>TBCAS-2011-Sep-0136-Reg.R1<

1

Abstract— An electrochemical detection system specifically

designed for multi-parameter real-time monitoring of stem cell

culturing/differentiation in a microfluidic system is presented. It

is composed of a very compact 24-channel electronic board,

compatible with arrays of microelectrodes and coupled to a

microfluidic cell culture system. A versatile data acquisition

software enables performing amperometry, cyclic voltammetry

and impedance spectroscopy in each of the 12 independent

chambers over a 100kHz bandwidth with current resolution down

to 5pA for 100ms measuring time. The design of the platform, its

realization and experimental characterization are reported, with

emphasis on the analysis of impact of input capacitance (i.e.

microelectrode size) and microfluidic pump operation on current

noise. Programmable sequences of successive injections of

analytes (ferricyanide and dopamine) and rinsing buffer solution

as well as the impedimetric continuous tracking for 7 days of the

proliferation of a colony of PC12 cells are successfully

demonstrated.

Index Terms - electrochemical measurements, multichannel

potentiostat, microfluidics, stem cell monitoring, impedance

spectroscopy.

I. INTRODUCTION

ellular dynamics and the complex pathways that regulate

the differentiation of stem cells in vitro are currently of

primary interest in biology and medicine ‎[1]. In

particular, the use of pluripotent stem cells shows great

promise for the treatment of neurodegenerative disorders ‎[2].

A feasible approach to investigate stem cell differentiation

towards adult neural cells is the combination of several

complementary detection techniques. Due to the specificity of

fluorescent markers, optical microscopy is by far the preferred

tool in biological investigation. However, when studying

electrogenic cells, the microscope is naturally replaced by

Manuscript received September, 2011. This work was supported by the

EU FP7 EXCELL project under grant NMP4-SL-2008-214706.

M. Vergani, M. Carminati, G. Ferrari and M. Sampietro are with

Dipartimento di Elettronica e Informazione, Politecnico di Milano, Milano,

Italy (phone: +39.02.2399.3773, fax: +39.02.2399.3574, email:

vergani@elet.polimi.it, sampietr@elet.polimi.it).

C. Caviglia, A. Heiskanen, K. Zor, D. Sabourin, M. Dufva, M. Dimaki and

J. Emnéus are with the Department of Micro- and Nanotechnology, Technical

University of Denmark, Lyngby, Denmark.

E. Landini and R. Raiteri are with the Department of Biophysical and

Electronic Engineering, Università di Genova, Genova, Italy.

C. Comminges and U. Wollenberger are with the Institute for

Biochemistry and Biology, University Potsdam, Golm, Germany.

instrumentation with a direct electrical interface with cells,

such as patch-clamp ‎[3]-‎[4] to record the activity of ion

channels (current detection) and multi-electrode arrays to map

the propagation of extracellular potentials (voltage detection)

in cultured neural networks ‎[5]. Even in the case of non-

electrogenic cells, electrochemistry enables the detection of

molecules produced by cellular metabolism with suitable

(micro- and nano-molar) sensitivity ‎[6]. The powerful

combination of electrochemical detection techniques with the

availability of planar microelectrodes, enabled by

microelectronic fabrication technologies, provides sub-cellular

spatial resolution ‎[7]-‎[8].

The aim of this work is the realization of a versatile

platform that allows automatic monitoring of several cell

culture chambers and capturing detailed information of the

population status by combining optical and multiple electrical

detection techniques. For this purpose, a compact multichannel

bipotentiostat has been specifically designed to be integrated

in a microfluidic culture system having an array of planar

microelectrodes that allow for parallel screening of several cell

populations. This integrated solution is intended to bridge the

gap between microfluidic bioreactors with electrodes but

operated with bulky bench-top electrochemical instruments ‎[9]

and sophisticated ad-hoc systems oriented to chemo-physical

monitoring of stem cell cultures ‎[10] that are not designed to

encompass all electrochemical techniques such as

amperometry, voltammety and impedance spectroscopy.

The instrument has been designed to perform: (i) automatic

tracking of impedance (both at a single frequency and in the

full frequency range from 10mHz to 100kHz) in order to

monitor over time the adhesion and proliferation of a cell

population that affects the electrode-solution interface

impedance ‎[11], and (ii) selective electrochemical detection of

electro-active species, in particular catecholamines (a family of

neurotransmitters, such as dopamine, released by cells), which

is carried out using cyclic voltammetry or amperometry ‎[12].

The release of these molecules, contained in vesicles that are

fused with the cellular membrane, is called exocytosis. An

exocytotic event can be electrically detected using planar gold

microelectrode as reported in ‎[13]. The continuous, spatially

resolved and label-free monitoring of exocytosis is believed to

represent an early indicator of the differentiation into a neural

lineage. Furthermore, preliminary results ‎[14]-‎[15], indicate

that also electrochemical impedance spectroscopy (EIS)

Multichannel Bipotentiostat Integrated with a

Microfluidic Platform for Electrochemical Real-

Time Monitoring of Cell Cultures Marco Vergani, Marco Carminati, Member, IEEE, Giorgio Ferrari, Member, IEEE, Ettore Landini,

Claudia Caviglia, Arto Heiskanen, Clément Comminges, Kinga Zor, David Sabourin, Martin Dufva,

Maria Dimaki, Roberto Raiteri, Ulla Wollenberger, Jenny Emnéus and Marco Sampietro

C

>TBCAS-2011-Sep-0136-Reg.R1<

2

represents a promising label-free tool to detect morphological

changes during the stem cell differentiation process. Specific

requirements for stem cell monitoring include the capability to

apply a sinusoid of tens of µV (for EIS) and record the current

with a resolution of a few pA (for amperometry).

II. SYSTEM OVERVIEW

A. Architecture

As illustrated in Fig. 1, the complete experimental setup is

composed of a compact microfluidics/potentiostat assembly

that can be placed under a standard upright optical

microscope. The electronic board is tightly assembled with the

microfluidic system, while the peristaltic pumps ‎[16] are on

the side. Stringent volume constraints have consequently

guided the board layout, in particular in proximity to the

microscope objective. Fig. 2 shows the vertical cross-section

of the complete platform. The potentiostat is fabricated on a

0.8mm thick multilayer PCB (Fig. 3) assembled in contact

with the microfluidic platform used to culture the cells on a

microelectrode array (MEA). The sealing between the rigid

PMMA substrate of the microfluidic system and the MEA chip

is provided by a biocompatible silicon adhesive tape. The

contact pads of the MEA chip are vertically connected to the

potentiostat using an array of 48 spring-loaded pins for easy

and fast “plug and play” replacement of the disposable MEA

chip. Beyond the highly increased practical convenience for

the user, the miniaturization achieved by this mounting

(avoiding connection cables between the electrodes and the

potentiostat) makes it possible to minimize the impact of input

parasitic capacitance, and consequently keep the signal

resolution as good as possible.

The MEA, placed at the bottom of the cell culture chambers

(see inset 2 of Fig. 3), consists of a single silicon chip hosting

an array of 12 independent measuring sites for cell based

assays. Each measuring site is composed of an individually

addressable microfluidic cell culture chamber containing one

counter (CE), reference (RE) and a pair of interdigitated

working electrodes (WE) (see inset 1 of Fig. 3). The two WEs

(WEa and WEb) can be addressed independently and each is

composed of 12 fingers (10µm width and spacing, 500µm

length) covering an area of 6·104µm

2. All the 48 planar

electrodes of the array are made of gold, deposited on a silicon

dioxide substrate, patterned with a lift-off process and

passivated with a silicon nitride layer (500nm thick). The REs

have been fabricated using gold instead of standard materials

as Ag/AgCl because of its simpler fabrication process. The use

of pseudo-reference gold electrodes results in a reference

potential shift and in a less controlled drift, that is acceptable

during short amperometric measurements (max. ~10 min) [9].

A custom software, run by a laptop computer, controls the

instrument through a USB interface by means of a portable

commercial data acquisition unit (NI USB-6259 by National

Instrument). (Fig. 1) The potentiostat board is connected to

this acquisition unit through an intermediate interconnection

unit that allows matching the single cable (68 pins for power

supply, analog and digital signals) from the potentiostat to the

two cables required by the USB-6259, thus simplifying the

handling of the board mounted with the pumps on the

microscope stage.

B. Operating Modes

The system can work in two different operating modes: (i)

two-electrode mode, where the instrument applies a voltage

signal to the CE and reads the current signal from the WE, and

(ii) three-electrode mode, where the voltage signal is applied

to the electrode-electrolyte interface in a closed-loop manner,

by driving the CE and monitoring the potential at the RE, still

reading the current in WE. The first operating mode is mainly

used when performing impedance spectroscopy measurements,

whereas the second one is typically used when performing

amperometric and voltammetric recordings to avoid the

distortion due to the non linear voltage drop at the CE-

electrolyte interface. Several electrochemical measurements

can thus be performed on the array by changing the operating

mode and the voltage stimulation waveform.

For maximum versatility, impedance spectroscopy can be

performed in vertical or coplanar mode. In the first case, the

sinusoidal signal is applied to the CE and the current is read at

one WE, thus probing a single WE-solution interface for

standard Electrical Cell Impedance Sensing (ECIS) detection

‎[17]. In the second case, the signal is applied to one set of

fingers (WEb) of the interdigitated couple, thus measuring the

impedance between the two sets (a versus b) ‎[18].

In addition to these operating modes, WEa and WEb of

each pair of interdigitated electrodes can be biased at two

independent potentials (e.g. the oxidation and the reduction

potential of a specific target molecule). This feature facilitates

redox cycling experiments, where oxidation of a chemical

species occurs at one WE, following reduction of the oxidized

form at the other WE of the interdigitated structure. The

product of the reaction at one electrode becomes the source of

the reaction at the other electrode, and the target molecule

contributes to the current response with several electrons. This

has the overall effect of improving the current response of the

detection technique through signal amplification ‎[19].

III. MULTICHANNEL POTENTIOSTAT HARDWARE DESIGN

The potentiostat electronic board (see Fig. 4) is composed

of two fundamental blocks: the voltage generation circuit

(often referred to as the potentiostatic circuit) and the current

reading amplifiers. This architecture in which the voltage is

forced (and current is sensed) is the natural choice to perform

all the measurements with the same analog front-end.

A. Potentiostatic Circuit

The voltage generation circuit is schematically shown in

Fig. 5. The voltage of the REs is read by 12 buffers

implemented using the OPA4141 JFET input operational

amplifiers and they are used to protect the small RE from non

idealities, such as current leakage from the analog

multiplexers. Two analog multiplexers (DG406) select the

active RE and CE in the MEA. The desired voltage signal is

>TBCAS-2011-Sep-0136-Reg.R1<

3

applied to the non-inverting input of an operational amplifier

(OPA124) that drives the voltage of the active CE. Due to the

negative feedback, the same voltage is applied also to the non

inverting input of the operational amplifier, which has the

same potential as the active RE. The voltage generation circuit

has ±2V voltage range to match the potential windows of

standard redox reactions and has 3µV resolution. To ensure

these performances, the outputs of two DACs (±10V range, 16

bit resolution) have been summed with a different attenuation,

respectively by a factor of 5 and 100. The OPA4141 buffers

have a 10MHz closed loop bandwidth, while the OPA124 has

been chosen for its 1.5MHz gain-bandwidth product and 65°

phase margin. This choice maintains the stability of the

potentiostatic loop under the expected operating conditions.

B. Current Reading Amplifiers

The electronic board (see Fig. 4) implements four

independent current sensing channels and analog multiplexers

to address the 24 working electrodes. With respect to using a

dedicated transimpedance amplifier for every WE, our choice

reduces size so to accommodate the integrated system on the

stage of an optical microscope without paying for the added

load of the multiplexer ADG1234, that has been chosen for its

low stray capacitance to ground, not influencing the resolution.

The circuit details of each current reading channel are

shown in Fig. 6. Every transimpedance amplifier biases the

active WE at the desired voltage, while simultaneously reading

the current flowing through it. The inactive WEs are biased

through the ADG1234 at the same potential as the active one,

allowing the maintenance of the electrochemical steady state

conditions at the WE-liquid interface and leading to a fast

response when they are selected. The ADA4817 JFET input

operational amplifier has been chosen to implement the

transimpedance stage due to its low noise and dynamic

performance (equivalent noise sources of 4nV/√Hz and

2.5fA/√Hz, 410MHz gain bandwidth product). The feedback

resistor and capacitor have been set to 100kΩ and 10pF,

respectively, to ensure a 100kHz bandwidth and a 400fA/√Hz

input referred current noise. The current amplifier has been

optimized for a maximum capacitive input load of 30nF, five

times the double layer capacitance of our WEs estimated on a

typical capacitance of 0.1pF/µm2 ‎[12] of a metal immersed in

phosphate buffered saline (PBS). A second amplification stage

(based on the instrumentation amplifier INA128 with a gain of

10) and a second-order filter complete the current

amplification chain. The 100kHz bandwidth, consistent with

the performance offered by commercial systems (ECIS Z by

Applied Biophysics and exCELLigence by Roche) is sufficient

to reach the resistive plateau given by the solution resistance

as for the fabricated electrodes in PBS the corner frequency

between the double layer capacitance (6nF) and the solution

resistance (~1kΩ) is at 24kHz. Beyond allowing the use of a

USB acquisition board, the choice to limit the bandwidth of

the current sensing amplifier enables the increase of the

feedback resistor up to 100kΩ, reducing the input referred

current noise at low frequencies down to 400fA/√Hz. The low-

pass (10kHz) filtered output (LPF) is used in amperometric

and voltammetric recordings to relax the sampling and

processing requirements for the acquisition board and PC.

C. Summary of Performance

The electronic performance of the multichannel potentiostat

is summarized in Table 1. The details of the experimental

characterization are reported in sec. V: here we anticipate that

all the design specifications have been achieved. For

voltammetry mode, the maximum scan rate (SR) has been

determined from the bandwidth of the current sensing

amplifier using the relation BW = 40V-1SR ‎[20] that relates

the delay of a single pole to a maximum potential error of

4mV.

The current resolution of 5pA has been measured using a

proper digital filter to limit the bandwidth of the current

sensing circuit at 5Hz. This performance translates into the

ability to perform amperometric detection of neurotransmitter

exocytosis from a cell population corresponding to current

pulses with amplitudes in the 800pA range and durations of

several seconds ‎[21]-‎[22].

IV. DATA ACQUISITION AND ANALYSIS SOFTWARE

In order to control and record data from the electronic

board, as well as to display and process the data in real-time,

with a single user interface, a custom software has been

developed in LabVIEW environment (National Instruments,

Austin TX). The software has a modular structure in order to

allow easy addition of new measurement techniques/protocols

as well as new analysis features. The modules have been

designed to be compatible with all DAQmx National

Instruments boards, requiring 8 analog acquisition channels, 4

analog generation channels, and 14 digital lines. The chosen

NI USB-6259 acquisition unit (see Fig. 1) allows a

multiplexed 1MSample/s acquisition-generation rate and is

connected to a laptop computer via USB in order to maintain

the “plug-and-play” architecture of the entire system.

The software is composed of different acquisition modules

which facilitate electrochemical measurements using four main

techniques: amperometry, cyclic voltammetry (both in single

mode and redox cycling mode), impedance measurements,

both in ECIS and coplanar configuration and both at a single

frequency (impedance tracking) or in a selectable frequency

TABLE 1

SUMMARY OF SYSTEM PERFORMANCE

Generation DC voltage range ±2V

Generation DC voltage resolution 3µV

Min. generation AC amplitude 30µV

Transimpedance gain 1MΩ

Max. input current ±10µA

Current resolution B~5Hz 5pA

Current resolution B~100kHz 2.3nA

Operating frequency range (EIS) 10mHz - 100kHz

Max. scan rate (CV) 2.5kV/s

Max. input capacitance 30nF

Power dissipation (supply voltage) 4W (±15V)

>TBCAS-2011-Sep-0136-Reg.R1<

4

range (impedance spectroscopy and tracking of the spectra

over time). Impedance is extracted from the acquired data

through a digital implementation of a phase-sensitive lock-in

detection algorithm ‎[23], taking advantage of the synchronous

generation and acquisition of analog samples enabled by the

board. Measurements using any of the techniques can be

conducted in an arbitrary sequence of electrodes in the array at

a freely defined time interval in order to automate time laps

experiments on cells in culture.

Since a calibration procedure is necessary to ensure high

accuracy in impedance spectroscopy by compensating

magnitude and phase errors (due to stray input impedance and

to the finite bandwidth of the conditioning circuits), an

automatic calibration circuit has been implemented. Whenever

necessary, a relay connects each input to previously

characterized impedance (a RC cell) mounted on the board,

then a full spectrum is recorded and the correction factors,

calculated for each frequency from discrepancy with respect to

the ideal spectrum, are stored.

All recorded data are displayed during the measurement and

some data analysis can be performed on-line in order to help

the user to monitor the behavior of the cultured cells in real-

time. For example, the user can select an equivalent circuit

from a set of pre-programmed equivalent electrical circuits in

order to fit automatically each of the recorded impedance

spectra. The fitting algorithms, implemented as compiled C#

libraries in the LabVIEW code for maximum speed of

execution, provide the characteristic electrical parameters of

the electrode-cell-solution interfaces.

V. EXPERIMENTAL RESULTS

A. Functional Characterization

The electronic instrumentation has been fully tested in terms

of bandwidth, gain, linearity and noise. Fig. 7 shows the

transfer functions of the four independent Current Reading

Amplifiers, measured with an HP4195A network/spectrum

analyzer. When operated in full bandwidth, they show a

constant gain (1MΩ) up to a frequency bandwidth above

100kHz whereas when the LPF output of Fig. 6 is used the

bandwidth is properly limited to about 10kHz. This second

option is important in increasing the resolution of the system

when speed is not a concern as in cyclic voltammetry. This can

be better understood by considering the input referred current

noise of the electronic platform for different capacitances of

the sensing electrode, measured with the HP4195A, reported

in Fig. 8. When loading the Current Reading Amplifier with

minimum capacitance (Cin = 56pF), we obtain the intrinsic

noise floor of the electronic board which is flat and below

500fA/Hz. The curve of Cin = 5.6nF represents the typical

operative situation when the WE of the array chip is connected

to the board. The scatter points in the graph are the values

expected from theoretical simulations in P-Spice based on the

voltage noise of the amplifier and on Cin. Because of the high

input capacitance that increases the noise spectral density at

high frequencies, the bandwidth limiter at 10kHz is very

effective in reducing the rms noise of the platform to a factor

of almost 30.

To assess the accuracy of the electronic section of the

platform and its bandwidth, an EIS measurement has been

performed on gold interdigitated electrodes immersed in PBS.

The impedance of the same sample by using our multichannel

platform has been compared to a stand-alone commercial

instrument (Agilent E4980A Precision LCR meter) as a

reference. Two impedance measurement modes have been

tested: the vertical (impedance of a single WE with respect to

the CE) and the coplanar (impedance of WEa with respect to

WEb) techniques. The comparison of experimental results by

the two instruments is presented in Fig. 9. In the frequency

range where both instruments can work, the same impedance

values have been recorded, thus proving the correct behavior

of the proposed multichannel potentiostat.

To further assess the capability of the instrument to address

each single electrochemical cell of an array and to handle

cyclic voltammetry measurements, the electrochemistry of the

interdigitated gold electrodes with standard redox mediators

has been sequentially tested. A self-assembled monolayer of

cysteamine (a short thiol molecule with an amine endgroup)

was deposited on the gold electrodes in order to get a

controlled and defined electrode surface, and then cyclic

voltammograms were recorded from all the 24 WEs versus the

on-chip gold RE when both immersed in a solution containing

10mM ferricyanide (K3Fe(CN)6) and ferrocyanide

(K4Fe(CN)6) in PBS. Fig. 10 depicts the voltammograms,

which indicate a very good reproducibility between all the

cysteamine modified electrodes.

B. Dynamic Electrochemical Measurements

The key feature of the platform to have the electronics

tailored onto a microfluidic network, thus resulting in an

integrated system for programmable on-line multi-

electrochemical measurements, has been investigated with a

sequence of amperometric tests. These comprised multiple

injections of a 62.5µM solution of ferricyanide in PBS into the

measurement chambers separated by injection of rinsing buffer

solution. During the experiment, the redox reaction was

continuously monitored. To this aim the cysteamine modified

WE were kept at -350mV with respect to the gold RE using

the three electrodes configuration. The amperometric current-

time trace generated by the system can be seen in Fig. 11. The

baseline was recorded while flushing PBS with 34µl/min flow

rate. Then the pump started to inject the ferricyanide solution

from a reservoir at the same flow rate. It takes about 12s for

the solution to reach the inlet of the microfluidic chamber.

This causes a delay before the current response (a negative

current signal due to reduction of ferricyanide) is observed.

The transient time, before reaching the saturation signal, is due

to the diffusion of the analyte in the chamber previously filled

with PBS buffer. After stopping the flow, the current response

significantly decreases due to purely diffusion-based mass

transfer. After starting a flow of PBS, the current response

reaches the previous value until the PBS completely rinse the

chamber removing the ferricyanide. As a result, the current

>TBCAS-2011-Sep-0136-Reg.R1<

5

response returns to the initial baseline. By repeating the same

experiment with different ferricyanide concentrations and

evaluating the magnitude of the current response with respect

to the PBS baseline, we have obtained the calibration curve of

Fig. 12, showing linear response with respect to concentration.

For each of the concentration values, the analyte was injected

three times in order to assess the repeatability of the

measurement, obtaining a deviation smaller than 1%. The

responses from 2 different WEs in the same chamber have

been recorded, giving very similar response with a deviation of

only 0.5%. The rms input noise in this experiment is

dominated by interferences caused by the peristaltic pumps

(see next section for further details) and can be quantified in

105pArms. Under these dynamic experimental conditions, the

detection limit is consequently of 1.6µM.

C. Analysis of the Noise Effects of Peristaltic Pumps

The sequence of amperometric tests has also allowed to

carefully explore the interplay between microfluidic operation

and electronic signal detection. The disturbing effect of the

peristaltic pumping on amperometric recordings is well known

and has previously been reported in the literature ‎[9]. To

highlight the effect of our peristaltic pumps, Fig. 13(a) depicts

a current signal resulting from the injection of 10mM

potassium ferricyanide in one microfluidic chamber. The

signal has been acquired with 10kHz bandwidth (from LPF,

see Fig. 6) at 40kHz sampling frequency to avoid aliasing

effects. Observing the black line signal of Fig. 13(a), the effect

of peristaltic pumping is evident: when the pumps are on

spikes are presents, when the pumps are off fluctuations

drastically reduce. Also by disconnecting the microfluidics

from the motors but still operating the motors, the fluctuations

reduce drastically as well, thus verifying that electromagnetic

interferences are not significantly contributing to the noise.

Instead, we attribute the main noise contribution to the

pulsating nature of the flow that produce a movement of

charged ions perturbing the electrode-electrolyte interface and

thus resulting in spikes of ionic current.

To investigate the effectiveness of heavy filtering in

reducing these spikes, a bandwidth reduction down to 5Hz

(using a simple mobile average digital filter with 100ms

period) has been applied. A value lower than 5Hz would have

affected the tracking of relevant biochemical signals. The

positive effect of the 5Hz filter in smoothing the curve can be

seen (gray curve) in Fig. 13(a). Nevertheless, the effect of the

pumps has not been eliminated as the residual rms noise can be

quantified in 105pArms. As a comparison, Fig. 13(b) shows the

system response to a 100pA input step after the same 5Hz

filtration when the signal is applied to the current sensing

amplifier through a 4.7nF input capacitance and the flow is not

playing a role. From Fig 13(b) a 5pArms noise level can be

quantified as the intrinsic current resolution of the system. The

residual noise contribution of the peristaltic pumps can be thus

considered the limiting noise level of our system when

operating dynamic measurements.

D. Impedimetric Tracking of Cell Proliferation

Dopaminergic PC12 cells have been used as a robust model

cell line for preliminary cell-based assays with the proposed

platform. The proliferation of PC12 has been continuously

monitored using automatic EIS tracking. To further

demonstrate the versatility of the instrument, commercially

available electrodes (8W2x1E array by Applied Biophysics)

have been employed. These arrays contain 16 gold disk

electrodes with 250µm nominal diameter, corresponding to a

WE area of 4.9·104µm

2, i.e. 5nF, compatible with the

instrument capability to handle a large input capacitance.

Before starting the experiment, the sterilized chambers were

coated with laminin to promote cellular adhesion. Dulbecco’s

Modified Eagle’s medium (F12/Glutamax) supplemented with

15% horse serum, 2.5% fetal bovine serum and 1%

penicillin/streptomycin was used as culture medium. About

2.5·105 cells have been initially seeded into each well and they

were cultured for 7 days (at 37° C, in a humidified atmosphere

with 5% CO2). During the first 3 hours after cell seeding,

impedance spectra were recorded every 13 minutes to observe

in more detail the cell settling and further on cell adhesion and

spreading. Then spectra were acquired every hour, while

taking a picture of the electrodes with a bright-field

microscope every 24 hours in correspondence with the change

of culture medium. 30 points were acquired in the frequency

range 100Hz-100kHz, each with an averaging time of 2s that

provided a sufficient noise reduction.

The amplitude of the applied sinusoidal signal was set to

only 215µV in order to limit the current flowing through the

cells below a safety threshold value (250nA for PC12, 1A

for HeLa) without the inconvenience of an external series

resistor, commonly adopted ‎[17]. Fig. 14(a) reports the

normalized impedance spectra taken during the experiment.

For each time instant, the normalized impedance is defined as

Znorm(t,f) = (|Z(t,f)| - |Z(0,f)|)/ |Z(0,f)|; where |Z(t,f)| is the

magnitude of the impedance spectrum acquired at the given

time instant. This method is commonly used to highlight

relative variations of the impedance magnitude with respect to

the impedance magnitude measured at the beginning of the

experiment (|Z(0,f)|). Consistently with what was reported in

‎[18], it has been possible to determine a peak frequency

(69kHz, approximately at the corner between the double layer

capacitive slope and the solution resistance plateau), where

impedance shows the highest sensitivity to the population

growth. The initial decrease of the impedance magnitude is

most probably due to the thermal transient occurred to the

system after it has been placed inside the incubator. The

increase in impedance magnitude at high frequency is

correlated with the electrode coverage fraction, as confirmed

by optical images shown in Fig. 14(b). It reaches a maximum

of 35% at confluence (t = 125 hour) and then slightly

decreases when some cells start to detach.

E. Dopamine Detection

In parallel to the demonstration of the platform capability to

host PC12 cultures for several days and monitor their growth

in a real-time and label-free manner, preliminary tests with

>TBCAS-2011-Sep-0136-Reg.R1<

6

cell-related analytes have been performed. In particular,

among various neurotransmitters secreted by cells, dopamine

is of major interest being a clear and early indicator of the

differentiation of stem cells into adult neural cells. As shown

in Fig. 15(a), the presence of dopamine can be detected by

means of impedance spectroscopy (VDC = 75mV). In this case,

performing a full spectrum from 10mHz to 100kHz of a

diluted solution of dopamine in PBS (5µM), a charge transfer

resistance of 140MΩ and a mass-transport Warburg term

become apparent at low frequency.

As impedance spectroscopy is not specific to the redox

reaction, being measured at a constant VDC potential, selective

detection of catecholamines can be performed by means of

cyclic voltammetry. In Fig. 15(b), the same concentration of

dopamine is detected: the oxidation and reduction peaks

(ΔI~2nA) are visible (at a scan rate of 100mV/s), with respect

to a background signal measured with buffer only (PBS). In

this measurement, the redox cycling mechanism achieved with

the pair of interdigitated working electrodes is also shown. The

bipotentiostat allows performing cyclic voltammetry both in

single mode and in cyclic mode. In the latter, WEb is scanned

while WEa is biased at the reduction potential (0V). A current

increase of a factor 1.5 is observed, supported by an extra

cycling current of -2nA measured at WEa. Further biological

experiments performed by means of the proposed platform,

such as the amperometric detection of the dopamine exocytosis

from a population of PC12 cells, are reported in ‎[22].

VI. CONCLUSION

A novel and compact 24-channel instrument tailored for an

array of planar microelectrodes for multi-parametric

quantitative electrochemical measurements has been presented.

The bipotentiostat is originally integrated with a microfludic

platform to allow automatic monitoring of cell cultures.

Though designed for a specific combination of ad-hoc

microfrabicated electrodes and microfludics, meant for opto-

electrical monitoring of neural stem cell cultures, it is suitable

for routinary use in biology investigations, provinding a user-

friedly software and being compatile with a wide range of

commercial arrays of electrodes. The experimental

characterization of the portable system confirmed the

achievement of the design specifications (100kHz bandwidth,

few pA resolution). Continuous operation during 7 days of

impedimetric tracking of PC12 proliferation and dopamine

detection in the µM range have been sucessfully demostrated.

REFERENCES

[1] F.H. Gage, “Mammalian neural stem cells,” Science, vol. 287, pp. 1433-

1438, 2000.

[2] O. Lindvall, Z. Kokaia and A. Martinez-Serrano, “Stem cell therapy for

human neurodegenerative disorders- how to make it work,” Nat. Med.,

vol. 10, pp. S42-S50, 2004.

[3] F.J. Sigworth and K.G. Klemic, “Microchip Technology in Ion-Channel

Research,” IEEE Trans. Nanobiosci., vol. 4, no. 1, pp. 121-127, 2005.

[4] P. Weerakoon, E. Culurciello, K. G. Klemic, and F. J. Sigworth, “An

Integrated Patch-Clamp Potentiostat With Electrode Compensation,”

IEEE Trans. Biomed. Circuits Syst., vol. 3, pp. 117-125, 2009.

[5] L. Berdondini, K. Imfeld, A. Maccione, M. Tedesco, S. Neukom, M.

Koudelka-Hep and S. Martinoia, “Active pixel sensor array for high

spatio-temporal resolution electrophysiological recordings from single

cell to large scale neuronal networks,” Lab Chip, vol. 9, pp. 2644-2651,

2009.

[6] R.M. Wightman, “Probing cellular chemistry in biological systems with

microelectrodes,” Science, vol. 311, pp. 1570-1574, 2006.

[7] I. L. Jones, P. Livi, M. K. Lewandowska, M. Fiscella, B. Roscic, A.

Hierlemann, “The potential of microelectrode arrays and

microelectronics for biomedical research and diagnostics,” Anal.

Bioanal. Chem., vol. 399, pp. 2313-2329, 2011.

[8] L. Li, X. Liu, W. A. Qureshi, and A. J. Mason, “CMOS Amperometric

Instrumentation and Packaging for Biosensor Array Applications,”

IEEE Trans. Biomed. Circuits Syst., vol. 5, pp. 439–448, 2011.

[9] J. Gottschamel, L. Richter, A. Mak, C. Jungreuthmayer, G. Birnbaumer,

M. Milnera, H. Brückl, and P. Ertl, “Development of a disposable

microfluidic biochip for multiparameter cell population measurements,”

Anal. Chem., vol. 81, pp. 8503-8512, 2009.

[10] X. Yue, E.M. Drakakis, M. Lim, A. Radomska, H. Ye, A. Mantalaris,

N. Panoskaltsis, and A. Cass, “A real-time multi-channel monitoring

system for stem cell culture process,” IEEE Trans. Biomed. Circuits

Syst., vol. 2, pp. 66-77, 2008.

[11] C. Spegel, A. Heiskanen, L.H.D. Skjolding, and J. Emnéus, “Chip based

electroanalytical systems for cell analysis,” Electroanalysis, vol. 20, pp.

680-702, 2008.

[12] A.J. Bard and L.R. Faulkner, Electrochemical methods: fundamentals

and application, 2001.

[13] C. Spégel, A. Heiskanen, J. Acklid, A. Wolff, R. Taboryski, J. Emnéus,

and T. Ruzgas, “On-Chip determination of dopamine exocytosis using

mercaptopropionic acid modified microelectrodes,” Electroanalysis,

vol. 19, pp. 263-271, 2007.

[14] P.O. Bagnaninchi and N. Drummond, “Real-time label-free monitoring

of adipose-derived stem cell differentiation with electric cell-substrate

impedance sensing,” PNAS, vol. 108, pp. 6462-6467, 2011.

[15] H.E. Park, D. Kim, H.S. Koh, S. Cho, J.-S. Sung, and J.Y. Kim, “Real-

time monitoring of neural differentiation of human mesenchymal stem

cells by electric cell-substrate impedance sensing,” J. Biomed. Biotech.,

p. 485173-485191, 2011.

[16] P. Skafte-Pedersen, D. Sabourin, M. Dufva, and D. Snakenborg, “Multi-

channel peristaltic pump for microfluidic applications featuring

monolithic PDMS inlay,” Lab Chip, vol. 9, pp. 3003-3006, 2009.

[17] I. Giaever and C.R. Keese, “Monitoring fibroblast behavior in tissue

culture with an applied electric field,” PNAS, vol. 81, pp. 3761-3764,

1984.

[18] L. Wang, L. Wang, H. Yin, W. Xing, Z. Yu, M. Guo, and J. Cheng,

“Real-time, label-free monitoring of the cell cycle with a cellular

impedance sensing chip,” Biosens. Bioelectron., vol. 25, pp. 990-995,

2010.

[19] M. Paeschke, U. Wollenberger, C. Köhler, T. Lisec, U. Schnakenberg

and R. Hintsche, “Properties of interdigital electrode arrays with

different geometries”, Anal. Chim. Acta, vol. 305, pp. 126-136, 1995

[20] R.M. Wightman and D.O. Wipf, “High-speed cyclic voltammetry,”

Accounts Chem. Res., vol. 23, pp. 64-70, 1990.

[21] A. Heiskanen, C. Spégel, J. Tønnesen, Z. Fohlerova, L. Goulart, J.

Hansen, M. Kokaia, M. Dufva, and J. Emnéus, “Development of a

microfluidic on-line culture system for combined electrochemical and

optical real-time detection of cellular processes,” Proc. of µTAS 2008,

San Diego (USA), vol. 2, pp. 1168-1171, 2008.

[22] K. Zór, M. Vergani, A. Heiskanen, E. Landini, M. Carminati, V.

Coman, I. Vedarethinam, P. Skafte-Pedersen, M. Skolimowski, A.

Martinez Serrano, M. Kokaia, T. Ramos Moreno, A. Ghio, W.E.

Svendsen, M. Dimaki, Zs. Kerestes, M. Adamovski, U. Wollenberger,

D. Sabourin, G. Ferrari, R. Raiteri, M. Sampietro, M. Dufva, J. Emnéus,

"Real-time monitoring of cellular dynamics using a microfluidic cell

culture system with integrated electrode array and potentiostat.," Proc.

of µTAS 2011, Seattle (USA), vol. 1, pp. 1532-1535, 2011.

[23] M. Carminati, G. Ferrari, and M. Sampietro, “Attofarad resolution

potentiostat for electrochemical measurements on nanoscale

biomolecular interfacial systems.,” Rev. Sci. Instrum., vol. 80, pp.

124701:1-10, 2009.

>TBCAS-2011-Sep-0136-Reg.R1<

7

Marco Vergani, born in 1985, received the Bachelor

degree cum laude from Politecnico di Milano (Italy) in

2007 and a Double Master of Science cum laude from

the same university and Politecnico di Torino in 2009.

In January 2010 he started a PhD program in

Information Engineering at Politecnico di Milano. He

is currently working on low-noise instrumentation for

electrical measurements on biological samples.

Marco Carminati, born in 1981, received B.Sc. and

M.Sc. in Electronic Engineering, both cum laude from

the Politecnico di Milano (Italy), in 2003 and 2005

respectively. In 2008 he spent a semester at MIT

working on BioMEMS and microfluidics. In 2009 he

completed his PhD in Electronics and Information

Science at DEI, Politecnico di Milano where he is

currently a post-doctoral fellow working in the field of

bio-chemical sensors and instrumentation.

Giorgio Ferrari, born in 1973, received the Master

degree and the Ph.D. degree from Politecnico di

Milano in electronics engineering in 1999 and 2003,

respectively. Since 2005, he has been an Assistant

Professor of electronics at Politecnico di Milano. The

focus of his research activities is on the development of

novel instrumentation to probe electrical properties of

materials and biosamples at the nanoscale. He has

coauthored more than 30 peer-reviewed publications

and three patents.

Ettore Landini, born in 1984, received his Master

Degree in Bioengineering in 2008 from the University

of Genova He is now taking his PhD in

Nanotechnologies at the Department of Biophysical

and Electronic Engineering (University of Genova). He

works mainly with the Atomic Force Microscope to

investigate the changes in the mechanical properties of

articular cartilage due to degenerative pathologies like

osteoarthritis.

Claudia Caviglia, born in 1985, received her Bachelor

degree in Biomedical Engineering from University of

Genova in 2008 and her Master degree cum laude in

Bioengineering from the same university in 2010. In

May 2011, she started a PhD program at DTU,

Nanotech. She is currently working on development of

safer non-viral gene transfection vectors.

Arto Heiskanen, born in 1965, received B.Sc in

biochemistry in 1987 at Åbo Akademi, Finland and

B.Sc. in chemistry in 2001 at University of the

Philippines. In 2004, he received M.Sc. and in 2009

PhD at the Department of Analytical Chemistry of

Lund University, on electrochemical monitoring of

cells. Currently, he is a Assistant Professor at DTU

Nanotech, focusing on the development of microfluidic

systems. He has co-authored 20 peer-reviewed

publications.

Clément Comminges, born in 1981, received his

Master degree in electrochemistry in 2004 from the

University Pierre et Marie Curie, Paris (France) and

Ph.D in electrochemistry in 2007 from the University

Paris 12 (France). Since 2010, he has a Postdoctoral

position in Potsdam University (Germany) in the group

of molecular enzymology. His research interests are the

sensitive electrochemical detection of biological

relevant compounds and biosensors.

Kinga Zór born in 1980, received B.Sc. from Babeş-

Bolyai University (Romania) in 2003. Currently she is

a PhD student at the Department of Biotechnology,

Lund University (Sweden). Her research is focused on

electrochemical (bio)sensors in miniaturized systems

for bioanalysis and on application of microfluidic on-

line cell culture systems for real-time monitoring of

cellular dynamics. During 2011 She spent a six months

period at DTU Nanotech.

David Sabourin, born in 1975, received both B.Sc. in

Biochemistry and B.A.Sc. in Chemical Engineering

from the University of Ottawa (Canada) in 1999. From

1999 to 2007 he held a position at i-STAT, a

manufacturer of microfluidic diagnostic devices. From

2007 to 2010 he completed his PhD at DTU Nanotech

developing a microfluidic platform with innovative

scalable interconnection and pumping solutions.

Martin Dufva, born in 1967, received master degree

in molecular biology in 1998 and PhD degree at the

medical faculty of University of Göteborg (Sweden) in

2001. He worked with DTU Nanotech since 2001 and

since 2005 holds a position as Associate Professor. His

research interests are microsystems for cell and

molecular biology research and actuation and

technologies for cancer and genetic diagnostics. He has

co-authored >50 publications.

Maria Dimaki received an M.Eng. from the National

Technical University of Athens, Greece, in 2000 and

an M.Sc. from the Imperial College, London in 2001.

She received her Ph.D. in 2005 from the Technical

University of Denmark. She is currently an associate

professor with the Nano-Bio Integrated Systems group

(NaBIS) at DTU Nanotech. Her research interests

involve the development of micro- and nanoelectrodes

in microfluidic systems for electrophysiological and

electrochemical measurements on neuron cultures.

Roberto Raiteri got his Master Degree in Electronic

Engineering in 1993 from the University of Genova

and his PhD in electronics in 1997 from University of

Trento. Since 2001 he is Assistant Professor in

Bioengineering at the University of Genova. His main

research interests deal with the use of Scanning Probe

Microscopy-based methods to characterize

biomolecules and biological tissues at the nanometer

scale. He is co-author of 32 peer-reviewed international

publications.

Ulla Wollenberger, received her PhD in Enzymology

from the Academy of Sciences of the GDR in 1984. In

1994 she joined the University Potsdam (Germany).

Currently she is group leader in the Department of

Molecular Enzymology and is working in the field of

bioelectrochemistry, biosensors and bioelectronic. She

published more than 120 peer reviewed papers and a

textbook.

Jenny Emnéus, born in 1961, received M.Sc. in 1986

and PhD in 1992 at Lund University, where she also

held the position of Associate Professor until 2007. In

2007, she became Professor at the Department of

Micro- and Nanotechnology, DTU, to lead research

focusing on development of microfluidic bioassay

systems. She has coordinated 6 European projects.

During her research career, she published over 100

peer-reviewed publications.

>TBCAS-2011-Sep-0136-Reg.R1<

8

Marco Sampietro, born in 1957, received his Master

Degree in Nuclear Engineering in 1982 from

Politecnico di Milano where he is now full professor of

Electronic Circuits and Devices. He is responsible for

the activities in high-sensitivity instrumentation for the

nanoscience. He is co-author of more than 150 peer-

reviewed international publications. He has been vice-

dean of the Faculty and the coordinator of many

national and international research projects and

scientific partner in 4 European projects.

Microfluidics

Multichannel Potentiostat

Acquisition unit

Array of cell culture chambers

USB

Control and analysis software

Fig. 1 Schematic view of the whole system architecture showing the

electronic board tailored onto the cell culture chambers and coupled to

microfluidics in a compact assembly placed under an upright microscope.

Spring

pins

Cells

Microscope

PMMA

Board

Silicon tape

Liquid inlet

MicroelectrodesChip

Screws

Contact pads

2mm

0.8mm

Fig. 2 Cross-section of the electronic board at the cell culture site integrated

with the array of planar microelectrodes and the microfluidic culture system.

Fig. 3 Photograph of the complete instrumentation platform. The board is

black to minimize autofluorescence. (Inset 1: enlargement of a sensing site of

the MEA; inset 2: window enabling microscope observations).

CEsREs

WE1-6,a

WE1-6,b

WE7-12,a

WE7-12,b

NI-USB

6259MEA

Current reading

amplifier

Potentiostatic

circuit

RE

WEa CE

Sensing site

ADC

DAC

Current reading

amplifier

Current reading

amplifier

Current reading

amplifier

WEb

Fig. 4 Multichannel potentiostat architecture comprising the potentiostatic

circuit driving the RE/CE and 4 current amplifiers connected to the 24 WEs

hosted in the MEA chip, through low-parasitics analog switches.

+

-To

CE+

-

200 Ω

1 kΩ

20 kΩ

From

RE

+

-

+

-

OPA4141

DG

406

VAC

VDC

100

V-

5

V- ACDC

OPA124

2/3 electrode

selector

DG

406

1.

.

.

.

12

1.

.

.

.

12

THS4032

Adder

Fig. 5 Detailed architecture of the potentiostatic circuit: the DC and AC

signal, generated by two DACs, are summed with different attenuation (5 and

100) to allow high accuracy. The potentiostatic loop is closed by OPA124.

1 kΩ

3.9 nF

10 kΩ

390 pF

100 kΩ

10 pF

+

-ADA4817

+

-

INA128

x 10

Full BW

LPFADG

1234

IinIin

VWE VWE

1.

.

.

.

.

6

Dout

Fig. 6 Detailed architecture of the current reading chain composed by the

input multiplexer, a transimpedance amplifier and INA-based second stage.

>TBCAS-2011-Sep-0136-Reg.R1<

9

100 1k 10k 100k 1M100

1k

10k

100k

1M

10 kHz Channels

Tra

nsim

pedance (

)

Frequency (Hz)

100 kHz Channels

Fig. 7 Transfer function of the four Current Reading Amplifiers operating at

full bandwidth (100kHz) or at increased resolution (10kHz).

100 1k 10k 100k100f

1p

10p

Input N

ois

e (

A/s

qrt

(Hz))

Frequency (Hz)

Cin = 56 pF

Cin = 560 pF

Cin = 5.6 nF

Fig. 8 Spectral density of the input referred current noise measured with

different input capacitances. Scatter points represent the values obtained by

theoretical simulations, while the measured values are plotted with continuous

lines. With a 5.6nF input capacitance, the capacitance resolution estimated

from this spectrum is 2fF (VAC = 10mV at 10kHz, 1s measurement time).

1 10 100 1k 10k 100k 1M100

1k

10k

100k

1M

10M

Agilent LCR:

Coplanar

Vertical

Multichannel Potentiostat:

Coplanar

Vertical

Imp

ed

an

ce

Ma

gn

itu

de

)

Frequency (Hz) Fig. 9 Test result of the impedance measurement. The impedance spectra

recorded with the designed multichannel platform overlap the measurement

performed with a stand-alone commercial instruments (Agilent Precision LCR

Meter E4980A) both for coplanar and vertical measurement technique.

-0.4 -0.2 0.0 0.2 0.4-4.0µ

-3.0µ

-2.0µ

-1.0µ

0.0

1.0µ

2.0µ

3.0µ

4.0µ

Cu

rre

nt

(A)

Voltage (V) Fig. 10 Cyclic voltammograms recorded on a whole array of WEs. A 10mM

solution of ferricyanide (K3Fe(CN)6) and ferrocyanide (K4Fe(CN)6) in PBS

was used as the electrolyte. The recordings were made at the scan rate of

50mV/s vs the on-chip gold RE.

0 20 40 60 80 100-12.5n

-10.0n

-7.5n

-5.0n

-2.5n

0.0

2.5n PBSStopPBS

Cu

rre

nt

(A)

Time (s)

Analyte

Fig. 11 Amperometric recording of the injection of a 62.5µM potassium

ferricyanide solution (flow rate 34µl/min, WE biased at -350mV vs gold, 5Hz

low pass digital filter).

100µ 1m 10m

10n

100n

1µ

WEa

WEb

Linear fit

Cu

rre

nt

(A)

Concentration (M)

Fig. 12 Calibration curve for the amperometric detection of potassium

ferricyanide injected at the flow rate of 34µl/min. Both WEs were biased at -

350mV vs gold RE.

>TBCAS-2011-Sep-0136-Reg.R1<

10

Fig. 14 One week tracking of the proliferation of a population of PC12 cells: (a) plot of normalized impedance spectra recorded over time and (b)

corresponding microscopic images taken at the indicated hour.

0 10 20 30 40-3.5µ

-3.0µ

-2.5µ

-2.0µ

-1.5µ

-1.0µ

-500.0n

0.0

500.0n

1.0µ

1.5µ

(b)

C

urr

ent

(A)

Time (s)

Pumps ON Pumps ON

(a)

0 10 20 30

0

50p

100p

150p

Curr

ent

(A)

Time (s)

Fig.13 (a) Amperometric tracking showing the disturbing effect of peristaltic

pumping (black line is the signal acquired at 10kHz bandwidth whereas the

grey line is the result of a 5Hz filtration). (b) Response of the system to a

100pA input current step. The acquired signal has been digitally filtered with

5Hz bandwidth.

10m 1 100 10k

1k

10k

100k

1M

10M

100M

-0.2 0.0 0.2 0.4

-4

-2

0

2

4

6

8

10Warburg

CPE

Rsol

RCT

Imp

ed

an

ce

Ma

gn

itu

de (

)

Frequency (Hz)

WEb Buffer Only

WEb Single

WEb Cycling

WEa Cycling

Cu

rre

nt

(nA

)

Potential vs. Ag/AgCl (V)

5M DA

(a) (b)

Fig. 15 Detection of 5µM of dopamine (DA) with (a) impedance spectroscopy

(VDC = 75mV, VAC = 10mV) and (b) cyclic voltammetry (100mV/s scan rate).

The redox cycling operation is also demonstrated as the current signal at WEa

is -2nA during the oxidation of dopamine.