Sunteți pe pagina 1din 24

Journal of

Electroanalytical
Chemistry
Journal of Electroanalytical Chemistry 566 (2004) 193–216
www.elsevier.com/locate/jelechem

A computational and experimental study of the cyclic


voltammetry response of partially blocked electrodes, part III:
interfacial liquid–liquid kinetics of aqueous vitamin B12s
with random arrays of femtolitre microdroplets of
dibromocyclohexane
Trevor J. Davies, Benjamin A. Brookes, Richard G. Compton *

Physical and Theoretical Chemistry Laboratory, Oxford University, South Parks Road, Grande-Bretagne, Oxford OX1 3QZ, UK
Received 16 September 2003; received in revised form 5 November 2003; accepted 7 November 2003

Abstract

The cyclic voltammetric response of electrodes modified with catalytically reactive microdroplets is modelled using finite dif-
ference simulations and a method is presented for the determination of kinetic parameters for the coupled heterogeneous reaction.
The method is first applied to investigate the liquid–liquid reaction between pure trans-1,2-dibromocyclohexane (DBCH) micro-
droplets, deposited on the surface of a basal plane pyrolytic graphite electrode, and vitamin B12s in aqueous solution. Second, cyclic
voltammetry on electrodes modified with microdroplets of DBCH diluted in dodecane is employed to determine the apparent bi-
molecular interfacial rate constant for the initial step in the DBCH(oil)jB12s (aq) reaction. The results are compared and contrasted
with a previous SECM/ITIES study of a similar reaction.
Ó 2003 Elsevier B.V. All rights reserved.

Keywords: Partially blocked electrodes; Finite difference methods; Modified electrodes; Microdroplets; Randomly arranged ensembles; Vitamin B12

1. Introduction ations the blocking material is of microscopic propor-


tions, i.e., the inhomogeneities and/or the distances
Charge transfer reactions at an electrode surface between them are small compared to the thickness of the
partially covered with an inert material display charac- diffusion layer. This results in non-linear diffusion in
teristic transient properties [1–3]. For example, in the regions close to the uncovered electrode (within the
case of cyclic voltammetry with a quasi-reversible redox diffusion layer), thus complicating the modelling pro-
couple, increasing the fractional coverage of inert ma- cess. Amatore et al. [1] were the first to consider this case
terial, h, causes an increase in peak to peak separation successfully and developed a one-dimensional model for
and (eventually) a decrease in peak height. Such systems an ensemble of microdisk electrodes dispersed under a
are not uncommon and it is relatively easy to modify or blocking film. By describing the observed voltammetric
intentionally ‘‘block’’ electrode surfaces [2–5]. Thus, a trends, their work resulted in a strategy for estimating
good quantitative description of partially blocked elec- the fractional coverage of the blocking film and the size
trodes is potentially very useful to a large number of of the active sites. In Part I of this series, we developed a
research workers. However, in many experimental situ- two-dimensional numerical method to describe the
transient response of a symmetric array of inert disks on
* an electrode surface (the exact inverse of the system
Corresponding author. Tel.: +44-1-865-275-413; fax: +44-1-865-
275-410.
studied by Amatore) [2]. In a subsequent paper, this
E-mail address: richard.compton@chemistry.ox.ac.uk (R.G. model was successfully extended to arrays of randomly
Compton). distributed disks and, more importantly, microdroplets

0022-0728/$ - see front matter Ó 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.jelechem.2003.11.026
194 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

[3]. Indeed, Part II concluded with an electrochemical cantly different from that for previous methods of
method for determining the average radius of inert mi- ITIES investigation by the fact that both electron
crodroplets deposited on an electrode surface. transfer and the coupled interfacial heterogeneous re-
A rather interesting situation arises when the block- action occur at the electrode surface.
ing material is not inactive. Such systems are created The contrasts between coupled homogeneous and
when electrodes are intentionally blocked with reactive heterogeneous reactions are interesting. In the latter,
material that can lead to coupled heterogeneous chemical electron transfer and the coupled heterogeneous chemi-
reactions. In the case of reactive microdroplets, e.g., oil cal reaction(s) occur exclusively at the electrode and
droplets on an electrode surface submerged in aqueous droplet surface, respectively. This contrasts the homo-
media, the accompanying chemical reaction(s) takes geneous analogues, where the coupled chemical reactions
place at the liquidjliquid (oiljwater) interface. Such re- occur within the diffusion layer. Mathematically, the
actions are important to study due to their wide rele- difference is even more apparent. For coupled homoge-
vance in a number of different research fields. For neous reactions, the kinetic terms are contained in the
example, catalytic electrochemical reactions in emulsi- diffusion equations whereas the corresponding hetero-
fied media, where the key reaction takes place at the geneous kinetic terms appear only in the boundary
liquidjliquid interface, 1 have proved to be a viable al- conditions. Thus, modifying a numerical method for
ternative to similar reactions in organic solvents [6,7]. different mechanisms is relatively simple for the hetero-
This is important industrially – as well as providing a geneous case when compared to the task involved in
cheaper alternative in which to carry out syntheses, modifying a coupled homogeneous simulation.
emulsions (especially sono-emulsions) are much The electrocatalytic reaction of vitamin B12r (the
‘‘greener’’ media. Furthermore, liquid–liquid systems Co(II) form of vitamin B12 ) with trans-1,2-dibromocy-
are important in our understanding of some funda- clohexane (DBCH) in certain organic solvents is a well-
mental biological problems. For example, Senda and co- documented example of a coupled homogeneous
workers [8] have used the nitrobenzenejwater interface chemical reaction which regenerates the electroactive
to model proton transfer across the biological mem- material [13]. The reaction pathway is thought to pro-
branejsolution interface. ceed via one of two mechanisms where the rate deter-
Previous studies of the rates of liquid–liquid reac- mining step is either an SN 2-type nucleophilic attack (A)
tions have involved both ‘‘direct’’ and ‘‘indirect’’ ap- or an E2-type elimination (B) [18]:
proaches. The ‘‘classical’’ electrochemical experiment is
that of Samec and Maracek [9], where a four-electrode CoðIIÞL þ e ¢ CoðIÞL Electrode
configuration is used to measure interfacial charge
CoðIÞL þ RBr2 ! RBr–CoðIIIÞL þ Br Solution
transfer. In the last decade, Bard and co-workers [10]

have developed a method based on scanning electro- RBr–CoðIIIÞL ! RBr þ CoðIIÞL Solution
chemical microscopy (SECM), where an UME is used CoðIÞL þ RBr ! CoðIIÞL þ R þ Br
 0
Solution
to probe charge transfer reactions directly at the in-
terface between two immiscible electrolyte solutions ðAÞ
(ITIES) [11]. A novel approach was that of Banks et al.
CoðIIÞL þ e ¢ CoðIÞL Electrode
[12], who, when working with particular sono-emul-
sions were able to relate bulk measurements to the CoðIÞL þ RBr2 ! Br–CoðIIIÞL þ Br þ R0 Solution
heterogeneous rate constant for their specific liquid–
Br–CoðIIIÞL þ CoðIÞL ! Br þ 2CoðIIÞL Solution
liquid reaction. In the following work we develop a
new approach based on the transient response of ðBÞ
electrodes modified with water-insoluble oil microdro-
0
plets in aqueous solutions. In particular we study the In the above pathways, RBr2 represents DBCH, R is
case where the droplet surface reacts with an electro- cyclohexene, Co(II)L is vitamin B12r and Co(I)L vi-
generated mediator in a catalytic pathway that leads to tamin B12s . Recently, we developed a heterogeneous
the regeneration of the original electroactive species. analogue of this system where microdroplets of
Using cyclic voltammetry we are able to probe such DBCH were deposited onto an electrode surface and
systems and obtain kinetic data on the coupled liquid– then immersed into an aqueous solution of vitamin
liquid reactions. The theoretical treatment is signifi- B12 [7]. The coupled chemical reaction between DBCH
and vitamin B12s occurs exclusively at the surface of
the microdroplet, and the reaction is heterogeneous.
1 In this paper we simulate the transient response at a
Although true for sono-emulsions, in bicontinuous microemul-
sions, for example, the reaction occurs between the liquid-adsorbed
partially blocked electrode where the block regenerates
surfactantjliquid interface. The liquidjliquid interface is, therefore, a electroactive material via a separate heterogeneous
model experimental system for surfactant based emulsions. reaction:
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 195

kf complicate the problem and the work reported in Part II


A þ e ¢ B Electrode
kb
ðCÞ confirmed that such a treatment was viable [3]. We
khet consider a simple one-electron redox reaction, described
B ! A Block
by Butler–Volmer kinetics, followed by an irreversible
The reaction occurring at the block in pathway (C) can heterogeneous reaction at the surface of the block that
be compared to the liquid–liquid reaction between B12s regenerates the electroactive material:
kf
and DBCH, in which Co(II)L is regenerated. It is the electrode : A þ e ¢ B
aim of this work to develop a method for determining kb
ðDÞ
khet
kinetic parameters of catalytic liquid–liquid reactions, block : B ! A:
like those in pathways (A) and (B), based on the vol-
tammetric response of a partially blocked electrode. Note that the heterogeneous reaction occurring at the
droplet surface could equally be described by reversible
kinetics. However, since reactions between super re-
2. Theory duced vitamin B12s and vicinal dibromides are generally
considered to be irreversible [13], we have avoided the
The underlying assumption in Parts I and II is that a extra complication of including another kinetic param-
partially covered electrode surface can be considered as eter in our simulations. In the above pathway khet is the
an ensemble of diffusion domains. These are indepen- heterogeneous reaction rate constant and the two-elec-
dently interacting unit cells of radius R0 through which tron transfer rate constants are given by
 
no diffusional flux passes. In our case the blocking ma- aF
0
EðtÞE0
terial takes the form of microdroplets which we assume kf ¼ k 0 e RT ð2Þ
to be hemispherical in shape with radius Rb . In reality,
the electrode area associated with each microdroplet ð
ð1aÞF EðtÞE00 Þ
kb ¼ k 0 e RT ð3Þ
would take the form of a Voronoi cell [14] but we ap-
proximate this to a circle of radius R0 . Fig. 1 illustrates a where k 0 is the standard electron transfer rate constant,
typical unit cell/diffusion domain which consists of a R is the molar gas constant, F is the Faraday constant, T
cylinder centred around the droplet/block. is the temperature, a is the transfer coefficient, EðtÞ is the
The microscopic coverage of each domain, h, is given potential at time t, E00 is the formal potential of the re-
by the ratio of areas of the block and the surface: dox couple and the number of electrons transferred is
 2 equal to 1. In our simulations E00 ¼ 0 V and a ¼ 0:5.
pR2b Rb
h¼ 2¼ : ð1Þ Since the electrochemical and chemical reactions oc-
pR0 R0 cur exclusively at the electrode and block surface, re-
spectively, mass transport is simply described by FickÕs
2.1. Mass transport equations second law. This is identical to the case of the simple
redox couple reported in Part I [2] and would be the
In the following mathematical description we will same for any reaction mechanism(s) occurring exclu-
ignore the height of the droplet and treat the surface of sively at the electrode and block surface. In the cylin-
the droplet as if it were flush with the electrode surface. drical polar coordinate system these are:
The inclusion of droplet height in the model would  2 
1 oa o o2 1 o
¼ þ þ a ð4Þ
Da ot oZ 2 oR2 R oR
 
1 ob o2 o2 1 o
¼ þ þ b ð5Þ
Db ot oZ 2 oR2 R oR
where a ¼ ½A=½Abulk , b ¼ ½B=½Abulk , and Da and Db are
the diffusion coefficients of the respective species. The
boundary conditions applicable to path D are
electrode Z ¼ 0; Rb 6 R 6 R0 ;
oa ob
Da ¼ kf a0  kb b0 ¼ Db ;
oZ oZ

oa
block Z ¼ 0; 0 6 R < Rb ; Da ¼ khet b0 ;
oZ
Fig. 1. Illustration of a single diffusion domain and the cylindrical ob
Db ¼ khet b0 ;
polar coordinate system employed in the modelling. oZ
196 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

axis of symmetry 0 < Z < 1; R ¼ 0;


oa ob
¼ 0; ¼ 0;
oR oR

unit cell edge 0 < Z < 1; R ¼ R0 ;


oa ob
¼ 0; ¼ 0;
oR oR
diffusion layer Z ! 1; 0 6 R 6 R0 ;
oa ob
¼ 0; ¼ 0; Fig. 2. Illustration of the mesh numbering system in the vicinity of the
oZ oZ electrode surface.
where a0 and b0 are the normalized concentrations of A
and B at Z ¼ 0. Note that the boundary condition for
the block is the only difference between pathway (D) and ou
unit cell edge 0 < z < 1; r ¼ 1; ¼0
the simple redox reaction previously discussed [2,3]. or
Following the procedure outlined in Part I [2], we nor-
ou
malize R, Z and t with respect to the parameter R0 : diffusion layer z ! 1; ¼0
0 6 r 6 1;
oz
R where kf0 ¼ kdl0 exp½aF ðEðtÞ  E00 Þ=RT , kb0 ¼ kdl0 exp
r¼ ð6Þ
R0 ½ð1aÞF ðEðtÞ  E00 Þ=RT , rb is the dimensionless block
Z radius (rb ¼ Rb =R0 Þ and u0 is the value of u at Z ¼ 0.
z¼ ð7Þ
R0
2.2. Finite difference formulation
Dt
s¼ 2: ð8Þ We solve Eq. (12) using a fully implicit finite differ-
R0
ence method with a geometrically expanding mesh
As a result, we can define three dimensionless parame- technique identical to that described in Part I [2]. The
ters: mdl (dimensionless scan rate), kdl0 (dimensionless only modification is in the implementation of the block
0
electron transfer rate constant) and khet (dimensionless boundary condition which concerns mesh points
heterogeneous rate constant) i ¼ 0; 1; . . . ; ðnb  1Þ: 2
FR20 oui 0
mdl ¼ m ð9Þ ¼ ðUi;0  1Þkhet
DRT oz
!
2
0
k R0 l2 Ui;2  ðl0 þ l1 Þ Ui;1 þ l1 ðl1 þ 2l0 ÞUi;0
kdl0 ¼ ð10Þ  0
D l0 l1 ðl1 þ l0 Þ

khet R0 /ðl0 þ l1 Þ2 Ui;1  /l20 Ui;2 þ khet


0
0
khet ¼ : ð11Þ ) Ui;0 ¼ 0
ð13Þ
D /ðl1 þ 2l0 Þl1 þ khet
Using the dimensionless parameters and assuming Da where / ¼ 1=l0 l1 ðl0 ¼ l1 Þ and the mesh numbering is
¼ Db , we can make the substitution u ¼ a ¼ 1  b into illustrated in Fig. 2. The modelling then follows the
Eqs. (4) and (5) to obtain just one equation that de- same method as described in Part I [2], where the di-
scribes the mass transport: mensionless current, w, is calculated using Eq. (14): 3
 2  Z 1 
ou o o2 1 o 2 oa
¼ þ þ u ð12Þ w ¼ pffiffiffiffiffi r dr: ð14Þ
ot oz2 or2 r or mdl rb oz z¼0
The relationship between the true current, I, and the
with the boundary conditions: dimensionless current, w, is given by
ou  0  rffiffiffiffiffiffiffiffiffiffiffi
electrode z ¼ 0; rb 6 r 6 1; ¼ kf þ kb0 u0  kb0 F 3 Dv
oz I ¼ ½Abulk Aelec w ð15Þ
RT
ou 0
block z ¼ 0; 0 6 r < rb ; ¼ ðu0  1Þkhet
oz
2
Misprint in [2]: The denominator in Eq. (13) contains an l1 term
ou which is missing in the corresponding equations in [2].
axis of symmetry 0 < z < 1; r ¼ 0; ¼0 3
or Misprint in [2]: The factor of 2 was omitted in Eq. (14) in [2].
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 197

where Aelec is the area of the electrode (or single diffusion in the reactive block voltammograms is similar to that
domain) and the number of electrons transferred is observed for coupled catalytic homogeneous reactions
equal to 1. and is essentially due to the same factors [13]. In the
Optimum values for the grid expansion factors and homogeneous analogue, any B species formed at the
initial grid spacing, determined via convergence testing electrode surface is oxidized back to A by the homoge-
0
0
over a range of khet , h; mdl and kdl , were identical to those neous mediator. Thus, upon scan reversal, there is no B
used in Part I for the case of an inert, rather than a present in the vicinity of the electrode surface and no
reactive, block [2]. reverse peak is seen. Our heterogeneous analogue is
slightly different in that the re-oxidation of B to A oc-
2.3. Simulation results curs only at the block surface. This implies that mass
transport of B from the uncovered electrode to the block
Initial simulations were carried out using parameters must be rapid or else we would see a reverse peak.
relevant to the experimental problem discussed below. The voltammograms for the reactive blocks appear to
In most simulated voltammograms, the dimensionless have a limiting current suggesting a constant flux of A is
current, w, is preferred over the true current, I, due to its being delivered to the uncovered electrode surface. On
wider relevance. For example, when dealing with linear closer inspection, there are actually gentle peaks at
sweep or cyclic voltammetry of a simple redox couple at E ¼ 0:32 V for khet ¼ 0:01 cm s1 and E ¼ 0:61 V for
macro electrodes, it is well known that the largest di- khet ¼ 10 cm s1 . However, the fact that the peak cur-
mensionless current is 0.4463 (where the mass transport rents are almost maintained suggests that block regen-
is linear diffusion only) [15]. In all simulations the tem- eration causes some kind of quasi-steady state to be
perature was 293 K, although slight variations made a achieved where solution depletion due to the one-elec-
negligible difference. tron reduction in the vicinity of the electrode surface is
compensated by the coupled heterogeneous reaction.
2.3.1. Inert vs. reactive Finally, the ‘‘peak’’ potential, Ep , appears to increase
Fig. 3 illustrates the simulated cyclic voltammograms with khet . Comparison with the homogeneous analogue
for three different heterogeneous rate constants, khet ¼ 0 is complicated because of mediator depletion.
(no reaction), 0.01 and 10 cm s1 , at a diffusion domain A similar situation would arise for a simple one-
with block radius Rb ¼ 2:5 lm and coverage h ¼ 0:5 electron redox couple and an array of collector–gener-
where the scan rate is 50 mV s1 . The other parameters ator electrodes with infinitely thin insulating layers. In
are listed in the figure legend. The effect of the catalytic this case the unit cell could be described as that in Fig. 1,
regeneration of species A is immediately obvious by the the only difference being that the block is now electrode
increased currents which increase with khet (for suitably material held at a constant potential, Ec . Varying the
low values of khet where the heterogeneous reaction is value of Ec would then be equivalent to varying khet and
not diffusion controlled). The absence of a reverse peak we would expect to see the same voltammetric trends.

2.3.2. Analysis of the concentration profiles


Fig. 4 illustrates concentration profiles taken at the
peak potentials for all three domains in Fig. 3. Note that
since h ¼ 0:5, rb ¼ 0:707 and the dark area around
r ¼ rb is due to the high density of mesh points required
to minimize the error caused by the singularity at r ¼ rb
and z ¼ 0. Again, the effect of the reactive block is im-
mediately obvious, especially in the case of the fast
heterogeneous reaction (khet ¼ 10 cm s1 ), where we
observe an almost vertical drop in a when E ¼ 0:61 V.
Immediately over the block, B is converted rapidly into
A, by the coupled heterogeneous reaction, so that the
concentration of A is ½Abulk and a ¼ 1. However, at the
unblocked electrode surface, the potential is such that A
undergoes a rapid one-electron reduction to B causing
½Az¼0 and a to equal zero. Hence, in the immediate vi-
cinity of the point r ¼ rb , we have a sharp drop in [A] as
we crossover from the blocked to the unblocked region.
Fig. 3. Simulated dimensionless current voltammograms for diffusion As we move away from the electrode surface, this abrupt
domains, where Rb ¼ 2:5 lm, k 0 ¼ 0:005 cm s1 , D ¼ 2:0  106 change evolves into a smoother profile. For the slower
cm2 s1 , m ¼ 0:05 V s1 , h ¼ 0:5 and khet ¼ 0; 0:01, and 10 cm s1 . heterogeneous reaction, khet ¼ 0:01 cm s1 , the concen-
198 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

Fig. 4. Dimensionless concentration profiles for the diffusion domains in Fig. 3, taken at E ¼ (i) )0.05 V, (ii) )0.32 V and (iii) )0.61 V, where
khet ¼ (a) 0 cm s1 , (b) 0.01 cm s1 and (c) 10 cm s1 .

tration of A over the block surface is less than the bulk from the uncovered electrode to the reactive block as
concentration and the gradients around the singularity implied by the absence of a reverse peak.
are much gentler. As observed in the differences between Fig. 4(i) and
In the case of inert blocks (khet ¼ 0 cm s1 ), mass (ii) for the ‘‘reactive’’ domains, the flux of A from the
transport to the electrode surface is supplied by diffusion region of the block to the uncovered electrode is not
arising from concentration gradients caused by the in- independent of potential. For example, it is not until the
creasing voltage. When the block is reactive we have a concentration of A across the uncovered electrode sur-
situation where the regeneration of A also leads to (quite face is zero that we observe a peak current. The reason
steep) concentration gradients. The concentration pro- for this can be attributed to two factors. First, looking at
file for khet ¼ 10 cm s1 might lead one to believe that the profiles for the khet ¼ 10 cm s1 domain, the large
transport due to block regeneration dominates and gradient at the singularity results in a massive flux of A
generally this is true for domains of high coverage and to the uncovered electrode surface which has a major
not too excessive scan rates, vide infra. The sharp drop influence on the overall current. This is illustrated in
in a at r ¼ ro results in a flux of A that greatly exceeds Fig. 5 where we have plotted ðoa=ozÞz¼0 (a variable that
that caused by the voltage increase and we observe a is proportional to the current density) taken at the peak
greatly enhanced current. For slower heterogeneous re- potential across the uncovered electrode surface for the
actions, the profiles are less abrupt and we observe a far (a) khet ¼ 0 cm s1 and (b) khet ¼ 10 cm s1 domains. In
lesser increase in current. Note also that the concen- the case of the inert block, the current density varies
tration profiles for species B are the inverse of those in little across the electrode surface. This contrasts the re-
Fig. 3 (u ¼ a ¼ 1  b). Thus, there is a large flux of B active block where we see a large decrease in ðoa=ozÞz¼0
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 199

tial where ½Az¼0 is greater than zero (remember that our


redox couple is quasi reversible, k 0 ¼ 0:005 cm s1 ). In
the case of the reactive blocks, reactant depletion in the
vicinity of the electrode surface is lessened – not only
does the reactive block supply a large flux of A, it does so
at distances close to the electrode surface. Therefore, the
steepest concentration gradients occur when ½Az¼0 ¼ 0,
after which we observe a more or less constant current
(‘‘steady’’ state). This also explains why we observe an
increase in Ep with khet (until we reach the diffusion
limited regime, vide infra). Faster coupled heteroge-
neous reactions can supply more A to the electrode
surface, meaning that we need to reach higher potentials
where our electrode kinetics are fast enough to con-
vert all the excess reactant at the uncovered electrode
surface.
On closer inspection of Fig. 4 two things become
apparent. First, in all three profiles, (a)–(c), the diffusion
layer thickness is approximately constant at each of the
potentials. Thus, it appears that a non-zero khet has little
influence on the diffusion layer thickness. Second, a

Fig. 5. Plots of ðoa=ozÞz¼0 (taken at the peak potential, Ep Þ against r


over the uncovered electrode surface for the khet ¼ 0 cm s1 (a) and
khet ¼ 10 cm s1 (b) domains in Fig. 3. The inset in (b) is a close-up of
the region near the r-axis. Note that rb ¼ 0.707.

as we move away from the singularity. For fast coupled


heterogeneous reactions where the value of a at the
blocked surface is always 1, the maximum flux and,
therefore, maximum current will occur at a potential
when [A] at the uncovered electrode surface is zero.
Second, because of the solution replenishment in the
vicinity of the electrode surface, bulk solution depletion
has a lesser influence on the voltammograms. For the
case of the inert block, we have an example of the well
documented ‘‘increasing rate of electron transfer vs.
depletion of bulk solution’’ [16]. As the potential in-
creases, ½Az¼0 decreases (which should lead to steeper
concentration gradients and therefore higher flux) but at
the same time the concentration of A in the vicinity of
the electrode surface is depleted (leading to shallower Fig. 6. Dimensionless concentration profiles in the immediate vicinity
concentration gradients). Thus, we have a trade off be- of the electrode surface for those profiles illustrated in (a) Fig. 4(a)(iii)
tween two opposing factors and observe a peak poten- and (b) Fig. 4(c)(iii).
200 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

‘‘kink’’ just above the surface of the electrode is visible according to the reasons discussed above we would ex-
in the concentration profiles for the reactive blocks. pect Ep to be independent of scan rate. The corre-
Fig. 6 illustrates a close-up of this region for Fig. 4(a)(iii) sponding concentration profiles (taken at E ¼ 0:61 V)
(khet ¼ 0 cm s1 , E ¼ 0:61 V) and (c)(iii) (khet ¼ 10 are displayed in Fig. 8. As expected, the diffusion layer
cm s1 , E ¼ 0:61 V). The magnitude of the concen- thickness decreases with increasing scan rates, but this is
tration gradient in Fig. 6(a) is approximately equal to the only useful information we can obtain from the
that of the section labelled ‘‘X ’’ in Fig. 6(b) and we can
associate this with the depletion of A at the electrode
surface leading to diffusion from the bulk solution. The
section labelled ‘‘Y ’’ in Fig. 6(b) is clearly steeper and
must arise from block regeneration. Thus, we have an
idea of the dominance of block regeneration over scan
rate in contributing to mass transport for this particular
domain. Also, it is clear to see that the effect of the block
is confined to a region close to the electrode surface. The
depth of region Y gives us an idea of how far regener-
ated reactant has diffused in the time taken to reach the
stated potential, which in this case is about 1.5 lm.

2.3.3. The effect of scan rate


So far we have considered only one scan rate, 50
mV s1 . What happens if we increase the rate of the
potential sweep? Fig. 7 illustrates linear sweep voltam-
mograms for the same domain in Fig. 4(c) where the
scan rates are 0.1, 0.5 and 2.0 V s1 . As observed for a
simple quasi-reversible redox reaction, the peak dimen-
sionless current, wmax , decreases as the scan rate in-
creases, although the relative decrease appears quite
large. In contrast to the inertly blocked case, the peak
potential, Ep is independent of scan rate (Ep ¼ 0:61 V
for all three scans). Even though the scan rate increases,
the potential at which the electrode kinetics are fast
enough to achieve ½Az¼0 remains the same. Therefore,

Fig. 7. Dimensionless current linear sweep voltammograms for a dif- Fig. 8. Dimensionless concentration profiles, taken at E ¼ Ep ¼ 0:61
fusion domain, where Rb ¼ 2:5 lm, h ¼ 0:5, k 0 ¼ 0:005 cm s1 , V, for the simulated voltammograms in Fig. 7, where m ¼ (a) 0.1 V s1 ,
D ¼ 2:0  106 cm2 s1 , khet ¼ 10 cm s1 and m ¼ 0:1, 0.5 and 2.0 V s1 . (b) 0.5 V s1 and (c) 2.0 V s1 .
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 201

bulk solution depletion decreases, resulting in steeper


gradients. Second, in the region between the kink and
the electrode surface, the profiles appear to be identical.
As mentioned above, the profile in this region can be
attributed solely to block regeneration. Unlike scan rate,
we cannot vary block regeneration so we would expect a
constant effect. More importantly, combining Eqs. (14)
and (15) to eliminate w, we can write the current as:
Z 1 
oa
I ¼ 2pR0 ½Abulk FD r dr ð16Þ
rb oz z¼0
in which we have used Aelec ¼ pR20 , the area of the dif-
fusion domain. The concentration profiles in Fig. 9 are
plots of a against r and z at a potential corresponding to
the peak current. Close to the electrode surface, all three
profiles are identical which suggests that the integral in
Eq. (16) is constant. Thus, according to Eq. (16), the
peak current will be independent of scan rate. This phe-
nomenon is illustrated in Fig. 10 where we have re-
plotted Fig. 7 replacing the y-axis with I=½Abulk instead
of dimensionless current (the current values appear low
because the voltammograms pffiffiare
ffi for a single diffusion
domain of radius R0 ¼ Rb = h ¼ 3:5 lm). The increase
in current as we increase the scan rate from 0.1 to 2.0
V s1 is tiny and illustrates what little effect scan rate has
on the voltammograms. The reason for this indepen-
dence of peak current with scan rate has already been
touched upon. Because block regeneration occurs ex-
clusively at the electrode surface, very large concentra-
tion gradients are developed such that the catalytic
reaction dominates mass transport to the whole of the
electrode surface. Scan rate has no influence on the
coupled heterogeneous reaction, and therefore, no in-
fluence on the concentration profiles in the immediate

Fig. 9. Dimensionless concentration profiles as illustrated in Fig. 8


where we have ‘‘zoomed in’’ on the region close to the electrode sur-
face, well within the diffusion layer.

larger profiles. On observing the close-ups in Fig. 9,


where we have focussed on the characteristic kinks, two
things become apparent. First, in the region after the
kink the concentration gradient increases with scan rate Fig. 10. The corresponding real current linear sweep voltammograms
which is what we would expect for a gradient induced by for those illustrated in Fig. 7. Bold: m ¼ 2:0 V s1 ; dashed: m ¼ 0:5
voltage increase – as the scan rate increases, the time for V s1 ; solid: m ¼ 0:1 V s1 .
202 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

vicinity of the electrode surface. Thus, the integral in Eq.


(16) is dominated by block regeneration and we observe
a peak current (‘‘limiting’’ current) which is independent
of scan rate.
Is this phenomenon true for all cases? The question
can be answered if we consider situations where block
regeneration might not dominate the mass transport to
all of the unblocked electrode surface. In other words,
when will increasing the scan rate cause an increase in
the peak current? Judging from the concentration pro-
files, scan rate will never influence mass transport near
the blockedjunblocked electrode boundary – the gradi-
ent due to block regeneration is simply too high. How-
ever, the outskirts of the diffusion domain are the least
affected by block regeneration, so scan rate could have
an influence on mass transport to this region of the
uncovered electrode surface. Therefore, we should ex- (a)
pect to see an increase in peak current with scan rate at
domains of low coverage, h, where the outskirts of the
domain are far from the block. As we increase the scan
rate, the time for regenerated material to diffuse from
the block to the outer regions of the domain decreases.
At the same time, the diffusion layer thickness decreases
and mass transport to the outer region of the electrode
should no longer be dominated by block R1 regeneration.
This is illustrated in Fig. 11 where rb ðoa=ozÞz¼0 r dr at
the peak potential (which according to Eq. (16) is pro-
portional to the peak current, Ip Þ is plotted against
coverage for a range of scan rates at a domain where
k 0 ¼ 0:005 cm s1 , D ¼ 2:0  106 cm s1 , khet ¼ 10
cm s1 and the block radius is (a) 2:5 lm or (b) 10.0 lm.
Overlaid on both (a) and (b) are the corresponding plots
for the inert domains (dashed lines). The figure is
slightly complicated by the increasing domain radius, (b)
R0 , for the decreasing h values but the trends are as Fig. 11. Plot of Ip =2pR0 ½Abulk FD against h for diffusion domains where
follows. At domains of high coverage, h > 0:5, block D ¼ 2.0  106 cm2 s1 , k 0 ¼ 0:005 cm s1 , m ¼ 0:05, 0.1, 0.5 and 2.0
regeneration dominates mass transport – not only do we V s1 , and Rb ¼ (a) 2.5 lm, and (b) 10 lm. The solid curves corre-
observe greatly increased currents but Ip is independent spond to khet ¼ 10 cm s1 and the dashed curves correspond to khet ¼ 0
cm s1 .
of scan rate. For 0:2 < h < 0:5 the outer regions of the
domains are too far from the block for it to have any
effect and we enter an intermediate region where diffu-
sion is also influenced by the increasing voltage. Thus, reduction of A to B is rapid), where R0 ¼ 20 lm,
we observe a significant increase in peak current with h ¼ 0:1, khet ¼ 10 cm s1 , k 0 ¼ 0:005 cm s1 , D ¼ 2:0 
increasing scan rate. Finally, for h < 0:1 the coverage is 106 cm s1 and m ¼ 2:0 V s1 , i.e., a domain where mass
so low that most of the uncovered electrode is ‘‘out of transport is not dominated by block regeneration. No-
range’’ from the influence of block regeneration and we tice that there is no characteristic kink in the profile and
enter a region where diffusion is dominated by the in- the concentration gradient out the outskirts of the do-
creasing voltage. Hence, we observe identical peak cur- main is what we would expect from an inert domain
rents to the inert counter part. The change on going with the same parameters. Fig. 12(b) illustrates the
from high to low coverage is more defined for the corresponding simulated cyclic voltammogram, along
Rb ¼ 10 lm situation because we are dealing with larger with that for the inert case. As observed, the effect of the
domains. Note that at low scan rates, m < 0:1 V s1 , for reactive block is minimal. Although the block supplies a
some domains (e.g., Rb ¼ 2:5 lm) block regeneration large flux of A at distances close to the electrode surface,
has a major influence on mass transport over the full it only does so for areas of the uncovered electrode surface
range of coverage. Fig. 12(a) illustrates the concentra- that are close to the blockedjunblocked boundary. For
tion profile, taken at E ¼ 0:6 V (a potential where instance, in Fig. 6 we approximated the ‘‘range’’ of
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 203

Fig. 13. Plot of Ip =2pR0 ½Abulk FD against h for diffusion domains where
Rb ¼ 2:5 lm, D ¼ 2:0  106 cm2 s1 , k 0 ¼ 0:005 cm s1 , and m ¼ 0:05,
0.1, 0.5 and 2.0 V s1 . The solid curves correspond to khet ¼ 0:01
cm s1 and the dashed lines correspond to khet ¼ 0 cm s1 .

region over which peak current is independent of scan


rate.

2.3.4. The effect of k 0


In the case of a given coverage with an inert block,
the peak dimensionless current, wmax , varies between
two limiting cases (reversible and irreversible electrode
kinetics) depending on the scan rate, m, and magnitude
of the electron transfer rate constant, k 0 [2]. The linear
Fig. 12. (a) Dimensionless concentration profile, taken at E ¼ Ep and
sweep voltammograms in Fig. 14(a) demonstrate the
(b) the corresponding linear sweep voltammogram (solid curve) for a effect of varying k 0 for a diffusion domain, where
domain where Rb ¼ 20 lm, h ¼ 0:1, khet ¼ 10 cm s1 , k 0 ¼ 0:005 Rb ¼ 2:5 lm, h ¼ 0:5, D ¼ 2:0  106 cm s1 , khet ¼ 10
cm s1 , D ¼ 2:0 106 cm2 s1 and m ¼ 2:0 V s1 . Overlaid as a dashed cm s1 , m ¼ 0:1 V s1 and k 0 ¼ 101 and 105 cm s1 . As
curve in (b) is the corresponding linear sweep voltammogram for the we move from fast (k 0 ¼ 10 cm s1 ) to slow
inert domain, i.e., khet ¼ 0 cm s1 .
(k 0 ¼ 0:00001 cm s1 ) electrode kinetics we observe an
expected change in the voltammogram shape, i.e., a shift
in the peak potential, but only a very slight decrease in
block regeneration as 1.5 lm (for the specified param- the peak dimensionless current: 7:330  7:314 ¼ 0:016,
eters). Therefore, when dealing with large domains of corresponding to a 0.2% decrease. For the same domain
low coverage, the area of the uncovered electrode that where the blocks are inert, Fig. 14(b), the decrease is
block regeneration will influence is small compared to much more significant: 0:414  0:312 ¼ 0:102, corre-
(1  hÞpR20 (the total uncovered area). Additionally, sponding to a 25% decrease. Similar trends are observed
because diffusion to and from the domain outskirts is for domains of high coverage (h > 0:4), providing the
approximately the same as that for the inertly blocked coupled heterogeneous reaction is fast enough (khet > 1
domain, we observe a reverse peak. The voltammogram cm s1 ).
for the reactive block in Fig. 12 also illustrates the ab- The reason for these observations has already been
sence of a ‘‘limiting current’’ which is observed for do- discussed – we observe a peak current when the poten-
mains of high coverage. This is to be expected for tial is such that the electrode kinetics are fast enough to
domains where block regeneration has little influence on achieve ½Az¼0 ¼ 0. The larger the electron transfer rate
the mass transport to the electrode surface. Fig. 13 il- constant, the lower is the required potential, hence as we
lustrates a similar plot to Fig. 11(a); the only difference increase k 0 we observe a decrease in Ep . The slight de-
is that khet ¼ 0:01 cm s1 . As seen, a slower catalytic crease in current is due to the increased depletion of A
reaction results in lower peak currents and a smaller that is not fully compensated by block regeneration.
204 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

same domain but with an inert block, which are overlaid


in Fig. 15. As seen, the pre-peak is almost identical to
the peak for the inertly blocked domain. In the con-
centration profiles displayed previously a severe con-
centration gradient at the blockedjunblocked electrode
boundary was labelled as the major source of the in-
creased currents. These profiles were all taken at po-
tentials where the reduction of A to B at the electrode
surface was rapid and able to ‘‘compete’’ with the cou-
pled chemical reaction. In the case of the pre-peak,
observed in Fig. 15, the potential is such that reduction
of A to B at the unblocked electrode surface is simply
not fast enough to generate a large concentration gra-
dient similar to those seen in previous profiles. There-
fore, we observe a peak where the mass transport is
dominated by the effect of the increasing voltage (scan
rate) with block regeneration playing a minor role. It is
not until we reach higher potentials that block regen-
eration plays a major role in mass transport and we
observe the larger peak currents. In the case of the
fastest electrode kinetics (k 0 ¼ 10 cm s1 ) a pre-peak is
not observed because rapid reduction is achieved at
much lower potentials so block regeneration plays its
role much earlier on in the scan. If we increase the scan
rate and/or increase the coverage then the pre-peak will
be shifted to more negative potentials but the main peak
(due to block regeneration) will stay at approximately
the same potential. Thus at higher values of m and h we
do not observe such anomalies.

2.3.5. The effect of khet


Fig. 16 illustrates linear sweep voltammograms for a
domain where Rb ¼ 3:0 lm, h ¼ 0:5 and khet is varied
between 101 and 105 cm s1 (the other parameters are
given in the figure legend). Similar plots are obtained for
Fig. 14. Dimensionless current linear sweep voltammograms for a
domains where block regeneration dominates the mass
single diffusion domain where khet ¼ (a) 10 cm s1 and (b) 0 cm s1 . In transport to the electrode surface (i.e., h > 0:4). There
both (a) and (b) D ¼ 2:0  106 cm2 s1 , Rb ¼ 2:5 lm, h ¼ 0:5, m ¼ 0:1 are two major features. First, the system appears to
V s1 , and k 0 ¼ 101 and 105 cm s1 . become diffusion controlled for khet > 103 cm s1 . Ini-
tially, this value may seem large in size; all electro-
chemists know that heterogeneous electron transfer
An interesting situation occurs when we vary k 0 in reactions are diffusion controlled when k 0 > 1–10
domains of low coverage (h < 0:05). Figs. 15(a)–(d) il- cm s1 , depending on the system parameters. However,
lustrate the simulated linear sweep voltammograms for a a comparison with Butler–Volmer kinetics is invalid. A
domain where h ¼ 0:02, m ¼ 0:02 V s1 , Rb ¼ 3:0 lm more detailed discussion and mathematical treatment
and k 0 ¼ 101 , 101 , 103 and 105 cm s1 . In all four can be found in Appendix A, but the general point is as
figures we have plotted the voltammograms for both the follows. In cyclic voltammetry, for example, diffusion
reactive (khet ¼ 100 cm s1 ) and inert (khet ¼ 0 cm s1 ) controlled electron transfer redox reactions are electro-
pffiffiffi The domain is relatively large ðR0 ¼
block. chemically reversible – an equilibrium is maintained at
Rb = h ¼ 17 lm) so mass transport will be influenced by the electrode surface at all times. The coupled hetero-
the increasing voltage. Hence, we do not observe a geneous reaction studied in this work is chemically ir-
‘‘limiting current’’ and as we decrease k 0 there is a de- reversible – there is no equilibrium.
crease in wmax . More interestingly, there appears to be a Second, as we increase khet , we observe a corre-
pre-peak that becomes more defined as the electrode sponding increase in Ep until we reach the diffusion
kinetics become slower. The origin of this peak can be controlled regime. The reason for this has already been
understood by considering the voltammograms for the touched upon. For domains where mass transport is
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 205

Fig. 15. Simulated voltammograms for a diffusion domain, where Rb ¼ 3:0 mm, h ¼ 0:02, D ¼ 2:0  106 cm2 s1 and m ¼ 0:02 V s1 . In each part the
two voltammograms are for a reactive (khet ¼ 10 cm s1 ) and inert (khet ¼ 0 cm s1 ) block. Parts (a)–(d) correspond to different electron transfer rate
constants: k 0 ¼ (a) 101 , (b) 101 , (c) 103 and (d) 105 cm s1 .

dominated by block regeneration we observe a peak electrode surface. By comparison with the homogeneous
current when ½Az¼0 ¼ 0. Faster coupled reactions sup- pathway for this reaction (pathway (A)), we can write
ply more A to the electrode surface meaning we need to the heterogeneous analogue as:
reach higher potentials where the electron transfer re- kf
action is fast enough to reduce all of the regenerated A. CoðIIÞLðaqÞ þ e ¢ CoðIÞLðaqÞ Unblocked electrode
kb

2.3.6. Summary of the factors discussed k1


RBr2 ðoilÞ þ CoðIÞLðaqÞ ! Br ðaqÞ þ CoðIIÞLðaqÞ
From the preceding discussion we can categorize the
voltammetry of catalytically reactive partially blocked þ RBr Drop surface ðEÞ
electrodes into three main areas, as given in Table 1.
k2
RBr þ CoðIÞLðaqÞ ! Br ðaqÞ þ CoðIIÞLðaqÞ
2.4. Relationship with experimental system
þ R0 ðoilÞ Drop surface or solution
The experimental system to be compared with the where the first chemical step (k1 Þ is the reaction with
theory described above is the catalytic regeneration of DBCH, the second step is the reduction of the RBr
0
aqueous vitamin B12r (Co(II)L) by microdroplets of 1,2- radical (k2 Þ and R is cyclohexene. A similar scheme can
trans-dibromocyclohexane (DBCH) immobilized on the be written for the E2 mechanism (pathway B). However,
206 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

 
d½CoðIÞL
D ¼ 2k1 ½CoðIÞLZ¼0 : ð20Þ
dZ Z¼0

Note that this is equivalent to saying RBr is at a steady


state. Hence, the relationship between khet (theory) and
k1 (experiment) is given by:
khet ¼ 2k1 : ð21Þ
A major difference between our scheme and the homo-
geneous analogue involves the organic mediator,
DBCH. In the homogeneous mathematical treatment,
the depletion of the mediator is accounted for in the
mass transport equations. However, in our model we
have assumed that depletion of DBCH at the droplet
surface has a negligible effect on the rate of the coupled
heterogeneous reaction. The validity of this assumption
will be discussed later.

Fig. 16. Dimensionless current linear sweep voltammograms for a 2.5. Accounting for the random distribution of microdro-
domain, where Rb ¼ 3:0 lm, h ¼ 0:5, k 0 ¼ 0:005 cm s1 , D ¼ plets
2:0  106 cm2 s1 , m ¼ 0:02 V s1 , and the heterogeneous rate con-
stant for the reactive block, khet , varies between 100 and 105 cm s1 .
The voltammetric response of a modified electrode is
the sum response of all the diffusion domains present (in
the experimental systems involved in this work there are
over 100,000 domains on the electrode surface). Because
since the two reaction pathways (A) and (B) are kinet-
microdroplet distribution is random, we require a pro-
ically indistinguishable [18], the results in this section are
tocol that allows us to predict the voltammetric re-
also applicable to the liquidjliquid equivalent of path-
sponse of an ensemble of randomly distributed diffusion
way (B). The overall flux of Co(I)L at the droplet sur-
domains.
face can be written as:
  When working with inert microdroplets in a previous
d½CoðIÞL paper [3], we used the peak dimensionless current, wmax ,
D ¼ flux due to first step
dZ Z¼0 as a variable for characterising the modified electrode.
þ flux due to second step ð17Þ In the work presented here we will use a similar method
to determine a value of k1 for the reaction in pathway
where we could write the flux due to the two individual (E). That is, for given values of D, khet ; k 0 and m, we will
steps as: focus primarily on the relationship between the volume
  first step of blocking material (DBCH microdroplets) and the
d½CoðIÞL 
D  ¼ k1 ½CoðIÞLZ¼0 ð18Þ peak dimensionless current, wmax .
dZ 
Z¼0
The following method is a modification of Model B1
  second step described in [3]. We assume that all droplets on the
d½CoðIÞL 
D  ¼ k2 ½CoðIÞLZ¼0 ð19Þ electrode surface are hemispherical with a constant ra-
dZ 
Z¼0 dius, Rb . The validity of these two assumptions, partic-
However, the rate of the second step depends entirely on ularly the monodisperse distribution, is discussed in Part
the rate of the first and because it is the first step which is II [3]. The number of droplets on the electrode surface,
rate determining we express Eq. (17) as: Nblock , can be calculated using

Table 1
Categorization of catalytically reactive partially blocked electrodes
Block regeneration dominates Both block regeneration and scan rate influence mass Mass transport dominated by the
mass transport transport to the electrode surface increasing voltage
Observe greatly enhanced Ip increases with m and k 0 Small increase in Ip and no ‘‘limiting’’
‘‘limiting’’ currents current observed
‘‘Ip ’’ (almost) independent A pre-peak is observed at low h and m Voltammetry (almost) identical to
of m and k 0 khet ¼ 0 domains
Ep depends on when [Az¼0 ¼ 0 Back peak observed in CV
No back peak observed in CV
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 207

3Vblock Fig. 17 illustrates a plot of simulated ðpNblock =


Nblock ¼ ð22Þ
2Rb Aelec Aelec ÞR20 P ðR0 Þwmax ðR0 Þ against R0 , where Vblock ¼
4:1  105 cm3 , k 0 ¼ 0:005 cm s1 , D ¼ 2:0  106 cm2
where Vblock is the volume of blocking material (a known s1 , khet ¼ 10 cm s1 , Rb ¼ 2:5 lm, m ¼ 0:1 V s1 and
experimental quantity) and Rb , the DBCH droplet ra- Aelec ¼ 0:2011 cm2 . From Eq. (26) it follows that the
dius, can be determined via experiments with an inert theoretically predicted current for the whole electrode is
redox couple (vide infra). The number of domains with given by the area under the curve. When working with a
radius R0 þ dR0 on the electrode surface, NR0 , is given by particular redox couple of known k 0 and D, for differ-
the total number of domains, Nblock , multiplied by the ent values of khet ; Vblock and m, we can simulate such
probability of the domainÕs occurrence: curves and thus generate a table of theoretical data
NR0 ¼ Nblock P ðR0 Þ dR0 ð23Þ which can be compared with those obtained experi-
mentally. Determining khet ðor 2k1 Þ is then a matter of
where P ðR0 Þ is given by
interpolation.
2
2pR0 Nblock pRA0 Nblock
P ð R0 Þ ¼ e elec : ð24Þ
Aelec
We then vary R0 between Rb and 3hR0 i. For R0 < Rb , the
3. Experimental
domain is completely blocked so wmax ¼ 0, and the
probability that domains of radius R0 > 3 < R0 > occur
3.1. Chemical reagents
is negligible. Given values of k 0 , D, m and khet , we can
convert to dimensionless variables and simulate a peak
All reagents were of the highest grade available
dimensionless current for each domain, wmax ðR0 Þ, in the
commercially and were used without further purifica-
R0 range:
  tion. These were dodecane (Aldrich, 99+%), trans-1,
0
wmax ðR0 Þ ¼ f h; mdl ; kdl0 ; khet : ð25Þ 2-dibromocyclohexane (Aldrich, GC), potassium hexa-
cyanoferrate(II) trihydrate (Lancaster, 99%), hydroxo-
The peak dimensionless current due to the whole elec-
cobalamin hydrochloride (vitamin B12a , Sigma, 98%),
trode is then given by the following integral:
Z potassium chloride (Fluka, AnalR), acetonitrile (Fisher,
pNblock 3hR0 i 2 dried and distilled) and ethyl acetate (Fisher, HPLC).
wmax ¼ R0 wmax ðR0 ÞP ðR0 Þ dR0 ð26Þ
Aelec 0 Water, with a resistivity of not less than 18 MX cm, used
to make the electrolyte and buffer solutions was taken
in which the ðpNblock =Aelec ÞR20 P ðR0 Þ term is used to weight
from an Elgastat system (USF, Bucks., UK). All solu-
each simulated wmax value according to the size and
tions were out gassed with oxygen-free nitrogen (BOC
quantity of each domain.
Gases, Guilford, Surrey, UK) for at least 20 min prior to
experimentation. All experiments were conducted at
22  3 °C. Experiments involving vitamin B12a were
conducted in a pH 2.5 phosphate buffer prepared with
an ionic strength of 0.2 M (0.1 M sodium dihydrogen-
phosphate and 0.1 M phosphoric acid (85 wt% solution
in water) both from Aldrich). At this pH, the Co(II)
complex (vitamin B12r Þ exists in its ‘‘base off’’ form,
which is easier to reduce than the ‘‘base on’’ form [17].
Vitamin B12a has a cobalt(III) centre and undergoes
two consecutive one-electron reductions to B12r (Co(II))
and B12s (Co(I)). In this work we were interested only in
the B12r =B12s couple and so to avoid the complication
caused by the presence of B12a ,a bulk electrolysis of B12a
to B12r was performed before each experiment as un-
dertaken by Rusling and co-workers [18]. The electrol-
ysis was stopped after about 4 h during which ca. 1.5
equivalents of the required charge had passed. After the
Fig. 17. Simulated curve of ðpNblock =Aelec ÞR20 P ðR0 Þwmax ðR0 Þ vs. R0 for pre-electrolysis the solution had a brown colour, indi-
an electrode modified with ‘‘reactive’’ blocks of surface radius cating the presence of vitamin B12r [19] and linear sweep
Rb ¼ 2:5 lm and khet ¼ 10 cm s1 . The other parameters are m ¼ 0:1 voltammetry with a 3.0-mm diameter glassy carbon
V s1 , D ¼ 2:0  106 cm2 s1 , k 0 ¼ 0:005 cm s1 , Aelec ¼ 0:20 cm2 and
Vblock ¼ 4:1  105 cm3 . The area under the curve is the predicted peak
electrode showed the absence of the B12a –B12r reduction
dimensionless current, wmax , for a modified electrode with the stated peak. A constant flow of nitrogen was kept over the B12r
parameters. solution at all times.
208 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

3.2. Instrumentation

Electrochemical experiments were under taken in a


conventional three-electrode cell, employing a 5.1 mm
diameter basal plane pyrolytic graphite electrode (bppg,
Le Carbone Ltd., Sussex, UK), a platinum wire counter
electrode and a saturated calomel reference electrode
(Radiometer, Copenhagen, Denmark). Electrochemical
data were recorded using a commercial computer con-
trolled potentiostat (AUTOLAB PGSTAT30, Eco-
Chemie, Utrecht, The Netherlands).
The bppg working electrode was modified with mi-
crodroplets of DBCH by solvent evaporation of a 2–10
lL aliquot of DBCH + acetonitrile stock solution. Sim-
ilarly, modification with microdroplets of a mixture of
DD and DBCH were obtained via evaporation of a 4 lL
aliquot of DD + DBCH + ethyl acetate solution – it was
Fig. 18. Cyclic voltammograms for the oxidation of 2.04 mM potas-
found that volumes in excess of 4 lL did not stay con-
sium hexacyanoferrate(II) trihydrate aqueous solutions containing 0.1
fined to the bppg surface. The electrode was cleaned M potassium chloride at a 5.1 mm diameter basal plane pyrolytic
immediately prior to experimentation by rinsing with graphite electrode modified with 0, 120, 300 and 600 nmol of DBCH.
acetone and water and the surface was renewed by The scan rate was 0.2 V s1 .
polishing on carborundum paper (P1000 grade, Acton
and Bormans, Stevanage, UK).
ther species in the redox couple, the voltammetry dis-
3.3. Computation playing all the characteristics of inert partially blocked
electrodes [1–3]. Analysis of voltammograms obtained
The voltammetric response of reactive diffusion do- from the naked electrode with DigiSimÓ gave a diffusion
mains was simulated using the same protocols described coefficient, D ¼ 5:6  0.3  106 cm2 s1 , and electron
in Part I [2]. Concentration profiles were obtained via transfer rate constant, k 0 ¼ 0:01  0.002 cm s1 , for the
Matlab 5.1.0. Fe(CN4 3
6 )/Fe(CN6 ) redox couple, both values in

4. Results and discussion

Experiments with DBCH and vitamin B12r were split


into two sections. First, the bppg electrode was modified
with varying amounts of pure DBCH in order to obtain
a rate constant for the DBCHjB12s liquid–liquid reac-
tion. Second, to mimic the experiments performed by
Rusling and co-workers [18], the bppg electrode was
modified with various ratios of DBCH and dodecane
(an inert oil).

4.1. Electrodes modified with pure DBCH

4.1.1. Determination of DBCH droplet radius


As mentioned in Section 2.5, before performing any
‘‘reactive’’ experiments with DBCH-modified electrodes,
for the purpose of data analysis it is necessary to obtain
a DBCH droplet radius, Rb . This was achieved by re-
cording the voltammetry of a basal plane electrode Fig. 19. Simulated relationship (according to Model B1 Þ between wmax
modified with microdroplets of DBCH in a solution of and Rb for a 5.1 mm basal plane electrode modified with 4.1  105
cm3 of inert blocking material at four different scan rates (0.1, 0.2, 0.5
ferrocyanide (a medium in which DBCH is inert).
and 1.0 V s1 ), where D ¼ 5:6  106 cm2 s1 and k 0 ¼ 0:01 cm s1 .
Fig. 18 illustrates cyclic voltammograms at a DBCH- The dashed lines represent the experimental data for DBCH blocking
modified bppg electrode in 2.04 mM ferrocyanide + 0.1 in ferrocyanide solution and the dots are the best agreement between
M KCl. The microdroplets appear not to react with ei- theory and experiment.
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 209

agreement for that determined previously on carbon


electrodes [20]. With values for D, k 0 and Ip for different
scan rates and DBCH blocking volumes, we were able to
simulate the electrode response according to Model B1
[3]. Fig. 19 illustrates the theoretically predicted varia-
tion of peak dimensionless current with droplet radius at
four different scan rates where Vblock ¼ 4:1  105 cm3 ,
Aelec ¼ 0:2011 cm2 , D ¼ 5:6  106 cm2 s1 and
k 0 ¼ 0:01 cm s1 . Overlaid as dashed lines are the ex-
perimental results – the peak currents for each scan rate
were converted into dimensionless form using Eq. (15).
The points where the dashed and solid lines cross are
highlighted in the figure and correspond to the best
agreement between theory and experiment. From a
number of plots like Fig. 19, we were able to determine
the (model B1 Þ droplet radius as Rb ¼ 2:5  0.5 lm.

4.1.2. Reaction with vitamin B12s


Initial experiments with DBCH modified electrodes
gave poor reproducibility at low block volumes, i.e.,
Vblock < 2  105 cm3 corresponding to around 150
nmol. For this reason, subsequent experiments were
performed on electrodes modified with Vblock > 3  105
cm3 DBCH. To obtain values of D and k 0 for the
B12r =B12s redox couple necessary for the simulations,
cyclic voltammograms with a naked electrode were re-
corded over a range of scan rates. Subsequent analysis
with DigiSimÓ gave D ¼ 2:0  0:1  106 cm2 s1 and
k 0 ¼ 0:005  0:002 cm s1 , which agrees well with values
previously determined [18].
Fig. 20(a) illustrates cyclic voltammograms recorded
at 0.05 V s1 in 1.18 mM vitamin B12r , where the elec-
trode is naked and modified with 300 nmol of DBCH
(corresponding to Vblock ¼ 4:1  105 cm3 ). The reduc-
tion of vitamin B12r to B12s in pH 2.5 phosphate buffer at
a bppg electrode occurs quite close to solvent break-
down leading to scans where peaks are observed just
before the potential becomes too negative. However, the
differences between the two scans are clear and very
much like those observed for the simulated voltammo- Fig. 20. (a) Cyclic voltammograms for the reduction of 1.18 mM B12r
solution in aqueous pH 2.5 phosphate buffer at a 5.1 mm diameter
grams in Fig. 3. Not only do we observe the absence of a bppg electrode modified with 0 (naked) and 300 nmol of DBCH. The
back peak, but also the current at the modified electrode scan rate was 0.05 V s1 . (b) Cyclic voltammograms for the reduction
is greatly enhanced with what appears to be a limiting of 1.21 mM B12r solution in aqueous pH 2.5 phosphate buffer at a 5.1
plateau before we enter solvent breakdown. All this mm diameter bppg electrode modified with 0 (naked) and 6:4  105
suggests that we are dealing with a fast coupled heter- cm3 dodecane. The scan rate was 0.2 V s1 .
ogeneous reaction where mass transport is dominated by
block regeneration. The catalytic reaction consumes
DBCH and, as expected, repetitive scans show decreased consistent with the predicted response for a droplet ra-
peak currents, the decrease being more significant for dius of approximately 4–5 lm. This suggests that the
the lower scan rates. The influence of this droplet de- presence of any oxygen dissolved in the microdroplets
pletion is discussed later. had a negligible effect on the observed voltammetry.
To confirm that the catalytic increase in current was Also, scans recorded with a DBCH-modified electrode
solely due to DBCH, voltammograms were recorded in phosphate buffer (i.e., no B12r Þ showed no appreciable
with similar block volumes of pure dodecane, an ex- direct reduction of DBCH within the potential window.
ample of which is illustrated in Fig. 20(b). These gave a Fig. 21 illustrates voltammograms recorded at four
decrease in current, when compared to the naked results, different scan rates in 1.18 mM vitamin B12r , where the
210 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

modified electrodes in Fig. 21, where khet ¼ 10, 100 and


1000 cm s1 . The corresponding relationships between Ip
and m are displayed in Fig. 22(b). In both (a) and (b) the
experimental results are overlaid as circles. As observed
in Fig. 22(b), the simulated results suggest Ip should be
(more or less) independent of m at all three values of khet
considered. This is a similar situation to that discussed
in Section 2.3.3 but the cause is slightly more compli-
cated. For example, Fig. 10 in which we observed a peak
current–scan rate independence concerned a single dif-
fusion domain where mass transport was dominated by
block regeneration. Our modified electrode surface is an
ensemble of domains, not all of which show such be-
haviour. To consider the effect of having a range of
domains (each with Rb ¼ 2:5 lm), we need to take into
account their relative numbers. This is illustrated in
Fig. 23 where we have plotted P ðR0 Þ against h for the
modified bppg electrode in Fig. 21(a) (Vblock ¼ 4:1 
105 cm3 ). Notice that domains with h > 1:0 do have a
probability of occurring. This is the basis of how our
model takes into account droplet overlap and is dis-
cussed in more detail in Part II [3] (in an attempt to
focus on the 0 < h < 1 region we have omitted the rest
of the distribution function which tends to zero for
higher values of hÞ. Referring to Fig. 11(a) in Sec-
tion 2.3.3, we can separate Fig. 23 into three regions:
wmax ¼ 0 corresponding to h P 1; 0:4 < h < 1 where
block regeneration dominates mass transport to the
electrode surface; and h < 0:4 where the voltammetry is
influenced by scan rate. The block volume is such that
domains in the h < 0:4 region contribute a small pro-
portion to the total number of domains on the electrode
surface. Hence, the voltammetric response will be
Fig. 21. Linear sweep voltammograms for the reduction of 1.18 mM dominated by domains in the region 0:4 < h < 1 (re-
B12r solution in aqueous, pH 2.5, phosphate buffer at a 5.1 mm di- member that domains in the region h  1 have no vol-
ameter bppg electrode modified with (a) 300 and (b) 600 nmol of tammetric response), each of which has a peak current
DBCH. The scan rates are labelled on the figures (0.05, 0.1, 0.5 and 1.0
which is independent of scan rate. If the major contri-
V s1 ).
bution to the total current comes from a collection of
domains which posses an Ip –m independence, the ob-
bppg electrode is modified with (a) 300 nmol and (b) 600 served peak current for the whole electrode should be
nmol DBCH. Only the forward scans are plotted (to independent of scan rate. In addition, the recorded
avoid over crowding) and the m ¼ 0:05 V s1 scan is voltammograms should show little variance with m, as
printed in bold to give a clearer representation. As we seen in Fig. 10. Increasing the block volume results in an
increase Vblock , the quality of the scans decreases but a even smaller contribution from the h < 0:4 region so we
peak/limiting current is still visible. At the highest cov- should expect a similar response from the electrode
erage, increasing the scan rate appears to have little ef- blocked with 600 nmol DBCH.
fect on the recorded voltammograms. Similarly, in However, the predicted Ip –m independence is observed
Fig. 21(a) the relative increase in peak/limiting current only at the higher coverage (ignoring the 0.2 V s1 re-
on going from 0.05 to 1.0 V s1 is extremely small sult) and high scan rates, suggesting that the peak cur-
compared to what one would expect from a m1=2 de- rents for the low scan rate data are too low. This
pendence, for example. discrepancy between theory and experiment probably
With knowledge of Rb , D, k 0 ; Aelec and Vblock , the results from our initial assumption that the reactivity of
theoretically predicted peak dimensionless current, wmax , the block is unaffected by DBCH depletion. Table 2 lists
can be calculated for chosen values of khet and m via the the approximate percentage depletion of DBCH for
approach described in Section 2.5. Fig. 22(a) illustrates both sets of (forward) scans in Fig. 21. The DBCH de-
the simulated relationship between wmax and m for the pletion is calculated as follows. Since the electrode is
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 211

Fig. 22. Simulated variation of (a) wmax with scan rate and (b) Ip with scan rate for the modified electrodes in Fig. 21 where khet ¼ 101 , 102 and 103
cm s1 . Part (i) corresponds to the electrode modified with 300 nmol of DBCH and (ii) corresponds to the 600 nmol modification. The experimental
results are plotted as circles.

heavily blocked and the catalytic currents are greatly at low scan rates we observe a lower Ip than expected
enhanced, we can approximate the current as being en- due to the currents being limited by DBCH diffusion
tirely due to the reduction of regenerated B12r . In within the droplet, a factor unaccounted for in our
pathway (A), the reaction of one DBCH molecule re- modelling. Upon increasing the scan rate, the depletion
sults in the generation of two B12r species, so the moles is less and the DBCH diffusion layer has a shorter time
of DBCH involved in the reaction is equal to half the in which to grow. Therefore, the effect of DBCH de-
moles of electrons transferred. This allows us to calcu- pletion becomes negligible and we observe the predicted
late the loss of DBCH (we know how much is deposited Ip –m independence. At higher blocking volumes the area
before the scan) which leads to a percentage depletion. of uncovered electrode is smaller, leading to lower cur-
The numbers in Table 2 suggest that the percentage rents, and the total microdroplet surface area is in-
depletion is not significantly high. However, we need to creased. Both of these factors lessen the effect of DBCH
remember that the reaction occurs only at the droplet depletion and we obtain results that show better agree-
surface (the liquidjliquid interface) – it is heterogeneous. ment with theory over a larger range of scan rates.
Therefore, a diffusion layer, similar to that produced in In the light of these findings, further experiments
a simple potential step experiment, will be formed as the were conducted at scan rates in the range 0:5 6 m 6 4:0
scan progresses – the longer the scan (i.e., the lower the V s1 with similar blocking amounts. Analysis of these
scan rate) the larger this diffusion layer becomes. Hence, results also suggested a value of khet to be 150 < khet <
212 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

4.2. Electrodes modified with a mixture of DBCH and


dodecane

This section examines the voltammetry of electrodes


modified with diluted DBCH oil droplets, the major
constituent of the drop being dodecane. Therefore,
comparisons of experimental results with simulations
lead to the determination of an effective heterogeneous
rate constant, k1eff , for the liquid–liquid reaction. We
are interested in the variation of k1eff with the DBCH
concentration.

4.2.1. Determination of DBCH/dodecane droplet radius


Following the same method described in Section
4.1.1, the inert blocking of dodecane and DBCH was
investigated with solutions of 50–70 mM dodecane and
1.5–4.1 mM DBCH in ethyl acetate, corresponding to
microdroplet DBCH concentrations of 0.096 6
Fig. 23. The distribution of diffusion domains on a 5.1 mm diameter [DBCH] 6 0.32 M. In all cases the droplet radius was
bppg electrode surface modified with 4.1  105 cm3 of blocking ma-
determined to be Rb ¼ 4:5  0:5 lm.
terial, where the individual block radius is 2.5 lm.

4.2.2. Reaction with vitamin B12s


Fig. 24 illustrates voltammograms recorded at a bppg
250 cm s1 . According to Eq. (21), this suggests that the electrode modified with 4 ll of 70 mM dodecane and 3.1
rate constant for the liquid–liquid reaction between mM DBCH solution (corresponding to [DBCH] ¼ 0.19
DBCH and B12s is 75 < k1 < 125 cm s1 . M and Vblock ¼ 6:5  105 cm3 ) in a 1.15 mM solution of
pH 2.5 vitamin B12r . The difference between the vol-
4.1.3. The double peak anomaly tammograms recorded at ‘‘high’’ (1.0, 2.0 and 3.0 V s1 )
The quasi-reversible kinetics of the B12r =B12s redox and ‘‘low’’ (0.1 V s1 ) scan rates agrees with that dis-
couple suggest that a double peak, as discussed in cussed earlier. Although our reactant has been diluted,
Section 2.3.4, should be observed for domains of low about 5% of the volume applied is DBCH. Therefore, as
coverage at low scan rates. However, attempts to ob- an initial estimate, we would not expect the effective
serve this anomaly with modified electrodes of a suit- heterogeneous rate constant to decrease by much more
able coverage were unsuccessful, primarily due to two than an order of magnitude, a value still high enough for
reasons. First, a modified electrode is an ensemble of block regeneration to dominate mass transport to the
diffusion domains, which in our case vary in size. Even electrode surface. Also, the block volume we are work-
at low block volumes, there will still be domains of ing with is large enough for domains that do show an
higher coverage (i.e., domains for which no double Ip –m dependence to have a negligible contribution. At
peak is observed) that contribute to the overall vol- the higher scan rates we observe this expected behaviour
tammetry. Because the observed current is a sum of – the voltammetry is almost independent of the rate of
that for all the domains on the electrode surface, the the potential sweep. At the low scan rates, the effect of
double peak signal, due to the small range of domains DBCH depletion is even more apparent for these mod-
we are interested in, could be lost. Second, at low ified electrodes. Not only is the current smaller than
block volumes and low scan rates the effect of droplet expected, a clear forward peak and slight back peak are
depletion is large. also observed. On closer inspection of the 0.1 V s1

Table 2
Approximate percentage depletion of DBCH (up to )1.1 V vs. SCE on the forward scans) on the modified electrodes in Fig. 21
m/V s1 300 nmol 600 nmol
DBCH depletion/nmol Depletion/% DBCH depletion/nmol Depletion/%
0.05 11 3.7 8.1 1.3
0.1 6.1 2.0 4.0 0.67
0.5 1.4 0.47 0.83 0.14
1.0 0.73 0.24 0.41 0.07
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 213

Fig. 26. Dependence of the effective heterogeneous rate constant, k1eff ,


on the concentration of DBCH in dodecane.
Fig. 24. Cyclic voltammograms for the reduction of 1.15 mM B12r
solution in aqueous, pH 2.5, phosphate buffer at a 5.1 mm diameter
bppg electrode modified with 6.5  105 cm3 of 0.19 M DBCH in
dodecane. The illustrated voltammograms correspond to scan rates of
and 3.0 V s1 . The high block volume results in vol-
0.1, 1.0, 2.0 and 3.0 V s1 . tammetry dominated by domains of high coverage and
we observe only a slight increase in Ip with m over the
range shown. Using the high scan rate experimental data
eff
and these simulated curves, the khet (and therefore k1eff Þ
values were determined via interpolation, as illustrated
in Fig. 25 for the m ¼ 2:0 V s1 scan, for a range of
DBCH concentrations. The results are illustrated in
Fig. 26 and suggest a linear relationship between k1eff and
[DBCH], the gradient of which is 21 M1 cm s1 and can
be equated to the apparent bimolecular rate constant
(k ¼ k1eff /[DBCH]). In a similar experiment, Rusling and
co-workers [18] investigated the liquid–liquid reaction
between aqueous B12r and DBCH diluted in benzonit-
rile. Using the SECM/ITIES method described by Bard
and co-workers [11], they determined effective hetero-
geneous rate constants for different concentrations of
DBCH in benzonitrile. Their results also indicated a
linear relationship with a value of 3.0 M1 cm s1 for the
apparent bimolecular rate constant. The difference (al-
most an order of magnitude) between the two apparent
bimolecular rate constants could be due to a number of
reasons. First, benzonitrile and dodecane are signifi-
Fig. 25. Simulated variation of the peak current with k1eff for the cantly different organic solvents: benzonitrile is polar
modified electrodes in Fig. 24, where the scan rates are 1.0, 2.0 and 3.0
V s1 . The dashed line is an example of the interpolation procedure by
whereas dodecane is not. Thus, in the two liquids, the
which we determine k1eff . solvation of the reactants and transition state will vary,
leading to different Gibbs energy of activation values for
voltammogram, a slight shoulder can be seen before the the two systems. Second, our work has clearly shown
reductive peak. Similar features have been observed in that droplet depletion can have a limiting effect on the
homogeneous media and are associated with reactant observed current. If it is neglected to model diffusion on
(DBCH) depletion [24]. Thus, when the electrode is both sides of the ITIES one must work with system pa-
modified with ‘‘diluted’’ reactant, the limiting effect of rameters where the effect of reactant depletion is negligi-
diffusion within the droplet is enhanced. ble. Since our approach is based on transient
eff
Fig. 25 illustrates the simulated variation between khet voltammetry we were able to do this easily by operating
and Ip for the modified electrode in Fig. 24 at 1.0, 2.0 at high scan rates. Once we had observed the predicted
214 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

Ip –m independence we could be certain our currents were Appendix A. Can we compare values of khet and k0
not limited by diffusion within the microdroplets. corresponding to diffusion limitation?
However, the SECM/ITIES technique used by Bard and
co-workers [11] and Rusling and co-workers [18] is Consider the following coupled heterogeneous elec-
based on steady-state measurements. Currents are re- trode reaction where electron transfer at a planar macro
corded at times when there will be well-established dif- electrode is preceded by an irreversible heterogeneous
fusion layers on both sides of the ITIES. Indeed, Unwin reaction:
and co-workers [21,22] have investigated the conse- khet
quence of reactant depletion in the second phase on such A!B
steady state measurements at the ITIES. Their work [22] kf ðFÞ
B þ e ¢ C:
suggests that the concentrations used by Rusling and co- kb
workers [18] falls into the region where the constant
After t ¼ 0, the potential is such that the reduction of B
composition assumption is invalid, leading to an un-
to C is rapid and limited by the irreversible heteroge-
derestimate of the actual bimolecular rate constant. In-
neous reaction. Therefore, the flux of B to the electrode
terestingly, the concentrations of DBCH required for
surface will be equal to the flux of A and we can write:
the constant composition assumption to be valid would  
result in SECM currents limited by diffusion of B12r I o½B
¼ DB ¼ khet ½Ax¼0 ðA:1Þ
rather than the rate of the liquid–liquid reaction [18,22]. FAelec ox x¼0
where DB is the diffusion coefficient of B and the other
symbols have already been defined. With B and C ini-
5. Conclusions tially absent, the equations that describe such a system
are:
In this paper we have seen how the principles devel-
oped in Parts I and II for the simulation of uncompli- diffusion equation:
 2 
cated electron transfer reactions at an electrode partially o½A o ½A
covered with inert blocking material can be extended ¼D ðA:2Þ
ot ox2
further to model irreversible liquid–liquid reactions at
the surface of microdroplets dispersed over an electrode boundary conditions:
surface. The transient technique developed in this work t¼0 x ½A ¼ ½Abulk ðA:3Þ
has a number of advantages over current probes for
reactions occurring at the interface between two im- t>0 x!1 ½A ¼ ½Abulk ðA:4Þ
miscible electrolyte solutions (ITIES). Not only are the  
experimental requirements relatively simple, compared o½ A 
t>0 x¼0 D ¼ khet ½Ax¼0 ðA:5Þ
to scanning electrochemical microscopy (SECM) for ox x¼0
example, but in theory we can measure very fast liquid– where D is the diffusion coefficient of species A, [Abulk is
liquid reactions before being limited by diffusion. Fur- the bulk concentration of A, and [Ax¼0 is the concen-
thermore, when studying fast liquid–liquid reactions tration of A at the electrode surface. A Laplace trans-
(khet > 1 cm s1 ) at high block volumes, we can easily formation of Eq. (A.2) and the first two boundary
identify results where the current is limited by diffusion conditions leads to:
in the second phase, a factor unaccounted for in our
½Abulk 1=2
modelling. When determining heterogeneous (bimolec- ½A ¼ þ f ðsÞeðs=DÞ x ðA:6Þ
ular) rate constants, this enables us to use the constant s
composition assumption with some degree of confi- where f ðsÞ is a factor to be determined and we have
dence, unlike in the SECM/ITIES method where reac- transformed from t to s space. The Laplace transform of
tant depletion in the second phase can often be the final boundary condition (Eq. (A.5)) allows the de-
problematic [21,22]. termination of f ðsÞ:
Hi ½Abulk
f ðsÞ ¼  ðA:7Þ
sðHi þ s1=2 Þ
Acknowledgements
where Hi ¼ khet =D1=2 . Substituting for f ðsÞ in Eq. (A.6)
The authors thank the EPSRC for studentships for gives:
T.J.D. and B.A.B. We further thank Russell Evans, Jay ½Abulk Hi ½Abulk eðs=DÞ
1=2
x
Wadhawan, Michael Hyde, Robert Jacobs and Phillip ½A ¼  ðA:8Þ
s sðHi þ s1=2 Þ
Greene for their assistance regarding experimental pro-
cedures, and Prof. J.F. Rusling for initiating our interest Now, we evaluate the Laplace transform of Eq. (A.1) to
in vitamin B12s chemistry. obtain an expression for the transformed current:
T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216 215

! kf
o½A FAelec khet ½Abulk A þ e ¢ B: ðGÞ
I ¼ FAelec D ¼ ðA:9Þ kb
ox s1=2 ðHi þ s1=2 Þ
x¼0
The current for this arrangement is given by
and the current is given by the inverse transform:  
  I ¼ FAelec kf ½Ax¼0  kb ½Bx¼0 ðA:12Þ
2
I ¼ FAelec khet ½Abulk eHi t erfc Hi t1=2 : ðA:10Þ
where kf and kb are defined in Eqs. (2) and (3), respec-
The diffusion limited current for this system, Id;i , is given tively. Following a similar procedure to that described
by the Cottrell equation [15]: above, we can express the current as [15]:
2  
FAelec D1=2 ½Abulk I ¼ FAelec kf ½Abulk eHr t erfc Hr t1=2 ; ðA:13Þ
Id;i ¼ : ðA:11Þ
p1=2 t1=2 where Hr ¼ ðkf þ kb Þ=D, assuming the diffusion coeffi-
cients of the two species are equal. The diffusion con-
We can compare the above irreversible heterogeneous
trolled current for the potential step experiment, Id;r is
kinetics with electron transfer kinetics by considering a
given by [15]:
simple potential step experiment on a one-electron redox
couple, where the potential applied is not high enough FAelec D1=2 ½Abulk
Id;r ¼ ðA:14Þ
to cause rapid reduction and B is initially absent: p1=2 t1=2 ð1 þ hÞ

Fig. 27. Simulated voltammetric response for the ‘‘potential step’’ experiments described in the text over (a) 10 ls and (b) 1 s, where the kinetic rate
constants vary between 103 and 101 cm s1 . Plots labelled (i) correspond to the irreversible heterogeneous reaction. Plots labelled (ii) correspond to
the electron transfer reaction where E ¼ 0:005 V vs. E00 . In both simulations T ¼ 298 K and D ¼ 1:0  105 cm2 s1 . The diffusion limited response
for each situation is plotted as a dashed line.
216 T.J. Davies et al. / Journal of Electroanalytical Chemistry 566 (2004) 193–216

where species at a rotating disc electrode, Harland and


ð
F EE00 Þ Compton [23] demonstrated that the chronoampero-
h¼e RT : ðA:15Þ metric response became diffusion limited around
khet ¼ 1000 cm s1 . They went on to investigate the ki-
Fig. 27 illustrates the voltammetric response for both
netics of bromine adsorption onto poly-crystalline
systems described in the above text over (a) short
platinum and determined the irreversible heterogeneous
(t ¼ 10 ls) and (b) long (t ¼ 1 s) time scales. The plots
rate constant to be ca. 220 cm s1 [23]. Therefore, the
labelled ‘‘(i)’’ correspond to the irreversible heteroge-
fact that our partially blocked electrode system also
neous reaction, whereas ‘‘(ii)’’ corresponds to the elec-
becomes diffusion controlled just after khet ¼ 1000
tron transfer reaction. The diffusion limited response is
cm s1 is not unreasonable.
overlaid as a dashed curve. In both parts (a) and (b) it is
clearly observed that the electron transfer reaction be-
comes diffusion controlled significantly earlier than the
irreversible heterogeneous reaction, for the same k 0 and
References
khet values. It could be argued that the figures differ
because they correspond to two slightly different ex- [1] C. Amatore, J.M. Saveant, D. Tessier, J. Electroanal. Chem. 147
perimental systems. However, that is the main point – a (1983) 39.
direct comparison between values of diffusion limited k 0 [2] B.A. Brookes, T.J. Davies, A.C. Fisher, R.G. Evans, S.J. Wilkins,
and khet is inappropriate because of the way diffusion K. Yunus, J.D. Wadhawan, R.G. Compton, J. Phys. Chem. B 107
limited is ‘‘defined’’ for the two types of reaction. In the (2003) 1616.
[3] T.J. Davies, B.A. Brookes, A.C. Fisher, K. Yunus, S.J. Wilkins,
case of Butler–Volmer electron transfer, the reaction P.R. Greene, J.D. Wadhawan, R.G. Compton, J. Phys. Chem. B
becomes diffusion controlled once equilibrium is 107 (2003) 6431.
achieved at the electrode surface. This is certainly not [4] K. Tokuda, T. Gueshi, H. Matsuda, J. Electroanal. Chem. 101
the case for irreversible heterogeneous kinetics where the (1979) 29.
[5] M. Etman, E. Levart, D. Schumann, J. Electroanal. Chem. 101
reaction becomes diffusion limited when mass transport
(1979) 141.
can no longer supply enough reactant. This is high- [6] J.F. Rusling, Pure Appl. Chem. 73 (2001) 1895.
lighted in the difference between the derivations of Eqs. [7] T.J. Davies, C.E. Banks, B. Nuthakki, J.F. Rusling, R.R. France,
(A.11) and (A.14). In the Cottrell equation (A.11) we use J.D. Wadhawan, R.G. Compton, Green Chem. 4 (2002) 570.
the boundary condition t > 0, ½Ax¼0 ¼ 0, whereas the [8] T. Ohkouchi, T. Kakutani, M. Senda, Bioelectrochem. Bioenerg.
25 (1991) 81.
Nernst boundary condition is employed for reversible
[9] Z. Samec, V. Marecek, J. Electroanal. Chem. 200 (1986) 17.
electron transfer reactions. Therefore, as well as having [10] For a recent review on SECM see: A.L. Barker, C.J. Slevin, P.R.
different equations that describe the voltammetric re- Unwin, J. Zhang, Surf. Sci. Ser. 95 (2001) 283.
sponse ((A.10) and (A.13)), we also have different [11] C. Wei, A.J. Bard, M.V. Mirkin, J. Phys. Chem. 99 (1995)
equations that describe the diffusion limited response 16033.
((A.11) and (A.14)). [12] C.E. Banks, O.V. Klymenko, R.G. Compton, Phys. Chem. Chem.
Phys. 5 (2003) 1652.
Fig. 27 demonstrates that a diffusion limited k 0 value [13] T.F. Connors, J.V. Arena, J.F. Rusling, J. Phys. Chem. 92 (1988)
corresponds to a diffusion limited khet value which is 2810.
within an order of magnitude larger. For example, in [14] J. Braun, M. Sambridge, Nature 376 (1995) 655.
Fig. 27(a) the k 0 ¼ 10 cm s1 response appears diffusion [15] A.J. Bard, L.R. Faulkner, Electrochemical Methods, second ed.,
Wiley, New York, 2001.
limited from the very start, whereas that for khet ¼ 10
[16] J.C. Eklund, A.M. Bond, J.A. Alden, R.G. Compton, Adv. Phys.
cm s1 does not. To get the same response from the ir- Org. Chem. 32 (1999) 1.
reversible kinetics, the value of khet needs to be increased [17] D. Lexa, J. Saveant, Acc. Chem. Res. 16 (1983) 235.
to 20 cm s1 . This method of comparison is for a system [18] Y. Shao, M.V. Mirkin, J.F. Rusling, J. Phys. Chem. B 101 (1997)
where mass transport is described by linear diffusion. In 3202.
[19] G.N. Schauzer, Angew. Chem., Int. Ed. Engl. 15 (1976) 417.
the reactive partially blocked electrode experiment de-
[20] R.L. McCreery, in A.J. Bard (Ed.), Electroanalytical Chemistry,
scribed within this paper, mass transport is described by vol. 17, Marcel Decker, New York, 1990, p. 221.
convergent diffusion. Upon increasing the mass-transfer [21] A.L. Barker, J.V. Macpherson, C.J. Slevin, P.R. Unwin, J. Phys.
coefficient, not only do the diffusion limited regimes Chem. B 102 (1998) 1586.
occur at higher values of k 0 and khet , but it would also be [22] A.L. Barker, P.R. Unwin, S. Amemiya, J. Zhou, A.J. Bard, J.
reasonable to expect an increase in the difference be- Phys. Chem. B 103 (1999) 7260.
[23] R.G. Harland, R.G. Compton, J. Colloid Interf. Sci. 131 (1989)
tween diffusion limited khet and diffusion limited k 0 val- 83.
ues. Indeed, in a theoretical treatment for an electron [24] J.F. Rusling, T.F. Connors, A. Owlia, Anal. Chem. 59 (1987)
transfer preceded by adsorption of the electroactive 2123.

S-ar putea să vă placă și