Minimally invasive silicon probe for electrical impedancemeasurements in small animals

A. Ivorra a, R. Gomez a, N. Noguera a, R. Villa a, A. Sola b, L. Palacios c, G. Hotter b,J. Aguilo a,*

a Centro Nacional de Microelectonica (IMB-CSIC), Campus UAB, E-08193 Bellaterra, Barcelona, Spainb Department of Medical Bioanalysis, Instituto de Investigaciones Biomedicas, IIBB-CSIC, IDIBAPS, Barcelona, Spain

c Department of Physiology, Faculty of Biology, University of Barcelona, Barcelona, Spain

Received 11 July 2002; received in revised form 4 June 2003; accepted 25 June 2003

Biosensors and Bioelectronics xxx (2003) 1�/9

www.elsevier.com/locate/bios

ARTICLE IN PRESS

Abstract

It is commonly accepted that electrical impedance provides relevant information about the physiological condition of living

tissues. Currently, impedance measurements are performed with relatively large electrodes not suitable for studies in small animals

due to their poor spatial resolution and to the damage that they cause to the tissue. A minimally invasive needle shaped probe for

electrical impedance measurements of living tissues is presented in this paper. This micro-probe consists of four square platinum

electrodes (300 mm�/300 mm) on a silicon substrate (9 mm�/0.6 mm�/0.5 mm) and has been fabricated by using standard Si

microelectronic techniques. The electrodes are not equally spaced in order to optimise the signal strength and the spatial resolution.

Characterisation data obtained indicate that these probes provide high spatial resolution (measurement radius B/4 mm) with a

useful wide frequency band going from 100 Hz to 100 kHz. A series of in vivo experiments in rat kidneys subjected to ischemia was

performed to demonstrate the feasibility of the probes and the measurement system. The impedance modulus and phase were

measured at 1 kHz since this frequency is sufficiently low to permit the study of the extracellular medium. The extracellular pH and

K� were also simultaneously measured by using commercial miniaturised Ion Selective Electrodes. The induced ischemia period (45

min) resulted in significant changes of all measured parameters (DjZj�/65%; DpH�/0.8; DK��/30 mM).

# 2003 Elsevier B.V. All rights reserved.

Keywords: Bio-impedance; Micro-probe; Silicon needle; Rat kidney; Induced ischemia

1. Introduction

It has been demonstrated that electrical impedance is

a useful parameter to determine the physiological

condition of living tissues (Grimnes and Martinsen,

2000; Rigaud et al., 1996). Some pathologies induce

changes in essential tissue parameters, such as the extra-

intracellular volume ratio or the ionic composition,

which are reflected as changes in the passive electrical

properties (electrical bio-impedance). One of these

pathologies is ischemia (lack of blood supply and

consequently lack of oxygen). Animal cells regulate

their volume by a Na� pump-mediated mechanism. In

the case of a decrease in the metabolic energy caused by

ischemia, this mechanism fails to regulate the Na�

equilibrium and an osmotic pressure imbalance is

created between the intracellular and the extracellular

media. As a result, the extracellular water penetrates

into the cell with a subsequent cell swelling (Flores et al.,

1972) and an extracellular medium shrinkage which is

manifested as a decrease in the electrical conductivity of

the tissue at low frequencies.Ischemia is induced during certain surgical practices

or during the cold storage of grafts for transplantation.

Its influence on the cells determines the further viability

of the tissue and, therefore, any parameter able to

monitor its evolution, such as bio-impedance, could

become in medical practice a useful tool to follow the

changes induced by the therapeutic management. More-

over, the use of bio-impedance measurements could be

* Corresponding author. Tel.: �/34-93-594-7700; fax: �/34-93-580-

1496.

E-mail address: jordi.aguilo@cnm.es (J. Aguilo).

0956-5663/03/$ - see front matter # 2003 Elsevier B.V. All rights reserved.

doi:10.1016/S0956-5663(03)00204-5

useful for biomedical researchers and, in this sense, the

possibility to experiment with small animals is very

attractive.

The impedance probe used in most past in vivo

studies consists of a linear array of four metallic needle

electrodes placed at a constant inter-electrode separa-

tion distance. The outer electrodes are employed to

inject an AC current into the tissue while the resulting

potential is differentially measured across the inner

electrodes (Rush et al., 1963). Although the feasibility

of this kind of probe has been widely demonstrated, it

implies some important practical drawbacks that restrict

its use: (a) the fabrication process results in large

tolerances because of the critical positioning and align-

ment of the electrodes, (b) the damage caused to the

tissue is considerable since each probe causes four

punctures, (c) the presence of a conductive layer (e.g.

blood) on top of the tissue under study shunts the

electrodes and seriously disturbs the measurements

(Steedijk et al., 1993) and (d) the strong dependence of

apparent resistance on insertion depth (Tsai et al., 2000)

makes crucial the probe fixation to the tissue. Because of

these facts, bio-impedance in vivo studies have been

mostly limited to moderate size animals and the few

studies that have been carried out with small animals

made use of different impedance probes with important

practical limitations (Jossinet et al., 2001; Raicu et al.,

1998).

In this paper, a needle shaped probe with four aligned

platinum electrodes on its surface is presented (Fig. 1).

This configuration not only lessens the limitations of the

classical plunge probe but provides another interesting

advantage: the probe can be easily fabricated by using

standard microelectronic technologies which implies

high reliability, low costs and the possibility to integrate

other sensors and electronics on the same probe. That

kind of geometry has been employed previously for

neural activity recording or stimulation (Blum and

Charles, 1988; Drake et al., 1988; Yoon et al., 2000),

but, as far as the authors know, it has not been applied

yet for impedance measurements. Thus, new design

strategies had to be considered in order to optimisethe probe performance for bio-impedance measure-

ments.

As it has been pointed out, ischemia induces tissue

changes that can be sensed by bio-impedance measure-

ments. Here, in order to demonstrate the feasibility of

these novel impedance probes to carry out in vivo

studies with small animals, a series of in vivo experi-

ments in rat kidneys subjected to ischemia was per-formed. Other two parameters known to be influenced

by the ischemia: the extracellular pH (Dmochowski and

Couch, 1966; Moller et al., 2000) and the extracellular

potassium concentration (Mayevsky et al., 2001; Reeves

and Shah, 1994) were also monitored and their results

and practical constraints were compared with those

obtained with the impedance measurements.

2. Materials and methods

2.1. Bio-impedance probes

2.1.1. Probe design

The starting point for the probe design was the four-

electrode structure with constant inter-electrode separa-tion distance (IESD). For this structure, the voltage

drop at the inner electrodes is inversely related with the

IESD (Steedijk et al., 1993) and, therefore, it is desirable

to make the probe as long as possible to enhance signal-

to-noise ratio. However, since the needle shaft length

was restricted in order to minimise the tissue damage

and to obtain high spatial resolution, a modification was

investigated: to bring the inner electrodes near to theouter electrodes (Ivorra et al., 2001). It can be deduced

intuitively that this strategy increases the voltage drop

Fig. 1. (a) Impedance probe after wafer sawing. The platinum electrodes are placed at non-constant inter-electrode separation distance to optimise

the probe performance. (b) The packaging covers the bonding pads and reduces the needle shaft to 9 mm.

A. Ivorra et al. / Biosensors and Bioelectronics xxx (2003) 1�/92

ARTICLE IN PRESS

for a given current and resistivity. Fortunately, it is also

beneficial in terms of spatial resolution.

By applying the same methods used by Suesserman

and Spelman (1993) the signal strength (voltage drop,V ) is easily related with the injected current (I), the

sample resistivity (r ), the separation distance between

the outer and the inner electrodes (a ) and the separation

distance between the inner electrodes (b ) for an isotropic

and uniform infinite medium:

V �1

4pr

b

a(a � b)I (1)

This expression shows what was expected intuitively:

it is possible to improve the voltage signal withoutincreasing the length occupied by the electrodes (2a�/b).

Concerning the spatial resolution, it is possible to

analytically derive, by using the image method (Robil-

lard et al., 1979), an expression that relates the apparent

resistivity (measured resistivity, r ?) with the infinite

extent medium resistivity (r ) when there exists a

medium transition at a distance x from the non-

constant IESD electrode array (see Fig. 2):

r?�r

�1�K

G?

G

�(2)

G��

1

a�

1

a � b

�(3)

G?��

1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4x2 � a2

p �1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

4x2 � (a � b)2

q�

(4)

K�r2 � r1

r2 � r1

(5)

where K (‘‘reflection coefficient’’) depends on the

sample resistivity (r1) and the resistivity of the boundary

medium (r2).

The r ?/r ratio versus the transition distance with a

non-conductive medium (K�/1) is represented in Fig. 2

for different electrode distances and a constant array

length. It can be observed that effective measurement

volume is reduced when a is reduced. Thus, bringing the

inner electrodes closer to the outer electrodes also

enhances the spatial resolution of the probe.

From what has been said above, it could be deduced

that the best situation would be to place the inner and

the outer electrodes at the minimum possible separation

distance. However, it must be taken into account that

the most sensitive volume of the probe will be the

volume surrounding the separation between an inner

and an outer electrode. In the case that this distance is

reduced too much, the impedance measurements will

depend on the relative position of these electrodes in the

cellular structure of the tissue and not on the homo-

geneous properties of the tissue surrounding the probe.

Hence, a separation distance quite larger than the cell

size is recommended in order to avoid the heterogeneity

of the living tissue at cellular level. In the designed

probe, the minimum separation distance between elec-

trodes is 300 mm.

The electrode�/tissue, or electrode�/electrolyte, inter-

face impedance determines the performance of any

tissue impedance measuring system, even if the four-

electrode method is applied (Pallas-Areny and Webster,

1993). This undesired impedance causes measurement

errors, especially at low frequencies, that could be

understood as noise or signal distortion. In order to

minimise these errors, the electrode�/tissue impedances

should be reduced and matched as much as possible.

Since the conductance of these interfaces is directly

related to the area, the most evident way to reduce their

impedance is to enlarge the electrode surface. In this

sense, the square shape is a good choice because it

makes good use of the shaft length given for each

electrode. In the vicinity of the square corners the

current density will not be uniform and non-linear tissue

electrical properties could become manifested, however,

since the tissue impedance measurement covers

a greater volume and the injected current is very low

(B/10 mA), this effect will unlikely influence the

measurements.

About the probe materials it must be said that the

technological process was a decisive factor in regard to

the substrate material (Si) and the isolating materials

(SiO2 and Si3N4). The chosen material for the electrodes

was platinum because of its biocompatibility and low

electrode�/electrolyte interface impedance compared

with other materials such as stainless steel or silver.

Furthermore, the Pt electrodes can be coated with a

porous layer of ‘‘black platinum’’ which largely reduces

the impedance (Geddes, 1972).

Fig. 2. Estimated r ?/r (apparent resistivity/infinite extent medium

resistivity) of a conductive medium bounded by a non-conductive

medium located at x .

A. Ivorra et al. / Biosensors and Bioelectronics xxx (2003) 1�/9 3

ARTICLE IN PRESS

2.1.2. Probe fabrication

The fabrication of the bio-impedance probes was

carried out at the Centro Nacional de Microelectronica

(CNM) clean room facilities. The technological processconsists of two photolithographic steps starting from a

thermal oxidation to grow a thick field layer (800 nm)

on 4-in. (�/10 cm) P-type �100� Si wafers with a

nominal thickness of 525 mm. The first photoresist layer

is applied and patterned on the wafer surface in order to

pattern a double Ti/Pt layer (30�/150 nm) by using the

so-called lift-off technique. Then, two Low Pressure

Chemical Vapour Deposited (LPCVD) layers of SiO2

and Si3N4 (300�/700 nm) acting as passivation layers are

deposited and patterned using the second photolitho-

graphic level to open the electrodes and the bonding

pads.

After the clean room processes, the wafer is sawed by

successive parallel and oblique cuts which result in a

significant amount (�/500) of needle shaped probes

(Fig. 1a). Then, each probe is fixed on a tiny PrintedCircuit Board (PCB) with gold contacts and wires

connected to the electrodes through the PCB by wedge

bonding (Fig. 1b). The packaging process ends with

complete covering of the PCB with an epoxy resin (H77

from Epoxy Technology, Billerica, MA, USA).

2.1.3. Electrode platinisation

A well known method to reduce the impedance of the

platinum electrodes is to electrochemically coat them

with a porous layer of black platinum as described by

Geddes (1972). At low frequencies, this method can

reduce the interface impedance ten or more times.However, the resulting black platinum surface is fragile

and becomes easily detached when the electrode is

inserted into the tissue. Here, the original technique

was modified in order to improve the mechanical

stability: after a first electroplating using a solution

containing platinum chloride (0.025 N hydrochloric

acid�/3% platinum chloride�/0.025% lead acetate) the

electrodes are mechanically cleaned with tissue paperand a second electroplating with the same solution is

performed. The mechanical cleaning seems to remove

the slightly attached platinum deposits and only the

strong attached deposits remain on the electrodes acting

as anchors for the second electroplating. Up to a point,

this method is equivalent to the one described by

Marrese (1987), in which ultrasound agitation is used

while the electroplating is performed.

2.1.4. Impedance probe characterisation

The characterisation of the probes was performed by

using a commercial impedance analysis system (SI 1260,Solartron Analytical from The Roxboro Group plc,

Cambridge, UK) after the fabrication process has been

completed. Dry inter-electrode and electrode-pad im-

pedance measurements were performed to obtain para-

site capacitances and parasite resistances.

To characterise the electrode�/electrolyte interface

impedance, the probe was immersed in physiologicalsaline solution (0.9% NaCl, resistivity at 298 K�/71.3 Vcm) (Oehme, 1991) and impedance spectroscopy was

obtained. For each electrode couple, a frequency scan

from 10 Hz to 1 MHz was performed at a constant

voltage amplitude of 10 mV. These tests were done

before and after the series of in vivo experiments to

verify the stability of the black platinum deposit on the

electrodes.With the aid of a front-end to enhance the input

properties of the SI 1260 (Gersing, 1991), four-electrode

measurements in physiological saline solution from 10

Hz to 1 MHz were also performed in order to assess the

useful frequency band.

Equation (2) was deduced under the assumption that

the electrodes are point-sized and this is not the case

here since the electrode sizes are comparable to theseparation distances (Fig. 1). For this reason, an

experimental set up was created to study the accuracy

of the model: the probe was introduced in a large

container filled with 0.9% NaCl and r was measured at

1 kHz, then, a PVC block (resistivity much larger than

the saline solution, K�/1) attached to a micro-positioner

was displaced along the perpendicular axis to the silicon

probe and the r ? values were obtained for variousdistances (x ) and compared with those computed with

the model.

2.2. Potassium and pH probes

To measure the pH and K� extra-cellular levels,

commercial miniaturised Ion Selective Electrodes (ISE)

and glass reference electrodes were acquired (references

800, 601 and 401 from Diamond General DevelopmentCorp., Ann Arbor, MI, USA).

The response of these electrodes was checked in lab

using known solutions before the in vivo experiments.

Especial effort was made to test sensibility, linearity,

cross-interferences and drifts for the expected measure-

ment ranges (pH from 6 to 8 and K� from 1 to 40 mM).

2.3. Instrumentation for in vivo tests

As it is described later on, an in vivo test to validate

the measurement frequency for impedance measure-

ments was performed using Solartron SI1260 and a

front-end (Gersing, 1991). However, since this commer-

cial system is not portable and it was desired to perform

multiple measurements of multiple parameters during

the in vivo studies, it was necessary to develop a custom-made instrumentation system (Gomez et al., 2001)

whose main features are summarised in Table 2. During

all the in vivo studies the system was programmed to

A. Ivorra et al. / Biosensors and Bioelectronics xxx (2003) 1�/94

ARTICLE IN PRESS

acquire the samples from all the active channels at each

10 s.

2.4. Experimental procedures

2.4.1. Validation of the measurement frequency

At frequencies below 1 Hz and at low current

densities, the electrical bio-impedance is mainly related

with the extracellular medium since the cell membranesbehave as dielectric layers with very low conductance

(Grimnes and Martinsen, 2000). In the case that the

conductivity of the extracellular medium do not change

significantly, this fact can be exploited to study the

extracellular volume changes. Unfortunately, due to the

electrode�/electrolyte interface impedance, it is practi-

cally impossible to work at these frequencies without

important measurement errors. In some cases, however,it is possible to use higher frequencies and consider that

the results are very close to those obtained at a very low

frequency. Generally, 1 kHz is considered to be a

sufficiently low frequency to study the extracellular

medium (Osypka and Gersing, 1995) because, at this

frequency, most of the current does not penetrate into

the cells. At the same time, this frequency is not too low

to induce measurement errors because of the electrode�/

electrolyte interface impedance. Thus, a priori this

frequency seemed a proper choice but, taking into

account the singularity of the tissue under study in the

related experiments, impedance spectroscopy was used

to test the suitability of this frequency. It was tested to

check if at this frequency it would produce results

similar to those obtained from studies at lower fre-

quency (�/1 Hz). The test is based on the fact that mostliving tissues exhibit relaxation processes that can be

modelled by the Cole�/Cole equations (Grimnes and

Martinsen, 2000), and from that model it is possible to

obtain an impedance value at 0 Hz (R0):

Z�R��DR

1 � (jvt)a; DR�R0�R� (6)

Following the same procedure that is described in thenext section, the right kidney of a rat was subjected to

acute ischemia (60 min) followed by a reperfusion period

(25 min) while the impedance was monitored using the

needle probe and the SI1260�/font-end system. The

impedance values were obtained from 10 Hz to 1 MHz

at every 50 s. With the resulting data, the Cole

parameters for the low frequency relaxation were

automatically fitted with ad hoc developed MatlabTM

routines. Then, after verifying a good Cole fitting, it was

possible to compare the R0 value with impedance

modulus at higher frequencies.

2.4.2. In vivo studies

The study was conducted under the supervision of the

IIBB ethics commission and conformed to the EU

guidelines for handling and care of laboratory animals.Male Wistar rats (Ifa Credo, Spain) weighting 250�/300

g were anaesthetised with sodium pentobarbital (50 mg/

kg). Indwelling polyethylene cannulas (PE-50, Clay

Adams, Sparks, MD) were inserted through the left

carotid artery into the aorta (Popovic and Popovic,) for

blood sampling and saline infusion (1 ml/100 g per h).

The trachea was intubated by using polyethylene tubing

(PE-240), and ventilation was maintained using aHarvard animal respirator. PaCO2 values were kept

between 4.7 and 5.4 by ventilatory control while PaO2

was controlled between 14 and 20 kPa by adequate

oxygen�/air mixture control. The abdominal area was

covered with saline soaked gauze at 310 K (37 8C) and a

plastic cover to minimise dehydration of the exposed

tissues. Animals were constantly exposed to radiant

heat, maintaining abdominal and kidney surface tem-peratures at 3099/1 K.

To induce kidney ischemia and reperfusion, lapar-

otomy was performed and the left renal pedicle was

dissected and occluded with a non-traumatic microvas-

cular clamp. Rats were randomised into two groups:

sham-operated (Control group, C, n�/5) and those

subjected to 45 min ischemia followed by 60 min

reperfusion (I/R group, n�/5).The impedance probes were orthogonally inserted

into the kidney by direct puncture. The penetration

depth of the centre of the electrode array was 5 mm. The

probe was kept into the tissue by its own shape and no

bleeding was appreciated.

The shape and volume differences between the

kidneys from the series of rats induce sample volume

differences that are manifested as differences in theimpedance modulus. For this reason, the impedance

modulus measurements were normalised to their initial

value.

ISE were inserted with the help of a cutting edge butt

needle allowing controlled deep puncture (bleeding was

negligible). The sensitive tip lain at the boundary area

between outer medulla and cortex. The reference

electrode was placed into the abdominal cavity. All theelectrodes were held by loose supports in order to keep

constant the angle of the tip.

The sensitivity of the pH and K� ISEs was calibrated

before each experiment using two isotonic solutions of

known pH and K� concentrations (sol. 1: 1 mM K, pH

8.78; sol. 2: 4 mM K, pH 6.74) prepared from a NaCl

buffer as described by Cosofret et al. (1995). The ISEs

were kept in these calibration solutions for more than anhour in order to reduce the stabilisation time after the

insertion into the living tissue. Baseline arterial plasma

pH and K� measurements were used to set the time-

zero values of tissue pH and K�.

A. Ivorra et al. / Biosensors and Bioelectronics xxx (2003) 1�/9 5

ARTICLE IN PRESS

3. Results and discussion

3.1. Bio-impedance probes characterisation

The results from the impedance probe characterisa-

tion after the manufacturing are summarised in Table 1.The electrode�/electrolyte interface impedance (Fig. 3)

becomes very high at frequencies below 100 Hz and that

can involve important tissue impedance measurement

errors, especially in a heterogeneous tissue where each

electrode can have completely different interface im-

pedances. On the other hand, at frequencies beyond 100

kHz, the capacitive coupling of the wires (including the

coaxial wires from the probes to the instrumentation) isstrongly manifested (Fig. 4). Thus, it can be considered

that the useful frequency band goes from 100 Hz to 100

kHz.

The interface impedance test after the series of in vivo

experiments resulted in minor changes (impedance

modulus increase at 100 Hz B/20%) confirming the

stability of the black platinum deposit.

The results from the spatial resolution characterisa-tion are shown in Fig. 5. It can be observed that the

experimental measurements follow the predicted values

by using the model. Therefore, the model works

properly although the electrodes cannot be considered

as point electrodes. From these results it is possible to

provide a value for the spatial resolution: 4 mm. That is,

any medium disruption beyond 4 mm from the centre of

the probe will cause measurement errors below 1%.The use of microelectronic materials and fabrication

processes reduces production costs and offers the

possibility to integrate signal conditioning electronics

or additional sensors. This could be used, for instance,

to expand the useful frequency band by integrating

voltage buffers on the silicon. Furthermore, a tempera-

ture sensor could be integrated on the same substrate to

isolate changes caused by physiological reasons from

those caused by temperature changes.

3.2. Validation of the measurement frequency

After inducing ischemia by arterial clamping, the R0

increased from 1.7 to 12.5 kV in 60 min. The impedance

modulus at 1 kHz followed this evolution with a reducedsensitivity resulting in a 25% of maximum relative

difference. This fact could suggest the use of lower

frequencies to reduce this difference, however, it must be

taken into account that measurement errors are ob-

served for frequencies below 300 Hz and that the range

of possible interface impedances is amazingly wide.

Therefore, in order to ensure quality for all the

measurements it seems that 1 kHz is a proper choice.Fig. 6 shows the results obtained from this experi-

mental test before and after 10 min of induced ischemia

and the superimposed Cole fittings. The Cole models

were adjusted for low frequencies (B/10 kHz) since the

relaxation arc presented distortion for higher frequen-

cies due to other tissue relaxation constants and because

of the needle parasitic capacitances.

Table 1

Summarised results from the probes characterisation

Parameter Conditions Minimum Typical Maximum

Electrode-pad resistance: (connection resistance) TA�/298 K

I�/ 1050 V 1200 V 1300 VV�/ 900 V 1000 V 1100 VV�/ 850 V 1000 V 1050 VI�/ 600 V 700 V 800 VInter-electrode capacitance TA�/298 K 5 pF 6 pF

Inter-electrode impedance modulus in saline solution TA�/298 K, 0.9% NaCl, VOSC�/100 mVp

10 Hz 5 kV 8 kV 25 kV100 Hz 3.5 kV 5 kV 7 kV1 kHz 3.6 kV 3.8 kV 4 kV10 kHz 3.4 kV 3.5 kV 3.6 kV100 kHz 3.2 kV 3.3 kV 3.4 kV

Cell constant (k�/r /R ). R , measured resistance; r , resistivity TA�/298 K, 0.9% NaCl, VOSC�/100 mVp 0.32 cm

Spatial resolution ErrorB/1% 4 mm

Table 2

Instrumentation system main features

Bio-impedance meter

Number of channels 10

Ground isolation impedance 50 pF

Oscillator frequency 100 Hz to 125 kHz

Injected current amplitude B/5 mA

CMRR at 1 kHz 88 dB

Input common impedance �/50 MV//7pF

Input differential impedance �/50 MV//2pF

Ion meter (voltmeter )

Number of channels 16

Ground isolation impedance 50 pF

Input impedance �/10 TV

A. Ivorra et al. / Biosensors and Bioelectronics xxx (2003) 1�/96

ARTICLE IN PRESS

3.3. Induced ischemia in rat kidneys

Fig. 7 shows the plots of mean values for normalised

impedance modulus, phase angle, tissue pH and inter-

stitial potassium ion concentration. It can be observed

that renal vascular occlusion induced immediate and

rapid changes in monitored variables. Impedance phase

and pH decreased (Fig. 7B and C), and impedance

modulus and K� increased (Fig. 7A and D). Rapid

initial changes in the four parameters were followed with

a declining rate change as ischemia progressed.

Reperfusion produced an increase in kidney pH as

well as in phase (Fig. 7C and B) and a drop in kidney

potassium and impedance modulus (Fig. 7D and A).

Post-ischaemic impedance modulus experienced a sharp

decay, reaching a minimum within the first 10 min of

reperfusion, but with second increase episode, with peak

values preceding a slow decrease. It is important to note

that the impedance do not return to its original values

after the reperfusion. This could indicate that some kind

of permanent damage to the tissue is not detected by the

measurement of pH or K�.

For low-frequencies (B/10 kHz), the impedance

changes that are exhibited by organ tissues during the

course of ischemia are mainly attributed to cell swelling,

which narrows the extracellular space, and the gap

junctions closure (Gersing, 1998). The presence of gap

junctions in the kidney is much more insignificant than

those in other organs (e.g. heart or liver). Thus, it is

quite reasonable to consider that the impedance at low

frequencies is mainly modified by the extracellular�/

intracellular volume changes. This assumption agrees

with previous histologic observations of rat kidney

during ischemia (Flores et al., 1972). In those observa-

tions, cell swelling is noted when the tissue is made

hypoxic. This cell swelling results in a decreasing

extracellular volume which in turn results in an increas-

ing impedance modulus (the current path is narrowed)

Fig. 3. Inter-electrode impedance modulus (a) and phase (b) measured in NaCl 0.9%.

Fig. 4. Measured impedance modulus (a) and phase (b) of a NaCl 0.9% solution using the impedance probe (four-electrode measurements).

Fig. 5. Expected and experimental r ?/r values for the implemented

silicon probe.

A. Ivorra et al. / Biosensors and Bioelectronics xxx (2003) 1�/9 7

ARTICLE IN PRESS

and in a decreasing impedance phase (current is forced

to go through dielectric membranes and, therefore, the

reactive part is increased).

From these results (Fig. 7) it is important to note that

the scatter of the impedance modulus is significantly

lower than the scatter of the pH and K� values.

Furthermore, the impedance values are stabilised

much faster (5 min) than the pH and K� readings

(20�/30 min) after the probes are inserted into the tissue

(not shown in Fig. 7). These facts reinforce the useful-

ness of the impedance measurements for experimental or

clinical applications.

The real time monitoring of living tissues with thesenovel impedance probes can become a useful tool for

biomedical researchers and physicians, not only for

rapid detection of ischemia, but also for following the

changes induced by the therapeutic management. The

electrical bio-impedance monitoring offers some impor-

tant advantages when compared with other parameters:

on-line monitoring, robustness, no calibration is needed,

no inherent drift and fully biocompatible materials.

4. Conclusion

A miniaturised four-electrode probe for electrical bio-

impedance measurements has been developed by usingmicroelectronic materials and fabrication processes. It

consists of four square platinum electrodes (300 mm�/

300 mm) on a silicon substrate (9 mm�/0.6 mm�/0.5

mm). The electrodes are placed at non-constant inter-

electrode distance in order to optimise the signal-to-

noise ratio and the spatial resolution.

The results from the characterisation indicate that

these probes provide high spatial resolution (measure-ment radius B/4 mm) with a frequency band from 100

Hz to 100 kHz.

Fig. 6. Nyquist plot evolution from a kidney rat subjected to ischemia

(frequency sweeping plots before and after 10 min of induced

ischemia). The modelled values (k) have been adjusted to the

experimental values (I) for the arc observed at low frequencies.

Fig. 7. Evolution of the measured tissue parameters in kidneys subjected to 45 min of renal vascular occlusion and 60 min of reperfusion. (A)

Normalised modulus impedance. (B) Phase angle. (C) Tissue pH. (D) Potassium ion concentration. Results are expressed as mean9/standard

deviation (n�/5).

A. Ivorra et al. / Biosensors and Bioelectronics xxx (2003) 1�/98

ARTICLE IN PRESS

The capability of the probe to measure bio-impedance

in small animals has been demonstrated. Furthermore,

the obtained in vivo results reinforce the usefulness of

the bio-impedance as a marker of the tissue condition.

Acknowledgements

This work has been supported by the European

Commission through projects ESPPRIT-LTR-23485,

IST-1999-13047 and QLK6-CT-2000-00064 and by the

Spanish government through projects TIC98-1634-CE,TIC2000-2486-CE, FISS 01/1691 and SAF 2000-3090-

CE.

References

Blum, N.A., Charles Jr, K.K., 1988. Multi-microlectrode probe for

neurophysiologial experiments. In: Proceedings of the Fourteenth

Annual Northeast Bioengineering Conference, Durham, NH,

USA, pp. 15�/18.

Cosofret, V.V., Erdosy, M., Timothy, A.J., Buck, R.P., 1995.

Microfabricated sensor arrays sensitive to pH and K� for ionic

distribution measurements in the beating heart. Analytical Chem-

istry 67 (10), 1647�/1653.

Dmochowski, J.R., Couch, N.P., 1966. Electrometric surface pH of the

ischemic kidney and the effect of hypothermia. Journal of Surgical

Research 6 (1), 45�/48.

Drake, K.L., Wise, K.D., Farraye, J., Anderson, D.J., BeMent, S.L.,

1988. Performance of planar multisite microprobes in recording

extracellular single-unit intracortical activity. IEEE Transactions

on Biomedical Engineering 35 (9), 719�/731.

Flores, J., DiBona, D.R., Beck, C.H., Leaf, A., 1972. The role of cell

swelling in ischemic renal damage and the protective effect of

hypertonic solute. The Journal of Clinical Investigation 51, 118�/

126.

Geddes, L.A., 1972. Electrodes and the Measurement of Bioelectric

Events. Wiley-Interscience/Wiley, New York, pp. 32�/35.

Gersing, E., 1991. Measurement of electrical impedance in organs-

measuring equipment for research and clinical applications.

Biomedizinische Technik 36 (1-2), 6�/11.

Gersing, E., 1998. Impedance spectroscopy on living tissue for

determination of the state of organs. Bioelectrochemistry and

Bioenergetics 45, 145�/149.

Gomez, R., Noguera, N., Ivorra, A., Villa, R., Aguilo, J., Millan, J.,

Lopez, J., Palacios, L., Sola, A., Hotter, G., 2001. Instrumentation

system for in vivo organ studies. In: Proceedings of 24th Interna-

tional Semiconductor Conference, Sinaia (Romany), 9�/13 Octo-

ber, vol. 1, pp. 261�/264.

Grimnes, S., Martinsen, O.G., 2000. Bioimpedance and Bioelectricity

Basics. Academic Press.

Ivorra, A., Aguilo, J., Millan, J., 2001. Design considerations for

optimum impedance probes with planar electrodes for bioimpe-

dance measurements. In: Proceedings of 24th International Semi-

conductor Conference, Sinaia (Romany), 9�/13 October, vol. 1, pp.

269�/272.

Jossinet, J., Moulin, V., Matias, A., McAdams, E.T., 2001. In vivo

spectrometry of dunning prostate tumours. In: Proceedings of the

XII Conference on Electrical Bio-Impedance, Oslo (Norway), 17�/

21 June, pp. 285�/288.

Marrese, C.A., 1987. Preparation of strongly adherent platinum black

coatings. Analytical Chemistry 59 (1), 217�/218.

Mayevsky, A., Nakache, R., Luger-Hamer, M., Amran, D., Sonn, J.,

2001. Assessment of transplanted kidney vitality by a multipara-

metric monitoring system. Transplantation Proceedings 33, 2933�/

2934.

Moller, H.E., Gaupp, A., Dietl, K.-H., Buchholz, B., Vestring, T.,

2000. Tissue pH in human kidney transplants during hypothermic

ischaemia. Magnetic Resonance Imaging 18, 743�/751.

Oehme, F., 1991. In: Gopel, W., Jones, T.A., Kleitz, M., Lundstrom,

J., Seiyama, T. (Eds.), Sensors, A Comprehensive Survey, Chemical

and Biochemical Sensors, Part I, vol. 2. VCH Verlagsgesellschaft

mbH, Weinheim, pp. 314�/332.

Osypka, M., Gersing, E., 1995. Tissue impedance spectra and the

appropriate frequencies for EIT. Physiological Measurement 16,

A49�/A55.

Pallas-Areny, R., Webster, J.G., 1993. AC instrumentation amplifier

for bioimpedance measurements. IEEE Transactions on Biomedi-

cal Engineering 40 (8), 830�/833.

Raicu, V., Saibara, T., Irimajiri, A., 1998. Dielectric properties of rat

liver in vivo: a noninvasive approach using an open-ended coaxial

probe at audio/radio frequencies. Bioelectrochemistry and Bioe-

nergetics 47, 325�/332.

Reeves, W.B., Shah, S.V., 1994. Activation of potassium channels

contributes to hypoxic injury in proximal tubules. Journal of

Clinical Investment 94 (6), 2289�/2294.

Rigaud, B., Morucci, J.-P., Chauveau, N., 1996. Examples of

applications. Critical Reviews in Biomedical Engineering 24 (4-6),

323�/351.

Robillard, P.N., Poussart, D., 1979. Spatial resolution of four

electrode array. IEEE Transactions on Biomedical Engineering

BME-26 (8), 465�/470.

Rush, S., Abildskov, J.A., McFee, R., 1963. Resistivity of body tissues

at low frequencies. Circulation Research XII, 40�/50.

Steedijk, P., Mir, G., Van Del Velde, E.T., Baan, J., 1993. The four-

electrode resistivity technique in anisotropic media: theoretical

analysis and application on myocardial tissue in vivo. IEEE

Transactions on Biomedical Engineering 40 (11), 1138�/1148.

Suesserman, M.F., Spelman, F.A., 1993. Quantitative in vivo mea-

surements of inner ear tissue resistivities: I. In vitro characteriza-

tion. IEEE Transactions on Biomedical Engineering 40 (10), 1032�/

1046.

Tsai, J.-Z., Cao, H., Tungjitkusolmun, S., Woo, E.J., Vorperian, V.R.,

Webster, J.G., 2000. Dependence of apparent resistance of four-

electrode probes on insertion depth. IEEE Transactions on

Biomedical Engineering 47 (1), 41�/48.

Yoon, T.H., Hwang, E.J., Shin, D.Y., Park, S.I., Oh, S.J., Jung, S.C.,

Shinn, H.C., Kim, S.J., 2000. A micromachined silicon depth probe

for multichannel neural recording. IEEE Transactions on Biome-

dical Engineering 47 (8), 1082�/1087.

A. Ivorra et al. / Biosensors and Bioelectronics xxx (2003) 1�/9 9

ARTICLE IN PRESS