Journal of Electroanalytical Chemistry 530 (2002) 89
/95

www.elsevier.com/locate/jelechem

Extracting nucleation rates from current
time transients

Part II: comparing the computer-fit and pre-pulse method
/

M.Y. Abyaneh a,*, M. Fleischmann b

a

Transformation Studies Research Group, School of Natural and Environmental Sciences, Coventry University, Coventry CV4 7DY, UK
b
Bury Lodge, Duck Street, Tisbury SP3 6LJ, UK
Received 30 June 1999; received in revised form 11 December 2000; accepted 24 May 2002

Abstract
Nucleation rate constants, over a range of deposition potentials, are obtained for the electrocrystallisation of nickel onto a
vitreous carbon electrode using the pre-pulse method. This method is based on the analysis of the early stages of two transients, one
obtained as a result of the application of a single potential step, and the other resulting from the application of two potential steps
(the first as a pre-pulse potential of large amplitude and short duration followed by another at which the nucleation rate is to be
determined). A ?-values thus obtained are compared with those estimated from the fit of the transients (each recorded at a single
potential step at which A ? is to be determined) to the generalised equation derived for the model of growth of right-circular cones.
The values obtained by the two methods are shown to differ from one another by about a factor of 20. Though it has been possible
to obtain, with the application of two potential steps, transients which display, in the early stages of deposition, a rising current
proportional to the square of the deposition time, it is shown to be impossible to describe adequately the behaviour of these
transients in terms of instantaneous formation and subsequent growth of centres for the case of the deposition of nickel onto a
vitreous carbon electrode. In the light of the above findings the suitability of the pre-pulse method as a means of estimating A ? has
therefore been questioned in this example of electrocrystallisation. It has been shown that, even for electrocrystallisation processes
where the pre-pulse method is a valid method of estimating A ?, this method is, strictly speaking, applicable only to the deposition
processes obtained at the lower range of potentials (at which A ? is 20 times less than the product of the radial growth rate and the
square root of the density of nucleation sites). The inability of classical and atomistic theories of nucleation to arrive at plausible
values for critical nucleation size has once again been demonstrated. # 2002 Elsevier Science B.V. All rights reserved.
Keywords: Electrocrystallisation; Nucleation; Growth; Stochastic models

1. Introduction
It is now well established [1
/6] that, to obtain kinetic
information about the initial nucleation and growth of
metal deposits, experimental current
/time transients
(CTTs) can be analysed. The rate of outward growth
is readily available from the steady-state value of the
current. Difficulties arise when values for nucleation and
lateral growth rates, as quantities isolated from one
another, are to be obtained. To overcome these, two
electrochemical techniques are currently employed;
hereafter referred to as the pre-pulse and the computer-fit methods. In this paper the suitability of these two

* Corresponding author
E-mail address: m.abyaneh@coventry.ac.uk (M.Y. Abyaneh).

methods, as means of estimating nucleation rates for the

deposition of nickel onto a vitreous carbon electrode, is
explored.

1.1. The pre-pulse method

The pre-pulse method was the first electrochemical
technique employed in estimating nucleation rate constants1. It is based on the observation that a given
increase in the applied potential results in a much higher
1
We note that this method was introduced for two reasons: firstly,
to facilitate the determination of the rate constants of crystal growth
[19] at a defined number of nuclei (i.e. to make this determination
independent of the kinetics of nucleation); secondly to determine the
rate of such nucleation processes [20
/22].

0022-0728/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 0 2 2 - 0 7 2 8 ( 0 2 ) 0 0 9 7 3 - 7

M.Y. Abyaneh, M. Fleischmann / Journal of Electroanalytical Chemistry 530 (2002) 89
/95

90

increase in the rate of nucleation than the corresponding

increase expected for the rate of crystal growth. Hence,
it must be possible, at least in principle, to apply a
potential pulse, the amplitude of which is so large as to
impose the instantaneous conversion of all available
sites into nuclei, but, the duration of which is so short as
to prevent the impingement of growth units. If this pulse
is immediately followed by the application of yet
another potential, at which the nucleation rate is to be
determined, the resulting CTT would be expected to
display the characteristics predicted for the instantaneous formation and growth of nuclei. The equation
which describes these characteristics prior to the impingement of growth units on one another, is given by
j

gzFk?pM 2 k2 N0 2
t
r2

(1)

where j (A cm 2) is the current density, zF (C mol 1) is

the charge transferred per deposition of one mole, M (g
mol 1) and r (g cm3) are the mass and density of the
deposit, k and k ?, both in mol cm 2 s 1, are the lateral
and outward growth rates, N0 (cm 2) is the number of
preferred sites per unit area of the substrate electrode
and g is the numerical constant specific to the particular
growth form (i.e. for right-circular cones g /1 and for
hemispheres g/2). Thus, from the slope of j against t2,
plotted for the early stages of the rising transient, a value
for k2N0 can be obtained.
However, to obtain a value for the nucleation rate
constant, A ? (s1), another CTT is recorded, this time
without the application of the pre-pulse. In this instance,
the appropriate equation is given by [1]
j

gzFk?pM 2 k2 A
3r2

t3

(2)

where A (cm 2 s1) is the hypothetical nucleation rate,

i.e. that rate that would ensue if nucleation and growth
were not restricting the number of available sites. From
the slope of the plot of j versus t3, this time, k2A is
found. The rate constant, A?(A=N0 ); can be obtained
from the ratio of k2A to k2N0.
Central to the pre-pulse method is the presumption
that a limited number of preferred sites for nucleation
exists, and hence all these sites can be converted to
nuclei before the application of the second potential
provided that a sufficiently large-amplitude pre-pulse of
short-duration is applied. To establish the characteristics of the required pre-pulse, it is common practice to
increase the amplitude of a chosen pre-pulse in steps,
while keeping its duration constant, and to record the
transients obtained at the second potential2. This step2
An alternative procedure is to increase the duration of a chosen
pre-pulse of fixed amplitude [19].

wise increase in the amplitude results in the conversion

of progressively more sites into nuclei, which leads to an
increase in the rate of coverage. This is observed
experimentally as an increase in the slope of the rising
transient, as well as a progressive shift in the characteristics of the early stages of the rising current from the
j 8/t3, Eq. (2), to j 8/t2 behaviour, Eq. (1). These
observations are expected to continue, until a pre-pulse
of a given amplitude is reached beyond which no
alteration in the characteristics of the recorded CTT is
observed. In other words, at pre-pulses of amplitudes
greater than this, it becomes impossible to form any
more nuclei, i.e. all preferred sites are occupied at this
pre-pulse.
1.2. The computer-fit method
This second method is based on the nonlinear
regression fit [7] of a recorded transient (obtained at
the potential for which the nucleation rate is to be
determined) to a theoretically derived transient equation, the nucleation rate constant of which appears on
its own and as a quantity isolated from the lateral
growth rate, k . Thus far, only two equations have
been derived within which the nucleation and lateral
growth rate constants can be viewed in isolation from
one another. These are the generalised equations
describing (i) the growth of right-circular cones [8] and
(ii) the growth of hemispheroids [9]. In this paper, we
use the equation derived for the growth of right-circular
cones



pM 2 k2 A
j  zFk? 1exp 
A?r2


2t
2
2
2

t   2  2 exp(A?t)
(3)
A? A?
A?
It has been shown in the previous paper of this series
[10] that nucleation rates, obtained from the fit of this
equation to the recorded transients, differ from those
obtained using the equation derived for the growth of
hemispheroids by at most a factor of 3.

2. Experimental
The circular end section of vitreous carbon rods,
sealed into glass, were polished with successively finer
grades of alumina (Banner Scientific Ltd.) down to 0.05
mm using a microcloth. The electrodes were first rinsed
with high speed pouring tap water and cleaned with
triply distilled water.
To record transients, vitreous carbon electrodes were
initially maintained at 0.0 V (SCE) and then subjected to
a potential step of /0.4 V (using a Hi-Tek potentiostat
type DT2101). This sequence of potential steps ensured

M.Y. Abyaneh, M. Fleischmann / Journal of Electroanalytical Chemistry 530 (2002) 89
/95

the reduction of surface oxidised species before the

application of the appropriate working potentials.
Transients obtained without the application of a prepulse were recorded when, following this sequence of
potential steps, a further potential step to the appropriate working potential was applied. However, the prepulse transients were recorded when the sequence of
potential steps was followed by, first the application of a
pre-pulse potential of short duration (by a Hi-Tek
waveform generator (PPR1)) and then, by the appropriate working potential at which the nucleation rate
was to be determined. Nickel deposits were dissolved
anodically and the electrodes were polished with 1.0 and
0.05 mm alumina, rinsed and thoroughly cleaned before
the next experimental sequence.
Measurements were made in conventional threecompartment cells using a graphite counter electrode,
a vitreous carbon working electrode and a saturated
calomel reference electrode at 2959/1 K. Solutions used
for the deposition of nickel were Watts type baths
(containing 0.85 mol dm 3 NiSO4 /7H2O, 0.15 mol
dm 3 NiCl2 /6H2O and 0.58 mol dm 3 H3BO3). All
reagents used were of AnalaR grade and solutions were
prepared from triply distilled water. Glassware was
cleaned in chromic acid and repeatedly washed, finally
being left in triply distilled water.

3. Obtaining nucleation rate constants from the pre-pulse

method
As previously stated, central to the pre-pulse method
is the presumption that a limited number of preferred
sites for nucleation exist. The most striking feature of
the results was our inability to attain a pre-pulse
potential beyond which the slope of the rising transient
remained constant. The early stages of the transients
exhibited a gradual change from j 8/t3 to j 8/t2
behaviour with the step-wise increase in the amplitude
of the applied pre-pulse. But, rather than the rising slope
remaining constant at this stage, there was a further
increase in the slope, accompanied by a gradual shift to
a linear relationship between current and time. This
gradual shift towards linearity in the early stages of the
deposition indicates that, during the application of the
pre-pulse, a number of growth forms had already
impinged on one another. To avoid this, the amplitude
of the pre-pulse potential was reduced to a value where,
for the early stages of the transient, a j 8/t2 relationship
was again observed. Keeping the amplitude of this prepulse potential constant, further experiments were conducted in which the duration of each pulse was
progressively prolonged. Again the same pattern of shift
was observed. These observations show that, at least for
the deposition of nickel onto a vitreous carbon elec-

91

Fig. 1. Typical behaviour of CTTs recorded for electrocrystallisation

of nickel onto a vitreous carbon electrode.

trode, it is impossible to define an appropriate pre-pulse

for use in this technique3.
Despite this hindrance, attempts were made to obtain
values for nucleation rate constants over a range of
potentials. To facilitate this, a pre-pulse of given
amplitude and duration (/1.5 V for 20 ms) was selected
for which the recorded transients during the early stages
of growth exhibited a j 8/t2 relationship. Though this is
a somewhat unsatisfactory compromise, it seems a
convenient way of arriving at A ?-values.
Fig. 1 represents the typical appearance of the CTTs
recorded during the electrocrystallisation of nickel onto
a vitreous carbon electrode at the lowest attainable
deposition potentials. At these potentials, the rate of
growth is controlled by the rate of charge transfer at the
electrode j electrolyte interface. The transient within the
time region 0 5/t 5/t1 is attributed to the formation of a
monolayer of deposit [2,3], which is then followed by
nucleation and three-dimensional growth of crystals on
top of the underlying layer. It can be seen that Eqs. (1)
and (2) do not account for the time delay, t1, or the
initial current density, j0, arising mostly from the
evolution of hydrogen on the surface of the monolayer.
After modification, these equations give rise, respectively, to
j  j0 

gzFk?pM 2 k2 N0
r2

t2

(4)

and
j  j0 

gzFk?pM 2 k2 A
3r2

(tt1 )3

(5)

3
It is of some interest that the first example of electrocrystallisation
examined in this way [19] conformed exactly to the ideal pattern
required for the preformation of nuclei. The rates of electrocrystallisation at the second, lower, potentials reached a limit both
with the increase of amplitude of the preformation pulses at fixed
duration and with the extension of the duration at fixed amplitudes. It
is also of interest that electrocrystallisation can be examined both in
overpotential ranges where there is otherwise no nucleation as well as
in ranges where one observes progressive nucleation in the absence of
preformation of nuclei.

M.Y. Abyaneh, M. Fleischmann / Journal of Electroanalytical Chemistry 530 (2002) 89
/95

92

These modifications are seen to complicate the

procedure for determining the nucleation rate constant
A ?, in that it can no longer simply be found by the ratio
of the slope of j versus t3 to that of j versus t2. In order
to determine a value for A ?, it is necessary to write Eqs.
(4) and (5), respectively, in the forms of


gzFk?pM 2 k2 N0 1=2
(j j0 )1=2 
t
(6)
r2

Having arrived at this equation, it is now possible to

fit the data obtained in the region t1 5/t 5/t2, Fig. 1, to
this equation, thus enabling the estimation of A ? as well
as other parameters.

5. Comparison of nucleation rates derived from both

methods

and
(j j0 )1=3 



gzFk?pM 2 k2 A 1=3

3r2



gzFk?pM 2 k2 A 1=3
3r2

t1

(7)

Hence A?(A=N0 ) can be obtained from the ratio of

the cube of the slope of (j/j0)1/3 versus t to the square of
the slope of (j/j0)1/2 versus t. Using this method A ?values were obtained over a range of applied potentials.

4. Obtaining nucleation rate constants from the

computer-fit method
The first three-dimensional growth model (the equations for which have been used in the analysis of the
recorded transients) is the model of growth of rightcircular cones [11]. Eq. (3) describes the transient
behaviour of conic growth forms over all possible
nucleation rates. This equation holds true irrespective
of whether nuclei are to be formed at a selected number
of preferred sites, N0( /A /A ?), or whether no such
preferred sites exist. It can be seen from Eq. (3) that the
rate constant for nucleation, A ?, appears on its own,
isolated from that of lateral growth, k . This particular
attribute allows values of A ? to be estimated; the
possible errors arising from the computer fit of recorded
transients to Eq. (3) are fully discussed in the previous
paper [10].
However, the equation must be modified to account
for the time delay, t1, as well as for the current density,
j0, arising from the simultaneous hydrogen evolution
reaction on the existing monolayer. The modified
equation is given by [2,3]




2(t  P4 ) 2
j P1 P2 1exp P3 (tP4 )2 
 2
P5
P5

2
 2 exp(P5 (tP4 ))
(8)
P5
where P1 /j0, P2 /zFk ?, P3 
P5 /A ?.

M 2 k2 A
A?r2

To compare nucleation rate constants, two main

nucleation theories, namely the classical and the atomistic theory, can be employed. The classical theory for
heterogeneous nucleation of hemispheres relates the rate
constant for nucleation A ? to the overpotential h by the
equation
ln A? ln n

8pM 2 s3 1
3r2 z2 F 2 kT h2

(9)

where n (s 1) is a frequency factor, s (J cm 2) is the

interfacial energy, h (V) is the overpotential, k (J K 1)
is the Boltzmann constant and T (K) is the temperature.
Fig. 2 represents the plots of ln A ? against 1/h2. We have
calculated the overpotential on the basis that the rest
potential is /0.47 V; a value that is approximately the
same as the thermodynamic reversible potential. As can
be seen, the nucleation rates obtained from both
methods differ from one another by about a factor of
20, on average. The maximum error ascribable to the
use of Eq. (8) is 300% [10] and is far from being
sufficient to account for the observed difference of
2000%. Despite the difference in the observed nucleation
rates, the slopes of both plots are similar (/0.31 and /
0.50 V2). The interfacial energy for nickel, calculated
from Eq. (9), using the slopes of the plots in Fig. 2, is
found to lie between values of 541 and 635 mJ m 2. The
critical radius of the circular base of nuclei, rc, can then
be found from

; P4 /t1 and
Fig. 2. Plots of ln A ? vs. 1/h2, (***) data obtained from the computerfit method, (""") data obtained from the pre-pulse method.

M.Y. Abyaneh, M. Fleischmann / Journal of Electroanalytical Chemistry 530 (2002) 89
/95

Fig. 3. ln A ? vs. h , plotted for the same data as in Fig. 2.

rc 

2Ms
rzF h

(10)

is
An unacceptably small value for rc of about 1 to 2 A
obtained [7] despite a good estimate for the interfacial
energy. This is not, however, the first time that such a
small value for rc has been found. Other data (first
obtained by Fleischmann and Thirsk [12]) have indicated even smaller values. These results are clearly
inconsistent with the classical theory of nucleation
which predicts that a finite number of atoms are needed
before a nucleus can grow spontaneously. To account
for such observations, the atomistic model of nucleation
has been proposed [13,14].
According to the atomistic theory, the rate of nucleation, A ?, is related to the overpotential, h , by [14]
(i  1  b)zF h
ln A?ln n?
RT

(11)

where n ? (s1) is a frequency factor, i* is the critical

number of atoms within a nucleus, b is the transfer
coefficient and R (J mol 1 K 1) is the gas constant.
Fig. 3 shows the plots of ln A ? against h with slopes of
16.45 and 26.87 V 1. If one assumes b to be 1/2, the
minimum number of atoms necessary for the formation
of a nucleus is estimated to lie between 0.2 and 0.6
atoms, values that are clearly unacceptable. These
findings indicate that the atomistic theory is also
incapable of estimating feasible values for the critical
nucleation size; a new way of calculating the critical size
of a nucleus is introduced in part IV of this series [15].

6. Discussion
Some of the short-comings of both electrochemical
methods have already been alluded to. For example, the
exact morphology of the nuclei has not been ascertained
and it is therefore possible that errors of up to a factor
of three [10] are created with the use of Eq. (8). In the

93

case of the pre-pulse method (as applied to the electrocrystallisation of nickel onto a vitreous carbon electrode) we can even go so far as to challenge the
fundamental precepts on which this method is based.
One such precept is the presumption that the behaviour of the transients, following the application of a
pre-pulse, is solely due to the growth of centres already
formed at the pre-pulse stage. This implies that the
behaviour of the pre-pulse transients can be completely
accounted for in terms of the instantaneous nucleation
and growth of centres. It is noted that, without
exception, all of the transients obtained with the prepulse (/1.5 V amplitude and 20 ms duration) exhibited
a j 8/t2 behaviour in the early stages. However, we were
unable to obtain a reasonable fit of these transients to
any of the existing equations derived for the instantaneous nucleation and growth of a variety of forms
[9,11,16,17]. For example, for the growth of rightcircular cones (accounting also for the simultaneous
evolution of hydrogen) the equation for instantaneous
nucleation and growth is given by [3,11]
j  P1 P2 [1exp(P3 t2 )]

(12)

Fig. 4 is a typical example of the best fit of the prepulse transients to Eq. (12). A very poor fit is observed
with the estimated standard deviation of 10.11 mA
cm 2. Parameter P4 is now intentionally introduced
into Eq. (12) to demonstrate further the inappropriateness of the instantaneous nucleation model.
j  P1 P2 f1exp[P3 (tP4 )2 ]g

(13)

Should P4 assume values that greatly differ from zero,

as a result of the reiteration process, then instantaneous

Fig. 4. The rising portion of the pre-pulse transient, recorded for

electrocrystallisation of nickel onto a vitreous carbon electrode at a
potential of /0.75 V (SCE) following a pre-pulse of /1.5 V for the
duration of 20 ms. Solution composition: 0.85 mol dm 3 NiSO4 /
7H2O, 0.15 mol dm3 NiCl2 /6H2O and 0.58 mol dm 3 H3BO3.
( */ */) the best fit of Eq. (12) to the transient data (***). Estimated
parameters: P1 /31.839/7.71 mA cm2; P2 /517.749/28.78 mA cm 2;
P3 /2.67/10 59/3.4/10 6 s 2. Estimated standard deviation/
10.11 mA cm 2.

94

M.Y. Abyaneh, M. Fleischmann / Journal of Electroanalytical Chemistry 530 (2002) 89
/95

Fig. 5. The best fit ( */ */) of Eq. (13) to the same portion of the
recorded transient (***) as shown in Fig. 4. Estimated parameters:
P1 /49.409/6.40 mA cm 2; P2 /457.639/20.31 mA cm 2; P3 /
4.12/10 59/6.20 /10 6 s 2; P4 /24.809/6.68 s. Estimated standard deviation/6.72 mA cm 2.

nucleation would be recognised as an invalid description

of the transient behaviour.
Fig. 5 is the best fit of the same experimental transient
(shown in Fig. 3), this time to Eq. (13). It is noted that
an improved estimated standard deviation of 6.72 mA
cm 2 is observed only at the expense of P4 assuming a
value of about 25 s. This estimated value for P4 suggests
that nuclei are formed instantaneously 25 s after the
application of the pre-pulse, an estimate which is clearly
absurd. These findings warrant further investigation as
to the cause of these discrepancies and will be explored
in the next paper of this series [18].
Amongst the obvious general short-comings of the
pre-pulse method (such as the difficulty of achieving the
same polish for the surface of the vitreous carbon before
the commencement of the two required experiments, one
with and one without the application of a pre-pulse, as
well as experimental flaws so far identified in this paper)
we are faced with another disadvantage. Nucleation
rates can only be obtained at potentials for which the
early portion of the rising transient (without a pre-pulse)
displays a j 8/t3 correlation, that is, when [9,10]
p
Mk A=A?
(14)
A?5
20r
Therefore, strictly speaking, for those values of A ? in
the range
p
p
Mk A=A?
60Mk A=A?
5 A?5
(15)
20r
r
the computer-fit method remains the only electrochemical method of measuring A ?. For values greater than
p
60Mk A=A?=r nucleation is considered to be instantaneous [9,10].
Our inability to arrive at a limiting slope for the rising
pre-pulse transients may indicate that a vitreous carbon
substrate does not present a finite number of preferred

nucleation sites and consequently nuclei can form at any

position on the substrate. The maximum possible
number of nuclei (per unit area of the substrate) which
can form at each potential is therefore given by 1/ar2c [8],
where a is a packing factor and rc is the critical radius of
the base of the cones at that particular potential. As the
critical size is expected to decrease with an increase in
the applied potential, the maximum possible number of
nuclei which can form at the pre-pulse increases, leading
to an increase in the slope of the pre-pulse transients.
Consequently, it seems appropriate to assume that
some of the nuclei during the pre-pulse are below the
critical size necessary for growth at the second potential.
Hence, while those nuclei exceeding this critical size
grow, nuclei which did not achieve this size, during the
application of the pre-pulse, would have either dissolved
or become inactive. In the next paper of this series [18],
transient equations based on the above assumptions are
derived. The characteristics of these derived equations
are then compared with the behaviour of the recorded
pre-pulse transients.

7. Conclusions
1) The pre-pulse transients (recorded during the electrocrystallisation of nickel onto a vitreous carbon
electrode), the early stages of which displayed a j 8/
t2 correlation, are found to manifest further increases in their maximum slope with corresponding
increases in the pre-pulse amplitude.
2) Equations derived for the instantaneous nucleation
and growth of centres do not fit the pre-pulse
transients.
3) Nucleation rates obtained from the pre-pulse
method are 20 times greater than the corresponding
values obtained with the computer-fit
pmethod.
4) Strictly speaking, when Mk A=A?=20r 5A?5
p
60Mk A=A?=r; the computer-fit method remains
the only electrochemical technique for estimating
A ?.
5) Both the classical and the atomistic theories fail to
estimate reasonable values for the critical nucleation
size.

8. List of symbols
A
A?
F
i*
j
j0

a hypothetical nucleation rate (cm 2 s 1)

nucleation rate constant (s1)
the Faraday (C mol 1)
the critical number of atoms in a nucleus
current density (A cm 2)
the current density before the commencement of 3D
growth (A cm2)

M.Y. Abyaneh, M. Fleischmann / Journal of Electroanalytical Chemistry 530 (2002) 89
/95

rate of lateral growth (mol cm2 s 1); Boltzmanns

constant (J K 1)
k ? rate of outward growth (mol cm 2 s 1)
M molar mass (g mol 1)
N0 density of preferred sites (cm 2)
P1 a parameter representing j0 (A cm2)
P2 a parameter representing zFk ? (A cm 2)

P3 a parameter representing

pM 2 k2 A 1
(s )
A?r2

P4
P5
rc
R
t1
T
a
b
r
g

a parameter representing the time delay t1 (s)

a parameter representing A ? (s1)
critical radius of the base of nuclei (cm)
gas constant (J mol1 K 1)
a time delay (s)
temperature (K)
a packing factor
the transfer coefficient
density of the deposit (g cm 3)
a numerical constant specific to a particular growth
form
n a frequency factor (s1)
n ? another frequency factor (s1)
s interfacial energy (J cm 2)
h the overpotential (V)

References
[1] M. Fleischmann, H.R. Thirsk, in: L.P. Delahay, C.W. Tobias
(Eds.), Advances in Electrochemistry and Electrochemical Engineering, vol. 3, Interscience, 1963.

95

[2] M.Y. Abyaneh, M. Fleischmann, Trans. Inst. Met. Fin. 58 (1980)

91.
[3] M.Y. Abyaneh, M. Fleischmann, J. Electroanal. Chem. 119
(1981) 197.
[4] M.Y. Abyaneh, M. Berkem, M. Fleischmann, Trans. Inst. Met.
Fin. 60 (1982) 114.
[5] M.Y. Abyaneh, J. Hendrikx, W. Visscher, E. Barendrecht, J.
Electrochem. Soc. 129 (1982) 2654.
[6] M.Y. Abyaneh, A. Tajali Pour, Trans. IMF 72 (1994) 19.
[7] M.Y. Abyaneh, Ph.D. thesis, Southampton University, 1980.
[8] M.Y. Abyaneh, M. Fleischmann, Electrochim. Acta 27 (1982)
1513.
[9] M.Y. Abyaneh, J. Electroanal. Chem. 387 (1995) 29.
[10] M.Y. Abyaneh, J. Electroanal. Chem. 530 (2002) 82.
[11] R.D. Armstrong, M. Fleischmann, H.R. Thirsk, J. Electroanal.
Chem. 11 (1966) 208.
[12] M. Fleischmann, H.R. Thirsk, Trans. Faraday Soc. 51 (1955) 71.
[13] R. Kaischev, Annuaire Univ, Sofia, Faculte Physicomathematique, Livre 2 (Chimie), 42 (1945/46) 109.
[14] A. Milchev, S. Stoyanov, R. Kaischev, Thin Solid Films 22 (1974)
255.
[15] M.Y. Abyaneh, Part (IV) of this series, submitted for publication.
[16] M.Y. Abyaneh, Electrochimim. Acta 27 (1982) 1329.
[17] M.Y. Abyaneh, J. Electroanal. Chem. 209 (1986) 1.
[18] M.Y. Abyaneh, J. Electroanal. Chem. 530 (2002) 96.
[19] M. Fleischmann, M. Liler, Trans. Faraday Soc. 54 (1958) 1370.
[20] (a) R. Kaishev, B. Mutaftschiev, Bull. Inst. Phys. Chem. (Sofia) 1
(1950) 137;
(b) R. Kaishev, B. Mutaftschiev, Bull. Inst. Phys. Chem. (Sofia) 4
(1954) 105;
(c) R. Kaishev, B. Mutaftschiev, Bull. Inst. Phys. Chem. (Sofia) 5
(1955) 77.
[21] (a) A. Sheludko, G. Bliznakev, Bull. Inst. Phys. Chem. (Sofia) 2
(1951) 227;
(b) A. Sheludko, G. Bliznakev, Bull. Inst. Phys. Chem. (Sofia) 3
(1952) 61.
[22] G. Tammann, Kristalliseren und Schmelzen, Leipzig, 1903; Die
Aggregatzustande, Leipzig, 1922.