The kinetics of the Cu2+/Cu+ redox couple in deep eutectic solvents

Page 1

T

Da

b

a

ARRAA

KDIECCI

1

ldaptbe

pttw31oii

ecpe

0d

Electrochimica Acta 56 (2011) 4942–4948

Contents lists available at ScienceDirect

Electrochimica Acta

journa l homepage: www.e lsev ier .com/ locate /e lec tac ta

he kinetics of the Cu2+/Cu+ redox couple in deep eutectic solvents

avid Lloyda, Tuomas Vainikkaa, Lasse Murtomäkia,∗, Kyösti Kontturi a, Elisabet Ahlbergb

Aalto University, Department of Chemistry, Kemistintie 1, PO Box 16100, 00076, Aalto, FinlandUniversity of Gothenburg, Department of Chemistry, SE-412 96 Göteborg, Sweden

r t i c l e i n f o

rticle history:eceived 31 January 2011eceived in revised form 28 March 2011ccepted 29 March 2011vailable online 7 April 2011

a b s t r a c t

Kinetics of electron transfer of the Cu(I)/Cu(II) redox couple at a platinum electrode has been studied withchronoamperometry, cyclic voltammetry and impedance spectroscopy in a deep eutectic solvent consist-ing of choline chloride and ethylene glycol. At 25 ◦C, the reaction was found to be quasi-reversible with arelatively high rate constant k0 of 9.5 ± 2 × 10−4 cm s−1, and a charge transfer coefficient ˛ of 0.25 ± 0.05.Diffusion coefficients for the Cu(I) and Cu(II) complexes were determined to be 2.7 ± 0.1 × 10−7 and1.5 ± 0.1 × 10−7 cm2 s−1, respectively. The viscosity of the electrolyte was 41 ± 3 mPa s. The temper-

eywords:eep eutectic solvent

onic liquidlectrochemical kineticsopperyclic voltammetry

mpedance spectroscopy

ature dependency was also investigated. The activation energy of mass transfer was found to be27.7 ± 1 kJ mol−1 and that of electron transfer 39 ± 7 kJ mol−1. Speciation of the Cu(I) and Cu(II) com-plexes was determined using UV–VIS spectroscopy, and the prevailing Cu(I) complex was found to be[CuCl3]2− and that of Cu(II) [CuCl4]2−.

© 2011 Elsevier Ltd. All rights reserved.

. Introduction

Room temperature ionic liquids (RTILs) are molten salts that areiquid at 20 ◦C [1]. The ions are assumed to show a high degree ofissociation. Due to the wide selection of cations and anions nowvailable it is possible to design liquids with a range of physicalroperties. A property of particular interest in the field of elec-rochemistry is the window of electrochemical stability. This haseen shown to be exceptionally large for some ionic liquids [1] andxpands the range of metals which can be electrodeposited.

Electrodeposition of metals from ionic liquids became an area oferiodic activity from the 1930s onwards [2]. The first major break-hrough was the development of aprotic haloaluminate eutectics inhe 1950s by Hurley and coworkers [3]. A further improvementas the development of the aluminium chloride and 1-methyl-

-ethylimidazolium chloride (AlCl3 and EMIM-Cl) system in the980s. This mixture forms a RTIL by formation of either AlCl4−

r Al2Cl7− complexes and has two, relatively wide low melt-ng regions. The biggest drawback to chloroaluminate systemss their reactivity with water, which drastically limits possible

Abbreviations: RTIL, room temperature ionic liquid; EMIM-Cl, 1-methyl-3-thylimidazolium chloride; Tf2N, bis(trifluoromethyl)sulfonylamide; ChCl, cholinehloride; DES, deep eutectic solvent; CPE, constant phase element; CA, chronoam-erometry; IS, impedance spectroscopy; CV, cyclic voltammetry; RDE, rotating disclectrode.∗ Corresponding author. Tel.: +358 9 470 22575; fax: +358 9 470 22580.

E-mail address: [email protected] (L. Murtomäki).

013-4686/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.oi:10.1016/j.electacta.2011.03.133

applications. From the 1990s onwards attention has shifted to theuse of other anions, for instance bis(trifluoromethyl)sulfonylamide[N(CF3SO2)2

−], which result in liquids that are stable in the pres-ence of air and water.

Deep eutectic solvents (DES) are a subclass of ionic liquidsformed by complexation of the anion in a halide salt by meansof hydrogen bonding. The prototypical salt is choline chloride(ChCl or (2-hydroxyethyl)trimethylammonium chloride) and typi-cal hydrogen bond donors are either primary amines [4], diols [5] orcarboxylic acids [6]. DES based on these combinations show a sig-nificant capacity to dissolve metal chlorides, making them simple touse for studies of metal complex electrochemistry and electrode-position in particular [7]. The primary advantages of DES are theextremely low cost of the precursors and their general biodegrad-ability.

DESs are inferior to other RTILs on a number of important points.Firstly, they have a much narrower window of electrochemicalstability, which limits the range of metals that can be deposited.Secondly, they are hygroscopic, which does not appear to lead toelectrolyte decomposition, but does complicate the reproducibleperformance of measurements in anything other than a carefullycontrolled atmosphere. Thirdly, they can exhibit significant volatil-ity if the hydrogen bond donor molecule is, for instance, ethyleneglycol.

One of the biggest problems in the reporting of DES propertiesis the wide difference between melting point and fusion tempera-ture, sometimes up to 125 ◦C [6]. Since a DES is formed by mixingthe two components at elevated temperature and then cooling,

Page 2

ica Acta 56 (2011) 4942–4948 4943

tira

ticlcIie

cqbtdi

Ftptptii

2

icbftsrip

2

oirfa

O

Tu

k

k

wc

10−2

10−1

100

−1.1

−1

−0.9

−0.8

−0.7

−0.6

−0.5

−0.4

−0.3

α = 0.70

α = 0.60

α = 0.50

α = 0.40

α = 0.30

Ψ

I p,a

0/I p,

c0

D. Lloyd et al. / Electrochim

his can lead to some confusion on whether a particular DES isn fact in a metastable region and over what time scale it willemain there. Certain DES can therefore defy simple classifications RTIL.

The most commonly reported figures of merit for ionic liquids,hat are relevant to electrochemistry, are conductivity and viscos-ty. As shown recently by Abbott et al. [8], one potential drawback toomparing ionic liquids on this basis is that the anion of the ioniciquid may form complexes with dissolved metal species. In thatase the electroactive species will vary from one liquid to the next.n any case, conductivity and viscosity measurements provide nonformation on the kinetics of electrode processes, which can bequally limiting.

The combination of methods such as cyclic voltammetry,hronoamperometry and impedance spectroscopy offer a means touantitatively determine the kinetics of electrochemical processes,oth at the electrode and in the electrolyte. This work demonstrateshe use of these methods to quantify the Cu2+/Cu+ redox couple in aeep eutectic solvent and compares the results with previous work

n ionic liquids and in water.The Cu2+/Cu+ system is of interest for a number of reasons.

irstly, in most electrolyte systems copper ions will be oxidisedo Cu2+ in the presence of oxygen. Hence, the first step in any cop-er electro-deposition process will often be the reduction of Cu2+

o Cu+. Secondly, the Cu2+/Cu+ redox couple is used in industrialrocesses, such as Outotec’s HydroCopper process for the produc-ion of copper from sulfide ores [9]. The application of ionic liquidsn processes of this kind offers significant possibilities for processnnovation due to their unique and tuneable properties.

. Theory

The review paper by Abbott et al. [7] offers the first exper-mental data hinting at reversible behaviour of the Cu2+/Cu+

ouple in a DES. This behaviour has been further demonstratedy Abbot et al. in a recent paper on copper electrodepositionrom a DES [10]. In the results presented below, this is showno be approximately true in the limiting case of exceptionallylow measurements. Under other conditions the system is quasi-eversible and while this kind of behaviour is less simple to analyset does offer the possibility to determine kinetic data for electroderocesses.

.1. Cyclic voltammetry

The first approach used to determine kinetic parameters is basedn the method of Nicholson [11]. This method utilises the increasen peak separation with scan rate to determine the heterogeneousate constant, k0, for a simple one electron transfer, as shown belowor the general case of an oxidised species O (corresponding to Cu2+)nd a reduced species R (corresponding to Cu+).

+ e− kf�kb

R (1)

he rate constants, kf and kb are approximated in the usual way

sing the Butler–Volmer kinetics [12].

f = k0 exp(−˛f (E − E0′)) (2)

b = k0 exp((1 − ˛)f (E − E0′)) (3)

here k0 is the heterogeneous rate constant, ˛ the transfer coeffi-ient, f = F/RT, all other symbols have their usual meaning.

Fig. 1. Dependency of the ratio of peak currents on ˛ and , simulated data(DO = 1.5 × 10−7 cm2 s−1, DR = 2.8 × 10−7 cm2 s−1, k0 = 9.5 × 10−4 cm s−1, T = 298.15 K,� was varied between 0.01 and 50 V/s).

Assuming that all current observed is due to the Faradaic processgiven in Eq. (1) and that linear diffusion holds, the fluxes of the twocomponents at the interface can be related to the observed currentby Fick’s first law, as usual:

i

FA= −DO

∂CO

∂x

∣∣∣∣x=0

= DR∂CR

∂x

∣∣∣∣x=0

(4)

where DO and DR are the diffusion constants, CO and CR the con-centration of oxidised and reduced species, i.e. Cu2+ and Cu+,respectively, all other symbols have their usual meaning.

Nicholson postulated that if the switching potential is chosento be sufficiently large with respect to the cathodic peak, then thepeak separation is largely independent of the transfer coefficient,˛, provided it lies within the range of 0.3–0.7. This was confirmedin the simulations performed in this work only for the case of highvalues of , as shown in Section 4.2. For such a case it is then possi-ble to tabulate the dependency of the peak separation on the extentof reversibility. Nicholson introduced the dimensionless number as a measure of reversibility in cyclic voltammetry:

� = (DO/DR)˛/2k0√�DOf�

(5)

where � is the scan rate and all other symbols have their usualmeaning.

If diffusion coefficients have been determined, either from cyclicvoltammetry within the fully reversible regime or using a suitablechronoamperometric technique, k0 can be estimated by measuringcyclic voltammograms at various scan rates and fitting the observedvariation in peak separation to tabulated values.

Although the peak separation is a poor indicator of the trans-fer coefficient ˛, the ratio of the peak currents shows a strongdependency. Simulated data is shown in Fig. 1, based on realisticparameters taken for the redox system presented in this work. Forboth the anodic and cathodic peak currents, I0p,a and I0p,c, the currentis taken from a zero baseline.

By determining this ratio over a range of scan rates it is possi-ble to also estimate ˛ and k0. This method is particularly suited toresistive electrolytes, such as ionic liquids, since the error in thepeak separation, due to poorly compensated solution resistance,can mimic quasi-reversible kinetics whereas the ratio of currentsis less sensitive. It should be noted that the theoretical ratios shown

here are applicable in other quasi-reversible systems only in a gen-eral sense, accurate estimation of ˛ is best achieved by performingsimulations which take in to account the specific switching poten-tial of the experiment relative to E0′

. Simulations were performed

Page 3

4 ica A

utm

2

tbrlc[fb

R

Z

IAtsa

C

wr

wu

C

C

Tcf

R

Z

Ktd

3

si9

50 ◦C by cycling the shear rate between 0.01 and 100 s−1. This wasfollowed by temperature ramp measurements at a fixed shear rateof 100 s−1.

0

0.5

1

1.5

2

Abs

orba

nce/

a.u.

Cu(II) complexCu(I) complex

944 D. Lloyd et al. / Electrochim

sing the PDEPE function in Matlab with relative and absoluteolerances of 10−6 and 10−12, respectively. A logarithmically spaced

esh of 100 elements was utilised.

.2. Impedance spectroscopy

In the case of a quasi-reversible single electron transfer reac-ion, the experimentally determined impedance spectrum cane fitted to the Randles’ equivalent circuit. This well knownesult is theoretically predicted by Eqs. (2)–(4) in the case ofinear diffusion through a layer of infinite thickness. A typi-al example of a theoretical derivation is provided by Gabrielli13], who gives a generalised description for the charge trans-er resistance, Rct, and Warburg impedance, Zw as shownelow.

ct = RT

n2F2A.

1˛kfCO,DC + (1 − ˛)kbCR,DC

(6)

w = Rct√jω

(kf√DO

+ kb√DR

)(7)

n Eq. (6), j is the imaginary unit, ω the angular frequency of theC excitation and CO,DC and CR,DC are the surface concentrations of

he oxidised and reduced species at the applied DC potential. Theurface concentration can be related with the bulk concentrationss [12]

R,DC − CbR = �(Cb

O − CO,DC); � =√DO

DR(8)

here CbO and Cb

R are the bulk concentrations of the oxidised andeduced species. Noticing that

iDC

nF= kbCR,DC − kfCO,DC (9)

here iDC is the DC current, the surface concentrations can be eval-ated from Eqs. (8) and (9) as

R,DC = CbR + �Cb

O + ��˛iDC/Fk0

1 + �� ≈ CbR + �Cb

O

1 + �� (10)

O,DC = �(CbR + �Cb

O) − �˛iDC/Fk0

1 + �� ≈ �(CbR + �Cb

O)

1 + �� ;

� = exp[nF

RT(E − E0′

)]

(11)

he approximate form is generally applicable in cases where theurrent density is low. Eqs. (6) and (7) can then be expressed asollows:

ct = RT

n2F2A· 1

�CbO + Cb

R

· 1k0

· 1 + ���1−˛ (12)

w = RT

n2F2A· 1

�CbO + Cb

R

· 1√DO

· (1 + ��)2

�˛· 1 − j√

2ω(13)

inetic parameters can be estimated by fitting these equationshrough values of circuit elements derived by fitting of impedanceata.

. Experimental

Measurements were performed between 25 ◦C and 50 ◦C in 5 ◦Cteps using a 20 mM solution of CuCl2 (Alfa Aesar, 99.95%, ultra dry)n a DES consisting of ethylene glycol (Sigma–Aldrich, anhydrous,9.8%) and choline chloride (Alfa Aesar, 98+%). This combination

cta 56 (2011) 4942–4948

is commonly referred to as ethaline. Choline chloride was recrys-tallised from absolute ethanol (99.5%, Altia Corp.), filtered and driedin the antechamber of the glovebox overnight under vacuum, allother chemicals were transferred directly in to the glovebox andused as received. The components of the DES were combined in a2:1 molar ratio using the method described by Abbott et al. [5]. Cu(I)electrolyte was prepared by performing comproportionation of theCu(II) electrolyte with copper powder (J.T. Baker, >99%) to yield a40 mM solution of the Cu(I) complex. Electrolytes were preparedand transferred in to an airtight cell inside an argon-filled glovebox(VAC Mo-5, 1.09 ppm H2O, O2 < 2 ppm).

The working electrode was a 5 mm Pt disc with a Teflonsheath (Pine Research Instrumentation AFE3T050PT) polished byfirst sanding with carbimet paper (600 grit, Buehler) and then1 and 0.05 �m alumina paste on a microcloth (Buehler). A plat-inum wire (diameter 0.4 mm) was used as a pseudo referenceand the counter electrode was a platinum foil (A = 3 cm2). Theuse of a platinum quasi reference has already been reported byAbbott et al. [14]. All electrodes were ultrasonicated in distilledwater and absolute ethanol prior to use. The temperature of theelectrolyte was directly monitored using a glass encased k-typethermocouple.

Electrochemical measurements were performed using aSolartron 1286 potentiostat controlled by a National Instru-ments 6211 multifunction DAQ. The entire system was controlledusing the Matlab data acquisition toolbox and custom Mat-lab software. The impedance response was determined bydigital correlation. Since the electrolyte showed significantresistance, IR compensation was required during cyclic voltam-metry. This was typically 95% of the value determined usingimpedance spectroscopy. Equivalent circuits were fitted to theimpedance data using the Zview software (version 2.3, ScribnerAssociates).

UV–VIS spectra of the Cu(II) and Cu(I) solutions were deter-mined by diluting both samples to a concentration of 0.5 mM andmeasuring in a 1 cm path length quartz cuvette with a Varian Cary50 UV–VIS spectrophotometer.

Viscosity measurements were performed using an Anton PaarPhysica MCR 301 rheometer with a 25 mm diameter plate geometry(Anton Paar PP25/TG, P-PTD 200-56, 0.5 mm gap) incorporating aprotection hood and solvent trap to reduce contamination of thesample. Initially the rheological behaviour was studied at 25 ◦C and

200 250 300 350 400 450 500Wavelength/nm

Fig. 2. UV–VIS spectra of Cu(II) and Cu(I) complexes in ethaline electrolyte, bothdetermined with 0.5 mM of the corresponding copper complex.

Page 4

D. Lloyd et al. / Electrochimica Acta 56 (2011) 4942–4948 4945

−0.6 −0.5 −0.4 −0.3 −0.2 −0.1 0 0.1−2

−1.5

−1

−0.5

0

0.5

1

1.5

2

(j/ν1/

2 )/m

A c

m−

2 V−

1/2 s

1/2

E vs. Pt quasireference/V

Fig. 3. Typical results for cyclic voltammetry at 25 ◦C, current is shown normalisedby the square root of the scan rate. Thirty five scan rates were measured and analysedfor each temperature. The data with the smallest peak separation was determinedaot

4

4

nllc

Flca

4

idtasloCmto

mi

vttitrpdtF

10−1

100

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

α = 0.10

α = 0.15

α = 0.20

α = 0.30

Ψ

ΔEpe

ak/V

theoretical behaviourobserved behaviourreversible limit

Fig. 4. Fitting of peak separations to simulated values (DO = 1.5 × 10−7 cm2 s−1,DR = 2.8 × 10−7 cm2 s−1, k0 = 9.5 × 10−4 cm s−1, �was varied between 0.01 and 50 V/s,T = 298.15 K).

10−2

10−1

100

−1.7

−1.5

−1.3

−1.1

−0.9

−0.7α = 0.40α = 0.35α = 0.30α = 0.25α = 0.20α = 0.15

α = 0.10

Ψ

I p,a

0/I p,

c0

Fig. 5. Theoretical correlation between ratio of peak currents and for variousvalues of ˛ (DO = 1.5 × 10−7 cm2 s−1, DR = 2.8 × 10−7 cm2 s−1, k0 = 9.5 × 10−4 cm s−1,

t a scan rate of 10 mV/s, the data with the largest at 50 V/s. For the sake of claritynly every fifth measurement is shown. The arrow indicates the direction in whichhe cathodic peak shifts with increasing scan rate.

. Results

.1. UV–VIS spectroscopy

The UV–VIS spectrum of the Cu(II) solution, shown in Fig. 2, wasearly identical to that reported by Abbot et al. [10] with a single

arge peak at 288 nm and two smaller peaks at 245 and 404 nm. Theocation of these peaks demonstrates the presence of the [CuCl4]2−

opper complex, as reported by Amuli et al. [15].The UV–VIS spectrum of the colourless Cu(I) solution, shown in

ig. 2, was relatively featureless with only one peak at 271 nm. Theocation of this peak corresponds well with the [CuCl3]2− copperomplex in concentrated chloride solutions, as reported by Sharmand Millero [16].

.2. Cyclic voltammetry

Typical experimental results for cyclic voltammetry are shownn Fig. 3. The data has been acquired at 25 ◦C and normalised byividing with the square root of the scan rate. The general evolu-ion of the voltammogram shape with scan rate agrees well with

quasi-reversible system. If the platinum working electrode isubjected to a copper deposition/stripping cycle, this appears toead to a change in surface structure resulting in the formationf sites where a low valence number product can be adsorbed.yclic voltammetry results acquired at a surface modified in thisanner are extremely difficult to interpret. Hence, it is necessary

o avoid potentials sufficiently negative for copper deposition toccur.

Heterogeneous rate constants were determined by using theethod of Nicholson [11]. The results of one such fitting are shown

n Fig. 4.The Nicholson dataset, correlating the peak separation to the

alue of , was recalculated using numerical simulation at eachemperature. As can be seen in Fig. 4, ˛ is not 0.5 and appearso be around 0.15. As shown in the theory section above, theres a simple extension to the method of Nicholson that utiliseshe ratio of the peak anodic and cathodic currents at lower scanates to determine ˛. Measurements at lower scan rates should

rovide more accurate results since the complicating effects ofouble layer charging and residual uncompensated solution resis-ance are less pronounced. The results of this analysis are shown inig. 5.

� was varied between 0.01 and 50 V/s, T = 298.15 K, indicated by solid lines) andobserved experimental data (indicated by squares).

Fig. 5 suggests that ˛ is between 0.25 and 0.2. Calculation of requires estimation of the diffusion coefficients of Cu(II) andCu(I), which was achieved by performing separate chronoampero-metric measurements with Cu(II) and Cu(I) electrolytes and fittingthe results to the Cottrell equation. Diffusion coefficients for bothspecies were also estimated using cyclic voltammetry data acquiredat the slowest scan rate (10 mV/s) using the Randles-Sevcík equa-tion.

4.3. Impedance spectroscopy

Impedance spectroscopy data could only be fitted to a Randles’circuit within ±0.1 V of the E0′

. A typical measurement and fittingare shown in Fig. 6 below. The Randles’ circuit used required a con-stant phase element (CPE) in the place of the traditional capacitor;the phase of the CPE was close to 0.9 in all measurements.

Once the impedance data has been fitted to the Randles’ circuitthe resulting values for the charge transfer resistance and the War-burg impedance can in turn be fitted to Eqs. (6) and (7). Examplesof such a fitting process are shown in Figs. 7 and 8 below.

The Warburg impedance, due entirely to diffusion in the elec-trolyte, agrees perfectly with the theoretical model presentedabove at all temperatures measured.

Page 5

4946 D. Lloyd et al. / Electrochimica Acta 56 (2011) 4942–4948

50 100 150 200 250 300 350 4000

50

100

150

200

250

ZRe

/Ω

−ZIm

/Ω

Fig. 6. Impedance spectrum collected at 40 ◦C and a potential of −0.075 V vs. Pt quasireference. Fitting of Randles’ circuit indicated by solid lines, observed experimentaldata indicated by squares.

−100 −75 −50 −25 0 25 50 75 100 1250

1000

2000

3000

4000

5000

6000

7000

8000

E − E0´/mV

Z W/Ω

fittingexperimental data

tpBnp

F

3.05 3.1 3.15 3.2 3.25 3.3 3.35 3.4−6.9

−6.7

−6.5

−6.3

−6.1

← [CuCl4]2−

← [CuCl3]2−

1000/T / K−1

log 10

(DO/c

m2 s−1

)

Fig. 7. Example of fitting of Warburg impedance at 40 ◦C.

Rct, the charge transfer resistance, showed agreement with theheoretical model over a narrower range of potentials at all tem-eratures measured. This may be due to the simplicity of the

utler–Volmer formulation of electrode kinetics becoming pro-ounced at larger overpotentials, the polycrystalline nature of thelatinum electrode or homogenous reactions involving the com-

−100 −75 −50 −25 0 25 50 75 100 1250

50

100

150

200

250

300

350

E − E0´/mV

Rct/Ω

fittingexperimental data

ig. 8. Example of fitting of charge transfer resistance at 40 ◦C, ˛was 0.3 ± 0.02.

Fig. 9. Diffusion coefficient of Cu(II)Cl42− and Cu(I)Cl32− complexes in ethaline asa function of temperature (solid lines = Arrhenius equation fits, � = potential step,�= cyclic voltammetry, + = impedance spectroscopy).

plexes. Simultaneous fitting of both W and Rct resulted in estimatesfor k0, ˛, DO and DR.

4.4. Viscosity measurements

Varying the shear rate did not lead to any change in the appar-ent viscosity of the electrolyte, hence the material is Newtonianwithin the dynamic range of the instrument. When cycling the tem-perature between 100 and −40 ◦C at a rate of 5 ◦C/min no phasechange was observed at any point during the cooling cycle. Duringthe heating cycle crystallisation was observed to initiate at −24 ◦Cand melting at 33 ◦C. Similar behaviour has been reported for otherionic liquids, for instance by Choudhury et al. for EMIM-Tf2N [18]. At100 ◦C significant condensation of evaporated ethylene glycol wasobserved on the solvent trap. The dynamic viscosity of ethaline is41 ± 3 mPa s at 25 ◦C.

5. Discussion

5.1. Mobility of Cu(I) and Cu(II) in ethaline

The three electrochemical methods employed in this workresulted in the diffusion coefficients shown in Fig. 9. All three meth-ods agree well with each other and when double measurementswere performed. The experimental data could also be readily fit-ted to the Arrhenius equation. Activation energies are reported inTable 1.

The measurements show that the rate of mass transport canbe more than doubled by increasing the temperature from 25to 50 ◦C. Based on the Arrhenius fitting, DO is estimated to be1.25 ± 0.1 × 10−7 cm2 s−1 at 20 ◦C. This value is roughly half ofthat reported by Abbott et al. [10] at the same temperature,although it should be noted that they did not appear to start fromflat concentration profiles (a prerequisite for using the Randles-Sevcík equation), the Cu(II) concentration used in that work is fivetimes higher (0.1 M) and the initial source of Cu(II) used there isCuCl2·2H2O.

The diffusion coefficient of the Cu(I)Cl32− species was alsodetermined using the same three techniques, albeit at fewertemperatures. The diffusion coefficients determined found werearound twice that of the Cu(II) species over the range of tempera-

tures used. As shown in Table 2, this is entirely normal behaviourfor copper complexes. This difference may be attributable to ion-pairing between the doubly charged Cu(II) complex and a cholinecation.

Page 6

D. Lloyd et al. / Electrochimica Acta 56 (2011) 4942–4948 4947

Table 1Activation energies of diffusion, heterogeneous kinetics and viscous flow plus estimated system parameters at 25 ◦C.

Method Activation energies, kJ mol−1 Parameters at 25 ◦C ˛

DCu2+ DCu+ k0 DCu+ (107 cm2 s−1) DCu2+ (107 cm2 s−1) k0 (104 cm s−1)

Cyclic voltammetry 27.6 ± 0.7 30.0 ± 1.4 40 ± 6 2.7 ± 0.06 1.45 ± 0.02 11 ± 3 0.2 ± 0.05Impedance spectroscopy 24.0 ± 0.9 28.0 ± 1.7 38 ± 7 2.6 ±Chronoamperometry 27.7 ± 1.4 29.5 ± 1.5 2.8 ±Viscometry Ea,viscosity = −26.3 ± 1.2

3.05 3.1 3.15 3.2 3.25 3.3 3.35 3.4−3.2

−3

−2.8

−2.6

−2.4

1000/T / K−1

log 10

(k0 /c

m s

−1)

Fig. 10. Heterogeneous rate constant of Cu(II)/Cu(I) in ethaline as a function of tem-perature (solid line = Arrhenius fit of impedance data, dashed line = Arrhenius fit ofC

5

vgutdTwt

v0ttipk

remarkable result given the rich variety of complex chemistriesexpected in the various copper systems and may suggest that ion

TS

V data, �= cyclic voltammetry, + = impedance spectroscopy).

.2. Kinetics of Cu(I)/Cu(II) couple in ethaline at a Pt electrode

The values of k0 determined by analysis of impedance and cyclicoltammetric data are shown in Fig. 10. Although both methodsive similar values for k0, the adherence of the experimental val-es to Arrhenius kinetics is poor compared to that seen for massransport. This may indicate that the reaction cannot be adequatelyescribed by a model based on a single quasi-reversible process.he high value of k0 suggests a relatively low activation energy,hich is plausible given the similarity in structure between the

wo complexes.The values of ˛ determined by peak ratio analysis of cyclic

oltammetry data and impedance spectroscopy were between.2 and 0.3, respectively, for all temperatures. Further investiga-ion of this system over a wider range of temperatures or with aechnique better suited to investigation of homogeneous kinet-cs, such as AC-voltammetry or square wave voltammetry, may

rovide better understanding of the heterogeneous/homogeneousinetics.

able 2elected kinetic parameters for Cu(II)/Cu(I) system in various electrolytes at 25 ◦C.

Author Methoda Solvent

Kiekens et al. [21] RDE Water, 0.5 M KClStepnik-Swiatek and Malyszko [22] RDE Water, 1 M Ca(ClO4)2

Tierney et al. [23] CA Chloroaluminate

Nanjundiah and Osteryoung [24]CV + RDE Chloroaluminate: basicCV + RDE Chloroaluminate: acidic

Vainikka [25] CV BMP-Clx-(Tf2N)1−x

Chen CV + RDE EMIM-Clx-(BF4)1−x

Abbott [10] CV Ethaline

This workIS

EthalineCACV

a CA = chronoamperometry, IS = impedance spectroscopy, CV = cyclic voltammetry, RDE

0.05 1.5 ± 0.04 8 ± 2 0.3 ± 0.020.1 1.6 ± 0.1

= 41 ± 3 mPa s

5.3. Activation energies

Table 1 shows that the activation energies for mass transportare significantly larger in ethaline than the 20 kJ mol−1 typicallyreported for aqueous systems [9,17]. The activation energy of vis-cous flow was estimated to be −26.3 ± 1.2 kJ/mol, which agrees wellwith the values presented in Table 1 below. They are comparableto the viscous flow activation energies reported by Okoturo andVanderNoot for a variety of Newtonian ionic liquids, which spana range of 21–30 kJ mol−1 [19]. In practical terms large values ofthe activation energy means that the viscous nature of a DES orionic liquid can be mitigated to an extent by using elevated tem-peratures, since the rate of mass transfer will increase more quicklythan in a similar aqueous system.The activation energy for the het-erogeneous rate constant is at the threshold of 40 kJ mol−1 that isconsidered indicative of a chemical reaction [9], a comparison isdifficult due to a lack of comparable data. The value of k0 itselfis relatively high, hence the system is often described as beingreversible.

5.4. Comparison with other electrolyte systems

The calculated values of the diffusion coefficients, k0 and ˛ areshown in Table 2 below, with additional data taken from the litera-ture for comparison. Where necessary, the Handbook of electrolytesolutions [20] has been used to normalise diffusion coefficients to25 ◦C.

In general the methods used in this work produce results whichare similar to those reported in other ionic liquids, the exceptionbeing the value of k0, which is appreciably larger. Attention is drawnto the fact that diffusion coefficients tend to be 20–50 times lowerin ionic liquids than water, as would be expected from viscositymeasurements. A common theme across all electrolyte systems isthat the diffusion coefficient of the Cu+ species is consistently afactor of two higher than that of the Cu2+ species. This is a rather

association is a common feature across a range of electrolytes forcopper complexes.

DCu+ 107 cm2 s−1 DCu2+ 107 cm2 s−1 k0 104 cms−1 ˛

120 60 100 0.53130 553.63.5 1.5 4.2 0.656.5 3.01.0 0.5 1.12.3 1.5 1.8 0.45

2.42.6 ± 0.05 1.5 ± 0.04 8 ± 2 0.3 ± 0.022.8 ± 0.1 1.6 ± 0.12.7 ± 0.06 1.45 ± 0.02 11 ± 3 0.2 ± 0.05

= rotating disc electrode.

Page 7

4 ica A

6

DcCfdtsTi

wiietcs

scifllctp

A

Da

[

[[

[

[

[

[[[

[[

[

[

[23] B.J. Tierney, W.R. Pitner, J.A. Mitchell, C.L. Hussey, G.R. Stafford, J. Electrochem.Soc. 145 (1998) 3110.

[24] C. Nanjundiah, R.A. Osteryoung, J. Electrochem. Soc. 130 (1983) 1312.[25] T. Vainikka, Masters’ thesis, Helsinki University of Technology, Helsinki,

2008.

948 D. Lloyd et al. / Electrochim

. Conclusions

This work has demonstrated that in a number of import aspectsES behave similarly to chloroaluminates and modern ionic liquidsomposed of bulky discrete anions. The rapid formation of stableu(I) and Cu(II) chlorocomplexes suggests that a large amount of

ree chloride is present, which in turn implies a high degree ofissociation of choline chloride. This is also reinforced by the massransport parameters and activation energies reported, which alsohow a high degree of agreement with other ionic liquid systems.his provides further empirical justification to consider DES asonic liquids.

The determination of mass transport properties shows that,hen properly applied, any of the electrochemical technique used

n this paper works. However, in the case of techniques other thanmpedance spectroscopy the effect of solution resistance shouldither be electronically compensated or, at the very least, taken ino consideration during experiment design. The heterogeneous rateonstant and diffusion coefficient of each species are remarkablyimilar across a wide range of ionic liquids and copper complexes.

A quasireversible single electron transfer process has beenhown to provide an adequate description of the Cu(II)/Cu(I) pro-ess. By characterising the behaviour of electrochemical systemsn this way it is possible to quantitatively compare the effect of,or instance, adding complexants or selecting a different ioniciquid. The stability of low oxidation state metal complexes in ioniciquids is a feature of particular interest. When combined with highathodic efficiencies this indicates great potential for these systemso more than double the energy efficiency of electrodepositionrocesses.

cknowledgements

This work was funded by TEKES and the Outokumpu foundation.L would like to thank Sönke Schmachtel M.Sc. for his considerablessistance in development of the Matlab software.

cta 56 (2011) 4942–4948

References

[1] N.V. Plechkova, K.R. Seddon, Chem. Soc. Rev. 37 (2008) 123.[2] L.F. Audrieth, A. Long, R.E. Edwards, J. Am. Chem. Soc. 58 (1936) 428.[3] F.H. Hurley, T.P. Wier, J. Electrochem. Soc. 98 (1951) 203.[4] A.P. Abbott, G. Capper, D.L. Davies, R.K. Rasheed, V. Tambyrajah, Chem. Com-

mun. (2003) 70.[5] A.P. Abbott, R.C. Harris, K.S. Ryder, J. Phys. Chem. B 111 (2007) 4910.[6] A.P. Abbott, D. Boothby, G. Capper, D.L. Davies, R.K. Rasheed, J. Am. Chem. Soc.

126 (2004) 9142.[7] A.P. Abbott, K.J. McKenzie, PCCP 8 (2006) 4265.[8] A.P. Abbott, G. Frisch, K.S. Ryder, Annu. Rep. Prog. Chem. Sect. A 104

(2008) 21.[9] M. Lundström, Chalcopyrite dissolution in cupric chloride solutions (Ph.D.

diss.), Helsinki University of Technology, Helsinki, 2009.10] A.P. Abbott, K. El Ttaib, G. Frisch, K.J. McKenzie, K.S. Ryder, PCCP 11 (2009)

4269.11] R.S. Nicholson, Anal. Chem. 37 (1965) 1351.12] A.J. Bard, L.R. Faulkner, Electrochemical Methods, 2nd ed., Wiley, New York,

2001.13] C. Gabrielli, Identification of Electrochemical Processes by Frequency Response

Analysis, Solartron Analytical, Farnborough, 1998.14] A.P. Abbot, J.C. Barron, K.S. Ryder, D. Wilson, Chem. Eur. J. 13 (2007)

6495.15] C. Amuli, M. Elleb, J. Meullemeestre, M. Schwing, F. Vierling, Inorg. Chem. 25

(1986) 856.16] V.K. Sharma, F.J. Millero, J. Solution Chem. 19 (1990) 375.17] T. Kekesi, M. Isshiki, J. Appl. Electrochem. 27 (1997) 982.18] A.R. Choudhury, N. Winterton, A. Steiner, A.I. Cooper, K.A. Johnson, CrystEng-

Comm 8 (2006) 742.19] O.O. Okoturo, T.J. VanderNoot, J. Electroanal. Chem. 568 (2004) 167.20] V.M.M. Lobo, Handbook of Electrolyte Solutions, Elsevier, Amsterdam,

1989.21] P. Kiekens, R.M.H. Verbeeck, E. Temmerman, Microchim. Acta 2 (1981)

29.22] B. Stepnik-Swiatek, J. Malyszko, Monatsch. Chem. 113 (1982)

715.