New trends in ion-selective electrodes - Electrochemical Sensors, Biosensors and Their Biomedical Applications - page 81

where E I and E J are the potentials in the separate solutions of ion I with activity a I and
ion J with activity a J , correspondingly (see Eq. (5)). In the FIM the calibration curve
for the primary ion is recorded in a background of interfering ion at fi xed activity
( a J (BG)). The detection limit is determined for this curve ( a I (DL)), as a cross-section
of the two linear segments of the calibration curve (see Fig. 4.5), and the selectivity
coeffi cient is calculated according to Eq. (8):

a DL

BG

( )

pot

I

log K (8)

IJ

zz /

a ( )

I J

J

Both methods, SSM and FIM, ideally should give identical results, but they both rely
on the assumption that ISEs responds to interfering ions with Nernstian slope, which
often is not true. There are several possible reasons for the non-Nernstian behavior: high
discrimination of the interfering ions by the sensor, when the potential is dictated by
the constantly released primary ions from the membrane; electrolyte coextraction into
membrane phase and loss of membrane permselectivity; and non-classical response of
a sensor. The non-Nernstian response towards either primary or interfering ions makes
the Nikolski selectivity coeffi cients biased, often by many orders of magnitude, and thus
meaningless; in such cases no numeric characteristic of selectivity becomes available.
Other methods were suggested to avoid the required Nernst slope limitation, including
the so-called matched potential method (MPM); however, most of these empirical meth-
ods strongly depend on experimental conditions and are of limited practical signifi cance.
Recently the method for unbiased selectivity coeffi cients determination was pro-
posed based on the concept of the SSM [29]. The classic procedure for ISE prepara-
tion involves sensor conditioning in a primary ion solution prior to use to ensure stable
and reproducible behavior. Furthermore, the sensor inner fi lling solution often contains
relatively high concentrations of the primary ion as well. These issues may prevent
sensors showing Nernstian response slopes to interfering ions due to the release of the
highly preferred ions into the sample-membrane interfacial layer, where these ions
dictate partly or completely the electrode response. However, if fresh sensors, never
having had contact with preferred ions, are used for calibrations towards interfering
ions, a Nernstian sensitivity can be achieved even for strongly discriminated ions.
In order to observe such response slopes the membranes are prepared with the ion
exchanger in the form of one of the interfering ions and sensor conditioning and inner
fi lling solutions are also free of the primary ions. Once the Nernstian response curves
for a series of discriminated ions are recorded, which fulfi lls the requirement for the
Nikolski selectivity coeffi cients, the sensor is reconditioned in a primary ion solution
and the conventional calibration is performed. The calculation of the unbiased selectiv-
ity coeffi cients is then made according to Eq. (7). The proposed method allows one to
signifi cantly widen the applicability of selectivity coeffi cients and, which is also very
important, provides thermodynamically meaningful sensor characteristics. Fortunately,
the unbiased true selectivity coeffi cients often have much more favorable values for the
same sensor compositions. This fact calls for the revision and reconsideration of many
previously proposed ionophores.