+ All Categories
Home > Documents > A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell...

A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell...

Date post: 15-Feb-2019
Category:
Upload: vuongkhanh
View: 226 times
Download: 0 times
Share this document with a friend
9
IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 38, NO. 12, DECEMBER 2003 2297 A CMOS Chip for Individual Cell Manipulation and Detection Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro, Luigi Altomare, Andrea Leonardi, Marco Tartagni, Member, IEEE, and Roberto Guerrieri Abstract—Manipulation of populations of living cells on an individual basis is essential for the investigation of complex interactions among cells. We present a new approach to the integration on silicon of dielectrophoretic actuators and op- tical sensors that allow us to carry out this task. The device presented in this paper is an 8 8 mm chip implemented in a two-poly three-metal 0.35- m CMOS technology, featuring 102,400 actuation electrodes, arranged in an array of 320 320, 20 m 20 m microsites each comprising addressing logic, an embedded memory for electrode programming, and an optical sensor. The chip enables software-controlled displacement of more than 10,000 individual living cells, allowing biologists to devise complex interaction protocols that are impossible to manage otherwise. The manipulation does not damage the viability of the cells, so that this approach could be a unique extension to the techniques already available to biologists. Index Terms—Biological cells, CMOS, dielectrophoresis, intelli- gent actuators, integrated sensors, lab-on-a-chip, optical Sensors. I. INTRODUCTION A NALYTICAL methods are critical to a wide range of industry sectors, from pharmaceutical research to the agri-food business, from environmental control to diagnostics, to name a few. In the last ten years, the field of laboratory methods for biology and chemistry has been shaken by a revolution which is reshaping the way research and analysis are carried out. Much research effort in biology, chemistry, and engineering has been recently aimed at pursuing the advantages of minia- turization for cheaper, better, and faster sample analysis. Micro Total Analysis Systems ( TASs) were envisioned in the late 1980s [1] as miniaturized, highly integrated chemical analysis systems. The early efforts regarded the microfluidic problems related to the motion of liquid samples in micromachined chan- nels which built on the experience of capillary electrophoresis. In the late 1990s, the advent of DNA microarrays, propelled by genomic research, captured the attention of researchers and investors alike. Although the field was generally indicated as that of biochips, the word “lab-on-a-chip” (LOAC) entered the jargon to differentiate between passive microarrays and Manuscript received April 18, 2003; revised July 7, 2003. This work was sup- ported by the European Community 5th Framework Programme under Contract IST-2001-32437, and by the Italian MIUR-PRIN 2000. N. Manaresi and G. Medoro are with Silicon Biosystems s.r.l., 40125 Bologna, Italy. A. Romani, L. Altomare, A. Leonardi, M. Tartagni, and R. Guerrieri are with the Advanced Research Center on Electronic Systems (ARCES), University of Bologna, 40123 Bologna, Italy. Digital Object Identifier 10.1109/JSSC.2003.819171 microanalytical systems sporting some degree of integration, programmability, or microfluidic capabilities. So far, the aim has been mainly the speedup of DNA amplifi- cation and detection and other molecular analyzes from prepro- cessed samples. Many technologies for fluid motion, DNA am- plification, detection, and other analytical functions have been miniaturized and have become mainstream techniques in bio- logical laboratories. Beside these advances on molecular analysis, there has been an emerging and still unmet need for LOAC which are able to deal with cells. In fact, cell-analysis protocols must be car- ried out in a number of fields, from the sample preparation for biomolecular analysis, to drug screening. Since microorganisms and cells are mostly electrically neutral, dielectrophoresis (DEP) is well suited to their ma- nipulation. DEP [2] is the physical phenomenon whereby neutral particles, in response to a spatially nonuniform electric field , experience a net force directed toward locations with increasing or decreasing field intensity according to the physical properties of particles and medium. In the first case, the force is called positive dielectrophoresis (pDEP), while in the second case it is called negative dielectrophoresis (nDEP). Several approaches for microorganism manipulation have been developed based on DEP. In [3], both pDEP and nDEP are used to precisely displace cells in a microchamber formed between two facing glass chips with elongated electrodes. However, cells get in contact with device surfaces and tend to stick to them. A solution is to levitate cells while manipulating them. Since maxima of the electric field cannot be established away from the electrodes, stable levitation is possible only with nDEP force. Hence, the use of closed nDEP cages has been proposed. In [4], three-dimensional (3-D) structures of electrodes located at the vertexes of a cube are used for such purpose. The main drawback is that fluid flow is required to lead cells into and out of the DEP cage and electrode alignment in three dimensions is necessary. In [5], traveling waves are combined with nDEP to move cells in a microchamber without fluid flow. However, it is difficult to precisely position cells, as needed by multistep experimental protocols, due to the fact that the cell speed depends on the type of cell. As far as sensing is concerned, approaches such as optical [6] or fluorescent labeling (e.g., as used in FACS) have been proposed. Their main drawbacks are that they normally require bulky and expensive equipment [7], are characterized by com- plex sample preparations and are thus not suited to miniatur- ization. This explains the emerging interest in electrical sensing approaches such as those based on impedance measurement [8]. 0018-9200/03$17.00 © 2003 IEEE
Transcript
Page 1: A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell Manipulation and Detection Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro,

IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 38, NO. 12, DECEMBER 2003 2297

A CMOS Chip for Individual Cell Manipulationand Detection

Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro, Luigi Altomare, Andrea Leonardi,Marco Tartagni, Member, IEEE, and Roberto Guerrieri

Abstract—Manipulation of populations of living cells on anindividual basis is essential for the investigation of complexinteractions among cells. We present a new approach to theintegration on silicon of dielectrophoretic actuators and op-tical sensors that allow us to carry out this task. The devicepresented in this paper is an 8 8 mm2 chip implemented ina two-poly three-metal 0.35- m CMOS technology, featuring102,400 actuation electrodes, arranged in an array of 320 320,20 m 20 m microsites each comprising addressing logic, anembedded memory for electrode programming, and an opticalsensor. The chip enables software-controlled displacement of morethan 10,000 individual living cells, allowing biologists to devisecomplex interaction protocols that are impossible to manageotherwise. The manipulation does not damage the viability of thecells, so that this approach could be a unique extension to thetechniques already available to biologists.

Index Terms—Biological cells, CMOS, dielectrophoresis, intelli-gent actuators, integrated sensors, lab-on-a-chip, optical Sensors.

I. INTRODUCTION

A NALYTICAL methods are critical to a wide range ofindustry sectors, from pharmaceutical research to the

agri-food business, from environmental control to diagnostics,to name a few. In the last ten years, the field of laboratorymethods for biology and chemistry has been shaken by arevolution which is reshaping the way research and analysisare carried out.

Much research effort in biology, chemistry, and engineeringhas been recently aimed at pursuing the advantages of minia-turization for cheaper, better, and faster sample analysis. MicroTotal Analysis Systems (TASs) were envisioned in the late1980s [1] as miniaturized, highly integrated chemical analysissystems. The early efforts regarded the microfluidic problemsrelated to the motion of liquid samples in micromachined chan-nels which built on the experience of capillary electrophoresis.In the late 1990s, the advent of DNA microarrays, propelledby genomic research, captured the attention of researchers andinvestors alike. Although the field was generally indicated asthat of biochips, the word “lab-on-a-chip” (LOAC) enteredthe jargon to differentiate between passive microarrays and

Manuscript received April 18, 2003; revised July 7, 2003. This work was sup-ported by the European Community 5th Framework Programme under ContractIST-2001-32437, and by the Italian MIUR-PRIN 2000.

N. Manaresi and G. Medoro are with Silicon Biosystems s.r.l., 40125Bologna, Italy.

A. Romani, L. Altomare, A. Leonardi, M. Tartagni, and R. Guerrieri are withthe Advanced Research Center on Electronic Systems (ARCES), University ofBologna, 40123 Bologna, Italy.

Digital Object Identifier 10.1109/JSSC.2003.819171

microanalytical systems sporting some degree of integration,programmability, or microfluidic capabilities.

So far, the aim has been mainly the speedup of DNA amplifi-cation and detection and other molecular analyzes from prepro-cessed samples. Many technologies for fluid motion, DNA am-plification, detection, and other analytical functions have beenminiaturized and have become mainstream techniques in bio-logical laboratories.

Beside these advances on molecular analysis, there has beenan emerging and still unmet need for LOAC which are ableto deal with cells. In fact, cell-analysis protocols must be car-ried out in a number of fields, from the sample preparation forbiomolecular analysis, to drug screening.

Since microorganisms and cells are mostly electricallyneutral, dielectrophoresis (DEP) is well suited to their ma-nipulation. DEP [2] is the physical phenomenon wherebyneutral particles, in response to a spatially nonuniform electricfield , experience a net force directed toward locationswith increasing or decreasing field intensity according to thephysical properties of particles and medium. In the first case,the force is called positive dielectrophoresis (pDEP), while inthe second case it is called negative dielectrophoresis (nDEP).Several approaches for microorganism manipulation have beendeveloped based on DEP. In [3], both pDEP and nDEP areused to precisely displace cells in a microchamber formedbetween two facing glass chips with elongated electrodes.However, cells get in contact with device surfaces and tend tostick to them. A solution is to levitate cells while manipulatingthem. Since maxima of the electric field cannot be establishedaway from the electrodes, stable levitation is possible onlywith nDEP force. Hence, the use of closed nDEP cages hasbeen proposed. In [4], three-dimensional (3-D) structures ofelectrodes located at the vertexes of a cube are used for suchpurpose. The main drawback is that fluid flow is required tolead cells into and out of the DEP cage and electrode alignmentin three dimensions is necessary. In [5], traveling waves arecombined with nDEP to move cells in a microchamber withoutfluid flow. However, it is difficult to precisely position cells, asneeded by multistep experimental protocols, due to the fact thatthe cell speed depends on the type of cell.

As far as sensing is concerned, approaches such as optical[6] or fluorescent labeling (e.g., as used inFACS) have beenproposed. Their main drawbacks are that they normally requirebulky and expensive equipment [7], are characterized by com-plex sample preparations and are thus not suited to miniatur-ization. This explains the emerging interest in electrical sensingapproaches such as those based on impedance measurement [8].

0018-9200/03$17.00 © 2003 IEEE

Page 2: A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell Manipulation and Detection Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro,

2298 IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 38, NO. 12, DECEMBER 2003

The possibility of investigatingindividualcell interactions ona large scale would open up great possibilities for tasks such asdrug screening, cell separation and analysis. This paper presentsa standard CMOS microsystem [9], implementing the movingDEP-cages approach (envisaged in [10]), toindividually detectand manipulate more than 10,000 cells in a parallel fashion. Tothe best of our knowledge, this is the first device that allows pro-grammable manipulation of individual particles with no need forfluid flow nor micromachining, combined with embedded op-tical detection. The paper is organized as follows. Section II re-ports an overview of the objectives of the biochip system, as re-lated to some examples of applications. Design constraints andspecifications for the manipulations and detection of particlesare also outlined. In Section III, the architecture of the CMOSchip is described along with details of the circuit implementa-tion. Test results are then reported in Section IV, while Section Vdraws some conclusions.

II. DESIGN OBJECTIVES ANDSPECIFICATIONS

A. Design Approach

The approach followed in the development of this micro-system for cell analysis was based on the following points.

Platform Approach—The objective was to create a flex-ible platform which could be used to carry out various dif-ferent analytical protocols by just changing the softwareand reagents. Although this is a well established concept inelectronic design, it is an innovative and challenging fea-ture to implement in LOAC.Smart Chips by Active Substrates—Although the fabrica-tion process of active silicon chips is much more complexthan the microfabrication processes commonly used forpassive biochips (e.g., with simple microchannels etched inglass), the availability of transistors affords massive paral-lelism, enabled by I/O multiplexing, and integrated detec-tion.Use of Standard CMOS—The use of commonly availablefabrication processes without micromachining options hasseveral advantages. It is possible to vary the shape, connec-tion and number of microchambers of the device by simplychanging the microfluidic packaging on top of the siliconchip, instead of requiring a new mask set. The availabilityof numerous foundries means the possibility of choosingthe best trade off between minimum resolution and fabri-cation cost in a wide range of processes. Scalability: whileit is possible to handle cells with this prototype, more ad-vanced technologies would enable the design of chips tohandle individual bacteria or viruses.

A sketch of the device is shown in Fig. 1. A microchamberis defined by the chip surface and a conductive-glass lid. Thechip surface implements a two-dimensional array of microsites,each consisting of a superficial electrode, embedded sensors andlogic. The electrode array is actually implemented with CMOStop-metal and protected from the liquid by the standard CMOSpassivation, not shown in the figure. Since the chip is disposable,we are not concerned with long-term reliability issues, so thatstandard passivation is good enough in this perspective.

Fig. 1. Sketch of the biochip section.

A closed DEP cage in the spatial region above a micrositecan be created by connecting the associated electrode and themicrochamber lid to a counterphase sinusoidal voltage,while the electrode of the neighboring microsites is connected toan in-phase sinusoidal voltage [10]. A field minimum isthus created in the liquid, corresponding to a DEP cage in which,depending on its size, one or more particles can be trapped andlevitated. By changing, under software control, the pattern ofvoltages applied to the electrodes, DEP cages can be indepen-dently moved around the device plane, thus grabbing and drag-ging cells and/or microbeads across the chip. Particles in thesample can be detected by the changes in optical radiation im-pinging on the photodiode associated with each microsite.

Implementation of the moving DEP-cages approach with theproposed CMOS chip enables one to achieve the key featureswhich are summarized in the following.

Single-Cell Addressing and Selection—Thanks to thesmall pitch of the electrodes, single cells can be individu-ally trapped in separate cages and independently movedon the device.Grab-and-Drag Motion—Particle position is digitally con-trolled step-by-step in a deterministic way, by applyingcorresponding pattern of voltages to the array which setthe position of the DEP cages. This feature is difficult toachieve with motion techniques based on fluid flows or ontraveling-wave DEP [5]. This difference may be compared

Page 3: A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell Manipulation and Detection Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro,

MANARESI et al.: CMOS CHIP FOR INDIVIDUAL CELL MANIPULATION AND DETECTION 2299

to that existing between step motors and asynchronouselectric motors.Embedded Monitoring—Use of an active substrate imple-mented with microelectronic semiconductor technologyallows us to integrate an array of optical sensors to detectthe position and possibly the status of all particles insidethe device. This information can be used to provide mean-ingful feedback on device operations. Thus, the devicehas the potential to be used without bulky and expensiveexternal microscopes and cameras. This will be important,in perspective, for portable LOAC.Massive Parallelism—Thousands of cells can be concur-rently and independently moved and detected thanks to thelarge number of electrodes.Contactless Movement—The closed DEP cage allows par-ticles to be suspended in a contactless manner, thus helpingthe prevention of cell adhesion to sensor surfaces.Robustness—Using an array instead of microchannelsallows one to: 1) alleviate clogging problems that arecommon with cells in microchannel devices and 2) befault tolerant with respect to cells that are stuck, so thatnew routing paths can be devised for other cells.

B. Examples of Applications

This chip will enable several experiments unaffordable withexisting techniques, with applications ranging from diagnosticsto drug discovery.

For example, a microbead coated (according to known art)with antibodies for a known cell receptor, could be mated (bymerging them in a cage) with an unknown cell. By pulling themapart with a controlled force (i.e., separating their cages) onemay test whether the molecules coating the bead match the re-ceptors on the cell surface (they remain stuck together), thusidentifying the cell itself (diagnosis). On the other hand, using aknown cell line and a large number of beads each coated with adifferent compound of uncertain activity, one may detect whichof these compounds binds to the unknown receptors on the cellsurface (drug screening).

Another protocol which could be implemented on this plat-form is cell sorting by label-free separation. As it was demon-strated in [11], exploiting the differences of dielectrophoreticresponse as a function of the frequency of the applied ac elec-tric field, it is possible to selectively move one population ofcells. This approach may be of interest for example in the sepa-ration and fractionation of cell populations for which molecularmarkers are not available.

Another possibility is to tap the wide range of availablefluorescent markers developed for established cell separationmethods like fluorescent activated cell sorters (FACS). Usingthese legacy techniques, one may label cells with appropriatefluorescent molecular markers and use conventional fluores-cence microscopes to identify and tag the cells on the array.Separation could then be carried out relying on the possibilityto selectively move a set of cages (according to the marker ofthe trapped cell) toward a separate microchamber, where theycould be flushed out and recovered or further analyzed. In thiscase the advantage on FACS machines would be the possibilityto work on small cell loads. In fact FACS typically require few

millions of cells as a minimum, while the proposed chip maystart with samples of few thousands cells and still be able torecover a small percentage of cells of interest. This would beimportant for example in the analysis of small biopsies.

C. Design Constraints and Specifications

The device has been optimized for handling eukaryoticcells (such as the lymphocytes found in blood) in the rangeof 20–30 m. A design guideline, derived from analysis ofsimulation results on the horizontal DEP forces as a functionof particle size with respect to the electrodes [15], suggeststhat the electrode pitch should be similar to the cell size. Sincethe counterphase electrode must be surrounded by in-phaseelectrodes to create a DEP cage, the periodicity of the cages,as well as the attraction basin, is actually two electrodes.Accordingly, two cells may fit into one cage. Yet we are ableto manipulate them individually since two cells originally inone cage can be segregated into two different cages by simplyenlarging and then dividing the cage. Larger particles can alsobe handled by increasing the width of the cage to encompassmore than one electrode.

Increasing the number of electrodes on the array one may in-crease capacity (number of cells in the input sample) and se-lectivity, i.e., possibility to select a smaller percentage of cells.However, silicon cost increases with chip size. Accordingly, thetotal number of cages was chosen to be greater than ten thou-sands. On one hand, this is satisfactory to recover a significantnumber of cells (10–100) which may be present in low per-centage (0.1-1%) in the starting sample. On the other hand, chipsize is, thus, still acceptable.

The time constants for cell motion due to DEP forces arerelatively slow (about one second or more to make a 20-mstep). This relaxes timing constraints for array programming,as well as for sensing frame rate.

For the choice of the most appropriate CMOS technology thefollowing considerations were taken into account. Since DEPforce is proportional, under certain assumptions, to the squareof the applied voltages [15], the supply voltage should be aslarge as possible, as this will limit actuation voltages.

As opposed to conventional IC designs, the lower the resolu-tion of the technology, the lower the cost of the chip, since diesize is set from other specifications. Scaling beyond the pointwhere the required number of transistors fits in the micrositearea does not improve neither cost nor performance. In fact,scaling is just required if one wants to manipulate smallercells, like individual bacteria (typically 1–3 m) or viruses(100–300 nm).

III. CHIP ARCHITECTURE

The LOAC architecture is based on a two-dimensional arrayof microsites, the purpose of which is to: 1) generate the electricfield necessary to create dielectrophoretic cages and 2) detectthe presence of single particles or clusters trapped in cages byusing optical sensing. Each microsite consists of an actuationelectrode, implemented with a top metal plate, and underlyingembedded circuitry for programming and detection. Micrositescan be addressed in a random access mode by means of row

Page 4: A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell Manipulation and Detection Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro,

2300 IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 38, NO. 12, DECEMBER 2003

Fig. 2. Chip architecture.

and column decoders, for both actuation and sensing. The pres-ence of particles is detected by photodiodes, embedded in thesubstrate, that measure signal variations from uniform light im-pinging on the chip surface.

The mode of operation consists of three phases:program-ming, actuation, and sensing. During programmingthe actu-ation pattern is stored in the microsites of the array to deter-mine DEP cage number, displacement, and shape. In theactu-ation phase each electrode is energized by either an in-phase

or counterphase sinusoidal actuationvoltage signal, according to the programmed patterns. Duringthe sensingphase, the actuation voltages are halted, to avoidcoupling to the photodiode and readout, and the optical imageof the array is grabbed. Thus,actuationandsensingphases arealways kept nonoverlapping. On the contrary,programmingandactuationmay be concurrent:actuationpatterns can be changedin real timewhile electrodes keep energizing the system.

The general architecture of the chip is sketched in Fig. 2where the main blocks are as follows:

• array of microsites, composed of 320320 elements of20- m pitch;

• 9-bit static column/row decoders for random access;• bias generator block;• readout circuit block.

Referring to Fig. 2, the row and columncircuits provide the logic signals to program and

read out each microsite. The word is used for eitheraddressing the microsites and setting the startup configuration.The microsites addressing is performed by sampling the address

on the rising edge of theCASandRASsignals for thecolumn and row decoder, respectively. The configuration wordis sampled on the rising edge of the “configuration strobe”CONFSand is used to set the bias current for both readoutblock and cells, as well as the gain of the readout circuit.

A. Microsite Circuit

The schematic of the microsite circuit is reported in Fig. 3.Vertical and horizontal labels refer to signals generated by thecolumn and row decoders, respectively.

The actuation circuit is composed of two complementary passtransistors controlled by a 1-bit memory element. During theprogrammingphase, the microsite is addressed andWRITE isactivated, thus, the metal 3 electrode can be switched to either

or by programming with the signal the memoryelement addressed byROWWandCOLW. The metal plate settingis kept by the memory until a new programming phase occurs.

Electrode peak current is due to capacitive coupling withneighboring electrodes and lid. In the worst case it is about5 A, for an electrode programmed at surrounded by

electrodes and working at 10 MHz, and produces anegligible 10-mV voltage drop through the pass transistors( k ).

During thesensingphase the actuation signals are halted,avoiding spurious coupling with the pixel readout. Thanks to theparticle inertia and fast readout, cells keep their position in levi-tation within the microchamber. The right side of the schematicof Fig. 3 shows the sensing circuit, consisting of a CMOS ac-tive-pixel sensor (APS) [13], implemented with a 217 mwell-junction photodiode placed underneath the 1.2-m-widegap that separates each electrode with its right neighbor. Thesensor array is read out row-wise by addressing each micrositeand asserting theSENSEsignal so thatROWSandCOLSare ac-tivated.

B. Readout Circuit

The readout amplifier, shown in Fig. 4, is a fully differen-tial charge integrator implemented using a high-swing foldedcascode switched capacitor (SC) operational amplifier with SCcommon-mode feedback. The input capacitanceis fixed tofour times a unit capacitor fF, while the feedback ca-pacitance is implemented as a bank of four unit capacitorseach in series with CMOS switches. The switches are set byusing the startup configuration word so as to fix variable gainsin the set of .

After the integration time, the APS output of the se-lected microsite is sampled bySIG1 and is stored on the cor-responding . While PHI1 is still high, theRESETsignal isactivated, and the reset voltage is sampled duringSIG2 on theother . Subtracting the reset voltage allows us to compensatefor and fixed pattern noise of photodiode and readout fol-lower transistor, with a correlated double sensing scheme [14].

Since cells are almost transparent, the signal is a small vari-ation on top of a larger voltage swing. Thus, increasing chargeamplifier gain in order to boost the sensitivity would take theoutput to saturation. To avoid this, a fixed charge is subtractedfrom the input by means of to keep the output within therange. The output differential voltage is provided duringPHI2according to the following relationship:

Page 5: A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell Manipulation and Detection Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro,

MANARESI et al.: CMOS CHIP FOR INDIVIDUAL CELL MANIPULATION AND DETECTION 2301

Fig. 3. Microsite circuit schematic.

Fig. 4. Readout circuit schematic.

In other words, the use of actually doubles the outputrange: the differential output voltage can then even be negative(which would otherwise be impossible since the pixel reset

voltage is always higher than the integratedvoltage ). The doubled output swing can, thus, beused to increase the charge integrator gain .

Page 6: A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell Manipulation and Detection Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro,

2302 IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 38, NO. 12, DECEMBER 2003

Fig. 5. Chip in microfluidic packaging: the die is mounted directly on the PCBwith chip-on-board technique, while a conductive-glass lid is glued to the chipto define the microchamber.

Fig. 6. Chip photograph.

Fig. 7. Comparison of embedded sensor (top left) and optical microscope (topright) images of 50-�m polystyrene beads.

IV. TEST RESULTS

Fig. 5 shows the chip in the microfluidic package, describedin [16]. After a conductive-glass lid with SU8 walls is gluedto the chip to define the microchamber, the die is mounted di-rectly on the printed circuit board (PCB) with chip-on-boardtechnique.

Fig. 8. Individual manipulation of a 50-�m polystyrene bead in water, 3.3 Vat 800-kHz phases.

Fig. 9. Manipulation and detection of 50-�m polystyrene beads. (a) Actuationpattern. (b) Microscope image. (c) Embedded optical sensors image.

The lid is spaced about 85m from the chip surface. Lidheight is not critical. The strength of the DEP-cage mostly de-pends from its height above the array: since near the array fieldgradients are stronger, the lower the cage the stronger the force.In turn, cage height increases with lid height and decreases withlid-voltage amplitude. For example, results from simulationsshow that increasing lid height from 50 to 80m while in-creasing lid-voltage amplitude from 3.3 to 6.6 V, cage heightis approximately the same, i.e., about 15m. Considering acell radius of about 10 m this is somewhat a lower bound inorder to prevent cell contact with the array. Also, for a givenlid-voltage amplitude, DEP-force sensitivity to microchamber

Page 7: A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell Manipulation and Detection Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro,

MANARESI et al.: CMOS CHIP FOR INDIVIDUAL CELL MANIPULATION AND DETECTION 2303

Fig. 10. (a)–(d) Mating and (d)–(f)separation of K562 tumor cells in 280-mM mannitol in water, 3.3 Vpp at 500-kHz phases. For the sake of simplicity, cageelectrodes are enclosed by dashed lines.

height is low. For example, keeping a 3.3-Vlid voltage, in-creasing lid height as above from 50 to 80m (i.e., 60%),the horizontal DEP force decreases, due to the variation of cageheight, of less than 50% only. It is thus apparent that interde-vice variations of microchamber height (e.g.,10% due to SU8thickness variations), or intradevice nonplanarity of the lid, havea negligible impact on DEP-cage strength.

Epoxy resin is used for bonding-wire protection. The fluidicinlet is provided by a capillary connected to an opening in themicrochamber lateral wall and sealed with a drop of insulatingglue. A drop of conductive glue is used to electrically connectthe lid electrode to the PCB. The die photograph is shown inFig. 6, where the main blocks are identified.

After a sample (50-m polystyrene beads in water) hasbeen flushed into the microchamber, the beads are randomlydistributed. Fig. 7 shows the corresponding image acquiredwith the embedded sensors.

Following the introduction of the sample, DEP cages are ac-tivated ( 3.3 V . Since in this case the bead diam-eter is more than twice the electrode pitch, the cage is set to 22electrodes. Three snapshots of the selective motion of one beadare reported in Fig. 8(i)–(iii). The top line shows the images ac-quired by the microscope while the bottom line displays the cor-responding programming pattern, where gray and white squaresindicate the electrodes receiving and , respectively(the lid is always proportional to ). The time for particlesto complete one step is approximately two seconds. In agree-ment with simulations, we observed that lateral forces actingon particles (hence speed), get stronger by increasing the lidvoltage peak-to-peak amplitude. Since this voltage is providedthrough the PCB, it is not limited by the chip supply voltage,and can be set two to three times as large as the array phases

, i.e., 6.6 or 9.9 V . As explained before, simula-tions show that the higher the lid-voltage amplitude with respect

Fig. 11. Manipulation of clusters ofSaccharomyces cerevisiae: 280-mMMannitol buffer, 3.3 V at 1-MHz phases.

to the array voltages, the lower the height of the DEP cage. Yet,this is difficult to verify from the microscope images. Thus thelid is typically set to 6.6 V .

Fig. 9 shows how, after applying a pattern implementing anarray of DEP cages, the microbeads are arranged correspond-ingly, and how they can be detected with the embedded opticalsensors.

Fig. 10 shows the mating and separation of two K562 tumorcells. Two cages trapping the selected cells [Fig. 10(a)] are firstmerged into a three-electrode cage [Fig. 10(b)], which is thenshrunk to a single electrode cage[Fig. 10(c) and (d)], forcing the

Page 8: A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell Manipulation and Detection Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro,

2304 IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 38, NO. 12, DECEMBER 2003

TABLE ICHIP SPECIFICATIONS

two cells to get in contact. The cage is then enlarged [Fig. 10(e)]and the cells lose their contact, until they are finally separatedagain in two distinct cages [Fig. 10(f)].

Smaller particles can also be manipulated, although morethan one is normally trapped in the same cage, as reported inFig. 11, which shows clusters of yeast.

Chip performance and specifications are summarized inTable I.

V. CONCLUSION

The device implemented in this project holds the promise ofbeing an enabling technology for the development of a range ofinnovative protocols in cell biology, infeasible with existing an-alytical techniques, including: 1) capacity of performing in par-allel a large number of experiments on individual cells; 2) pos-sibility to detect and isolate rare cells from a very small sample;3) possibility to selectively deliver controlled amounts of com-pounds to target cells’ and 4) possibility to investigate in realtime the dynamics of cell response to chemicals and to cell–cellinteractions.

Experimental results on the basic capabilities of individualmanipulation and detection of particles have been presented.The programmability of the system, afforded by the use of amicroelectronic substrate, makes this device a flexible platformto create different analytical protocols just by changing softwareand reagents, while sharing the same hardware.

ACKNOWLEDGMENT

The authors acknowledge A. Fuchs, D. Freida, I. Chartier ofCEA, and P. Marche and C. Villiers of INSERM, for their valu-able suggestions on microfluidic and biological aspects. We aregrateful to R. Gambari of the University of Ferrara for the help

and advice with tumor cells. S. Ronconi did an excellent jobhelping in the design and setup of the test board.

REFERENCES

[1] A. Manz, J. C. Fettinger, E. Verpoorte, H. Ludi, H. M. Widmer, and D.J. Harrison, “Micromachining of monocrystalline silicon and glass forchemical analysis systems. A look into next century’s technology or justa fashionable craze?,”Trends Anal. Chem., vol. 10, pp. 144–149, 1991.

[2] X.-B. Wang, Y. Huang, F. F. Becker, and P. R. C. Gascoyne, “A uni-fied theory of dielectrophoresis and travelling wave dielectrophoresis,”J. Phys. D, Appl. Phys., vol. 27, pp. 1571–1574, 1994.

[3] J. Sueheiro and R. Pethig, “The dielectrophoretic movement and po-sitioning of a biological cell using a three-dimensional grid electrodesystem,”J. Phys. D, Appl. Phys., vol. 31, pp. 3298–3305, 1998.

[4] T. Muller, G. Gradl, S. Howitz, S. Shirley, Th. Schnelle, and G. Fuhr,“A 3-D microelectrode system for handling and caging single cells andparticles,”Biosens. Bioelectron., vol. 14, pp. 247–256, 1999.

[5] Y. Huang, X.-B. Wang, J. A. Tame, and R. Pethig, “Electrokinetic be-havior of colloidal particles in travelling electric fields: Studies usingyeast cells,”J. Phys. D, Appl. Phys., vol. 26, pp. 1528–1535, 1993.

[6] M. S. Talary and R. Pethig, “Optical technique for measuring the pos-itive and negative dielectrophoretic behavior of cells and colloidal sus-pensions,”Proc. IEE—Sci. Meas. Technol., vol. 14, no. 5, Sept. 1994.

[7] A. Y. Fu, C. Spence, A. Scherer, F. H. Arnold, and S. R. Quake, “A mi-crofabricated fluorescence-activated cell sorter,”Nat. Biotechnol., vol.17, pp. 1109–1111, Nov. 1999.

[8] S. Gawad, L. Schild, and Ph. Renaud, “Micromachined impedance spec-troscopy flow cytometer for cell analysis and particle sizing,”Lab. on aChip, vol. 1, pp. 76–82, 2001.

[9] N. Manaresi, A. Romani, G. Medoro, L. Altomare, A. Leonardi, M.Tartagni, and R. Guerrieri, “A CMOS chip for individual cell manip-ulation and detection,” inISSCC 2003 Dig. Tech. Papers, vol. 487, Feb.2003, pp. 192–193.

[10] G. Medoro, N. Manaresi, M. Tartagni, and R. Guerrieri, “CMOS-onlysensors and manipulator for microorganizms,” inProc. IEDM, Dec.2000, pp. 415–418.

[11] G. Medoro, N. Manaresi, M. Tartagni, L. Altomare, A. Leonardi, andR. Guerrieri, “A lab-on-a-chip for cell separation based on the moving-cages approach,” presented at the Eurosensors XVI, Prague, Czech Re-public, Sept. 2002.

[12] G. Medoro, A. Leonardi, L. Altomare, N. Manaresi, M. Tartagni, andR. Guerrieri, “A lab-on-a-chip for cell detection and manipulation,” inProc. IEEE Sensors Conf., June 2002, pp. 472–477.

[13] S. Mendis, S. Kemeny, B. Pain, C. Staller, Q. Kim, and E. Fossum,“CMOS active pixel image sensors for highly integrated imaging sys-tems,”IEEE J. Solid-State Circuits, vol. 32, pp. 187–197, Feb. 1997.

[14] S. Mendis, S. Kemeny, and E. Fossum, “A 128� 128 CMOS active pixelimage sensor for highly integrated imaging systems,” inProc. ElectronDevices Meeting, 1993, pp. 583–586.

[15] A. Leonardi, G. Medoro, N. Manaresi, M. Tartagni, and R. Guerrieri,“Simulation methodology for dielectrophoresis in microelectroniclab-on-a-chip,” inProc. Modeling and Simulation of Microsystems,Apr. 22–25, 2002, pp. 96–99.

[16] M. Rabarot, J. Bablet, M. Ruty, I. Chartier, and C. Dubarry, “Thick SU8lithography for BioMEMS,” presented at the Photonics West, Microma-chining and Microfabrication, SPIE Conf., San Jose, CA, Jan. 2003.

Nicoló Manaresi (S’94–M’99) received the degree(cum laude) in electrical engineering and computersciences and the Ph.D. degree from the University ofBologna, Bologna, Italy, in 1993 and 1999, respec-tively.

From 1993 to 1995 and from 1997 to 2000, hewas with the University of Bologna as a Consultantto ST Microelectronics in the field of analog ICsand sensors design. In 1996, he spent one year asa Research Assistant at the Swiss Federal Instituteof Technology, Zurich, Switzerland. In 1999, he

cofounded Silicon Biosystems s.r.l., Bologna, and has since served as its CEO.He is the coauthor of more than 30 scientific papers and co-inventor of eightEuropean and U.S. patents.

Page 9: A cmos chip for individual cell manipulation and detection ... · A CMOS Chip for Individual Cell Manipulation and Detection Nicoló Manaresi, Member, IEEE, Aldo Romani, Gianni Medoro,

MANARESI et al.: CMOS CHIP FOR INDIVIDUAL CELL MANIPULATION AND DETECTION 2305

Aldo Romani received the Laurea degree (cumlaude) in electrical engineering in 2001 from theUniversity of Bologna, Bologna, Italy, where heis currently working toward the Ph.D. degree inelectrical engineering and computer science.

Since 2001, he has been with the Advanced Re-search Center on Electronic Systems (ARCES), Uni-versity of Bologna. His research interests focus on re-configurable logic applications and systems for ma-nipulation and detection of biological objects.

Gianni Medoro received the degree (cum laude) inelectrical engineering in 1999 from the Universityof Bologna, Bologna, Italy, where he is currentlyworking toward the Ph.D. degree.

His research interests include dielectrophoresisand microelectronic bio-manipulators and sensors.In 1999, he cofounded Silicon Biosystems s.r.l.,Bologna.

Luigi Altomare received the degree (cum laude) inbiological sciences and the Ph.D. degree from theUniversity of Bologna, Bologna, Italy, in 1996 and2000, respectively.

From 1996 to 1997, he was with the Department ofPharmaceutical Sciences, and later with the Depart-ment of Genetics, University of Bologna. He then re-turned to the Department of Pharmaceutical Sciencesto work on genetic engineering of bifidobacteria. Be-ginning in June 1997, he spent 19 months in Ger-many, first at the Institute of Biotechnologies at the

Forschungszentrum Juelich, Juelich, Germany, then with the Department of Mi-crobiology and Biotechnology, Ulm University, Ulm, Germany. Since 2000, hehas been with the Advanced Research Center on Electronic Systems (ARCES),University of Bologna, Italy.

Andrea Leonardi received the degree in electricalengineering in 2000 from the University of Bologna,Bologna, Italy, where he is currently working towardthe Ph.D. degree.

His research interests include dielectrophoresissimulators.

Marco Tartagni (M’99) received the Laurea degreein electrical engineering and the Ph.D. degree inelectrical engineering and computer sciences fromthe University of Bologna, Bologna, Italy, in 1993and 1998, respectively.

In 1992, he joined the Department of ElectricalEngineering, California Institute of Technology,Pasadena, as a Visiting Student, and beginning in1994, he was a Research Fellow there, working onvarious aspects of analog VLSI for imaging pro-cessing. Since March 1995, he has been an Assistant

Professor in the Department of Electronics, University of Bologna, where hedesigned and tested low-noise optical and capacitive sensors. He is currentlyinvolved in research on sensors aimed at implementing a hybrid technologywhere biomolecules are self-assembled on a microelectronic substrate.

Roberto Guerrieri received the degree in electricalengineering and the Ph.D. degree from the Universityof Bologna, Bologna, Italy.

He is currently Associate Professor of electricalengineering at the University of Bologna. Since1986, he has been visiting the Electrical Engineeringand Computer Science Department, Universityof California, Berkeley, and the Department ofElectrical Engineering, Massachusetts Instituteof Technology, Cambridge. During his scientificactivity, he has authored more than 80 published

papers in various fields, including numerical simulation of semiconductordevices, numerical solution of Maxwell’s equations, and parallel computationon massively parallel machines. In recent years, his work has been focusedon integrated silicon systems to solve various problems, such as optical andcapacitive smart sensors, integrated digital circuits for speech and video pro-cessing, and analog circuits for fuzzy controllers. In 1998, he became Directorof the Laboratory for Electronic Systems, a joint venture of the University ofBologna and ST Microelectronics, for the development of innovative designsof systems-on-chip.

Dr. Guerrieri was awarded the Best Paper Award of the IEEE for his work inthe area of process modeling in 1992.


Recommended