Influence of fabrication procedure on the electrochemical performance of Ag/AgCl reference electrodes

Article abstract of DOI:10.1016/j.electacta.2011.08.089

The influence of several parameters in the preparation procedure of thermal-electrolytic Ag/AgCl electrodes on the resulting electrode performance has been studied. In particular, we report the effect on electrode performance of subtle variations in the preparation of silver oxide paste used for electrode manufacture, in thermal annealing conditions employed and in the procedure for electrochemically converting a fraction of the electrode from silver to silver chloride. Scanning electron microscopy and electrochemical impedance spectroscopy have been used to study the characteristics of the electrodes produced. This work reveals a correlation between the electrochemical behaviour and surface physical characteristics - in particular electrode porosity. The outputs of this study have positive implications for improving the accuracy and comparability of primary pH measurement.

Full text of DOI:10.1016/j.electacta.2011.08.089

Electrochimica Acta 56 (2011) 10009–10015
Contents lists available at SciVerse ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Influence of fabrication procedure on the electrochemical performance of
Ag/AgCl reference electrodes
Daniela Stoicaa, Paul J. Brewerb,∗, Richard J.C. Brownb, Paola Fisicaroa
a Department of Biomedical and Inorganic Chemistry, Laboratoire National de Métrologie et d’Essais, 1 Rue Gaston Boissier, 75015 Paris, France
b Analytical Science Division, National Physical Laboratory, Teddington, Middlesex TW11 0LW, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 5 August 2011
Received in revised form 24 August 2011
Accepted 24 August 2011
Available online 31 August 2011
The influence of several parameters in the preparation procedure of thermal–electrolytic Ag/AgCl elec-
trodes on the resulting electrode performance has been studied. In particular, we report the effect on
electrode performance of subtle variations in the preparation of silver oxide paste used for electrode
manufacture, in thermal annealing conditions employed and in the procedure for electrochemically
converting a fraction of the electrode from silver to silver chloride. Scanning electron microscopy and
electrochemical impedance spectroscopy have been used to study the characteristics of the electrodes
produced. This work reveals a correlation between the electrochemical behaviour and surface physical
characteristics – in particular electrode porosity. The outputs of this study have positive implications for
improving the accuracy and comparability of primary pH measurement.
Keywords:
Ag/AgCl reference electrode
Primary pH determination
Uncertainty
© 2011 Elsevier Ltd. All rights reserved.
constant, ꢀC0l is the activity coefficient of chloride ions and mCl is
1. Introduction
the molality of chloride added to the solvent.
The procedure for primary pH measurement is based on a poten-
tiometric cell, commonly referred to as a Harned cell. The cell is
designed in order to avoid any liquid junction. This is the only
system accepted by International Union of Pure and Applied Chem-
istry (IUPAC) and the Bureau International des Poids et Measures
(BIPM) for primary pH measurements. The primary potentiometric
pH determination is based on the measurement of the difference
in potential between a hydrogen-ion-sensing platinum-hydrogen
electrode and a Ag/AgCl reference electrode in a buffer solution
under test. Chloride ions at known concentrations are added to
this buffer solution to enable the potential difference at zero chlo-
ride concentration to be obtained by extrapolation. The relationship
used for pH calculation is given in Eq. (1).
Ag/AgCl reference electrodes are frequently used in electro-
chemistry as they possess numerous beneficial characteristics such
as electrochemical reversibility, high exchange current, low polar-
ization (when passing small currents) and highly reproducible
potential. The final characteristic is the main reason [1] why
the liquid-junction free Ag/AgCl electrodes are used as reference
electrodes for primary pH measurements. These electrodes are gen-
erally only produced by National Metrology Institutes (NMIs) as
they require meticulous fabrication procedures and specific exper-
tise. The current market is too small for commercial production.
For primary pH measurements the typical expanded uncertainty
is usually no more than 0.005 pH [2]. Generally, the contribution
of the standard potential (E0) of the Ag/AgCl electrode is about
half of the final estimated uncertainty [3,4], and electrodes agree-
ing within 60–100 ␮V should be used in order to meet the target
uncertainty. The standard potential is an intrinsic characteristic of a
reference electrode. According to the IUPAC recommendations, the
determination of the standard potential should be made in an aque-
ous 0.01 molal HCl solution [5]. The 1978 IUPAC report invokes a
maximum inaccuracy on E0 values of about 0.2 mV using the generic
preparation methods suggested.
ꢀꢁ
ꢂ
Emeasured − E0
RT/F ln(10)
pH = −log(asH) = lim
m
→0
Cl
ꢃ
+ log(mCl
)
+ log(ꢀC0l
)
(1)
where asH is the activity of hydrogen ions in the solvent, Emeasured
is the measured potential (corrected to 1 atm partial pressure of
hydrogen), E0 is the standard potential of reference electrode, R
is the molar gas constant, T is the temperature, F is the Faraday
Ag/AgCl electrodes may be grouped into three types, according
to the preparation method [6]. Provided that a suitably optimised
protocol is followed, all electrode types exhibit good precision,
accuracy and reliability, with bias potentials within the confidence
interval required for primary standard measurements. It is essential
that the same electrode type is used both in the determination of
∗
Corresponding author. Tel.: +44 20 8943 6007.
E-mail address: paul.brewer@npl.co.uk (P.J. Brewer).
0013-4686/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2011.08.089

10010
D. Stoica et al. / Electrochimica Acta 56 (2011) 10009–10015
E0 and in the parallel emf measurement of the IUPAC-prescribed
Harned cell [5]. This study investigates the thermal–electrolytic
type as it has been used more extensively than any other [7–9].
Brewer and Brown [10] have shown that microstructure plays an
important role in electrode performance. The porosity of the elec-
trode is one of the most important aspects of its microstructure. A
porous electrode presents a high efficiency surface enabling high
exchange current densities at equilibrium, thereby resulting in a
highly reproducible reference potential. However, a high degree of
porosity also increases the probability of creating a mixed potential
as deeper solution penetration may allow contact of the electrolyte
with the Pt wire skeleton supporting the Ag/AgCl. Moreover, impu-
rities may also be attracted to this Pt/Ag interface during annealing
as a result of it being a high free energy surface [11]. This may have
a significant and variable impact on the potential exhibited by the
electrodes and should be avoided if possible. Therefore to achieve
highly repeatable Ag/AgCl electrodes it is imperative to control
their degree of porosity. A number of parameters in the prepara-
tion of Ag/AgCl electrodes are believed to influence the porosity [1].
These parameters include the purity of materials used for the Ag2O
paste preparation (presence of impurities such as bromide have a
detrimental effect), the efficiency of the washing process carried
out to remove ionic impurities from the Ag2O paste, the storage
solutions used and the conditions employed for converting Ag2O
to Ag.
2.1.2. Thermal annealing
Successive layers of Ag2O were applied in order to obtain a
sphere of material on the platinum wire (99.999%, Goodfellow, UK).
This sphere was then annealed in order to obtain Ag. Complete ther-
mal decomposition of the Ag2O to Ag and O2 occurs at 400 ◦C [13].
Both procedures used temperatures in excess of this value. For L2,
each electrode was annealed for 15 min at 450 ◦C. For N2, electrodes
were heated for 30 min at 100 ◦C (in order to first remove water
from the Ag2O paste) and then at 500 ◦C for 2 h.
2.1.3. Anodisation
The anodisation process used a three-electrode cell in which
the potential of the working electrode (WE–the Ag/AgCl electrode
under test) was controlled relative to a pseudo-reference electrode
(RE – platinum flag). During the electrolysis, the current through the
cell was kept constant using a Potentiostat (Keithley Model 228A
for L2 and EGG Princeton Applied Research Model 263A for N2). A
current was passed between the WE and the counter electrode (CE –
also a platinum flag). Approximately 15% of the Ag was converted to
AgCl using L2 and N2 respectively. In the process, Ag undergoes an
anodic oxidation to AgCl, at the surface exposed to the electrolyte,
and forms an AgCl layer. The process took place in aqueous HCl
solution (1 M for L2 and 0.1 M for N2). For L2, the current was fixed
at 10 mA, the anodisation time being calculated for each electrode
individually, according to the total mass of Ag in each electrode, by
using Faraday’s law (constant current electrolysis):
It therefore seems plausible that the electrode microstructure
and hence electrode performance can be controlled by manip-
ulating certain parameters in the preparation procedure. We
demonstrate for the first time, the influence of several key param-
eters in the electrode preparation procedure (Ag2O paste, thermal
decomposition conditions and anodisation) on the porosity charac-
teristics of the electrodes produced. This work is a significant step
towards developing an optimum methodology for making repro-
ducible Ag/AgCl electrodes.
Q = I · t = n · F
(2)
where Q is the total electric charge passed through the electrolyte,
I is the current applied, t is the duration of the electrochemical
process, n is the amount of Ag converted to AgCl and F is the Faraday
constant. For N2, Eq. (2) was used to calculate the applied current
required for a fixed anodisation time of one hour.
2.2. Electrochemical impedance spectroscopy
2. Experimental
Electrochemical impedance spectroscopy (EIS) measurements
used the same three electrode configuration. These measurements
were carried out at open circuit potential at different frequencies
using a potentiostat with an integrated impedance analyser (N-stat,
Ivium Technologies). The measurements were carried out in 0.01 M
HCl solution under potentiostatic control, with a signal amplitude
of 10 mV. The frequency was varied from 100 kHz to 0.1 Hz with 10
points per decade.
Two electrode preparation methods, in use at the Labora-
toire National de Métrologie et d’Essais (LNE) and the National
Physical Laboratory (NPL) have been studied. The preparation
of thermal–electrolytic electrodes involves three main stages,
detailed in Fig. 1: (i) synthesis of Ag2O paste, (ii) thermal decom-
position of Ag2O to form metallic Ag (thermal annealing) and (iii)
conversion of Ag to AgCl by anodisation.
To assist later discussion of the data, L1 and N1 are used to rep-
resent protocols employed for the production of Ag2O paste at LNE
and NPL respectively. L2 and N2 represent protocols employed for
thermal annealing and anodisation at LNE and NPL, respectively.
2.3. Scanning electron microscopy
Scanning electron microscopy (SEM) measurements were made
with a field-emission scanning electron microscope (FE-SEM), Carl-
Zeiss-Supra 40. An accelerating voltage of 10 kV was used with a
working distance of 3.7 mm and a final lens aperture of size 30 mm.
Secondary electron images were acquired using an in-lens detector.
2.1. Electrode fabrication
2.1.1. Preparation of Ag2O paste
AgNO3 and NaOH (Sigma Aldrich, 99.9999% and 99.998%,
respectively) were used for the synthesis of Ag2O. In order to
obtain homogeneous Ag2O crystals in the desired size range
(0.1–15 microns) [12] a solution of NaOH with a concentration
of 5 M for L1, and 1 M for N1, was added dropwise to a vigor-
ously stirred AgNO3 solution. Since nitrate anions are more easily
removed from the Ag2O precipitate by a washing process than
hydroxide ions, nitrate anions were kept in excess during the reac-
tion. Ag2O paste produced using method L1 was aged for 1 month
during which time it dried out and became substantially more vis-
cous, whereas paste produced using N1 was used immediately after
preparation.
3. Results and discussion
3.1. Potential measurements during the anodisation process
The evolution of the potential during electrode anodisation pro-
vides a useful tool to interpret the physico-chemical processes
occurring. Typically the final anodisation voltage is greater than
the initial value. This is due to the formation of AgCl, which is a
substantially non-conductive substance, resulting in an increase in
resistance. The effective conductivity of such a layer depends on
the electrode porosity, as this influences the available surface area
and the degree to which the electrolyte is able to penetrate through

D. Stoica et al. / Electrochimica Acta 56 (2011) 10009–10015
10011
Fig. 1. Preparation steps for thermal–electrolytic Ag/AgCl electrodes. The steps involved in L1 and L2 can been seen on the left-hand and those involved in N1 and N2 on the
right-hand side. L2 involves 7 successive layers of Ag2O whereas N2 only 3 layers.
the non-conductive silver chloride layer and maintain contact with
the conductive silver underneath the surface.
of the electrode surface with a layer of AgCl requiring a much higher
voltage to continue passing a constant current.
Fig. 2 shows the potential during the anodisation process for
several Ag/AgCl electrodes prepared by the two institutes. The
final voltage of Ag/AgCl electrodes prepared at LNE (Fig. 2a) is
smaller than those prepared at NPL (Fig. 2b). Moreover, the vari-
ation in anodisation voltage with time of electrodes prepared at
each institute is vastly different, suggesting a contrast in electrode
porosity and AgCl distribution. The slightly upwardly curving pro-
file in Fig. 2a suggests that the AgCl formed has not fully covered
the available surface area by the end of the anodisation procedure.
This indicates the presence of a larger surface area for the LNE elec-
trodes compared to those prepared by NPL (Fig. 2b) and hence a
more porous structure. The S-shape profile in Fig. 2b is characteris-
tic of a less porous electrode with a lower available surface area. The
sudden rise in potential being attributed to the complete coverage
These findings are important as P. Brewer et al. [14] previously
showed that the porosity of Ag/AgCl electrodes is a parameter
that requires careful control to ensure good electrode compara-
bility. Since the anodisation is the last step in the preparation, it
is not clear which process (Ag2O paste preparation, thermal con-
version or anodisation) has the largest influence on the electrode
microstructure. Therefore we have investigated the influence of all
these parameters on the electrode porosity.
The influence of protocols L2 and N2 on the transient anodi-
sation potential of electrodes prepared from Ag2O paste from
methods L1 and N1 is shown in Fig. 3a and b respectively. In Fig. 3a,
different signatures are observed when the thermal annealing and
Fig. 3. Variation of the potential vs. Pt (pseudo-ref) measured during the anodisation
process. (a) Shows electrodes prepared from Ag2O paste using L1 with protocols L2
(filled triangles) and N2 at different viscosities (open triangles and filled squares). (b)
Shows the same with Ag2O paste prepared using N1 with protocols L2 (open circles)
and N2 with different Ag2O paste aging times (open squares and filled circles).
Fig. 2. Evolution of the potential vs. Pt (pseudo-ref) during the anodisation process
for several Ag/AgCl electrodes using (a) L1–L2 and (b) N1–N2.

10012
D. Stoica et al. / Electrochimica Acta 56 (2011) 10009–10015
Table 1
anodisation protocol (L2 and N2) is changed. Protocol L2 results
in a featureless transient, whereas N2 produces the ‘S’ shape pro-
file obtained in Fig. 2b. This suggests that changes in the thermal
conversion and anodisation processes alter the porosity of the final
structure of electrodes prepared from Ag2O paste using method
L1. In Fig. 3b, the ‘S’ shape profile is obtained for each of different
protocols (L2 and N2) employed. This suggests that Ag2O paste pre-
pared using method N1 will result in a structure with low porosity
regardless of the thermal conversion and anodisation procedures
employed.
Parameters determined by using complex impedance spectroscopy.
Electrode type
Porosity factora
Relaxation frequency (Hz)
˛
Ds
L1–L2
L1–N2
N1–L2
N1–N2
14
3
1.2
1
501.2
501.2
316.2
158.5
0.24
0.41
0.34
0.60
5.2
3.4
3.9
2.7
The porosity factor = Rp˝2−P2/R˝
, where RX˝ is the ohmic resistance of elec-
a
electrode
trode X, ˛ is related to constant phase element behaviour (a value of 1 indicating a
perfect capacitor), and Ds is the fractal nature of the electrode.
Fig. 3a shows two profiles from electrodes prepared with Ag2O
paste at different viscosities (open triangles–less viscous and filled
squares-more viscous). The data suggests that a more viscous Ag2O
paste produces an electrode with a more porous structure. Fig. 3b
shows two profiles from electrodes prepared from Ag2O with dif-
ferent aging times (filled circles and open squares) However the
similar profiles suggests no influence of the Ag2O paste age on the
porosity.
Mendizza et al. [15] suggested that the conductivity of the
deposited AgCl layer increases with increasing acidity in the solu-
tion. Based on the previous conclusions, the increased conductivity
of L2 electrodes cannot be exclusively explained by the acidity of
HCl solution. As Fig. 3 shows, the concentration of the HCl solution
acts mainly to influence the kinetics of AgCl formation: faster AgCl
formation occurs with more concentrated HCl solutions.
Hence we may conclude that the water content of the Ag2O paste
and its subsequent removal during annealing has an affect on the
porosity of the final electrode. It seems that the lower the water
content in the initial paste, the more porous the resulting electrode.
This may be a result of a porous structure already beginning to
form as the paste dries. The hydrated paste will not allow a pore
structure to form prior to annealing. Furthermore the more extreme
the annealing conditions the more rapid the removal of the water
during annealing and hence the more extensive the pore structure
created by this process: leading to more porous electrodes. These
conclusions support the SEM data.
3.2. Electrochemical impedance spectroscopy
Fig. 4 shows SEM measurements of the surface morphol-
ogy of some representative electrodes. The images confirm the
previous observations. Fig. 4a shows a porous structure for elec-
trodes prepared by method L1–L2. In contrast, a dense structure
is shown in Fig. 4c for electrodes prepared by method N1–N2.
A structure with intermediate porosity is shown for an elec-
trode prepared with the Ag2O paste protocol from one institute
(L1) and the thermal conversion and anodisation protocols from
another (N2).
Therefore we may hypothesise that the electrodes obtained
using L1 and L2 are characterized by a porous but inhomoge-
neous AgCl layer, with several regions having different AgCl content
(Fig. 4a). This is likely to result in poor electrochemical perfor-
mance. Despite some controversy over the mechanism by which
the AgCl film nucleates and grows on an anodized Ag surface,
there are some sources [16] that state that the concentration of
the HCl electrolyte may affect the loss of AgCl from the deposited
layer during the anodisation process. According to the well-known
common-ion effect, the solubility of AgCl will decrease with an
increase of chloride concentration in the electrolyte. Thus, in order
to limit the loss of AgCl from the deposited layer into the solution
and also to prevent the formation of AgCl2−, an HCl solution at or
below 1 M should be used [15]. Since this condition is met by the
protocols studied, the poor homogeneity of the AgCl layer in Fig. 4a
could be related to the nature of the anodisation process: i.e. using
a fixed anodisation time and a less concentrated electrolyte (N2)
may produce a more homogeneous layer of AgCl (Fig. 4b and c).
However, further work is required to investigate this.
Further SEM was carried out to study the effect of different
thermal annealing protocols (L2 and N2) on the microstructure of
electrodes fabricated from Ag2O paste prepared using L1.
Fig. 5a shows the image of an electrode prepared with annealing
protocol L2. The microstructure is considerably less homogeneous
than when annealing protocol N2 is used (Fig. 5b). In this case the
pores have a similar size of around 10 ␮m. The difference between
these structures can partially be explained with the fact that the
intermediate step in N2, which consists of annealing each electrode
for 30 min at 100 ◦C, allows for the removal of water molecules from
the Ag2O paste before increasing the temperature to 500 ◦C. A quick
and violent loss of water by annealing in one step may create larger
and more variable pores (as is the case with L2).
EIS measurements were performed to validate our earlier con-
clusions and further investigate the influence of the Ag2O paste,
thermal annealing conditions and anodisation protocol on the
microstructure of the resulting Ag/AgCl reference electrodes. Vari-
ables such as the distance between the CE and RE to the electrode
under test have been kept constant in order to minimise any arte-
facts in the data and allow for a quantitative interpretation of the
spectra.
The complex plane representations (Nyquist diagrams), display-
ing – ZIm vs. ZRe for different values of frequency, are shown in Fig. 6,
where ZIm is the imaginary component and ZRe is the real compo-
nent of the complex impedance. The spectra have been normalised
to the geometric area of the electrode. Three distinct regions can
be identified. Firstly the partial contribution of electrolyte and cell
configuration at high frequencies (the frequency range is too lim-
ited to reveal the entire arc). Secondly a semicircle at intermediate
frequencies when the system is under kinetic control, as a result of
an electrochemical charge transfer step at the electrode–electrolyte
interface. Lastly a straight, or slightly curved, line at low frequencies
which is characteristic of diffusion processes.
Since all the electrodes have been analyzed in the same HCl
solution, the shift of their respective ohmic resistance is due to the
difference in the active surface (real area), and thus is related to the
porosity. As the SEM images have shown (Fig. 4), the least porous
electrode structure is produced from N1–N2 and is shown in Fig. 6d.
The absence of a significant diffusion process is consistent with its
dense structure. The ratio between its ohmic resistance and those
of each of the other electrodes enables a determination of the rela-
tive porosity factor, indicating how many times the electrodes are
more porous than those produced using N1–N2. The porosity factor
calculated for electrodes in Fig. 6 is given in Table 1.
Fig. 7 shows the imaginary part of complex impedance versus
frequency, corresponding to the data displayed in Fig. 6. The data
shows an increase in magnitude of ZIm with a decrease in porosity
of the electrode. The frequency at which ZIm exhibits a maximum
value corresponds to the relaxation frequency related to the relax-
ation time of mobile charges. The shift of the relaxation frequency
towards higher values, observed for electrodes made from Ag2O
paste with L1 (Fig. 7a), is a consequence of the decrease in the relax-
ation time and indicates a faster movement of the charges, coherent
with a more porous structure associated with these electrodes.

D. Stoica et al. / Electrochimica Acta 56 (2011) 10009–10015
10013
Fig. 4. SEM images of Ag/AgCl electrodes after anodisation: (a) L1–L2, (b) L1–N2, (c) N1–N2. In (a), region 2 has twice as much AgCl content as region 1.
Fig. 5. SEM images of electrodes made from Ag2O prepared using L1. Images (a) and (b) were obtained before anodisation and correspond to annealing procedures L2 and
N2, respectively.
Due to their porosity, the systems cannot be modelled using the
idealized circuit elements. Thus, the use of distributed impedance
elements such as a constant phase element (CPE) are required [17].
The empirical CPE impedance function, ZCPE, is expressed according
to [18]:
where Z0 is a constant, j is the imaginary number and ␻ is the angu-
lar frequency (ω = 2␲f and f is the frequency). The exponent ˛ is
a dimensionless parameter, describing the CPE behaviour. For an
ideal capacitor with a smooth surface, ˛ is equal to 1. This param-
eter also expresses the deviation from an ideal system and can be
quantified by plotting the imaginary part of the impedance, ZIm, as
a function of frequency using a logarithmic scale (Fig. 8) [19].
ZCPE = Z0(jω)−˛
(3)
Fig. 6. Complex plane (Nyquist) representation of the impedance obtained at open circuit potential. (a), (b), (c) and (d) Represent electrodes prepared using L1–L2, L1–N2,
N1–L2 and N1–N2, respectively, between 100 kHz and 0.1 Hz.

10014
D. Stoica et al. / Electrochimica Acta 56 (2011) 10009–10015
Fig. 9. Voltage stability at equilibrium of investigated Ag/AgCl electrodes. Open
triangles, open squares, filled triangles and filled squares represent electrodes pre-
pared using L1–L2, L1–N2, N1–L2 and N1–N2, respectively. The measurements were
carried out in 0.01 molal HCl solution, under a nitrogen atmosphere, at room tem-
perature. A thermal electrolytic Ag/AgCl electrode was used as a defacto reference.
Each point represents a mean value of the potential over a 3 h measurement period.
Fig. 7. Imaginary part of impedance as a function of frequency. (a) shows elec-
trodes prepared from Ag2O paste using L1 with open triangles and open squares
representing protocols L2 and N2, respectively. (b) Shows electrodes prepared from
Ag2O paste using N1 with filled triangles and filled squares representing protocols
L2 and N2, respectively.
impedance. It has been shown that the larger the deviation from an
ideal capacitor, the more porous the electrode. Equally the larger
the real component of impedance at a given frequency, the less
porous the electrode. The results have confirmed previous obser-
vations made in Section 3.1 that the wetter the initial Ag2O paste,
and the less violent the removal of the water during annealing, the
denser and less porous the electrodes produced.
As the imaginary part of impedance is independent of the elec-
trolyte resistance, the slope is constant in the whole frequency
range. ˛ will then be equal to the negative slope. The values of ˛,
determined in the frequency domain from 104.5 to 103.3, are given
in Table 1.
The exponent ˛ may also be related to the fractal dimension of
the surface, Ds, by [20]:
1
˛ =
(4)
3.3. Relative voltage measurements
Ds − 1
The value of the standard potential (E0) depends, among other
things, on the quality of the Ag/AgCl interface with the electrolyte.
Experience shows that the main contribution, about 50%, of the
overall pH uncertainty is contributed by the E0. In order to reach
an overall expanded uncertainty for primary pH measurements of
less than 0.005 pH, it is necessary to reduce the contribution from
E0. The use of electrodes with similar morphology should there-
fore minimise the dispersion of the E0 values, improving the final
accuracy and minimising uncertainty.
The performance of the Ag/AgCl reference electrodes fabricated
with the different Ag2O paste, annealing and anodizing protocols
has also been studied by evaluating the stability of their potential at
equilibrium in 0.01 M HCl solution with respect to a defacto Ag/AgCl
reference. This method is commonly used for electrode selection
prior to Harned cell measurements of pH. Selection is based on the
dispersion of the individual voltage compared to the average value
of the group. The selection criteria is specific to each NMI, however,
in order to meet uncertainty requirements, the usual acceptable
range for measured electrode potentials is about <100 ␮V [21]. In
this analysis we will consider the mean potential of the group of
electrodes rather than just looking at the difference between the
highest and lowest result. The results of the potential measure-
ments are presented in Fig. 9, each point representing the mean
value over a 3 h measurement period.
The smaller ˛, the higher Ds, corresponding to an electrode
with a greater fractal character. The porosity as well as the micro-
structural defects observed on the surface of the dense electrodes
could be considered as a form of fractal geometry. A slightly higher
Ds value but a smaller porosity factor obtained for N1–L2 com-
pared to L1–N2 electrode could be explained by the presence of
macroscopic cracks on the electrode surface of the latter.
Hence we may conclude that impedance spectroscopy is a very
powerful method by which to characterise the Ag/AgCl electrodes.
The most useful parameters to measure are: the exponent defining
how far the capacitive element in the equivalent circuit devi-
ates from that of an ideal capacitor, and the real component of
The majority of electrodes satisfy the criterion employed for
pH measurements (most deviate <100 ␮V from the reference elec-
trode). As this approach uses another Ag/AgCl electrode from the
same batch as a defacto reference, it allows for a comparison
between the electrodes.
The mean potential difference of L1–L2 and L1–N2 electrodes to
the reference is vastly different which further confirms the sensitiv-
ity of Ag2O paste prepared using L1 to the annealing and anodising
protocols employed. This corroborates the findings in Fig. 3a. Elec-
Fig. 8. Imaginary part of the impedance as a function of the frequency (logarithmic
scale). Open triangles, open squares, filled triangles and filled squares represent
electrodes prepared using L1–L2, L1–N2, N1–L2 and N1–N2, respectively. The slope
for each plot was determined using data between 103.3 and 104.5 Hz.

D. Stoica et al. / Electrochimica Acta 56 (2011) 10009–10015
10015
trodes made from Ag2O paste prepared using N1 are shown to be
much less sensitive to the annealing and anodisation protocols (L2
and N2), with both N1–L2 and N1–N2 exhibiting very similar equi-
librium voltages. This again supports the results in Fig. 3b where the
profile of the anodisation transient was unchanged when a different
annealing or anodisation protocol was employed.
The difference in mean potentials between the four electrode
types could be explained by a difference in porosity, which will
affect the amount of AgCl formed and the access of the electrolyte
to impurities within the body of the electrode, and supporting Pt
wire [12]. Electrodes prepared using N1–N2 and N1–L2 show a sim-
ilar difference in potential to the reference. From earlier results
their porosity is expected to be similar. Electrodes prepared using
L1–N2 show a shift in mean potential to the reference. This can be
explained by a change in porosity (Fig. 4b) which shows a more
porous structure compared to electrodes prepared using L1–L2
(Fig. 4a). Electrodes prepared using L1-L2 show an even larger shift
in mean potential to the reference and an even more porous struc-
ture (Fig. 4c).
highlights the interest of using non-destructive electrochemical
methods for the evaluation of the electrode porosity. The complex
impedance spectroscopy appears to be an excellent investigative
tool, also allowing comparison of porosity with fractal geometry.
Further research should be carried out to study further the rela-
tionship between porosity and electrode performance. Organizing
an international comparison on this topic could be a possible way
to gain a better understanding of the effect of different electrode
preparation protocols on electrode performance. This would have
positive implications for pH measurements and may impact on the
use of Ag/AgCl electrodes in emerging new areas of electrochemical
research [22].
Acknowledgements
The research leading to these results has received funding from
the European Union on the basis of decision no. 912/2009/EC.
References
We may conclude that the absolute voltage characteristics of
Ag/AgCl electrodes produced are difficult to assess because of the
correlation with the voltage reference electrode used. Therefore the
salient criterion (apart from electrode slope which has not been
assessed here) is the relative positions of their mean voltages. This
gives some indication of the effect of each method on relative equi-
librium potential – which is often a surrogate indication of other
parameters such as porosity. This supports previous observations
that the drier L1 paste produces more porous electrodes, and seems
much more sensitive to the annealing and anodisation protocol as
compared to the wetter N1 paste which produces less porous elec-
trodes and seems to be much less sensitive to the annealing and
anodisation regime used.
[1] R.G. Bates, Determination of pH: Theory and Practice, 2nd ed., John Wiley &
Sons, 1973, Ch. 10.
[2] R.P. Buck, S. Rondinini, A.K. Covington, F.G. Baucke, C.M. Brett, M.F. Camoes,
M.J.T. Milton, T. Mussini, R. Naumann, K.W. Pratt, P. Spitzer, G.S. Wilson, Pure
Appl. Chem. 74 (11) (2002) 2169.
[3] P. Spitzer, Final report CCQM-K17: pH Determination on a Phthalate Buffer by
Harned Cell Measurements, 2003.
[4] R. Champion, C. Rivier, Revue Franc¸ aise de Métrologie 5 (1) (2006) 13.
[5] R.G. Bates, J.B. Macaskill, Pure Appl. Chem. 50 (1978) 1701.
[6] G.J. Janz, H. Taniguchi, J. Am. Chem. Soc. (1953) 397.
[7] M.J. Guiomar Lito, M.F. Camões, J. Solution Chem. 38 (2009) 1471.
[8] P.J. Brewer, R.J.C. Brown, Sensors 9 (1) (2009) 118.
[9] R.J.C. Brown, P.J. Brewer, D.J.L. Brett, Accred. Qual. Assur. 14
(2009) 139.
[10] P.J. Brewer, R.J.C. Brown, Sensors 10 (3) (2010) 2202.
[11] P.J. Brewer, D. Stoica, R.J.C. Brown, Sensors 11 (8) (2011) 8072.
[12] M-S. Chan, C. Hill, United States patent No 5565143, (1996).
[13] G.I.N. Waterhouse, G.A. Bowmaker, J.B. Metson, Phys. Chem. Chem. Phys. 3
(2001) 3838.
4. Conclusions
[14] P.J. Brewer, R.J.C Brown, F.B Gonzaga, X.Hongyu, W.Bing, P.Spitzer, P.T Jakobsen,
R. Champion, P. Fisicaro, D. Stoica, M. Ohata, L. Vyskocil, M. Máriássy, K. Pratt,
M. Monroy, A. Reyes, W. Kozlowski, E. Kardash, D. Ivanova, M.F. Camoes, H.
Lee, E. Hwang, CCQM P37.1. Preparation of Ag/AgCl Electrodes for Harned Cell
Measurements, NPL Report AS60, 2011.
[15] A. Mendizza, United States patent No 3004903, (1961).
[16] S. Jaya, T. Prasada Rao, G. Prabhakara Rao, J. Appl. Electrochem. 17 (3) (1987)
635.
[17] E. Barkousov, J.R. Macdonald, Impedance Spectroscopy: Theory, Experiment
and Applications, 2nd ed., John Wiley & Sons, NJ, 2005.
[18] D.D. Macdonalds, Electrochim. Acta 51 (2006) 1376.
[19] J.B. Jorcin, M.E. Orazem, N. Pébère, B. Tribollet, Electrochim. Acta 51 (2006)
1473.
[20] E. Garcia-Ochoa, J. Genesca, Surf. Coat. Technol. 184 (2004) 322.
[21] M. Máriássy, K.W. Pratt, P. Spitzer, Metrologie 46 (2009) 199.
[22] R.J.C. Brown, A.C. Keates, P.J. Brewer, Sensors 10 (11) (2010) 9982.
The accuracy of primary potentiometric results depends on
the structural characteristics of Ag/AgCl reference electrodes. For
achieving reliable pH measurements, some of the Ag/AgCl reference
electrode properties need to be well controlled. This task is very
difficult since many parameters have been identified as influenc-
ing the final electrode characteristics. In this context, the present
work has advanced the understanding of the influence of elec-
trode preparation technique on electrode morphology, especially
its porosity.
The role of the microstructure in affecting the behaviour and
response of the electrodes has been investigated, i.e. during anodi-
sation and prior to their use in the Harned cell. The present work