The New Wave of

Ion-Selective
Thanks to recent advances, ISEs are
tackling trace-level measurements.

few years ago, leading scientists advised their

younger colleagues not to work in the field of
potentiometric sensors. Researchers viewed po-
tentiometry as a mature field; all the basic facts
were well understood, and no possibilities existed
for significant new achievements.
In fact, although ion-selective electrodes (ISEs) had been de-
scribed for more than 50 analytes, their practical use was limited
to only a few fields. They were especially well established in clin-
ical chemistry for the measurement of K+, Na+, Ca2+, and Cl–
concentrations (activities) in whole blood, serum, or plasma. The
two main limitations were insufficient detection limits and selec-
tivities. Indeed, most potentiometric sensors had detection limits
in the micromolar range, and interfering ions were discriminated
by, at most, a factor of 103–104. Such detection limits and selec-
tivities were not acceptable for trace-level measurements, that is,
for assessing submicromolar activities in the presence of large
concentrations of interfering ions. Applications such as monitor-
ing heavy metals in drinking water or blood seemed like hope-
less ideas.
This situation has dramatically changed in the past few years.
Today, it is feasible to construct ISEs for a number of analytes
with detection limits down to low parts-per-trillion levels (10–10
M). Accurate monitoring of heavy metals in unspiked environ-
mental samples has been demonstrated and compared with rou-
tine analytical methods. This breakthrough certainly came as a
surprise to most researchers working with chemical sensors and
Eric Bakker has fueled renewed interest in pursuing active research in this
Auburn University field. Unlike most other analytical methods for trace-level deter-
minations, these advances were not achieved by improving the
Ernö Pretsch instrumental technique but by studying the underlying chemi-
Swiss Federal Institute of cal principles and eliminating experimental biases. In this article,
Technology (ETH Zürich) we describe how the detection limits and selectivities of poten-

420 A A N A LY T I C A L C H E M I S T R Y / A U G U S T 1 , 2 0 0 2
Electrodes

A U G U S T 1 , 2 0 0 2 / A N A LY T I C A L C H E M I S T R Y 421 A
(a) emf (b) emf

Ion-selective Reference Primary ion I+

electrode electrode
Interfering
Inner reference
(Ag/AgCl)
Inner
reference Reference
(Ag/AgCl) electrolyte
Diaphragm
Inner filling
Bridge
solution electrolyte
Ion-selective Diaphragm or
membrane Sample solution capillary
ai(DL)
log a
FIGURE 1. How electrodes work.
(a) Schematic of a potentiometric cell with an indicator and a reference electrode. (b) Calibration
curves for a primary cation (blue) and an interfering cation (red) of the same charge. The emf
shift between the two curves is a measure of the selectivity. For example, a shift of 178 mV
signifies a 3-orders-of-magnitude preference of the analyte ion over the interference. The back-
ground activity of interfering ions, aj(BG), defines the lowest possible potential (horizontal green
dotted line) that can be measured under these conditions. This potential represents the best
achievable detection limit of the primary ion ai(DL) in the presence of the background. This de-
tection limit is often not observed in practice because of transmembrane ion fluxes.
tiometric sensors have been improved by many orders of magni-
tude by simple chemical means.
Basics
Potentiometric sensors are essentially passive electrochemical de-
vices, in which changes in the electromotive force (emf) are mon-
itored under virtually zero current conditions. In so-called di-
rect potentiometry, the emf is ideally a function of the activity
of only one chosen sample ion so that it can be selectively as-
sessed in the presence of other ions. Achieving sufficient selec-
tivity to measure one ion in the presence of others has long been
the main focus of research in this field (1). The basic concept is
perhaps best illustrated with ISEs containing polymeric mem-
branes, which are assumed to behave as organic liquids of high
viscosity. Today, this class of sensors is by far the most versatile
and widely established (2).
If this membrane contains a lipophilic cation exchanger, its
countercation can freely exchange or equilibrate with the two
contacting aqueous solutions (Figure 1a). Assume that these
solutions contain the same hydrophilic electrolyte, M+A–, and
M+ is the countercation of the ion exchanger in the membrane.
A chemical driving force that depends on the activity of this ion
in both phases causes M+ to partition from the organic to the
aqueous phase. Because M+ carries a charge and no net current
can flow, this driving force is counterbalanced by the phase
boundary potential. It builds up spontaneously as M+ starts to
cross the interface and leads to interfacial charge separation.
The relationship between this interfacial potential, EPB, and the
ion activities on either side of the phase boundary is
RT k a (aq)
E PB = 1n M M (1) (1)
zF aM(org)
in which aM(aq) and aM(org) are the activities of the ion,
with charge z in the aqueous and organic phases, respec-
tively; kM is a constant that incorporates the standard
ion J+
Gibb’s free energy of ion transfer from the aque-
ous to the organic phase; and R, T, and F are the
gas constant, the absolute temperature, and the
Faraday constant, respectively.
Figure 2a illustrates this simple mechanism.
In the case shown here, the composition of the
membrane phase (aM(org) in Equation 1) remains
aj(BG) essentially unchanged as the concentration in the
sample phase is altered, and a response accord-
ing to the so-called Nernst equation is observed:
RT
E PB = K + 1naM(aq) (2)
zF
in which K is a constant for each M. If interfer-
ence from other sample ions is significant, the
membrane concentration of M+ changes, which
alters the observed emf (Equation 1). To obtain
highly selective ISEs on the basis of polymeric membranes, se-
lective complexing agents (ionophores) have been synthesized
and doped into these membranes. Their purpose is to selective-
ly incorporate only one type of ion into the membrane and to
suppress competition from interfering ions (3, 4). The synthe-
sis and application of novel ionophores were the primary fo-
cuses of research in this field for many years (5).
The relationships outlined earlier can be adequately described
with thermodynamic models. Indeed, the composition of the or-
ganic-phase boundary layer can be related to the composition
of the adjacent sample layer on the basis of ion-exchange and elec-
trolyte coextraction equilibria between the two phases and com-
plexation and ion-pair formation equilibria in the organic mem-
brane phase (2). Today, the theoretical background has matured
to a stage that it can be used to optimize the membrane compo-
sition and describe how ion sensors behave in mixed electrolyte
solutions. The lower detection limit, for example, is reached in
theory when interfering ions displace 50% of the ions to be mea-
sured from the membrane. As a consequence, one established
experimental protocol to determine the selectivity of the sensor
toward an ion calls for the measurement of the lower detection
limit in the presence of a constant ionic background (6). If the
detection limit is low, the selectivity toward the background ion
is good (Figure 1b).
Unfortunately, the thermodynamic treatment of ISE function
has apparently failed for dilute solutions. The observed emf has
often been affected by sample stirring, for example, and selectiv-
ities determined under such conditions have turned out to be
grossly underestimated. It is now well established that the ob-
served detection limit is often not dictated by changes in the
membrane concentration of M+, as expected from the thermo-
dynamic model alone. Rather, nonthermodynamic
processes may polarize the concentration of M+
in the contacting sample solution and give a
higher value at the sensing surface than in
the sample bulk. A thermodynamic treat-
ment is still valid for the local interfacial

422 A A N A LY T I C A L C H E M I S T R Y / A U G U S T 1 , 2 0 0 2
processes, even though the sensed concentration at the phase layer near the membrane. Because ordinary ion concentrations
boundary may deviate from the concentration in the bulk. This is are on the order of 10–2 M in the membrane phase, a concen-
very surprising. After all, potentiometry is assumed to be a static tration polarization of only 1% in the membrane produces a con-
method and should give thermodynamic information about a centration difference of ~10–7 M between the surface layer and
sample. How can we locally polarize the concentra- the bulk of the sample. The analyte will have this concentra-
tion if no current is passed across the cell? tion in the aqueous layer adjacent to the membrane even
if the bulk contains virtually no sample ions. Because
Ion fluxes in potentiometry the phase boundary concentration determines the
Generally, the sample and the inner solu- emf, it is the fundamental reason the lower detec-
tion do not have identical compositions. tion limits of innumerable polymer membrane
Because ions can diffuse across the organ- electrodes were reported to be on the order of
ic membrane phase according to various 10– 6–10–7 M (3).
mechanisms, this concentration asymme- The same process has also dramatically biased
try induces a process toward a true equi- the quantification of the selectivities of many
librium. This process is so slow that it can membrane electrodes. To quantify selectivities cor-
be ignored if the solution concentrations rectly, the measurements must be done under con-
of the relevant species are higher than ~1 µM. ditions in which only one of the ions is potential-de-
However, if lower concentrations are involved, termining. If the membrane were highly selective, the
transmembrane ion fluxes may have a decisive in- emf of a concentrated solution of an interfering ion would
fluence on the response of ISEs. According to the equilib- be lower than the emf of a 10–6-M solution of the analyte ion
rium model (Figure 2a), no gradients arise in the system. This (Figure 1b). In other words, the small amount of leached primary
model is very close to reality, even if the composition of the inner ions, not the discriminated interfering ions, determined, in many
solution is different from that of the sample. The ion-exchange cases, the measured selectivities (6).
property of the hydrophobic membrane largely blocks the incor-
poration of the aqueous cation and its counterion into the organ- How can ion fluxes be reduced?
ic phase. Furthermore, the high selectivity largely hinders the The key to the recent spectacular improvements toward lower
exchange of the analyte ion with an interfering ion. The key to un- detection limits and better selectivities has been the reduction of
derstanding the processes at low sample concentrations is the word transmembrane ion fluxes (9, 10). As shown in Figure 2b, opti-
“largely”. A very small fraction of the primary ions is exchanged by
interfering ions already at concentrations that are orders of magni-
tude larger than the lower detection limit (Figure 1b). (a)
Ideal potentiometric response
For clarity, the situation is represented in Figure 2b with ex-
aggerated gradients (only the unstirred layer of the sample is Sample Membrane Inner solution
+ +
– – + – + – + +
shown). The exchange of a small portion of the sample ions + –
+– + + +– – –+
–
(blue) with interfering ions (red) induces opposite gradients of + –– + + + + – + –
– + + – – –+ – +
the primary (blue lines) and interfering ions (red lines) in the – –
+ +– + +
– – –+
– + – + + –
membrane and in the contacting stagnant aqueous layer. These + + – – +
+ – – + – – +
gradients generate ion fluxes without net charge transport and
cause an elevated primary ion concentration at the sensor surface. (b)
Because this phase boundary concentration determines the emf, Near detection limit
these diffusion processes may cause a strong bias when trying to Stagnant layer Membrane Inner solution
+ + + – + – + +
measure lower detection limits and apparent selectivities. Such – – + –
+ + + +– – –+
elevated concentrations near the membrane surface were indeed – +
+ –– + + + + – + –
recently observed by scanning electrochemical microscopy (7 ). – + + – – –+ – +
– –
+ +– + +
– – –+
Fortunately, the phase boundary potential model described – + – +
+ – – –+ – + + – –+
above is also valid when the whole system is out of equilibrium + + –
because phase boundary processes are much faster than trans- FIGURE 2. Schematic of a sensing membrane with the sample
membrane ion fluxes. Because the thickness of the concentration and the inner solution for a symmetrical setup (a) in equilibrium
gradient in the stagnant aqueous layer and the membrane are and (b) in the steady state with zero-current ion fluxes.
comparable (Figure 2b), the ratio of the two gradients is rough- (a) The concentration of analyte ions (blue) in the membrane is defined by
ly equal to the ratio of the corresponding diffusion coefficients the amount of lipophilic anions, which cannot leave the membrane. (b) Inter-
(8). Diffusion coefficients in the membrane phase vary according fering ions (red) replace a small amount of analyte ions in the membrane on
the sample side. Because of the resulting gradients, analyte ions leach from
to polymer and plasticizer content but are ~1000-fold smaller
the membrane into the sample and induce an elevated concentration near
than those in water. Therefore, the concentration difference be- the membrane. At submicromolar sample activities (e.g., during trace-level
tween the two surface layers of the membrane is ~1000-fold measurements or selectivity determinations), these analyte ions can become
higher than between the bulk of the sample and in the aqueous potential-determining and cause gross errors.

A U G U S T 1 , 2 0 0 2 / A N A LY T I C A L C H E M I S T R Y 423 A
emf
Ca2+ Cd2+ Pb2+

59.2 mV 59.2 mV 59.2 mV

by excessively rapid diffusion through the membrane and exces-
sively slow diffusion in the aqueous phase. Any measure that re-
–14 –10 – 6 –2 –10 –8 –6 –12 –8 –4
log aCa log aPb
duces ion fluxes in the membrane is, therefore, beneficial (14,
log aCd
15). For example, membranes are more rugged if the diffusion
emf ClO4– coefficients are reduced by increasing their polymer content; if
NH4+ K+
the ionophore is covalently bound to the polymer backbone; if
the total amount of ions (i.e., the concentration of the ion ex-
changer in the membrane) is reduced; if the thickness of the
membrane is increased; if its effective surface is reduced relative
59.2 mV 59.2 mV 59.2 mV to water by adding inert lipophilic particles; or if the stirring of
the aqueous phase is increased (e.g., by using a flow-through sys-
–9 –7 –5 –3 –9 –7 –5 –3 –4 –6 –8 tem or a wall-jet configuration) (15, 16). By combining two of
log aNH4 log aK log aCIO4
these measures (i.e., adjusting the inner solution and making the
FIGURE 3. Response functions of some ISEs with recently membrane less sensitive to concentration gradients), scientists
improved lower detection limits. have designed potentiometric sensors that are adequate for mon-
itoring heavy metals such as lead in drinking water (17 ).

mal detection limits and unbiased selectivity behavior would, in
theory, be achieved by using an inner solution that exactly match-
es the sample. Here, “matching” means that the composition of
the inner solution is designed in a way that the same amount of
analyte ion is replaced by an interfering ion on either side of the
membrane. As shown earlier, diffusion of analyte ions from the
membrane to the sample generates less-than-optimal detection
limits and selectivities. On the other hand, if the inner solution
is designed so that too many analyte ions at the inner membrane
side are replaced by other ions, a flux of analyte is created from
the sample to the membrane. Also, this process is harmful be-
cause during low sample activities, it depletes sample ions near
the membrane and leads to a nonideal response (11).
Originally, the composition of the inner solution was adjust-
ed by using a hydrophilic complexing agent such as EDTA to
buffer the primary ion activity and a rather high concentration
of the interfering ion (10). On the basis of the selectivity behav-
ior of the membrane, researchers can predict the composition of
the inner solution in order to replace the required portion (typ-
ically <1%) of primary ions with interfering ions (12). For ions
such as alkali metal ions or many anions for which no appropriate
complexing agents are known, an ion exchanger can be added to
the inner solution to act as a nonselective ion buffer because of
the rather large capacity of the resin (13).
Because the inner solution is optimal for only one (typical)
sample composition, any concentration change in the sample will
induce an ion gradient in either direction. It is therefore also im-
portant to make ion-selective membranes as insensitive as pos-
sible against such gradients. It is astonishing at first glance that
problems like biased detection limits and selectivities are caused
Table 1. Direct potentiometric determinations
of Pb2+ in tap water (17 ).
ISE (µg/L) ICPMS (µg/L)
Unspiked tap water 0.71 ± 0.04 0.54 ± 0.07
2.0 µg/L 2.61 ± 0.06 2.80 ± 0.04
10 µg/L 11.5 ± 0.3 11.30 ± 0.06
20 µg/L 22.2 ± 1.2 22.0 ± 0.2
What’s going on now?
A series of ISEs have recently been developed with lower detec-
tion limits (Figure 3) that are far better than what was expected
only a few years ago (18). These ISEs can now compete with the
most sophisticated methods of trace analysis. More importantly
and parallel to these developments, a sound theoretical under-
standing of the basic mechanisms has been achieved, which makes
it possible to optimize the electrodes very efficiently (12, 17, 19).
Once their selectivity has been determined, the composition of
the inner solution can be predicted (12).
The most important practical applications seen today are heavy
metal analyses of drinking water, other environmental samples,
and biological samples, which demand high selectivity, rugged-
ness, and short response and calibration times. For lead in drink-
ing water, the detection limit should be at least 10 times lower
than the 15-ppb (7.2
10–8 M) action limit imposed by the U.S.
Environmental Protection Agency (20). In Figure 4b, the action
limit is marked by the color transition between the green and red
zones of the plot. The detection limit of 7.2
10–9 M must be
reached in the presence of 10–3–10–4-M levels of the abundant
interfering ions Na+, Ca2+, and H+.
Screening of several existing ionophores determined that the
basket-shaped, thioamide-derivatized calix[4]arene was the most
appropriate host molecule for lead because of its excellent selec-
tivity (21). In fact, the characteristics of the resulting Pb2+ ISE ex-
ceeded the earlier stated requirements, and alkali earth metal ions
were found to be discriminated by as much as 14 orders of mag-
nitude. This left some room for reducing the ion exchanger con-
centration in the membrane, which resulted in a further reduction
in ion flux and more rugged behavior, even though the selectivi-
ty was somewhat reduced. The detection limits in typical ionic
backgrounds for tap water were ~1 nM, and they were perfectly
in line with theoretical expectations. Because response times dur-
ing such low activities are rather long, calibration solutions in the
range 10–5–10–4 M were used, indicated by the colored points
in Figure 4b and coded with the same colors in Figure 4a. The
5-point calibration by standard addition required ~10 min. It
may seem dangerous to rely on calibration solutions that are
1000–10,000-fold more concentrated than the samples. How-

424 A A N A LY T I C A L C H E M I S T R Y / A U G U S T 1 , 2 0 0 2
ever, thanks to excellent reproducibility of the response slopes,
the observed uncertainties were acceptably small (Table 1).
Figure 4b shows a typical measurement of untreated tap water
at its native pH 8, indicated as the lower blue point on the cali-
bration curve. Clearly, the assessed level of lead (2
10–10 M) is
extremely low. The lead concentration of the same sample deter-
mined with inductively coupled plasma MS (ICPMS), indicated as
a filled circle on the calibration curve of Figure 4b, is an order of
magnitude higher than the one obtained with the potentio-
metric sensor. This is not surprising if one realizes that
potentiometric measurements assess free ion activities,
not total ion concentrations. At elevated pH, most
of the lead ions are complexed, mainly by carbonate,
and <10% remains in the free, uncomplexed form. It
is this form that the potentiometric sensor observes,
hence the lower value. If the sample is buffered to pH 4.0,
where lead is mostly in its free form, the values obtained by po-
tentiometry and ICPMS are indeed nearly identical (Figure 4b).
The unique characteristics of potentiometric sensors that allow
speciation measurements are further illustrated in Figure 4c. A
spiked tap water sample was adjusted to various pH values, and
the potentiometrically assessed lead values (open circles) agree
well with theoretical expectations on the basis of two carbonate
levels in the sample (dotted lines). Again, it can be seen that
buffering to pH 4 transforms all the lead complexes into their
free form, and potentiometric and atomic spectrometric methods
the original papers, if the biasing effect of ion fluxes is elimi-
nated. Very probably, such improvements will be found for many
other available systems.
A sound theoretical basis of the involved zero-current trans-
port processes has been developed and the main predictions have
been experimentally verified. A palette of measures is available
that can be used to optimize membranes. The theory can be di-
rectly applied to predict the required composition of an inner
electrolyte on the basis of the selectivity behavior of the mem-
brane and the composition of the sample. Improving the detec-
tion limits of future electrodes will now be more routine. Lastly,
the practical use of recently developed electrodes for trace-level
determination has been demonstrated. On the basis of this, we
expect that a series of potentiometric sensors with submicromo-
lar detection limits for use in environmental and clinical analyses
of trace metals will be described soon. Although the possibili-
ties are promising, clinical trace metal determinations by poten-
tiometry have not yet been the focus of research.
A very small external current can be used to compensate for
zero-current ion fluxes (16, 22). This method is attractive because
it is easier to alter the required current than the compositions of
(a) 29.6 mV (b) Tap water

emf
yield comparable results under these conditions (17 ).
emf

10 –9 M
10 –9 M
To compare total assessed lead levels determined by poten- pH 4
29.6 mV
tiometry and ICPMS, various samples were collected and buffered pH 8 ICPMS
to pH 4.0 before measurement. Table 1 shows the observed lead 5 x 10 –7 M 5 x 10 –7 M
concentrations for native and spiked samples and demonstrates
excellent correlation and acceptably small standard deviations for 0 5 10 15 20 25 –10 –9 –8 –7 –6 –5 –4
the potentiometric data. The standard deviations were obtained Time (min) log [Pb2+]
from measurements with three different sensors on the same sam-
ple. The determination of lead represents a prototype study and (c) 100
indicates that a procedure for environmental monitoring of other
% Uncomplexed lead

heavy metal ions for which ISEs with sufficient selectivities are
available (e.g., Cd2+, Ag+, Cu2+) is likely to be developed soon.
In the context of practical applications, it is important to re-
member that ISEs are quite unique in their capability to assess
free ion activities. This is in contrast to the information obtained
by other methods: Atomic spectrometry gives total concentra-
tions, and voltammetry yields concentrations of labile species
and complexes. Depending on the problem at hand, one or the
other value may be of primary interest. For speciation purpos-
es, a comparison of several methods or potentiometric measure-
ments under different conditions may be useful.
What holds the future?
It’s been almost four years since the first papers on improving de-
tection limits were published. The successes achieved since have
been rather impressive. A series of ISEs have been described,
showing how the lower detection limits of any polymeric mem-
brane electrode of sufficient selectivity can be improved. The se-
lectivities of a series of reinvestigated electrodes have been im-
proved by many orders of magnitude over those determined in
80
60
40
20
0
3 4 5 6 7 8 9
pH
FIGURE 4. Lead in drinking water.
(a) Time responses during a five-point calibration of two different samples
containing the indicated concentration of lead ions. (b) Lead in native tap
water at pH 8, after buffering to pH 4, and five calibration points with the
same activities, as in (a). The black-filled circle represents the concentration
determined by ICPMS analysis. The observed slope of 29.6 mV/decade is
expected, according to the Nernst equation (Equation 2) for a divalent ion
(ln 10
RT/2F = 29.6 mV). (c) Potentiometric responses of a drinking water
sample spiked with 10 ppb of lead at different pH values. Dotted lines are
the theoretically expected fractions of uncomplexed lead, taking into account
complexation equilibria with the specified levels of CO2– 3 (two assumed con-
centrations), HCO3–, OH–, SO42–, NO3–, HSO4– , and Cl–. Carbonate is the domi-
nating complexing ion.

A U G U S T 1 , 2 0 0 2 / A N A LY T I C A L C H E M I S T R Y 425 A
the membrane and the inner solution.
However, determining the required cur-
rent for an unknown sample is not a sim-
ple task, and further research is needed
to show the practical applicability of this
approach. Another related research en-
deavor is the development of ISEs without
inner solution to eliminate one major source of
transmembrane ion fluxes (23). This promises to References
eventually eliminate steady-state ion fluxes completely because only (1)
two phases would be involved (Figure 2). If successful, it should (2)
further improve detection limits by many orders of magnitude. (3)
Sensor miniaturization will be necessary for achieving low de- (4)
tection limits because an ordinary membrane contains ~10–7 moles (5)
of primary ions, which corresponds to a picomolar concentra-
tion in 10,000 L of sample. If a sensing membrane is placed di- (6)
rectly on a metal, there are two main concerns: how to guarantee (7)
a stable (redox) potential between the membrane and the metal (8)
(23–26) and how to prevent the spontaneous formation of a
water film between the two (27, 28). Active research on different (9)
possible solutions is being pursued, so the development of low- (10)
detection-limit, solid-contact electrodes is likely.
Another promising field of research that will benefit from (11)
these developments is the design of optical sensors with small
sensing volumes (29), whose response mechanism is closely re- (12)
lated to that of potentiometric sensors (2). However, steady-state (13)
ion fluxes are not anticipated here because these sensors rely on (14)
a two-phase equilibration process. Therefore, low detection lim- (15)
its should be more easily obtained than with classical ISEs (30). (16)
For most techniques, lower detection limits are either set by
interferences or by the physical limitations of the instruments. (17)
The ultimate low detection limit of potentiometry is still not
yet known. Highly selective systems exist and, in principle, they (18)
should be able to achieve such extremely low detection limits (19)
that they do not seem realistic. It is expected, but not yet shown
experimentally, that the small currents on the order of 100 fA (20)
required for zero-current potentiometric measurements will set
the limits, and these might be somewhere between picomolar
and femtomolar activities. (21)
Compared with other methods of trace analysis, potentiome-
try is an extremely inexpensive technique. Because it can com- (22)
pete with the most sophisticated methods, it is very
likely that we will encounter many practical applica- (23)
tions in the future. Potentiometric sensors are (24)
known to only minimally perturb the sample,
compared with most other analytical methods, (25)
and they should be especially well suited for in-
tegration into optimized fluidic and microfluidic (26)
systems for a variety of applications. (27)
We are grateful for financial support of our research from the Na- (28)
tional Institutes of Health (grant R01-GM59716), the Swiss National Science
Foundation, and Orion Research. (29)
Eric Bakker is a professor at Auburn University. His research interests (30)
include fundamental and applied aspects of potentiometric, voltam-
426 A A N A LY T I C A L C H E M I S T R Y / A U G U S T 1 , 2 0 0 2
metric, and optical sensors based on molecular recognition prin-
ciples. Ernö Pretsch is a professor at the Swiss Federal Institute
of Technology (ETH). His more recent research interests focus
on the improvement of lower detection limits for ISEs and the
development of computer programs for spectrum estimation
and automatic spectra interpretation. Address correspondence
about this article to Pretsch at pretsch@org.chem.ethz.ch.
Buck, R. P.; Lindner, E. Anal. Chem. 2001, 73, 88 A–97 A.
Bakker, E.; Bühlmann, P.; Pretsch, E. Chem. Rev. 1997, 97, 3083–3132.
Bühlmann, P.; Pretsch, E.; Bakker, E. Chem. Rev. 1998, 98, 1593–1687.
Bakker, E.; Bühlmann, P.; Pretsch, E. Electroanalysis 1999, 11, 915–933.
Umezawa, Y. Handbook of Ion-Selective Electrodes: Selectivity Coefficients;
CRC Press: Boca Raton, FL, 1990.
Bakker, E.; Pretsch, E.; Bühlmann, P. Anal. Chem. 2000, 72, 1127–1133.
Gyurcsányi, R. E.; et al. Anal. Chem. 2001, 73, 2104–2111.
Sokalski, T.; Zwickl, T.; Bakker, E.; Pretsch, E. Anal. Chem. 1999, 71,
1204–1209.
Mathison, S.; Bakker, E. Anal. Chem. 1998, 70, 303–309.
Sokalski, T.; Ceresa, A.; Zwickl, T.; Pretsch, E. J. Am. Chem. Soc. 1997, 119,
11,347–11,348.
Sokalski, T.; Ceresa, A.; Fibbioli, M.; Zwickl, T.; Bakker, E.; Pretsch, E.
Anal. Chem. 1999, 71, 1210–1214.
Ion, A. C.; Bakker, E.; Pretsch, E. Anal. Chim. Acta 2001, 440, 71–79.
Qin, W.; Zwickl, T.; Pretsch, E. Anal. Chem. 2000, 72, 3236–3240.
Zwickl, T.; Sokalski, T.; Pretsch, E. Electroanalysis 1999, 11, 673–680.
Ceresa, A.; Sokalski, T.; Pretsch, E. J. Electroanal. Chem. 2001, 501, 70–76.
Lindner, E.; Gyurcsányi, R. E.; Buck, R. P. Electroanalysis 1999, 11,
695–702.
Ceresa, A.; Bakker, E.; Günther, D.; Hattendorf, B.; Pretsch, E. Anal. Chem.
2001, 73, 343–351.
Bakker, E.; Pretsch, E. Trends Anal. Chem. 2001, 20, 11–19.
Morf, W. E.; Badertscher, M.; Zwickl, T.; de Rooij, N. F.; Pretsch, E.
J. Phys. Chem. B 1999, 103, 11,346–11,356.
Lead and Copper Rule Minor Revisions; EPA 815-F-899-010; U.S. Environ-
mental Protection Agency, 1999, http://www.epa.gov/safewater/
standard/leadfs.html.
Malinowska, E.; Brzózka, Z.; Kasiura, K.; Egberink, R. J. M.;
Reinhoudt, D. N. Anal. Chim. Acta 1994, 298, 253–258.
Pergel, E.; Gyurcsányi, R. E.; Tóth, K.; Lindner, E. Anal. Chem. 2001, 73,
4249–4253.
Cattrall, R. W.; Freiser, H. Anal. Chem. 1971, 43, 1905–1906.
Bobacka, J.; Lindfors, T.; McCarrick, M.; Ivaska, A.; Lewenstam, A.
Anal. Chem. 1995, 67, 3819–3823.
Cadogan, A.; Gao, Z.; Lewenstam, A.; Ivaska, A.; Diamond, D. Anal. Chem.
1992, 64, 2496–2501.
Fibbioli, M.; et al. Chem. Commun. 2000, No. 5, 339–340.
Fogt, E. J.; Untereker, D. F.; Norenberg, M. S.; Meyerhoff, M. E.
Anal. Chem. 1985, 57, 1995–1998.
Fibbioli, M.; Morf, W. E.; Badertscher, M.; de Rooij, N. F.; Pretsch, E.
Electroanalysis 2000, 12, 1286–1292.
Tsagkatakis, I.; Peper, S.; Retter, R.; Bell, M.; Bakker, E. Anal. Chem. 2001,
73, 6083–6087.
Lerchi, M.; Orsini, F.; Cimerman, Z.; Pretsch, E.; Chowdhury, D. A.;
Kamata, S. Anal. Chem. 1996, 68, 3210–3214.