J. Am. Chem. Soc.

1997, 119, 11347-11348 11347

Large Improvement of the Lower Detection Limit

of Ion-Selective Polymer Membrane Electrodes

Tomasz Sokalski,† Alan Ceresa,‡ Titus Zwickl,‡ and

Ernö Pretsch*,‡
Department of Organic Chemistry
Swiss Federal Institute of Technology (ETH)
CH-8092 Zürich, Switzerland
ReceiVed August 20, 1997
ReVised Manuscript ReceiVed October 8, 1997
The lower detection limits of ion-selective polymeric mem-
brane electrodes (ISEs) are in the micromolar range except when
sample ion activities are adjusted by using ion buffers, which
maintain low and constant activity via superimposed complex- Figure 1. Schematic measuring setup and concentration profiles in
ation or solubility equilibria.1 Therefore, they are unsuitable membrane at steady state. The potential difference between the aqueous
for many important applications. In contrast, detection limits solutions is a measure of analyte (M+) activities at the membrane
in the picomolar range have been achieved with optical sensors surfaces. In conventional systems (dashed line), slight disturbances
induce a flow of M+ toward the sample, thus biasing its activity at the
on the basis of similar technology and the same ionophores,2
corresponding membrane surface. This is prevented by the gradient
suggesting that there is no fundamental reason for the detection
(solid line) brought about in this work.
limit of analogous ISEs to be worse. Much lower values have
indeed been observed when buffering the sample analyte ion,3 thus indicating that additional processes must be involved.
thus supporting the hypothesis that primary ions leaching from Moreover, if the detection limit were solely defined by the
the sensor membrane determine the lower limit of the measuring transport of primary ions from the internal solution, it should
range of the ISE. This also explains the reduction or even lack be significantly lower for solid contact electrodes. This,
of response toward highly discriminated ions, which prevents however has not been reported so far.
the determination of true selectivity coefficients.4 Theoretical The basic idea of the present work is to prevent primary ions
(Nernstian) response and true selectivity coefficients can, from leaching into the sample by building up a concentration
however, be obtained by buffering the primary ion in the sample gradient in the membrane (Figure 1). The-thereby induced
without influencing the activity of the interfering ion.5 Recent constant flux of primary ions toward the inner compartment
experiments proved that, even for extremely discriminated ions, prevents their passing into the sample. The concentration
Nernstian ISE response is found without buffering the primary gradient is established by choosing an internal electrolyte with
ion as long as its contact with the electrode is avoided.6 low activity of the primary ion and sufficiently high activity of
Although such experiments are important for understanding the an interfering ion. Since any activity change in the inner
selectivity behavior, they do not contribute to extending the compartment would induce a potential change, the low activity
response range in analytical applications. of the primary ion was kept constant by using an ion buffer.
Since the response of polymeric membrane ISEs is, in The responses of two Pb2+ ISEs with the same membrane10
essence, a surface phenomenon governed by bulk equilibria,1b,7 but different inner reference systems11 are shown in Figure 2.
the influence of the aqueous sample composition on concentra- With the conventional inner filling solution, the detection limit
tions in the bulk membrane is negligible if the electrode response for Pb2+ is 4 × 10-6 M, no Nernstian response being observed
is Nernstian.8 Therefore, primary ions most likely leach into for the discriminated Na+ since it is not potential determining
the sample because, together with their counterions, they are under these conditions (Figure 2). On the other hand, with the
extracted from the internal electrolyte and transported across new internal solution, the linear Nernstian response range for
the membrane. Very recently, Mathison and Bakker showed Pb2+ is extended to about 10-11 M and the detection limit is
that such a transport indeed occurs.9 Their model correctly lowered to 5 × 10-12 M (Figure 2). At the same time, the
describes the influence of the composition of the internal response toward Na+ is now Nernstian between 10-2 and 10-3
electrolyte at high concentrations but does not hold for low ones, M, indicating that Na+ is the potential-determining ion at these
† On leave from Department of Chemistry, University of Warsaw, PL- concentrations. Thus, true selectivity coefficients can be
02093 Warsaw, Poland. determined.
‡ ETH Zürich.
(1) (a) Umezawa, Y. Handbook of Ion-SelectiVe Electrodes: SelectiVity (10) The composition of the membrane was 1.31 wt % (14.5 mmol/kg)
Coefficients; CRC Press: Boca Raton, Ann Arbor, Boston, 1990. (b) Bakker, of N,N,N′,N′-tetradodecyl-3,6-dioxaoctanedithioamide (ETH 5435), 0.82 wt
E.; Bühlmann, P.; Pretsch, E. Chem. ReV. In press. % (9.1 mmol/kg) of potassium tetrakis[3,5-bis(trifluoromethyl)phenyl]borate
(2) (a) Lerchi, M.; Bakker, E.; Rusterholz, B.; Simon, W. Anal. Chem. (KTpClPB), 62.5 wt % of bis(2-ethylhexyl) sebacate (DOS), and 35.3 wt
1992, 64, 1534-1540. (b) Lerchi, M.; Orsini, F.; Cimerman, Z.; Pretsch, % of poly(vinyl chloride) (PVC all Fluka) totaling 368.5 mg dissolved in
E.; Chowdhury, D. A.; Kamata, S. Anal. Chem. 1996, 68, 3210-3214. ca. 2.5 mL of tetrahydrofuran. By casting this solution into a glass ring of
(3) Schefer, U.; Ammann, D.; Pretsch, E.; Oesch, U.; Simon, W. Anal. Ø 53 mm, membranes of ca. 85 µm thickness were obtained.
Chem. 1986, 58, 2282-2285. (11) Conventional Pb2+ ISEs were prepared with Philips IS 561 electrode
(4) (a) Umezawa, Y.; Umezawa, K.; Sato, H. Pure Appl. Chem. 1995, bodies and inner filling solutions of 5 × 10-4 M PbCl2 and 5 × 10-2 M
67, 507-518. (b) Bakker, E. Electroanalysis 1997, 9, 7-12. MgCl2. For the construction of the new type of Pb2+ ISEs, the membrane
(5) Sokalski, T.; Maj-Zurawska, M.; Hulanicki, A. Mikrochim. Acta 1991, was glued to PVC tubing. The inner filling solution (10-3 M Pb(NO3)2 and
I, 285-291. 5 × 10-2 M ethylenediaminetetraacetic acid disodium salt (EDTA-Na2))
(6) Bakker, E. J. Electrochem. Soc. 1996, 143, L83-L85. was in contact with the reference element (Ag/AgCl in 3 M KCl) through
(7) (a) Pungor, E. Pure Appl. Chem. 1992, 64, 503-507. (b) Umezawa, a 1 M KCl bridge electrolyte. After the electrodes were conditioned in 10-3
K.; Lin, X. M.; Nishizawa, S.; Sugawara, M.; Umezawa, Y. Anal. Chim. M Pb(NO3)2 for >2 d, measurements were made with a Metrohm NET-
Acta 1993, 282, 247-257. (c) Buck, R. P.; Nahir, T. M.; Cosofret, V. V.; Titrino 721 at 22-23 °C in successively diluted stock solutions of 10-2 M
Lindner, E.; Erdosy, M. Anal. Proc. 1994, 31, 301-312. (d) Bakker, E.; Pb(NO3)2 from higher to lower concentrations. End potentials were taken
Nägele, M.; Schaller, U.; Pretsch, E. Electroanalysis 1995, 7, 817-822. when the drift was ,0.5 mV/min (after e15 min for concentrations down
(8) (a) Cobben, P. L. H. M.; Egberink, R. J. M.; Bomer, J. G.; Bergveld, to 10-6 M and 15-30 min for lower ones). Activities were obtained
P.; Reinhoudt, D. N. J. Electroanal. Chem. 1994, 368, 193-208. (b) considering the association of the cations with Ac-, OH-, and NO3-,13
Kakiuchi, T. Anal. Chem. 1996, 68, 3658-3664. activity coefficients were calculated according to Debye-Hückel, and
(9) Mathison, S.; Bakker, E. Anal. Chem. In press. diffusion potentials were corrected for according to Henderson.1

S0002-7863(97)02932-6 CCC: $14.00 © 1997 American Chemical Society

11348 J. Am. Chem. Soc., Vol. 119, No. 46, 1997 Communications to the Editor

The concentration of free Pb2+ in the buffered internal

solution, calculated13 at the measured pH 4.34, is 10-12 M and
that of Na+ (counterion of EDTA) is 10-1 M. From log KpotPbNa
) -4.7 (Table 1) it is obvious that aPb is much smaller than
KpotPbNa × a2Na in the internal electrolyte so that Na+ is the
potential-determining ion at the inner reference side.12 Hence,
it nearly quantitatively replaces Pb2+ at that boundary so that,
at steady state, a concentration gradient of Pb2+ (and a
corresponding inverse one of Na+) must build up in the
membrane. The importance of at least partial displacement of
primary ions at the inner membrane surface is also suggested
by the fact that when Pb2+ in the internal electrolyte is buffered
to 10-7 M with nitrilotriacetic acid in the presence of ca. 10-3
M Ca2+, no improvement of the lower detection limit is found.14
As shown in Table 1, Ca2+ is strongly discriminated (log KpotPbCa
) -8.6) so that at the concentrations used, Pb2+ remains the
dominating ion also at the inner membrane side.
Figure 2. Response of two Pb2+ ISEs with the same membrane10,11 The new simple approach described shifts the detection limit
but different internal electrolytes (22-23 °C). Conventional (empty to the picomolar region and thus doubles the assessable
symbols): 1:1 mixture of 10-3 M PbCl2 and 0.1 M MgCl2. New (full
logarithmic response range from 6 to 12 orders of magnitude.
symbols): 1 mL of 0.1 M Pb(NO3)2 in 100 mL of 0.05 M EDTA-Na2;
It can be expected that in the future not artifacts but interferences
measured pH 4.34. Calculated13 activities: 10-12 M Pb2+, 10-1 M Na+.
from ions will be the only limiting factor. The method is
Table 1. Logarithmic Selectivity Coefficients, log KpotPbM, for a generally valid for any ion that can be buffered to low activity
Pb2+ ISE10,11,15 with Conventional and Buffered Inner Electrolyte in the internal electrolyte of the ISE. Similar results have been
(cf. Figure 2) obtained with Ca2+ and Ag+ ISEs. For Ca2+, its concentration
log KpotPbM for Pb2+ ISE with inner solution in the inner solution was kept low by adding sulfuric acid as
an ion buffer. This method could be of interest in miniaturized
ion M conventional this work
ISEs. In all cases investigated, not only the detection limit but
H +
-2.2 -3.5 also the measured selectivities were greatly improved. The
Li+ -2.7 -5.1 robustness of the system and the influence of various membrane
Na+ -2.2 -4.7 parameters are currently being studied. The procedure described
K+
allows one to make full use of extremely selective ionophores
-2.2 -4.1
NH4+ -2.5 -4.1
Mg2+ -3.6 -9.4 and opens up new dimensions of ISE applications in trace metal
Ca2+ -4.0 -8.6 analysis for environmental and biochemical studies by eliminat-
Cd2+ 0.9 0.4 ing a fundamental limiting factor.
Sr2+ -4.2 -6.7
Ba2+ -4.1 -6.0 Acknowledgment. The authors acknowledge support from World
Mn2+ -4.1 -5.8 Laboratory (Lausanne), Eidgenössische Stiftung zur Förderung Sch-
Co2+ -4.3 -6.3 weizerischer Volkswirtschaft durch wissenschaftliche Forschung (Zü-
Ni2+ -4.2 -5.9 rich), Hitachi Ltd. (Tokyo), and Orion Research Inc. (Beverly, MA).
Cu2+ -0.5 -0.4 We thank Prof. E. Bakker for various discussions and for sending a
Zn2+ -3.7 -5.0 preprint of his paper9 and Dr. D. Wegmann for careful reading of this
manuscript.
In Table 1, logarithmic selectivity coefficients, log KpotPbM, JA972932H
are given for both electrodes. It can be seen that most values
are drastically lower with the new internal electrolyte, e.g., (13) (a) Sillén, L. G.; Martell, A. E. Stability Constants of Metal-Ion
Complexes, Special Publication No. 17; The Chemical Society: London,
KpotPbM for Mg2+ and Ca2+ by ca. 5 orders of magnitude, which 1964. (b) Sillén, L. G.; Martell, A. E. Stability Constants of Metal-Ion
agrees with the corresponding shift of the lower detection limit. Complexes Supplement No. 1, Special Publicaton No. 25; The Chemical
As theoretically expected,12 the maximal improvement of the Society: London, 1971.
(14) Bakker, E.; Willer, M.; Pretsch, E. Anal. Chim. Acta 1993, 282,
logarithmic selectivity coefficient toward monovalent ions is 265-271.
about half that for divalent ions, i.e., 2.5 logarithmic units. (15) Selectivities were measured by the separate solution method1 in 0.01
M nitrate solutions of Na+, K+, Ca2+, Mg2+, Mn2+, Co2+, Ni2+, Mn2+,
(12) Bakker, E.; Meruva, R. K.; Pretsch, E.; Meyerhoff, M. E. Anal. Cu2+, Zn2+, and Cd2+ buffered to pH 4.7 with a magnesium acetate buffer
Chem. 1994, 66, 3021-3030. (2.5 × 10-4 M Mg2+, 10-3 M total Ac-).