Electronics meets Biology

SensorsInterfacesHybridsNeuronal Networks

Silicon-based Biochemical Sensors · · · · · · · · · · · · · · · · · · · 165

Cell-Transistor-Hybrid Systems · · · · · · · · · · · · · · · · · · · · · · 171

Biological Neuronal Networks · · · · · · · · · · · · · · · · · · · · · · · 176

SQUIDs: The Ultimate Magnetic Sensors for Materials Characterization and Biomedical Diagnostics · · · · · · · · · · 181

163

164

Silicon-basedBiochemical Sensors

Silicon-based microelectronics represents the platform of ourmodern information technology. In recent years, silicon technolo-gy has been utilized to couple data processing systems to chem-ical and biological structures, integrating ion-selective materialsand simple biomolecules or even cells and cell systems.The main advantage of these (bio-)chemical sensors is the highsensitivity and selectivity of their chemical and biological com-ponent as well as the possibility of miniaturization down to thenanometer scale. (Bio-)chemical sensors have been developed asrugged and reliable devices for the rapid and quantitative detec-tion of specific analytes. For example, enzymes allow to monitorthe blood glucose concentration of diabetic patients, a pH elec-trode may adjust the proper fermentation routine for cheeseproduction and sensors and catalysts control the car pollution.(Bio-)chemical sensors constitute an interdisciplinary interfacebetween the environment and data processing systems. More-over, because these sensors can be designed in a modular con-cept, the combination of single sensors to sensor arrays is possi-ble.

We present some examples of new silicon-based (bio-) chemicalsensors, which have been developed in a collaboration betweenISG (FZJ) and the University of Applied Sciences Aachen (Jülichdivision):• capacitive field-effect sensors as a combination of ionophores

or enzymes and silicon technology,• a silicon-based multi-parameter hybrid ion-sensitive FET

(ISFET) module suitable for sensor arrays,• a beetle/chip biologically sensitive field-effect transistor

(BioFET) as a first step towards a bioelectronic device with extraordinary sensory abilities.

All described (bio-)chemical sensors utilize the field effect totransfer the detected (bio-)chemical information to an electricalsignal.

165

INTRODUCTION

APPROACH

Capacitive field-effect sensors combine the physical field effectas transducer principle and the chemical recognition mechanismof ion-selective electrodes. In this sensor the charge carrier dis-tribution at the interface between insulator and semiconductor(for example SiO2 and p-type Si) can be either controlled by anexternal dc and ac voltage (Fig. 1a) or a dc voltage combinedwith a probe light (Fig. 1b). Depending on the read-out principleof the capacitance of the space charge layer, the sensor is calledEIS (electrolyte-insulator-semiconductor) sensor or LAPS (light-addressable potentiometric sensor). The formation of charged

adsorbed species at the phaseboundary sensitive layer/electro-lyte results in an electrochemicalinteraction, which is in bothcases accompanied by a measurable surface potential.Usually, the sensor is operatedvs. a miniaturized Ag/AgCl refer-ence electrode. Within the lastyears, different sensitive layersand substrate combinationshave been developed in our

group: pH-sensitive layers consisting of Al2O3 or Ta2O5 allow todetermine pH variations with high accuracy in aqueous solutionsin the concentration range from pH 2 to pH 12. Here, the pulsedlaser deposition technique has been introduced as an innovativefilm preparation method. The average pH sensitivity of themicrofabricated pH sensors is about 56-58 mV/pH and the long-term stability extends to more than 3 years.The immobilization of additional enzyme layers (e.g., penicilli-nase, organophosphorus hydrolase, alliinase) on top of themicrofabricated pH sensor leads to a simple and successfulbiosensor concept. The enzymes take advantage of the catalytichydrolyzis of their biological substrate releasing H+ ions, whichcan be detected with high efficiency by the underlying pH-sensi-tive layer. Thus, the resulting sensor signal is a measure of thesubstrate concentration in the solution: the enzyme penicillinasecatalytically converts penicillin, while organophosphorus hydro-lase is sensitive towards pesticides, and alliinase detects garlic.

166

RESULTS

Fig. 1Sketch of a capacitive field-effectsensor for the determination ofions or biomolecules in aqueoustest samples. Various sensitivelayers and materials as well assubstrate configurations havebeen developed. The field-effectsensor can work as an EIS (elec-trolyte-insulator-semiconductor)structure (a) or as LAPS (light-addressable potentiometric sen-sor) device (b).

By using these three enzymes, the applications span the areasbiotechnology, environmental monitoring and biomedicine.Instead of biomolecules, specific ionophores supported in a PVC-based membrane as sensitive layer allow to determine variousmono- and divalent ions, like K+, Li+, Cs+, Mg2+ and Ca2+.The major task, however, in view of the combination of biomole-cules and microelectronics is the stable coupling of the enzymesonto the field-effect structures. Complex surface chemistry(immobilization by heterobifunctional cross-linking) or sophisti-cated physical binding, like the structuring of the sensor surfaceby etching, is developed. For example, we favour porous siliconwith its sponge-like structure, because the pore size and geome-try can easily be tailored for the attachment and anchoring ofbiomolecules, permitting the design of a three-dimensional (bio-)chemical sensor.

For the capacitive field-effect sensors, the signal is derived fromcapacitance-voltage measurements, as shown for the EIS sensorin Fig. 2a. When a substrate variation occurs in the test solution,e.g. from c1 to c2, the curves shift along the abscissa, accordingto the sensitivity of the transducer. The value of the shift givesthe variation in the concentration. For the LAPS device, one canobtain the same concentration-dependent shift, but with the distinction that the photocurrent serves as the sensor signal (Fig. 2b).The requirement to measure several parameters in complex mix-tures consequently leads to the development of miniaturizedsensor arrays or systems, like the µTAS (micro total analysis sys-tem) or “lab-on-a-chip” sensors. A sensor system for the multi-parameter detection can be built-up monolithically or hybrid.Different concepts for hybrid modules are discussed in literature.However, all modules contain both (bio-)chemical sensors andphysical sensors for the additional control of physical parame-ters as temperature, conductivity, flow rate or liquid level. Sincethe design of the physical sensors differs from that of the (bio-)chemical sensors, the complete sensor system needs amore complicated fabrication technology and read-out scheme.

167

Fig. 2Schematic sensor response ofan EIS sensor (a), a LAPSdevice (b) or an ISFET (c) withvarying ion concentrations inthe solution. The concentrationto be determined increasesfrom c1 to c2.

EIS(a)

LAPS(b)

ISFET(c)

In order to ease these demands, we have suggested a novelapproach for a multi-functional hybrid sensor module using thesame transducer principle and structure for both (bio-)chemicaland physical parameters. In this hybrid module, the (bio-)chemi-cal sensor can serve as physical sensor and thus, the number ofobtained (bio-)chemical and physical parameters exceeds thenumber of sensors. The setup is called a “high order system”.Fig. 3a shows the schematic configuration of such a hybrid mod-ule with a Ta2O5-gate ISFET as transducer. The module (Fig. 3b)consists of two ISFETs, or one ISFET and one EnFET (enzyme-

modified FET), and an ion generator (Au or Pt). On the basis ofthis module, it is possible to realize at least three (bio-)chemicalsensors (pH, ion and/or enzyme sensor), five physical sensors(flow velocity, flow direction, diffusion coefficient, temperature,liquid level) and one actuator (ion or gas generator). The multi-functionality of this array is achieved by different sensorarrangements and operation modes. Moreover, this array mini-mizes the influence of various disturbing factors by using a dif-ferential measuring set-up. For this ISFET-type array, the concen-tration-dependent sensor signal can be evaluated by the shift ofthe gate voltage at a fixed drain current (Fig. 2c).The beetle-chip BioFET is a completely new type of biosensor inwhich the antenna of an insect is coupled via an electrolyte to

168

Fig. 3The multi-parameter hybrid module enables the detec-tion of both (bio-)chemicaland physical quantities (a).Video microscopic picture ofthe developed hybrid sensor module consisting oftwo ISFET chips and an ion generator onto a printed circuit board before encap-sulation by an epoxy resin (b).

(a)

(b)

the gate of a FET (Fig. 4a). Such a bioelec-tronic interface possesses extraordinary sen-sory abilities, because the biological parthas been optimized by millions of years of evolu-tion. The organic compound detected by the beetle’s antennainitiates a recognition process that results in a net potential: ifodour molecules reach the antenna by diffusion, they are able tobind to specific odour-binding proteins in the neurones of theantenna. The complex of odour and protein then opens ionchannels in the cell membranes, yielding membrane potentials.Triggered by these protein-driven reactions, an electrical polar-ization develops, resulting in a sum dipole over the whole anten-na. As a consequence, the channel conductivity of the FET andthus its drain current is modulated (Fig. 4b). The more intensethe odour stimulus is, the stronger are the recorded signals.By using the antennae from selected insects (e.g., Coloradopotato beetle, steelblue jewel beetle), different odours such as Cis-3-hexen-1-ol, guaiacol and 1-octen can be detected.Plant damages by the attack of a potato beetle can be discrimi-nated by setting a threshold, which allows to distinguish

169

Fig. 4Beetle-chip sensor as an example of a bioelec-tronic field-effect sensor. The tip of the beetle’santenna contacts the gate of the FET via anelectrolyte; the resulting drain current changesare amplified (a). Characteristic drain currentchanges (abscissa) derived from different odourconcentrations (e.g., Cis-3-hexen-1-ol) as a function of time (b).

(a)

(b)

between healthy and damaged plants. Besides the detection ofplant diseases, an optimized use of pesticides and a more effi-cient food protection due to the determination of mycotoxine-producing organisms,e.g., becomes feasible. A BioFET based onthe antenna of the steelblue jewel beetle can specifically detectburning wood as in forest fires with unrivalled sensitivity and asuperior selectivity, as compared to semiconductor gas sensors.A “library” with different beetle species and their odour detec-tion spectra will broaden the field of application. It is a remain-ing challenge to isolate of the receptor molecules inside theantennae in order to fabricate these biosensors without the useof the beetles or their antenna. Then the entire molecular recog-nition centre will be reconstructed on the silicon microchip, lead-ing to the development of “nano-biosensors”.

The author thanks Y. Ermelenko, M. Keusgen, A. Mulchandani, A.Poghossian, P. Schroth, J. Schubert, S. Schütz, A. Steffen, M. Thustand T. Yoshinobu for technical support and valuable discussions.

(1) “Novel approaches to design silicon-based field-effect sensors”M.J. Schöning in: J.W. Schultze, T. Osaka, M. Datta (eds.) Electrochemical Microsystem Technologies, New Trends in Electrochemistry Vol. 2, Taylor & Francis, 2002,London New York, pp. 384-408.

(2) “(Bio-)chemical and physical microsensor array using an identical transducer principle”A. Poghossian, H. Lüth, J.W. Schultze, and M.J. SchöningElectrochim. Acta 47 (2001) 243-249.

(3) “Electrochemical methods for the determination of chemical variables in aqueous media”M.J. Schöning, O. Glück, and M. Thustin: J.G. Webster (ed.) The measurement, Instrumentation and Sensors Handbook, CRC Press, 1999, Boca Raton and Springer-Verlag Heidelberg, pp. 70/1-70/49.

Michael J. Schöning(ISG, FZJ, and University of Applied Sciences Aachen,Jülich Division)

170

ACKNOWLEDGEMENTS

REFERENCES

AUTHOR

171

Cell-Transistor-HybridSystems

Biological cells are able to receive, process, and transmit infor-mation. Connecting these cells to micro-electronic circuits opensup exciting new perspectives in bioelectronics, information tech-nology, medical engineering and in sensor development. Livingcells possess receptors of unmatched sensitivity that detectexternal signals of chemical nature (nutrients, hormones, neuro-transmitters, changes in proton- or ion-concentration, etc.) or physical stimuli as a change in temperature, light, mechanicalforce, or even electromagnetic fields. These input parameters areprocessed by the cells. The internal “machinery” of the cellincludes signal amplification cascades and logic connections ofhigh non-linearity, but the details remain to be unveiled. Theresulting output signal may generate many physiological reac-tions inside the cell, as the synthesis of specific molecules, achange in gene expression or the storage of certain substances.The output signals also allow the cell to communicate with itsenvironment and with other cells.

In order to provide selective long-term cell-transducer interfacesin vitro, microtechnology is used for the development of planararrays with large numbers of field-effect transistors or metalelectrodes in the size of the individual cells. These arrays usuallyconsist of a culture chamber with embedded chip. For metal-electrode arrays (MEAs), insulated conductor paths are patternedlithographically. Their opened metallic ends form the sensingelectrodes. In addition, field-effect transistor (FET) arrays havebeen developed to record the electrical signals from cells. Modi-fications of standard FET fabrication processes lead to deviceswith metal-free gate electrodes. A variation of these devices isthe so-called ion-sensitive field-effect transistor (ISFET). Its gatedielectric is modified to yield higher sensitivity for certain ions.Sufficient electrical coupling between the cell and the electrodefor extracellular signal recording is achieved only if a cell or apart of a cell is located directly on top of the electrode.

INTRODUCTION

APPROACH

Figure 1(a) shows a photograph of the complete device with theactive chip mounted onto a standard chip carrier. A mini-petridish is sealed to the chip to allow the cells to be cultured in con-tact with the FET gate electrodes. A magnified picture of thechip surface is given in Fig. 1(b). Our device integrates 16 tran-sistors in a 4x4 array. They share a common source but haveindividual drain connections for parallel read-out. The gate areasare covered by non-metallized SiO2 , optimized for high sensitivi-ty, low noise and low signal drift.

172

Fig. 1(b)Micrograph of the array of4x4 field-effect transistorsused in these studies

Fig. 1(a)Assembled and encapsulat-ed FET. The chip with the4x4 transistor array ismounted in the middle of amini cell culture dish.

Prior to cell seeding, the chip surface has to be functionalizedby a protein coating that promotes cell adhesion and survivaland, if necessary, allows for cell patterning. The employed strate-gies to achieve this goal range from mere physisorption fromsolution to complex supramolecular interfacial multi-layer archi-tectures.

The properly prepared chip surface can be used as a substratefor in-vitro cell culture. In particular we work with cardiacmyocytes (heart muscle cells) and primary neurons, both fromembryonic rats. The heart muscle cells offer the particular advan-tage that after random seeding they grow on the chip surface toa confluent monolayer. They establish both mechanical and elec-trical contacts between neighboring cells. Moreover, after a fewdays in culture they start to spontaneously contract and to fireaction potentials, which are at first completely random. Howev-er, eventually one cells becomes the “pacemaker” and triggersall others of that population to generate a synchronizedmechanical and electrical excitation pattern with a rather regu-lar repetition rate of approx. 1 Hz. The excitation wave propa-gates across the cell layerwith a velocity of approx.0.15 m/sec. The actionpotential of an individualcel, can be monitored by aclassical micropipetterecording unit and showsthe well known voltage-time profile given in Fig.2(a), resulting from time-dependent contributions ofNa+-, K+-, and Ca++-currentsacross the cell membrane.The patch pipette recordingshows an excellent signal-to-noise ratio. However, thecell does not survive thattreatment longer than a fewhours, and a multielectrode

173

Fig. 2Action potential recordings from a rat cardiac myocyte taken simultaneously with a glass micro-electrode (A) and a FET (B).

RESULTS

recording exceeding 3 electrodes/cell-culture at the same time isbarely possible. Recordings by our FET devices (one example isgiven in Fig. 2(b), which is the simultaneously measured event ofFig. 2(a)) show lower amplitude and a higher noise level due toa weaker coupling of the cell to the gate electrode as comparedto the patch pipette. However, they can be monitored for weeks,and with our present array design on 16 channels simultaneous-ly. As an example for the possible application of this cell-transis-tor hybrid system, we recently demonstrated the influence ofcertain drugs known to stimulate the frequency of the heartbeat. Isoproterenol (ISO), e.g., was shown to enhance the beatfrequency of our myocyte monolayer. A full dose-response curvecould be recorded within minutes. The limit for the responsedetection was shown to be at an ISO concentration of only10-11 mol/liter !After 4 days in culture, rat brain stem neurons randomly adheredto the substrate, with one neuron fortuitously positioned on topof a gate electrode (Fig. 3). It developed “healthy” dendrites andaxons, the “receivers” and “transmitters” for cell-cell communi-cation. These cells were sufficiently coupled to the gate elec-trode to allow for FET recording of an action potential that wastriggered by a patch electrode, but formed a totally random net-work structure.

174

Fig. 3Micrograph of a randomnetwork of brain stem neurons cultured on thesurface of a FET chip. Oneneuronal cell resides almostcompletely on the gateelectrode of the transistor(triple gate structure).

“Electrical recordings from rat cardiac muscle cells using field-effect transistors”C. Sprössler, M. Denyer, S. Britland, W. Knoll, A. Offenhäusser Phys. Rev. E 60 (1999) 2171-2176

“Cell-transistor hybrid systems and their potential applications”A. Offenhäusser and W. KnollTrends in Biotechnology 19 (2001) 62-66

“Cardiomyocyte-Transistor-Hybrids for Sensor Application”S. Ingebrandt, C.-K. Yeung, M. Krause, A. OffenhäusserBiosensors and Bioelectronics 16 (2001) 565

Sven Ingebrandt, Andreas Offenhäusser

175

REFERENCES

AUTHORS

176

Biological Neuronal Networks

The growth of neurons into networks of controlled geometry hasnumerous potential applications in cell-based biosensors, neuro-electronic circuits, neurological implants and pharmaceuticaltesting. It also permits fundamental biological studies aboutneuronal interactions. Within the neurons, differences in electri-cal potential encode information, and messages can be passedon to other neurons through chemical or electrical connections.In order to decipher the neural code, our aim is the constructionof networks of neurons in culture, with each neuron connectedto others through synapses. We intend to reproduce the connec-tivity of intact nervous systems. On the other hand, their inter-ruption leads to a loss of physiological function. It is assumedthat the messages passing across a synapse carry instructionsfor processing and possibly routing of the neuronal information.Therefore, we would like to test the effects of different connec-tivities and input messages on the neuronal output. These datapossibly allow to establish a basic system to model the in vivosituation of neuronal processing as closely as possible.

Different techniques can be used to produce surface patterns forcell guidance of high spatial resolution down to the micrometerscale. Examples are standard photolithographical patterning, acombination of laser ablation techniques with lithographicmasks or with topographic features. We have chosen the micro-contact printing (µCP), because it is a comparatively simple anduniversal method for patterning biomolecules. A stamp is pro-duced casting a silicone elastomer (polydimethylsiloxane, PDMS)in the desired pattern and then coated with a biomolecule solu-tion to be transferred. After contacting the “inked” stamp withthe substrate surface, the biomolecules are transfered to the pre-determined pattern.

Microstamps for the experiments were produced by photolithog-raphy and molding. Applying UV-photolithography, master

INTRODUCTION

APPROACH

stamps were produced out of spin coated thick photoresist lay-ers on silicon wafers. Polydimethylsiloxane (PDMS) microstampswere then fabricated curing the polymer on the master stamps(Fig. 1).

Inking took place by applying certain biomolecules as laminindissolved in phosphate buffer. The inked stamp was dried in asoft nitrogen air stream and immediately pressed onto the sub-strate.

Line and grid patterns were applied. These patterns have nodesfor the adhesion of cell bodies in equidistant steps of 100 µm inx-direction and 50 µm in y-direction. The nodes of the grid pat-terns are located at the intersections of the grid. The nodes ofthe line patterns are distributed along the lines. In order to studythe influence of the pattern geometry on the cellular growth,patterns with varying line size and node size were chosen (Fig.2).

177

Fig. 1Image of a PDMS stampused for the neuronal cellpatterning.

Fig. 2Fluorescence image of animprint of fluorescein-labeled biomolecules. The pattern consists of 5 µm wide linescrossing in 50 respective 100µm. The crossing points consistof nodes of 12 µm in diameter.

Dissociated neurons from rat embryos were seeded ontomicropatterned substrates. The substrates consisted of poly-styrene as a highly hydrophobic background onto which a pat-tern made of physiological proteins had been applied asdescribed before. The contrasting surface areas forced the cellsto align themselves along the pattern and to extend theirprocesses accordingly. We used grid patterns consisting of 4-6µm wide lines and nodes that were 12-14 µm in diameter. Inorder to avoid the deposition of serum proteins onto thehydrophobic background, which would reduce the contrastbetween the two areas, a serum-free medium was used. Asshown in Fig. 3, the aligning of the cell bodies onto the nodeswas achieved to a very high degree, as well as the guidance ofthe growth process along the lines. In the resulting network, anygiven cell could physically contact and thus form synapses withits direct neighbours on the pattern. Apart from morphological

and electrophysiological evidence, the neuronal identity of thecells was confirmed by antibody staining, using an antibodyagainst the neuron-specific cytoskeleton protein neurofilament-M. Our networks were stable for up to three weeks. After thistime, the cells tended to overgrow the pattern and form connec-tions across the non-permissive “forbidden” areas.

Electrophysiological recordings in the form of triple patch clampmeasurements were performed to test the synaptic connectivity.Two to three neighbouring cells were patched simultaneously inthe current clamp mode. An action potential was evoked in one

178

RESULTS

Fig. 3Cortical neurons on grid patterns after 13 days in culture. Next to the cells,the patch-clamp electrodesused for the electrophysiological recordings are visible.

cell by a depolarising current pulse and the resulting effect wasmonitored in the other cells. As shown before, small circuits ofthree cells communicating through two successive synapsescould be found. A signal evoked in the first cell (c1) was trans-duced through the first synapse to the second cell (c2), resultingin an excitatory postsynaptic potential (EPSP), while no effect onthe third cell (c3) was observed. An action potential evoked in c2induced an EPSP in c3 through the second synapse, but not inc1, whereas an action potential in c3 had no effect on the othercells. This indicates a unidirectional connection from c1 via c2 toc3 through two synapses in succession. In another recording(Figure 4), stimulation of c1 resulted in an EPSP in c2 that waslarge enough to trigger an action potential, which in turn causedan EPSP in c3. In this recording, the signal travelled via twosynapses through all three cells of the constellation.

In summary, our experiments demonstrate an excellent controlof cell growth on patterned substrates, permitting a detailedtracing of of the electrical signal propagation. This will enable usto study the cellular communication code in increasingly com-plex networks.

179

Fig. 4Functional coupling recordedfrom cortical neurons. Apresynaptic action potential(c1) induced a postsynapticaction potential (c2), whichinduced a postsynaptic depolarization (c3 ).The start of the postsynapticdepolarisation was delayedby a specific time to thepresynaptic spiking. Thisdelay, which is specific forchemical synapses, is mainlycaused by the time requiredfor transmitter release, trans-mitter diffusion and post-synaptic ion channel gating.

“Ordered networks of rat hippocampal neurons attached to sili-con oxide surfaces”M. Scholl, C. Sprössler, M. Denyer, M. Krause, K. Nakajima,A. Maelicke, W. Knoll, A. OffenhäusserJ. Neurosci. Meth. 104 (2000) 65

“Spot Compliant Neuronal Networks by Structure OptimizedMicro-Contact Printing”L. Lauer, Ch. Klein, A. OffenhäusserBiomaterials 22 (2001) 1925

“Electrophysiological recordings of patterned rat brain stem sliceneurons”L. Lauer, A. Vogt, C. K. Yeung, W. Knoll, A. OffenhäusserBiomaterials 23 (2002) 3123

Simone Böcker-Meffert, Andreas Offenhäusser,Angela Vogt

180

REFERENCES

AUTHORS

181

SQUIDs: The Ultimate Magnetic Sensors for Materials Characterizationand Biomedical DiagnosticsSQUIDs (Superconducting QUantum Interference Devices) arethe most sensitive magnetic field or flux detectors known. Forsimplicity, they can be described as analog-to-digital convertersof magnetic signals, quantized in units of the magnetic fluxquantum, Φ0 = 2.07 x 10–15 Wb. An external feedback circuit isused to cancel the applied flux. Thus the SQUID operates as anull detector. Such an analog device can resolve very small frac-tions of the flux quantum, down to 10-6 Φ0, permitting a mag-netic field resolution of the order of a few fT/√Hz. SQUIDs areespecially useful for low frequency magnetic signals. They areunmatched with respect to flux sensitivity. They permit thedetection of the weak magnetic fields generated by biologicalcurrents, including heart or brain activities. Such a high sensitivi-ty is achieved by the macroscopic coherence of the electronicwavefunction of the superconducting state. Therefore the SQUIDsensor has to be cooled below the critical temperature of thesuperconducting material.The discovery of the High Temperature Superconductors, espe-cially YBaCuO, led to the development of SQUIDs which can beoperated at a temperature of 77 K (liquid nitrogen) instead of4.2 K (liquid helium). They allow a significant simplification ofcryogenics with the ensuing reduction in investment and operat-ing cost, and the simplification of handling and maintenance,including an increased portability. Small, reliable cryocoolerswith a low level of vibration-induced und electromagnetic noise(e.g. Joule-Thomson or Pulse Tube) have been shown to be appli-cable. They constitute a practical cooling option for industrialSQUID applications.

Our sensors are based on radio-frequency (rf) SQUIDs. Thesedevices consist of one Josephson junction in a superconducting

INTRODUCTION

APPROACH

ring. A well defined grain boundary con-stitutes a weak link within an epitaxialYBaCuO layer. These grain boundary junc-tions are prepared first by ion beam etch-ing of a steep ditch into single crystalLaAlO3 or SrTiO3 substrates (1), see Fig.1. Then epitaxial YBCO thin films aregrown by laser ablation, forming the cru-cial step-edge grain boundary Josephsonjunction at the edge of the ditch (Fig. 2).The radio-frequency (rf) SQUIDs, consist-ing of single junction loops with a flux-focusing washer (Fig. 1), are photolitho-graphically patterned using chemicaletching. A 10 x 10 x 1 mm3 SrTiO3 sub-strate, acting as substrate resonator (2),

serves as the tank circuit for SQUID readout and is positioned inflip-chip geometry adjacent to the rf washer SQUID, as sketchedin Fig. 3. The substrate resonator is equipped with a YBCO thinfilm washer structure in order to adjust the resonance frequencyto approximately 650 MHz and to enhance the magnetic fieldsensitivity.

182

Fig. 1Schematic of the SQUID layout.

Fig. 3Principle of substrate resonator SQUID.

Fig. 2SEM of the step edge junction of a SQUID.

SQUID SensorFig. 4 shows a noise measurement of the SQUID magnetometerwith substrate resonator at 77K. The SQUID loop was 100 x 100µm2, this corresponds to a SQUID inductance of 150 pH.Because the flux concentrator (washer structure) uses the fullresonator substrate area of 10 x 10 mm2, we achieved a field-to-flux transfer coefficient ∂Β/∂φ of 3.2 nT/Φ0. A white SQUIDflux noise of 7.3 µΦ0/√Hz was found, corresponding to a fieldsensitivity of 24 fT/√Hz . In the low frequency range, a noise fig-ure of 38 fT/√Hz at 10 Hz and 83 fT/√Hz at 1 Hz is achieved (2).As the fabrication of the single layer SQUIDs is relatively simple,the substrate resonator concept offers a very practical andstraightforward method to obtain high sensitivity SQUID magne-tometers.

Eddy current non-destructive materials testingNonmagnetic metallic materials are usually tested by inducingeddy currents and measuring the magnetic field response. Thisinductive ac technique is based on a narrowband lock-in readoutscheme, resulting in noise suppression and allowing to evaluatethe quadrature component of the response field which containsinformation on excitation energy dissipation. In highly conduc-tive materials, probing of deeper defects mandates using lowerexcitation frequencies. Sensitivity of induction coils decreaseswith frequency, so that SQUIDs can offer a higher sensitivity ofdetection of very deep faults (3).

183

RESULTS

Fig. 4Flux noise and magneticfield noise of a substrateresonator rf SQUID incorpo-rating a YBCO step edgejunction.

As the construction of the megaliner Airbus aircraft progresses,testing procedures for extremely thick-walled structures areneeded. A bolted three-layer aluminium sample from EADS Air-bus, modeling the projected outer wing splice of the Airbus A380with a total thickness of 62 mm (Fig. 5), was measured with aJoule-Thomson machine-cooled SQUID. For sufficient currentpenetration into the layered aluminium sample, an excitationfield in the range of 10 – 40 Hz was applied. The small field vari-ations caused by the defects are superimposed on the currentdistortions and the corresponding field changes in the vicinity ofthe titanium bolts. Separation of these two contributions was

achieved by signal processing. Fig. 6 (top)shows the measured SQUID line scantraces (after liftoff correction and phaseoptimization) in the case of a titaniumbolt with an adjacent flaw (30 mm long,at a depth of 31 – 46 mm) and, for com-parison, an unflawed bolt connection. Theflawed bolt is easily identified. In contrast,it was impossible to distinguish betweenflawed and unflawed bolt in the corre-sponding measurement with a conven-tional commercial eddy current system(Fig. 6 bottom). Hence, the SQUID tech-nique constitutes a powerful non-destruc-tive method for the inspection of extreme-ly thick aluminium structures.

184

Fig. 6Signal traces of the AirbusA380 wing splice sample,measured with SQUID (top)and induction coil (bottom)in conjunction with eddycurrent excitation.

Fig. 5Projected structure of theAirbus A380 wing splice,where the aluminium wingand the carbon fiber rein-forced wing panel meet.

Biomagnetic diagnostics: MagnetocardiographySimilar to electrocardiography (ECG), magnetocardiography(MCG) extracts clinically relevant information from the biosig-nals generated by the human heart. The cardiac cellular actionpotential initiates intra- and extracellular ion currents, giving riseto a magnetic field above the chest. The basic advantages ofMCG over ECG is the high sensitivity to local myocardial cur-rents, undisturbed by contact resistances at tissue boundaries.The diagnostic value has been shown in a number of medicalstudies, e.g. on MCG diagnosis of coronary artery disease, onmyocardial infarction, on fetal MCG. As the magnetic signal ofthe human heart amounts to only about 100 pT (peak-to-peak)above the chest, a SQUID is required for MCG recordings. A1997 study of the Research Center Jülich together with the Med-ical Clinic of the Technical University Aachen showed that MCGdata allow to differentiate between a group of patients withcoronary artery disease in conjunction with ventricular tachycar-dia and a healthy control group, thus allowing risc stratificationof sudden cardiac death. This is significant because it is impossi-ble to separate the groups on the basis of ECG data.A major drawback of MCG, which probably has prevented MCGfrom finding widespreadclinical use so far, is thenecessity to conductMCG measurements in amagnetically shieldedroom. Recently, we devel-oped a first-order elec-tronic gradiometer incor-porating our noveldielectric substrate res-onator rf SQUIDs withhigh field sensitivity (2). After adaptation of the gradiometerbaseline to the specific environmental magnetic disturbances atthe measurement location, it was experimentally verified thatthe system is well suited for performing MCG measurements inan unshielded environment. Fig. 7 shows our portable SQUIDgradiometer system. The tripod holding the cryostat may be dis-mantled, the complete system fits into a travel suitcase. A typical

185

Fig. 7Portable SQUID gradiometer system for recording human magnetocardiograms in un-shielded enviromment.The tripod holding the cryostatmay be dismantled, the com-plete system fits into a travelsuitcase.

real-time MCG recording is displayed in Fig. 8. Our first-orderportable HTS gradiometer utilizing substrate resonator rf SQUIDsmay be upgraded to a commercial multichannel MCG system.

Biosensors: Magnetic nanoparticle detectionImmunological techniques, such as immunoassays, employ thehighly specific interaction between antigenes and antibodies inconjunction with markers for the detection and quantification ofspecific biomolecules. Typically, fluorochromes, enzymes, orradioactive compounds are used as markers. Each method hasintrinsic limits. High sensitivity fluorescence detection is veryinvolved. Therefore the sensitivity of enzyme techniques is notalways sufficient. The use of radioactive markers poses radiationhazards. Magnetic label-based bioassays have been identified asa very promising alternative. These markers comprise of an ironoxide core of a few tens or hundred nanometer diameter with abiocompatible surface coating. They are stable, non-toxic andthey can be manipulated by a magnetic field, thus allowing tosort them rather easily. The particles are superparamagnetic. Incase of monodispersed particles, their concentration in a testvolume can be quantitatively determined by AC susceptometry.In order to determine the achievable sensitivity for a given mag-netic field sensor, the minimum detectable magnetic momentwas calculated as a function of the sensor – particle distance(Signal-to-noise ratio of 1 for frequency of 50 kHz and band-width of 1 kHz). In the case of small particle-to-sensor spacingsof the order of a few mm, the sensitivity advantage of SQUIDsamounts to more than three orders of magnitude over an opti-

186

Fig. 8Typical MCG measurement recorded in unshielded environment. The real timesignal is displayed with avideo band-width of 30 Hz.

187

Fig. 9Calculated mimimum detectable magnetic moment and correspondingeffective particle by means of an optimized induction coil (black) or anoptimized HTS SQUID sensor (red), as a function of the distance sensor – particle.

Fig. 10AC susceptometry of 18 nl aqueous solution of Dynabead M-280magnetic particles (effective iron oxide diameter 1.4 µm) with aninduction coil. The solution was guided through a 150 µm diametercapillary close to the gap of a magnetic recording head, measuringthe synchronous response to a 45 kHz excitation field.

188

REFERENCES

AUTHORS

mized induction coil (see Fig. 9). Fig. 10 shows the result of aninduction coil measurement (using a magnetic recording head)of less than 36000 Dynabead M-280 magnetic particles (effec-tive iron oxide diameter 1.4 µm) in aqueous solution in a150 µm diameter capillary. Hence, with an optimized SQUIDmicroscope, which is currently developed, the detection of singlemagnetic particles of about 100 nm diameter is possible. Inaddition, the setup allows to employ the so-called relaxometrytechnique. This method developed at the PTB Berlin allows toseparate the contributions of bound and unbound magnetic par-ticles in solution by measuring the time transient of magnetiza-tion relaxation. The Néel relaxation (reorientation of the magne-tization vector inside the magnetic core) of the bound particlesis governed by a significantly larger time constant than theBrownian relaxation of unbound particles.

“SQUID Magnetometers”A.I. Braginski, H.-J. Krause, J. Vrba,in: “Handbook of thin film devices”, Ed.: M.H. Francombe,Volume 3, “Superconducting Film Devices”,Academic Press, San Diego (2000), pp. 149-225.

“Substrate resonator for HTS rf SQUID operation”Y. Zhang, J. Schubert, N. Wolters, M. Banzet, W. Zander and H.-J. KrausePhysica C 372-376 (2002) 282-286.

“Recent developments in SQUID NDE”H.-J. Krause and M. v. KreutzbruckPhysica C 368 (2002) 70-79.

Hans-Joachim Krause, Yi Zhang, Gregor I. Panaitov,Dieter Lomparski, Norbert Wolters, Jürgen Schubert,Willi Zander