24

Design of Biocompatible Ion Sensors

KEIICHI KIMURA Department of Applied Chemistry, Faculty of Systems
Engineering, Wakayama University, Wakayama, Japan

I. INTRODUCTION

Potentiometric ion sensors such as ion-selective electrodes are compact and convenient
tools for ion assay in solution. Specifically, much attention has been focused on neutral-
carrier-type ion sensors owing to their high ion selectivities. Solvent-polymeric membranes
are generally required as their ion-sensing membranes. A most popular ion-sensing mem-
brane material for neutral-carrier-type ion sensors is plasticized poly(vinyl chloride)
(PVC). The solvent-polymeric ion-sensing membranes, however, have some drawbacks.
The ion-sensing membranes need high volume percentages of a special plasticizer, which is
easy to exude from the membranes to aqueous (sample) phases and thereby to deteriorate.
Not only the plasticizers but also the neutral carriers are often toxic, and PVC itself is
becoming an issue in environmental science due to the chlorinated compound. Also,
plasticized-PVC membranes do not seem to be very biocompatible because they tend to
adsorb bio-organic compounds such as proteins on serum metal-ion assay. Sometimes,
neutral-carrier-type PVC-membrane ion-selective electrodes may give some negative errors
on metal-ion assay in biological systems. It is therefore quite difficult to apply ion sensors
based on neutral-carrier-type PVC membranes to in-vivo metal-ion assay such as in-situ
ion monitoring at a bedside. In another aspect, plasticized- PVC membranes are not very
suitable for ion-sensitive field-effect transistors (ISFETs) mainly due to the poor adhesion
to the FET gate surface.
Considerable efforts are therefore required for alleviating the toxicity and promoting
the biocompatibility of the neutral-carrier-type ion sensors, in order to obtain biocompa-
tible ion sensors that are applicable in-vivo ion sensing in biological systems, especially in
clinical systems. One of the efforts is in material selection for the support polymers of the
ion-sensing membranes. That is to say, materials that are compatible with biological
systems such as human blood should be applied as the membrane support. Another effort
is in chemical bonding of active materials such as neutral carriers to the membrane sup-
ports and avoidance of special plasticizers. They would diminish the toxicity of membrane
components considerably by preventing the exudation of neutral carriers from the mem-
branes to aqueous (sample) phases.
This chapter is concerned with the recent research concerning the design of biocom-
patible ion sensors, especially the results of our research group.

585

Copyright © 2001 Marcel Dekker, Inc.

 586 Kimura

II. MEMBRANE MATERIALS SELECTION

A. Poly(oxyethylene)-Containing Polymers
Poly(oxyethylene) (POE) is thought to have some blood compatibility by reducing protein
adsorption [1]. Even PVC, if it contains PEO, shows some effect on alleviating protein
adsorption in blood, and the polymer was therefore applied as an outer membrane for an
enzyme electrode [2]. Poly(urethane) derivatives containing oligo(oxyalkylene) units, which
are called segmented poly(urethane)s, are known to have some nonthrombogenic property,
so their use as a material for artificial human organs was investigated. Crosslinked poly-
ether-urethane was used as a coating agent to obtain blood electrolyte sensors [3]. A pH-
ISFET was mounted on a catheter, which was then dipped into a prepolymer of segmented
poly(urethane) for the coating. The coating polymer can pick up water and Hþ easily, so the
pH-ISFET can work in blood with some blood compatibility. Segmented poly(urethane)
was also applied as a membrane support for ion-selective electrode [4]. A poly(urethane)
prepolymer as well as an ion-sensing material was dissolved in tetrahydrofuran (THF) and
then the THF solution was cast on a glass substrate. The poly(urethane) membranes con-
taining a quaternary ammonium salt and a Naþ ionophore yielded ion-selective electrodes
for Cl and Naþ , respectively. The ion-sensing membranes based on poly(urethane) gave
better nonthrombogenic property than those based on PVC. Surface modifications of
segmented poly(urethane)-based ion-sensing membranes with crosslinked PEO brings
about marked nonthrombogenic property, that is, remarkably alleviated blood platelet
adsorption [5]. Potentiometric ion electrodes based on the membrane composite doped
with valinomycin and quaternary ammonium salts retained good response properties for
Kþ and anion sensors, essentially equivalent to the corresponding untreated membranes.

B. Poly(dimethylsiloxane)
An alternative membrane material to PVC for ion-selective electrodes is crosslinked
poly(dimethylsiloxane), which is generally called room temperature vulcanizing-type
(RTV) silicone rubber. Formation of crosslinked poly(dimethylsiloxane) (RTV silicone
rubber) consists of two processes: (1) the hydrolysis of active end groups (alkoxy, acetoxy,
oxime, amino, amido groups, and so forth) of dimethylsiloxane oligomers, and (2) cross-
linking by condensation of the active end groups with the resulting silanol groups. Silicone
rubber has several advantages: (1) easy fabrication at ambient temperature, (2) high elasticity
at ambient temperature, (3) excellent adhesive property, especially for inorganic materials,
(4) excellent electrical property, (5) chemical inactivity, and (6) low toxicity. Silicone rubber
is, therefore, an excellent candidate for membrane materials of ion sensors. The material was
first applied to heterogeneous precipitate-based membranes for silver ion-selective electrodes
[6]. Since the polymer material is highly elastic at ambient temperature, neutral-carrier-type
ion-sensing silicone-rubber membranes do not necessarily require special plasticizers to
endow a neutral carrier with mobility enough for ion exchanges. This eliminates deteriora-
tion of the sensing membranes caused by the plasticizer exudation. Most importantly, sili-
cone-rubber-membrane ion-selective electrodes are also expected to show some
biocompatibility on practical use in biological systems [7–11]. Moreover, the excellent adhe-
sive property of silicone rubber to inorganic materials is very advantageous on its use as ion-
sensing membranes for ISFET. These characteristics of silicone rubber become great advan-
tages on the application of the silicone-rubber-membrane ion sensors to clinical analyses.
Simple encapsulation or dispersion of neutral carriers in silicone rubber is an easy
way to fabricate silicone-rubber ion-sensing membranes based on neutral carriers.

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 587

However, there seems to be some drawback in the solubility or dispersibility of ion-sensing

material in silicone rubber. This is mainly because silicone rubber does not contain a large
quantity of plasticizer as the membrane solvent, in which neutral carriers can be dissolved
easily, unlike in plasticized-PVC ion-sensing membranes. This issue is serious, especially
with silicone-rubber membranes containing neutral carriers that show high crystallinity.
Valinomycin, a typical Kþ ionophore, seems applicable to silicone-rubber-based Kþ -selec-
tive electrodes [7,8,12–14]. Conventional crown-ether-based neutral carriers are also quite
soluble in silicone rubber.
Calix[4]arene derivatives carrying carbonyl linkages at the lower rim [15] are excellent
neutral carriers for plasticized-PVC membrane Naþ -selective electrodes [16–19]. We have
also designed high-performance Naþ -ISFETs based on calix[4]arene neutral carriers. Most
calix[4]arene neutral carriers are quite soluble in the plasticized membranes, but they are
not necessarily soluble and applicable in silicone rubber. Silicone-rubber-based ion-sensing
membranes possess relatively high electrical resistivity, which often brings about unstable
EMF readings. The issue for the poor dispersibility in membrane is serious especially in ion-
sensing membranes containing electrically neutral ion-exchanger, that is, neutral carriers.
Highly soluble calix[4]arene neutral carriers should, therefore, be designed for the applica-
tion in silicone-rubber membranes. Most calix[4]arene neutral carriers incorporating four
ester or amide linkages show high crystallinity and high melting points because of their
structural symmetry. The symmetrical structure results in comparably high molecular
cohesion, and therefore the calix[4]arene neutral carriers are generally not very easy to
dissolve in polymeric matrices, even in plasticized PVC. The solubility of calix[4]arene
neutral carriers is generally worse in silicone-rubber membranes than in plasticized-PVC
membranes. Incorporation of unsymmetrical structures into calix[4]arene neutral carriers
improves their solubility in silicone rubber. So, an unsymmetrical calix[4]arene neutral
carrier was designed which possesses three ethyl-ester linkages and an amide linkage
with longer alkyl chains at the phenoxy positions, (1) [20]. As compared to a symmetrical
calix[4]arene neutral carrier such as (2), the unsymmetrical one possesses only low symme-
try and is therefore amorphous. Two other calix[4]arene neutral carriers with different
unsymmetrical moieties were also synthesized for comparison with (1). Calixarene neutral
carrier (3), which carries an amide carbonyl linkage without a long aliphatic chain, turns
out to be an amorphous glass by melting, followed by cooling to room temperature.
Compound (4) has a dodecyl group as the long aliphatic chain and an ester carbonyl
linkage and is still unsymmetrical only with the aliphatic chain.

Copyright © 2001 Marcel Dekker, Inc.

 588 Kimura

Figure 1 shows a typical potential response of a Naþ -ISFET based on a silicone-

rubber membrane containing 10 wt% unsymmetrical calix[4]arene neutral carrier (1). A
Nernstian response to Naþ activity changes was attained in the range of 1  104 to 1 M
with the silicone rubber/(1) membrane. The potential response is very fast with a response
time (t90 ) of about 2 s for a tenfold Naþ -activity increase. However, silicone-rubber
membranes of symmetrical calix[4]arene neutral carrier (2), when applied for Naþ -
ISFETs, exhibit only poor response to Naþ activity changes. The silicone-rubber mem-
brane of (2) is very different in the ISFET response from the corresponding plasticized-
PVC membrane. This poor response of the silicone-rubber membrane (2) is derived mainly
from its low solubility in silicone rubber. Scanning electron microscopy (SEM) afforded
some information about the solubility of (2) in silicone rubber. Microcrystals of (2), which
were not very compatible with silicone rubber, were found on the silicone-rubber mem-
brane of (2), while the corresponding (1) membrane possesses a smooth surface. This
implies good compatibility of the unsymmetrical calix[4]arene neutral carrier with silicone
rubber. Calixarene neutral carrier (3) is also quite soluble in silicone rubber. In Naþ -
ISFETs with (3)/silicone-rubber membranes, a Nernstian response was attained in the
range of 1  104 to 1  101 M Naþ , but the sensitivity was decreased more or less at
higher Naþ activities. On the contrary, calixarene neutral carrier (4), although also unsym-
metrical, bears only poor solubility in the polymer matrix. Serious phase separation was
observed in the (4)–silicone-rubber membranes, thus resulting in poor ISFET sensitivity.
The high solubility of (1) in silicone rubber and thereby high ISFET sensitivity are, there-
fore, ascribable mainly to the unsymmetrization of calixarene neutral carriers by the
incorporation of an amide carbonyl group. There also seems to be some contribution of
the unsymmetrization based on the incorporation of a long aliphatic chain to the solubility
enhancement.
The highly dispersible calix[4]arene neutral carriers can also improve the durability
for neutral-carrier-type ion sensors. Time-course changes in both sensitivity (slope for Naþ
calibration graph) and selectivity (selectivity coefficient for Naþ with respect to Kþ ) were
followed in the Naþ -ISFETs based on ion-sensing membranes of silicone rubber–(1),
plasticized PVC–(1), and plasticized PVC–(2) (Fig. 2). Deterioration proceeded quite
quickly in the Naþ -ISFETs of plasticized PVC–(2); both the Naþ sensitivity and selectivity

FIG. 1 Potential response to Naþ activity changes of ISFETs based on silicone-rubber membranes
of calixarene neutral carriers (1)(*), (2) (*), (3) (&), and (4) (~). (From Ref. 20.)

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 589

FIG. 2 Time-course changes of sensitivity (a) and selectivity (b) for Naþ -ISFETs based on silicone
rubber – (1) (90/10 wt%) membrane (*), PVC–DOS–(1) (30/60/10 wt%) membrane (&), and PVC–
DOS–(2) (30/60/10 wt%) membrane (~). (DOS is 2-ethylhexyl sebacate.) (From Ref. 20.)

were drastically decreased in about 30 days. This occurred because the plasticizer and
neutral carrier (2) dissolved out from the membrane into the aqueous phases. In the Naþ -
ISFETs of plasticized PVC–(1) membranes, a marked deterioration started at about 50
days, that is, more gradually than in the plasticized PVC–(2) system. This implies that the
unsymmetrical calix[4]arene (1) is very superior to the symmetrical one (2) in solubility
even in plasticized PVC. Nevertheless, leaching of the plasticizer out from the plasticized
PVC–(1) membrane resulted in such ISFET deterioration. The Naþ -ISFETs containing
silicone rubber–(1) membranes, on the other hand, showed an excellent durability. In these
ISFETs, their high Naþ sensitivity and selectivity remained unchanged even after 120
days.
Silicone rubber is a polymer consisting of siloxane units, as mentioned above. It is
therefore expected that incorporation of oligosiloxane moiety to calix[4]arene neutral
carriers increases their affinity to silicone rubber and thereby to enhance their solubility
or dispersibility in the polymer matrix. Oligosiloxane-modified calix[4]arene neutral car-
riers have been designed which are unsymmetrical and symmetrical ones (5) and (6) [21]. In
the unsymmetrical oligosiloxane-modified calixarene neutral carrier (5), one may also
expect dispersibility enhancement derived from its structural asymmetry. Dispersibility
of oligosiloxane-modified calixarene neutral carriers (5) and (6) in silicone rubber was
first checked by microscopic observation. In the SEM photographs, the silicone-rubber
membranes of (5) exhibit smooth surface, which suggests high dispersibility of the oligo-
siloxane-modified calixarene neutral carriers in silicone rubber. This is also the case with
the (6)-containing silicone-rubber membranes. Membrane impedance (resistivity),
obtained by AC impedance measurements in Naþ aqueous solutions, also provided impor-
tant information concerning the membrane dispersibility of the calixarene neutral carriers

Copyright © 2001 Marcel Dekker, Inc.

 590 Kimura

FIG. 3 Comparison of membrane impedance among silicone-rubber-based ion-sensing membrane

containing calixarene neutral carriers (5) (*), (6) (*), and (2) (~). (From Ref. 21.)

(Fig. 3). The silicone-rubber membranes containing the oligosiloxane-modified calix[4]ar-

ene neutral carriers, especially (5), possess extremely low resistivity as silicone-rubber-type
ion-sensing membranes. On the contrary, high resistivity was found with the membrane of
(2), which is equivalent to or even higher in the resistivity than silicone-rubber membranes
not containing any neutral carrier. This fact strongly supports that both (5) and (6) are
highly dispersible in silicone rubber, thus undergoing active cation diffusion in the mem-
brane. Thus, the Naþ -ISFETs of (5) and (6) exhibit high sensitivity with a Nernstian slope
in a wide Naþ activity range, which is very similar to that for (1) as shown in Fig. 1. The
time–response profile indicated that the potential response is quite fast in the Naþ -ISFETs
based on the oligosiloxane-modified calixarene neutral carriers, and their response time
being about 1 s. The potential response of the Naþ sensors is fast enough for their practical
use. Figure 4 summarizes the selectivity coefficients for Naþ with respect to other ions in
the Naþ -ISFETs based on calixarene neutral carriers (1), (2), and (5). The Naþ sensors

FIG. 4 Selectivity comparison among silicone-rubber-membrane Naþ -ISFETs based on calixarene

neutral carriers (1), (2), and (5).

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 591

based on nonsymmetrical calixarene neutral carrier (1) and oligosiloxane-modified one (5)
are similar to that based on (2) in the Naþ selectivity against Kþ . The selectivity coefficient
for Naþ with respect to Kþ for the ISFET of (1) and (5) is about 3  103 , being sufficient
of Naþ assay in blood serum.
Silicone rubber is also applicable to membrane supports for neutral-carrier-type ion-
selective electrodes. The problem of membrane impedance is, however, more serious in the
ion-selective electrodes than in the ISFETs. High membrane impedance often leads sig-
nificantly to the low sensitivity and slow potential response in the ion-selective electrodes.
The silicone-rubber membrane based on oligosiloxane-modified calix[4]arene neutral car-
rier (5) still exhibits high performance when used for Naþ -selective electrodes [22]. Figure 5
shows a typical potential response for a silicone-rubber-membrane Naþ -selective electrode
based on calixarene neutral carrier (5), together with those for (2) and a well-known Naþ
neutral carrier, bis[(12-crown-4)methyl] dodecylmethylmalonate [bis(12-crown-4)] [23].
The silicone-rubber membrane of (5) afforded high electrode sensitivity with a Nernstian
slope in the wide Naþ activity range of 3  105 to 1 M. On the other hand, both the
silicone-rubber membrane electrodes based on (2) and bis(12-crown-4) possess only poor
sensitivity to Naþ activity changes and even anion response in the high NaCl concentration
range. Obviously, the poor dispersibility of neutral carriers (2) and bis(12-crown-4) in
silicone rubber and therefore the high electrical resistivity of their silicone-rubber mem-
branes lead to the low sensitivity of their Naþ -selective electrodes. A marked difference in
response time was also found among Naþ -selective electrodes with silicone-rubber mem-
branes of the three neutral carriers (Fig. 6). Fast response [2 s of response time (t90 Þ was
attained in the electrodes with silicone-rubber membranes of (5). The electrode response is,
however, sluggish in the silicone-rubber membrane system of (2) and bis(12-crown-4). The
response times in both the silicone-rubber membrane systems of (2) and bis(12-crown-4) are
in the order of 1 min. Thus, high dispersibility of oligosiloxane-modified calixarene neutral
carrier (5) in silicone rubber is reflected more remarkably in the performance of silicone-
rubber membrane Naþ -selective electrodes than the corresponding ISFETs.

C. Sol–Gel-Derived Membranes
Sol–gel chemistry enables us to design hybrid inorganic–organic materials, allowing us to
fabricate glasses and films with novel properties and functionalities [24]. The low reac-

FIG. 5 Potential response of silicone-rubber-membrane Naþ -selective electrodes based on neutral

carriers (5) (*), (2) (*), and bis(12-crown-4) (~). (From Ref. 22.)

Copyright © 2001 Marcel Dekker, Inc.

 592 Kimura

FIG. 6 Time-course changes of potential response for silicone-rubber-membrane Naþ -selective

electrodes based on neutral carriers (5), (2), and bis(12-crown-4) on changing Naþ concentration
from 1  103 to 3  103 M. (From Ref. 22.)

tion temperature on sol–gel processing facilitated incorporation of functional organic

compounds into inorganic materials. Sol–gel-derived membranes are therefore promising
candidates for sensing membrane materials for ion sensors mainly due to their physical
stability and chemical inactivity. Sol–gel-derived membranes encapsulating enzymes have
already been applied to amperometric biosensors [25,26]. For potentiometric ion sensors,
we have tried to apply sol–gel-derived membranes containing valinomycin and a bis(12-
crown-4) derivative to ion-sensing membranes for Kþ - and Naþ -ISFETs, respectively
[27]. Tetraethoxysilane (TEOS) is often used for fabrication of sol–gel-derived materials
by its hydrolysis followed by condensation of its resulted silanols. However, the sol–gel
processing of TEOS itself affords brittle glassy products, which may not be suitable for
ion-sensing membranes due to the poor mechanical strength. An idea to alleviate the
brittleness of sol–gel-derived membranes for neutral-carrier-based ion-sensing mem-
branes is to decrease the crosslinking degree by using mixtures of TEOS and diethox-
ydimethylsilane (DEDMS) as starting alkoxysilanes. Solutions containing DEDMS and
TEOS mixtures with DEDMS/TEOS ratios of 1, 2, and 3 and also a small quantity of
neutral carriers were, therefore, tested for the sol–gel-derived membranes for neutral-
carrier-type ISFETs. It takes too long to solidify the ion-sensing membranes in the
higher DEDMS/TEOS ratios than 3. The potential response to Kþ activity changes
for the Kþ -ISFETs based on valinomycin-encapsulated sol–gel-derived membranes
depends on the DEDMS/TEOS ratios (Fig. 7). Only poor sensitivity was attained in

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 593

FIG. 7 Potential response of Kþ -ISFETs based on valinomycin-encapsulated sol–gel-derived mem-

branes fabricated with initial DEDMS/TEOS ratios of 1 (~), 2 (&), and 3 (*). (From Ref. 27.)

the ISFETs of sol–gel-derived membranes fabricated with the lower DEDMS/TEOS

ratios. On the other hand, the Kþ -ISFETs of the membrane with a DEDMS/TEOS
ratio of 3 can afford a Nernstian response in the wide Kþ -activity range. The poor
sensitivity in the membrane systems with the lower DEDMS/TEOS ratios is probably
due to the high degree of crosslinking in their resulted sol–gel-derived membrane, which
in turn leads to the low mobility of the neutral carrier encapsulated in membrane.
The response time of the ISFETs also depends remarkably on the initial DEDMS/
TEOS ratios for sol–gel processing. The ISFETs with the initial DEDMS/TEOS ratios of
1 exhibited slow potential response and the response time (t90 ) is in the order of minute.
The higher DEDMS/TEOS ratios allowed rapid responses. The response time is ten or
more seconds for the membrane system with the ratio of 2 and only a few seconds for that
with the ratio of 3. The response times for the higher DEDMS/TEOS ratios, especially
with the ratio of 3, are acceptable for practical use of ion sensors. The initial DEDMS/
TEOS ratio of 3, therefore, seems to be suitable for the fabrication of sol–gel-derived
membranes encapsulating valinomycin, at the viewpoint of the membrane mechanical
strength, sensor sensitivity, and response time.
Sol–gel-derived membranes encapsulating a bis(crown ether) derivative [bis(12-
crown-4-ylmethyl) 2-dodecyl-2-methylmalonate] [28] were also fabricated with an initial
DEDMS/TEOS ratio of 3 for Naþ -ISFETs. The Naþ -ISFETs showed a Nernstian
response to Naþ activity changes in the activity range of 1  104 to 1 M and a short
response time of 2 s. Employment of a polythiophene interlayer improved the potential
instability and the lower detection limit in both of the Kþ - and Naþ -ISFETs based on the
sol–gel-derived membranes with the initial DEDMS/TEOS ratio of 3 (Fig. 8).
Selectivity coefficients values for Kþ - and Naþ -ISFETs with the optimized ion-sen-
sing membranes encapsulating valinomycin and bis(12-crown-4) are summarized in Fig. 9.
The selectivity coefficient for Kþ with respect to Naþ in the Kþ -ISFET is 2  104 and
that for Naþ with respect to Kþ in the Naþ -ISFET is 3  103 . The selectivity coefficient
values are similar to those for the ISFETs and ion-selective electrodes with the previous
membrane materials containing the same neutral carriers. The high sensitivity and selec-
tivity for the neutral-carrier-type ISFETs based on sol–gel-derived membranes can last for
at least 3 weeks.

Copyright © 2001 Marcel Dekker, Inc.

 594 Kimura

FIG. 8 Potential response of Kþ - and Naþ -ISFETs based on optimized sol–gel-derived membranes
of valinomycin (*) and bis(12-crown-4) (*) with interlayer of polythiophene [27]. The initial
DEDMS/TEOS ratio for sol–gel processing is 3.

Adsorption of blood proteins on the sol–gel-derived membrane containing a neutral

carrier was tested to obtain some information about biocompatibility of the ion-sensing
membranes and therefore the resulted ISFETs on blood ion assay. For comparison,
plasticized-PVC membranes containing the same neutral carrier were also employed for
the biocompatibility check. Comparison in the SEM photographs between a sol–gel-
derived membrane and its corresponding plasticized-PVC membrane indicates a striking
difference in the blood protein adsorption (Fig. 10). The sol–gel-derived membrane hardly
adsorbs blood proteins, whereas the plasticized-PVC one adsorbs a large amount of
proteins. This suggests that the sol–gel-derived ion-sensing membranes are superior to
the corresponding plasticized-PVC membranes in the biocompatibility. The adsorption
of organic components in blood on plasticized-PVC ion-sensing membranes often causes
some negative errors on blood ion assay by the resulted membrane ion sensors [8,29]. The
serious protein adsorption also makes it difficult to use plasticized-PVC membrane ion
sensors for in-vivo ion monitoring. Conversely, the alleviated protein adsorption on the

FIG. 9 Selectivity coefficients for Kþ - and Naþ -ISFETs based on optimized sol–gel-derived mem-
branes of valinomycin (a) and bis(12-crown-4) (b) with polythiophene interlayer. (From Ref. 27.)

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 595

FIG. 10 SEM photographs for sol–gel-derived (a) and plasticized-PVC (b) membranes encapsulat-
ing bis(12-crown-4) after protein adsorption test. (From Ref. 27.)

sol–gel-derived membrane suggests a possibility for intra-arterial use of the neutral-car-

rier-type ISFETs based on the membrane.

D. Other Biocompatible Materials

Heparin is often utilized for preventing blood coagulation. When ion-selective electrodes
are used in vivo, supplying heparin to the bloodstream suppresses fibrin formation on the
surface of the ion-sensing membranes [9]. The heparin treatment to patients on in-vivo
measurements, is however, a serious disadvantage. This induced the covalent bonding of
heparin to the ion-sensing membrane surface [30]. Heparin was bonded covalently to the
surface of cellulose triacetate (CTA) membrane effectively by carbonyldiimidazole activa-
tion of the cellulose hydroxyl groups. The heparin-immobilized membrane, which is bio-
logically active, was combined with an unmodified CTA membrane containing
valinomycin. The heparin modification by covalent bonding does not affect significantly
the original electrode property such as sensitivity and selectivity for the Kþ -selective
electrode.
Nitric oxide is recently becoming well known due to its biological action. Nitric
oxide exhibits antiaggregation property of blood platelet [31] as well as vasodilating
property [32]. Therefore, if ion-sensing membranes can emit low levels of nitric oxide
gradually, the ion sensor must be blood-compatible. For instance, piperizine–PVC copo-
lymer/N2 O2 (diazeniumdiolate) adducts can produce NO gas by its hydrolysis. The NO-
evolving polymer was dispersed as a solid particle in plasticized PVC or segmented poly(-
urethane) membranes containing tridodecylamine or valiomycin, in order to obtain blood-
compatible pH- or Kþ -sensing membranes [33]. The ion-sensing membranes are capable of
generating NO gas for several days on exposing aqueous samples and thereby hardly
adsorb blood platelets. As a result, excellent biocompatibility was attained for the ion-
selective electrodes without affecting their sensor performance. Alternatively, a low-mole-
cular-weight NO-evolving compound, N; N 0 -dimethylhexanediamine/NO adduct, was dis-
solved in plasticized-PVC membranes for providing blood compatibility [34,35]. The ion-
sensing membranes were fabricated as a multilayer film; an membrane containing the NO-
evolving material was sandwiched by two unmodified ion-sensing films containing Hþ or
Kþ carrier. NO generation by the hydrolysis of the material with water is slower and
therefore the blood compatibility can last a long time.

Copyright © 2001 Marcel Dekker, Inc.

 596 Kimura

Recently, remarkably high blood compatibility was realized in polymers containing

phosphorylcholine groups at the side chain (MPC) polymer), which are generally copoly-
mers of methacryloyloxyethyl phosphorylcholine and an alkyl methacrylate [36]. It was
proved that they possess higher thrombogenic property as compared to conventional
blood-compatible materials such as poly(dimethylsiloxane) and poly(urethane). They are
now being tested for their usefulness as artificial internal organs [37]. The MPC polymer is
also a material candidate for biocompatible ion sensors. MPC polymers were applied as
blood-compatible layers for amperometric sensors [38,39]. Enzyme-based glucose sensors
generally have a membrane on which glucose oxydase is immobilized on an H2 O2 ampero-
metric electrode. On glucose assay in blood samples, the enzyme membranes are apt to
adsorb proteins. The protein adsorption on the enzyme membrane often results in the
sensor deterioration. Coating of the enzyme membrane with an MPC polymer prevents
protein adsorption and thereby increases the sensor lifetime. We have applied MPC poly-
mers for biocompatible potentiometric ion sensors. Modification of neutral-carrier-based
PVC membranes by an MPC polymer has improved thrombogenic properties markedly,
with the original sensor property of the PVC membrane remaining unchanged [40].

III. CHEMICAL MODIFICATION WITH ION-SENSING COMPONENTS

A. PVC-Based Membrane
The exudation of neutral carriers and plasticizers from ion-sensing membranes results in
deterioration of the ion sensors and contamination in environmental and biological sys-
tems. In order to prevent the outflow of neutral carriers from ion-sensing membranes,
attempts have been made to immobilize neutral carriers to PVC membranes by covalent
bonding [41–43]. However, unless the neutral-carrier-type PVC membranes are plasticized
like the conventional PVC-based ion-sensing membranes, they do not seem to show very
good electrode properties, i.e., non-Nernstian response and slow potential response. It is
probably because PVC and the chemically modified PVC themselves do not allow enough
mobility for the chemically bonded neutral carriers to exchange cations due to their high
glass-transition temperatures (at least higher than room temperature). High contents of
membrane plasticizers in the PVC-based ion-sensing membranes may also lead to the
toxicity and deterioration of the resulting ion sensors due to their exudation to aqueous
sample solutions. It is still difficult for the plasticized-PVC membranes modified chemi-
cally with neutral carriers to be biocompatible.

B. Poly(dimethylsiloxane)
Poly(dimethylsiloxane) (silicone rubber) membranes are fabricated by hydrolysis of alkox-
ysilyl terminal groups of silicone-rubber precursors [oligo(dimethylsiloxane) derivatives
and crosslinking agents], followed by condensation. Covalent bonding of neutral carriers
carrying an alkoxysilyl group to silicone rubber is, therefore, feasible by simple reaction of
the silicone-rubber precursor with alkoxysilylated neutral carriers, as schematically shown
in Scheme 1 [44].
As neutral carriers for the chemical modification, 16-crown-5 and calix[4]arene deri-
vatives possessing a triethoxysilyl group (7) and (8) were designed for Naþ sensors.
Triethoxysilylethyl-16-crown-5(7) was then mixed with a silicone-rubber precursor for
the membrane fabrication accompanying covalent bonding of the neutral carrier.
Comparison of IR spectra before and after extraction of the nonbonded neutral carrier

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 597

SCHEME 1 Mechanism for chemical modification of poly(dimethysiloxane) (silicone rubber)

membranes with ion-sensing active material.

with benzene confirmed the covalent bonding of the 16-crown-5 derivative to silicone-
rubber membranes. In the IR spectrum for the silicone-rubber membranes fabricated
using the ethoxysilylated 16-crown-5, a significant peak assigned to C–H deformation
vibration for its 16-crown-5 ring was observed around 1350 cm1 even after benzene
extraction of (7) from the membranes for a whole day. It is obvious that a considerable
amount of 16-crown-5 moiety was bonded covalently to the silicone-rubber membranes
fabricated using (7). IR spectroscopy of the silicone-rubber membranes obtained using an
alkoxysilylated calixarene (8) also supported the covalent bonding of the calix[4]arene
neutral carrier to silicone-rubber membranes.

Copyright © 2001 Marcel Dekker, Inc.

 598 Kimura

The Naþ -selective electrodes of the silicone-rubber membranes containing (8) with-
out any other additive exhibited a very curious potential response, especially in the Naþ
activity changes between 1  105 to 1  102 M (Fig. 11). It is probably because the
chemical modification of silicone-rubber membranes with a calixarene neutral carrier
results in high membrane impedance. Sodium tetrakis[3,5-bis(trifluoromethyl)phenyl]bo-
rate (TFPB) was, therefore, used to improve the electrode response. The TFPB addition to
the silicone-rubber membranes reduces the membrane impedance to about one-third of the
initial values, thus the membranes allow high sensitivity with a Nernstian slope in the Naþ
activity range of 1  105 to 1 M. TFPB is still easy to dissolve out from the membrane
because it is not bonded covalently to silicone rubber. For further durability enhancement
and toxicity alleviation of the ion-sensing membranes, sodium triphenyl-4-[3-triethoxysi-
lyl)propyloxy]phenylborate (9), together with the calixarene neutral carrier, was also used
for the immobilization of an anion excluder on silicone-rubber membranes by covalent
bonding. A decreased membrane impedance was again attained in the silicone-rubber
membranes incorporating both calixarene neutral carrier (8) and anion excluder (9).
The potential response profile for the membrane system containing both (8) and (9) is
quite similar to that for the conventional membrane system of calixarene neutral carriers
and TFPB that are not bonded covalently, showing a Nernstian response to Naþ activity
changes in the wide activity range. Response time was also improved with Naþ -selective
electrodes based on the silicone-rubber membranes modified chemically by calixarene
neutral carrier (8) and an anion excluder TFPB or alkoxysilylated sodium tetraphenylbo-
rate (9). Another way to immobilize neutral carriers to silicone rubber by covalent bond-
ing may be copolymerization of a vinyl-modified polysiloxane and neutral-carrier vinyl
monomers induced by UV-light irradiation and the resulting neutral-carrier-based mem-
branes were applied to Kþ - and Naþ -sensitive field-effect transistors (ISFETs) [45–47].
The neutral-carrier-modified silicone-rubber membranes often experience anomalous elec-
trode response, unless the ion-sensing membrane contains a lipophilic salt to overcome
their high membrane impedance. The design of an oligomeric neutral carrier with a med-
ium molecular weight and its blending with a silicone rubber might alleviate the demerits
of the silicone-rubber membranes modified chemically by a neutral carrier. The medium-
molecular-weight neutral carrier may improve the poor mobility of the covalently bonded

FIG. 11 Potential response of Naþ -selective electrodes based on silicone-rubber membranes mod-
ified chemically by triethoxysilylated calix[4]arene (8) (*) without anion excluder; (*) with TFPB;
(&) modified chemically by triethoxysilylated tetraphenylborate (9) as well. (From Ref. 44.)

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 599

neutral carriers and its outflowing from the membrane owing to its relatively high mole-
cular weight. For obtaining a medium-molecular-weight calixarene neutral carrier,
calix[4]arene tetra-allyl ester was incorporated into an oligo(methylhydrosiloxane) to
yield oligosiloxane-modified calix[4]arene neutral carrier (10) [48]. The oligosiloxane back-
bone is expected to enhance the solubility or dispersibility of the resulting calixarene
neutral carrier in silicone-rubber membranes containing dimethylsiloxane repeating
units. Calixarene oligomer (10) carries two or three calixarene units and possesses a
molecular weight of several thousands. The Naþ -selective electrodes based on silicone-
rubber membranes containing the oligomeric calixarene neutral carrier (10) in 10 wt%
exhibit a Nernstian response to tenfold Naþ activity changes in the wide activity range.
Such a curious potential response as shown in Fig. 11 was not found in the silicone-rubber
membranes containing the oligomeric calixarene neutral carrier (10). This means that the
calixarene moiety itself is more mobile and it is easier to exchange Naþ in the silicone-
rubber membranes encapsulating an oligomeric calixarene neutral carrier than in those
modified chemically with a calixarene moiety. There is another significant difference in the
sensor response time between the membranes containing oligomeric calixarene neutral
carrier (10) and incorporating a calixarene moiety by chemical bonding. The response
time (t90 ) is no shorter than 3 min in the system of the calixarene-bound membranes. In
the present membrane system containing oligomeric calixarene neutral carrier (10), on the
other hand, the response time is below 1 min. Thus, the employment of the silicone-rubber
membranes containing oligomeric calixarene neutral carrier (10) instead of the calixarene-
bound membranes can improve the electrode potential response and the response time
owing to its enhanced mobility of the calixarene moiety in the membranes. Of course, the
oligomeric calixarene neutral carrier (10) is much more insoluble in water than monomeric
calix[4]arene neutral carriers and is still not easy to exude from the membrane to aqueous
phases.

Copyright © 2001 Marcel Dekker, Inc.

 600 Kimura

C. Sol–Gel-Derived Membranes
Appropriate fabrication of sol–gel-derived membranes encapsulating neutral carriers such
as valinomycin can afford an excellent type of neutral-carrier-type ion-sensing membranes
for ISFETs [27] as already mentioned. The simple encapsulation of neutral carriers in sol–
gel-derived membranes, however, has a drawback: the encapsulated neutral carriers are
still apt to exude from the membranes into aqueous sample solutions, which thereby
makes the resulting ion sensors less durable and more toxic. Incorporation of neutral
carriers to sol–gel-derived membranes by covalent bonding is desirable.
Sol–gel-derived membranes are generally made by hydrolysis of an alkoxysilane such
as tetraethoxysilane (TEOS), followed by condensation of the resulted silanol. So, an easy
way to incorporate a neutral carrier to sol–gel-derived membranes by covalent bonding is
sol–gel processing of the alkoxysilane together with an alkoxysilylated neutral carrier [49].
Triethoxysilylated 16-crown-5 and bis(12-crown-4) derivatives (7) and (11) were selected as
the crown-ether neutral carriers. The corresponding triethoxysilylated derivative of
sodium tetraphenylborate (9) was chosen as an anion excluder. Since sol–gel-derived
membranes made from solutions containing only TEOS are very brittle [27] sol–gel-
derived ion-sensing membranes were fabricated by using solutions containing TEOS
and diethoxydimethylsilane (DEDMS) with a 1:3 ratio. IR-spectral measurements indi-
cated that the neutral carrier and anion excluder can be covalently bonded to the mem-
branes to a great extent, by the present sol–gel processing using the alkoxysilyl-substituted
neutral carrier and anion excluder.
Typical profiles for the potential response of Naþ -ISFETs based on the sol–gel-
derived membranes modified chemically by alkoxysilylated 16-crown-5 (7) and bis(12-
crown-4) (11), together with anion excluder (9) show that both of the Naþ -ISFETs
have high sensitivity with a Nernstian response to Naþ activity changes in wide activity
ranges of 3  105 to 1 M (Fig. 12). The potential response is quite fast in the Naþ -ISFETs
in spite of the covalent bonding of the neutral carriers to the ion-sensing membranes, as
exemplified in the membrane system of the bis(12-crown-4) (11). Some mobility of the
chemically bonded neutral carriers can probably be maintained in the present sol–gel-
derived membranes. The response time (t90 Þ is several seconds for both of the Naþ -

FIG. 12 Potential response to Naþ activity changes of Naþ -ISFETs based on sol–gel-derived
membranes modified chemically by alkoxysilyl-substituted neutral carriers (7) (*) and (8) (*),
together with anion excluder (9). (From Ref. 49).

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 601

ISFETs based on the sol–gel-derived membranes of (7) and (11), being almost comparable
to that for similar ISFETs based on sol–gel-derived membranes simply encapsulating
bis[(12-crown-4-)ylmethyl]2-dodecyl-2-methylmalonate [27].
Selectivity coefficients for Naþ with respect to interfering cations which exist in
biological and environmental systems were determined for the Naþ -ISFETs based on
sol–gel-derived membranes incorporating neutral carriers (7) and (11) by covalent bonding
(Fig. 13). In the ISFETs of the 16-crown-5 membrane system, the Naþ selectivities against
Hþ , Liþ , NH4 þ , Mg2þ , and Ca2þ are quite good, but the selectivity against Kþ is not high
enough, especially for Naþ assay in biological systems such as blood. On the contrary,
high Naþ selectivity against Kþ as well as the other interfering ions was found in the
sensor of the bis(crown ether) membrane system. This ion sensor is of practical use in
biological systems from the standpoint of the ion selectivity.
An advantage of the ion-sensing membranes modified chemically by a neutral carrier
and an anion excluder over those simply encapsulating them is high durability of the ion-
sensing membranes and therefore of the resulted ion sensors. In the ISFETs of the mem-
branes simply encapsulating bis[(12-crown-4-)ylmethyl]2-dodecyl-2-methylmalonate and
NaTFPB, an abrupt decrease in the sensitivity was observed after 30 days (Fig. 14),
probably due to their easy extrusion from the membrane to sample solutions. In the
ISFETs of the membranes modified chemically by (11) and (9), on the other hand, the
high sensitivity of a Nernstian response was kept even after 100 days. The long lifetime for
the ISFETs of the membranes modified chemically by the neutral carrier and anion
excluder is desirable on practical applications. The high sensor durability also means
that the active materials for ion-sensing membranes hardly exude out from the membrane
to aqueous sample solutions during ion assay. This also implies the low toxicity of the ion
sensors for use in biological systems.
Quaternary alkylammonium salts are generally water-soluble surfactants. The sol–
gel-derived anion-sensing membranes encapsulating a quaternary alkylammonium salt,
especially with high contents, are easy to deteriorate due to the exudation of the cationic
site from the membrane to aqueous sample phases. Moreover, another issue concerning
the dispersibility of ammonium salts in sol–gel-derived membranes may happen when high

FIG. 13 Selectivity comparison between Naþ -ISFETs based on sol–gel-derived membranes mod-
ified chemically by alkoxysilyl-substituted neutral carriers (7) and (11). (From Ref. 49.)

Copyright © 2001 Marcel Dekker, Inc.

 602 Kimura

FIG. 14 Time-course changes of sensitivity (slope of calibration plots) for Naþ -ISFETs based on
sol–gel-derived membranes incorporating bis(12-crown-4) by covalent bonding using (11) (*) and
by encapsulation of bis[(12-crown-4-)ylmethyl] 2-dodecyl-2-methylmalonate (*). (From Ref. 49.)

contents of quaternary alkylammonium salt are applied to membranes. Chemical mod-

ification of sol–gel-derived membranes by an alkoxysilylated quaternary ammonium
chloride (12) was therefore attempted to improve the durability and dispersibility of
ammonium salts in the membranes [50]. The alkoxysilylated derivative of quaternary
ammonium salt can take part in the sol–gel reaction to allow the covalent bonding of
its ammonium sites to sol–gel-derived membranes. For the sol–gel-derived membranes
modified chemically by the anion exchanger, the initial TEOS/DEDMS ratio of 1:1 and
(12) content of 30 wt% are the best with respect to the sensor sensitivity and membrane
stability. The ISFETs exhibit a Nernstian response in the wide Cl -activity range (Fig. 15).
The high sensitivity is comparable to the anion sensors based on that of the corresponding
membranes simply encapsulating trioctylmethylammonium chloride. The response for the

FIG. 15 Potential response of anion ISFETs based on sol–gel-derived membranes encapsulating

trioctylmethylammonium chloride (*) and modified chemically by alkoxysilylated quaternary
ammonium salt (12) (*) to CI -activity changes. (From Ref. 50.)

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 603

anion sensors with sol–gel-derived membranes of (12) is somewhat slower as compared to

the sensors of quaternary-ammonium-encapsulating membranes. The response time of
>10 s, however, still allows practical use of the sensors based on the chemically modified
membranes without difficulty. The anion selectivity order for the ISFETs of (12) also
follows the Hofmeister series, being the same as those of quaternary-ammonium-encap-
sulating membranes (Fig. 16). The selectivity difference among the anions seems to be
more remarkable in the former anion sensors than the latter ones. SEM photographs
showed that the anion-sensing sol–gel-derived membranes modified chemically by (12),
are superior to the corresponding membranes simply encapsulating trioctylmethylammo-
nium chloride in the dispersibility of quaternary ammonium salts in membranes and the
protein-adsorption property.

IV. APPLICATIONS TO BIOLOGICAL ION ASSAY

A. Poly(dimethylsiloxane)
Applicability in biological ion assay is an important factor for biocompatible potentio-
metric ion sensors. Attempts were made to determine Naþ concentrations in human
blood sera by using silicone-rubber membrane Naþ -ISFETs based on (5) [Fig. 17(a)]
[29]. The found values for Naþ concentration in undiluted, 10-fold diluted, and 100-
fold diluted serum samples are in good agreement with the Naþ calibration plots. Even
in the undiluted serum samples, only a slight potential shift was observed from the
calibration. This indicates that the calixarene-based silicone-rubber-membrane Naþ -
ISFETs are reliable on serum Naþ assay. For comparison with the silicone-rubber
membrane, Naþ -ISFETs with corresponding plasticized-PVC membrane containing
(2) or (5) were also tested for the Naþ assay. The found values of Naþ concentration

FIG. 16 Selectivity coefficients for CI with respect to other anion ISFETs based on sol–gel-
derived membranes (a) encapsulating trioctylmethylammonium chloride (b) modified chemically
by (12). (From Ref. 50.)

Copyright © 2001 Marcel Dekker, Inc.

 604 Kimura

FIG. 17 Naþ calibration graphs and Naþ concentrations in sera obtained by Naþ -ISFETs based
on (a) silicone-rubber membrane of calixarene (5) and (b) plasticized-PVC membrane of (2). (*)
Naþ calibration plots: (~) found values for Serum I (145 mM Naþ and 4.1 mM Kþ ); (&) found
values for Serum II (124 mM Naþ and 6.5 mM Kþ ). (From Ref. 29.)

in the serum samples, obtained by the PVC-membrane Naþ -ISFET, deviate consider-
ably from the Naþ calibration [Fig. 17(b)]. The negative potential shift from the Naþ
calibration was observed in any of the serum samples. However, there is hardly any
significant deviation in the Naþ -concentration found value for the dummy samples
containing the same concentrations of Naþ and Kþ without any organic components
such as proteins. Obviously, the negative potential shift of the serum samples from the
calibration graphs is not ascribable to interference by Kþ and by organic components
(proteins) in sera. Also, for the Naþ assay in human urine samples, the calixarene-
based silicone-rubber membranes are superior to the corresponding plasticized-PVC
ones as ion-sensing membranes for Naþ -ISFETs. Thus, the silicone-rubber membrane
Naþ -ISFETs based on calixarene neutral carrier (5) are quite useful for Naþ assay in
human sera and urine, whereas the corresponding plasticized-PVC membrane ISFETs
are likely to cause significant errors on the assay.
Adsorption of blood protein on the silicone-rubber membranes containing a neutral
carrier was tested to obtain some information about biocompatibility of the ion-sensing
membranes and therefore the resulted ISFETs on Naþ assay in blood. Comparison in the
SEM photographs for a silicone-rubber membrane and its corresponding plasticized-PVC
membrane indicates a striking difference in the protein adsorption. The silicone-rubber
membrane hardly adsorbs blood proteins, whereas a large amount of proteins can be seen
on the plasticized-PVC membrane. This indicates that the silicone-rubber ion-sensing
membranes are superior to the corresponding plasticized-PVC membranes in the blood
compatibility. The adsorption of blood organic components on plasticized-PVC ion-sen-
sing membranes often causes some negative errors on blood Naþ assay by the resulted
membrane ion sensors. The serious protein adsorption also makes it difficult to use plas-
ticized-PVC membrane ion sensors for in-vivo ion monitoring. On the contrary, this is not

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 605

the case with the silicone-rubber membranes that can alleviate markedly the protein
adsorption.
The Naþ -selective electrodes based on silicone-rubber membranes modified chemi-
cally by (8) and (9), were also investigated for Naþ assay in control serum and urine [22].
The found values for the Naþ concentrations in both of the serum and urine samples are in
good agreement with their corresponding actual values with a relative standard deviation
of about 1%. These results suggest that the Naþ -selective electrodes based on silicone-
rubber membranes modified chemically by calix[4]arene neutral carrier (8) are reliable on
assay in human body fluid.

B. Sol–Gel-Derived Membranes
The sol–gel-derived membrane ISFETs based on bis(12-crown-4) and valinomycin were
applied for Naþ and Kþ assay in blood sera [27]. For reproducibility checking, the metal-
ion assay was carried out ten times for each sample, using two different control sera. The
ISFETs gave reliable data both in Naþ and Kþ assay. The coefficient of variation ranges
between 0.45 and 0.78%, thus indicating excellent reproducibility in the serum cation
assay with the ISFETs of neutral-carrier-type sol–gel-derived membranes. The results
for real blood sera for normal human for the cation assay showed relative errors within
2% for the Naþ assay and within 4% for the Kþ assay.
The sol–gel-derived membranes incorporating the alkoxysilylated bis(crown ether)
(11) by covalent bonding were tested for their thrombogenic property. Any significant
adsorption of blood platelets was hardly observed on the membranes, as is the case with
sol–gel derived membranes encapsulating a neutral carrier [27]. Such excellent thrombo-
genic property of the sol–gel-derived membranes modified chemically by a bis(crown
ether) suggests a high possibility of their application to Naþ assay in blood sera. The
results for Naþ assay in human blood serum with the Naþ -ISFETs are given in Fig. 18,
showing that the found values for Naþ concentrations are in good agreement with the
corresponding actual values.

FIG. 18 Correlation between actual and found values of Naþ concentrations on serum assay with
ISFETs based on sol–gel-derived membranes of bis(12-crown-4) (11). (From Ref. 49.)

Copyright © 2001 Marcel Dekker, Inc.

 606 Kimura

V. CONCLUSION

The design of biocompatible (blood compatible) potentiometric ion sensors was described
in this chapter. Sensing membranes fabricated by crosslinked poly(dimethylsiloxane) (sili-
cone rubber) and sol–gel-derived materials are excellent for potentiometric ion sensors.
Their sensor membrane properties are comparable to conventional plasticized-PVC mem-
branes, and their thrombogenic properties are superior to the PVC-based membranes.
Specifically, membranes modified chemically by neutral carriers and anion excluders are
very promising, because the toxicity is alleviated drastically. The sensor properties are still
excellent in spite of the chemical bonding of neutral carriers on membranes.
The biocompatible potentiometric ion sensors have been successfully applied for ion
assay in biological systems such as human blood. They might be used in in-vivo cation
assay in biological systems such as intra-arterial assay in the near future.

REFERENCES

1. Y. Mori, S. Nagaoka, H. Takiuchi, N. oguchi, H. Tanzawa, and Y. Noishiki. Trans. Am. Soc.
Artif. Intern. Organs 28:459 (1982).
2. M. Kyrolainen, S. M. Reddy, and P. M. Vadgama. Anal. Chim. Acta 353:281 (1997).
3. M. Karakelle and R. J. Zdrahala, U. S. Patent 4,816,130 (1989).
4. K. Hiiro, T. Tanaka, A. Kawahara, S. Wakita, M. Yamane, N. Yamamoto, and I. Yamashita,
Japanese Patent 63,229,356 (1995).
5. C. Espadas-Torre and M. E. Meyerhoff. Anal. Chem. 67:3108 (1995).
6. E. Pungor, J. Havas, K. Toth, and G. Mandarasz, French Patent 140,235 (1965).
7. W. Crowe, A. Mayevsky, and L. Mela. Am. J. Physiol. 233:C56 (1977).
8. H.-B. Jenny, C. Riess, D. Ammann, B. Magyar, R. Asper, and W. Simon. Mikrochim. Acta,
1980 II: 309 (1980).
9. M. E. Collison, G. V. Aebli, J. Petty, and M. E. Meyerhoff. Anal. Chem. 61:2365 (1989).
10. M. Mascini and G. Marrazzaa. Anal. Chim. Acta, 231:125 (1990).
11. G. Marrazza and M. Mascini. Electroanalysis 4:41 (1992).
12. J. Pick, K. Toth, and E. Pungor. Anal. Chim. Acta 64:477 (1973).
13. L. A. Mostert, P. Anker, H.-B. Jenny, U. Oesch, W. E. Morf, D. Ammann, and W. Simon.
Mikrochim. Acta 1885 I: 35 (1985).
14. O. H. LeBlanc and W. T. Grubb. Anal. Chem. 48: 1658 (1976).
15. A. Arduini, A. Pochini, S. Reverberi, R. Ungaro, G. D. Andreetti, and F. Ugozzoli.
Tetrahedron 42:2089 (1986).
16. D. Diamond, G. Svehla, E. M. Seward, and M. A. McKervey. Anal. Chim. Acta 204:223
(1988).
17. K. Kimura, T. Miura, M. Matsuo, and T. Shono. Anal. Chem. 62:1510 (1990).
18. R. J. Forster, A. Cadogan, M. T. Diaz, D. Diamond, S. J. Harris, and M. A. McKervey. Sens.
Actuators B 4:325 (1991).
19. F. Arnaud-Neu, G. Barrett, S. Cremin, M. Deasy, G. Ferguson, S. J. Harris, A. J. Lough, L.
Guerra, M. A. McKervey, M. J. Schwing-Weill, and P. Schwinte. J. Chem. Soc. Perkin Trans.
2 1992:1119 (1992).
20. K. Kimura, T. Matsuba, Y. Tsujimura, and M. Yokoyama. Anal. Chem. 64:2508 (1992).
21. Y. Tsujimura, M. Yokoyama, and K. Kimura. Electroanalysis 5:803 (1993).
22. Y. Tsujimura, M. Yokoyama, and K. Kimura. Anal. Chem. 67:2401 (1995).
23. T. Shono, M. Okahara, I. Ikeda, K. Kimura, and H. Tamura. J. Electroanal. Chem. Interfacial
Electrochem. 132:99 (1982).

Copyright © 2001 Marcel Dekker, Inc.

 Design of Biocompatible Ion Sensors 607

24. L. L. Hench and J. K. West, Chemical Processing of Advanced Materials, Wiley, New York,
1992.
25. B. C. Dave, B. Dunn, J. S. Valentine, and J. I. Zink. Anal. Chem. 22:1120A (1994).
26. J. Wang, P. V. A. Pamidi, and K. R. Rogers. Anal. Chem. 70:1171 (1998).
27. K. Kimura, S. Sunagawa, and M. Yokoyama. Anal. Chem. 69:2379 (1997).
28. K. Kimura, M. Yoshinaga, K. Funaki, Y. Shibutani, K. Yakabe, T. Shono, M. Kasai, H.
Mizufune, and M. Tanaka. Anal. Sci. 12:67 (1996).
29. Y. Tsujimura, M. Yokoyama, and K. Kimura. Sens. Actuators B 22:195 (1994).
30. K. A. Brooks, J. A. Allen, P. W. Feldhoff, and L. G. Bachas. Anal. Chem. 68:1439 (1996).
31. B. T. Mellion, L. J. Ignarro, E. H. Ohlstein, E. G. Pontecorvo, A. L. Hyman, and P. J.
Kadowitz. Blood 57:946 (1981).
32. M. W. Radomski, M. J. Palmer, and S. Moncada. Br. J. Pharmac. 92:639 (1987).
33. K. A. Mowery, M. H. Schoenfisch, and M. E. Meyerhoff. Polym. Mater. Sci. Eng. 79:33 (1998).
34. C. Espadas-Torre, V. Oklejas, K. Mowery, and M. E. Meyerhoff. J. Am. Chem. Soc. 119:2321
(1997).
35. K. A. Mowery, C. Espadas-Torre, V. Oklejas, and M. E. Meyerhoff. Polym. Mater. Sci. Eng.
76:562 (1997).
36. K. Ishihara, N. Nakabayashi, K.Nishida, M. Sakakida, and M. Schichiri. Chemtech 1993:19
(1993).
37. K. Ishihara and N. Nakabayashi. Artificial Organs. 19:1215 (1995).
38. C.-Y. Chen, E. Tamiya, K. Ishihara, Y. Kossugi, Y.-C. Su, N. Nakabayashi, and I. Karube.
Appl. Biochem. Biotechnol. 36:211 (1992).
39. K. Nishida, M. Sakakida, Y. Hashiguchi, M. Uehara, T. Uemura, T. Kajiwara, M. Schichiri,
K. Ishihara, and N. Nakabayashi. Jinko Zoki 22:1090 (1993).
40. S. Yajima, K. Suzuki, and K. Kimura, 6th National Meeting of the Chemical Society of Japan,
Abstr. No. 3E215, 1999.
41. J. Tietje-Girault, I. MacInnes, M. Schroder, G. Tennant, and H. H. Girault. Electrochim. Acta
35:777 (1990).
42. S. Daunert, and L. G. Bachas. Anal. Chem. 62:1428 (1990).
43. G. G. Cross, T. M. Fyles, and V. V. Suresh. Talanta 41:1589 (1994).
44. Y. Tsujimura, T. Sunagawa, M. Yokoyama, and K. Kimura. Analyst (London) 121:1705
(1996).
45. D. N. Reinhoudt, J. F. J. Engbersen, Z. Brzozka, H. H. v. d. Vlekkert, G. W. N. Honig, H. A.
J. Holterman, and U. H. Verkerk. Anal. Chem. 66:3618 (1994).
46. J. A. J. Brunink, R. J. W. Lugtenberg, Z. Brzozka, J. F. J. Engbersen, and D. N. Reinuoudt. J.
Electroanal. Chem. 378:185 (1994).
47. R. J. W. Lugtenberg, R. J. M. Egbernk, A. v. d. Berg, J. F. J. Engbersen, and D. N. Reinhoudt.
J. Electroanal. Chem. 452:69 (1998).
48. K. Kimura, Y. Tsujimura, M. Yokoyama, and T. Maeda. Bull. Soc. Chem. Jpn. 71:657 (1998).
49. K. Kimura, T. Sunagawa, S. Yajima, S. Miyake, and M. Yokoyama. Anal. Chem. 70:4309
(1998).
50. K. Kimura, H. Takase, S. Yajima, and M. Yokoyama. Analyst (London) 124:517 (1999).

Copyright © 2001 Marcel Dekker, Inc.