.G. Lee, J.-G. Kim, H. Park, T.J. Park, S.B. Lee,Y.-K. Han, Y.S. Huh, Enhanced pseudocapacitance of ionic liquid/cobalt hydrox-ide nanohybrids, ACS Nano 7 (2013) 2453–2460.

40] M.D. Stoller, R.S. Ruoff, Best practice methods for determining an electrodematerial’s performance for ultracapacitors, Energy Environ. Sci. 3 (2010)1294–1301.

41] D. Andre, M. Meiler, K. Steiner, C. Wimmer, T. Soczka-Guth, D.U. Sauer, Charac-terization of high-power lithium-ion batteries by electrochemical impedancespectroscopy. I. Experimental investigation, J. Power Sources 12 (2011) 8.

42] X.-G. Li, M.-R. Huang, W. Duan, Y.-L. Yang, Novel multifunctional polymersfrom aromatic diamines by oxidative polymerizations, Chem. Rev. 102 (2002)2925–3030.

43] W.R. Heineman, H.J. Wieck, A.M. Yacynych, Polymer film chemically modifiedelectrode as a potentiometric sensor, Anal. Chem. 52 (1980) 345–346.

44] B. Conway, Electrochemical Supercapacitors: Scientific Fundamentals andTechnological Applications, Kluwer Academic/Plenum, New York, 1999.

45] M. Bichat, E. Raymundo-Pinero, F. Béguin, High voltage supercapacitor builtwith seaweed carbons in neutral aqueous electrolyte, Carbon 48 (2010)4351–4361.

46] K. Fic, G. Lota, E. Frackowiak, Effect of surfactants on capacitance properties ofcarbon electrodes, Electrochim. Acta 60 (2012) 206–212.

47] D.N. Futaba, K. Hata, T. Yamada, T. Hiraoka, Y. Hayamizu, Y. Kakudate, O.Tanaike, H. Hatori, M. Yumura, S. Iijima, Shape-engineerable and highlydensely packed single-walled carbon nanotubes and their application as super-capacitor electrodes, Nat. Mater. 5 (2006) 987–994.

48] Y.-H. Lee, K.-H. Chang, C.-C. Hu, Differentiate the pseudocapacitance anddouble-layer capacitance contributions for nitrogen-doped reduced grapheneoxide in acidic and alkaline electrolytes, J. Power Sources 227 (2012) 300–308.

49] S. Ardizzone, G. Fregonara, S. Trasatti, Inner and outer active surface of RuO2

electrodes, Electrochim. Acta 35 (1990) 263–267.50] K.-H. Chang, C.-C. Hu, C.-Y. Chou, Textural, Capacitive characteristics of

hydrothermally derived RuO2·H2O nanocrystallites: independent control ofcrystal size and water content, Chem. Mater. 19 (2007) 2112–2119.

51] J.W. Kim, V. Augustyn, B. Dunn, The effect of crystallinity on the rapid pseudo-capacitive response of Nb2O5, Adv. Energy Mater. 2 (2012) 141–148.

52] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals and Applica-tions, Wiley, New York, 1980.

53] J. Yan, Z. Fan, T. Wei, W. Qian, M. Zhang, F. Wei, Fast and reversible surface redoxreaction of graphene–MnO2 composites as supercapacitor electrodes, Carbon48 (2010) 3825–3833.