Sensors and Actuators B: Chemicalyoksis.bilkent.edu.tr/pdf/files/15093.pdfanalytical, numerical and experimental approaches. The results demonstrate that reliable impedimetric particle

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical

journal homepage: www.elsevier.com/locate/snb

Focusing-free impedimetric differentiation of red blood cells and leukemiacells: A system optimization

Ismail Bilicana, Mustafa Tahsin Gulerb, Murat Serhatliogluc, Talip Kirindid, Caglar Elbukenc,*a Science and Technology Application and Research Center, Aksaray University, Aksaray 68100, TurkeybDepartment of Physics, Kirikkale University, Kirikkale 71450, TurkeycUNAM—National Nanotechnology Research Center, Institute of Materials Science and Nanotechnology, Bilkent University, 06800 Ankara, TurkeydDepartment of Mathematic and Science Education, Faculty of Education, Kırıkkale University, Kırıkkale, 71450, Turkey

A R T I C L E I N F O

Keywords:Impedance spectroscopyRed blood cellWhite blood cellLeukemiaMicrofluidic flow cytometry

A B S T R A C T

A focusing-free microfluidic impedimetric cell detection system is developed. The effect of the channel di-mensions, solution conductivity, excitation voltage, and particle size on impedimetric signal outputs were op-timized to increase the sensitivity of the system. Conventional microfabrication techniques were adapted toobtain low height, resealable microchannels. The geometry optimization was performed by a combination ofanalytical, numerical and experimental approaches. The results demonstrate that reliable impedimetric particledifferentiation can be achieved without any labeling or particle focusing. The system parameters were studiedand rule-of-thumb design criteria were provided. Finally, using the developed system, red blood cells and leu-kemia cells were experimentally detected and differentiated. Thanks to its simplicity, the focusing-free celldifferentiation system may find applications in several cellular diagnostic uses.

1. Introduction

Electrical impedance measurement systems have been used for thedetection of a wide range of biological materials such as microorgan-isms [1,2], tissue samples [3] and cell suspensions [4–7]. Following thedemonstration of electrical single cell detection by Wallace H. Coulterin 1950s, Coulter counter has become one of the most commonly usedresearch tool for particle detection and enumeration studies [8]. Im-plementation of this technique using micro fabricated structures wasfirst shown for detection of latex nanoparticles [9]. Coulter counter usesdirect current (DC) to measure the resistance change during the dis-placement of particles through a narrow orifice. Since the polymericparticles or cell membrane behaves as an insulation layer under DCelectric field, the resistance of the orifice changes due to the replace-ment of the conductive liquid. The variation in DC resistance is in ac-cordance with the particle size.

Recently, various designs have been recommended to increase theperformance of microfluidic Coulter counters [10–13]. Impedance flowcytometer (IFC) was introduced as an improved version of Coultercounter that employs AC electric field instead of a DC signal. Whileconventional Coulter counters only classify cells based on their size, IFCprovides detailed analysis of single cells [14–16]. Thus, it is being usedin some commercial clinical analysis devices for the purpose of full

blood count, white blood cell differentiation, nucleated red blood cellcount, and reticulocyte count [17].

Microfluidic-based impedance flow cytometer (μIFC) is an attractivetool due to its miniaturization capabilities, low-cost, low sample re-quirements, and on-chip, single-cell analysis capability. In μIFC devices,mostly coplanar electrodes are used and non-homogenous electric fieldcauses variation in the measured impedance signal due to the electricfield interaction of dispersed dielectric particles. The impedance mea-surement signal is position-dependent since each particle has a differenttrajectory and interacts with different intensity of electric field.Therefore, particle focusing techniques are used in μIFC to reach asingle train of particles and improve the measurement precision andresolution. Sheath flow supported hydrodynamic focusing [18–21],flow-induced inertial focusing [22] and viscoelastic focusing [23],dielectrophoresis [18,24–26] and surface acoustic waves [27–29] havebeen used to attain single train of particles. However these focusingmechanisms bring additional concerns to the cytometry system and theoptimal sample focusing mechanism remains elusive so far. For in-stance, commonly used sheath flow systems suffer from the use of ex-tensive amount of shield fluid; inertial focusing requires very high flowrates putting a high demand on the electrical sampling rate; viscoelasticfocusing is highly dependent on rheological properties of the carrierfluid; and active focusing methods add complexity to fabrication.

https://doi.org/10.1016/j.snb.2019.127531Received 8 August 2019; Received in revised form 5 November 2019; Accepted 2 December 2019

⁎ Corresponding author.E-mail address: [email protected] (C. Elbuken).

Sensors & Actuators: B. Chemical 307 (2020) 127531

Available online 07 January 20200925-4005/ © 2020 Elsevier B.V. All rights reserved.

T

Recently, some focusing-free μIFC studies were introduced, wheresignal processing algorithms and electrode wiring configurations areused to compensate the signal change due to positional variation of theparticles [30–34]. These studies were able to estimate the cross-sec-tional position of the particles along the channel. In this study, wepresent a focusing-free impedimetric detection system by only geo-metry optimization without using any additional signal processingstrategies.

Detection of cells is an important application of μIFCs. The firstsingle cell detection in μIFC was carried out by Ayliffe and his collea-gues [35]. Gawad et al. made a big step in electrical single cell analysisby demonstrating differentiation of micron-sized polystyrene beads[15] and blood cells [36]. The electrical properties of single cells canprovide cellular biomarkers to classify different cell types such astumor, stem and blood cells [37]. Quantification of different blood cellsand their subpopulations has clinical importance particularly for thediagnosis of the autoimmune disorders such as leukemia [38]. Chroniclymphocytic leukemia (CLL) is the most common type of leukemia inthe developed world; and the second most common one worldwide[39]. While leukemia detection requires detailed immune-phenotypingand blood smear tests [36], a recent study has demonstrated that thereis a strong correlation between lymphocyte level of the peripheral bloodand CLL [40,41].

In this study, we developed a focusing-free μIFC device with co-planar electrode configuration that can count and distinguish red bloodcells (RBCs) and lymphocyte white blood cells (WBCs). A poly-dimethylsiloxane (PDMS) microchannel layer was reversibly sealedonto a glass electrode layer to form the microfluidic chip. A lock-inamplifier was employed as the measurement unit. The effect of theexcitation voltage and the solution conductivity on the signal-to-noiseratio (SNR) were investigated. The optimization of the channel di-mensions (width and height) was also explained in detail. The systemwas calibrated with 3 and 6 μm diameter PS beads. Finally, RBCs andleukemia WBCs were investigated and differentiated. To our best, thepresented system is one of the simplest yet effective μIFC tool.

2. Materials and methods

2.1. Sample preparation

Spherical PS beads with diameter of 3 and 6 μmwere suspended in a1X-PBS buffer solution (106 particles/ml). Blood samples were fingerpricked from a healthy volunteer donor (50 μl) and mixed with EDTA (5μl) solution in an eppendorf tube. The sample was centrifuged at 5000rpm for 3 min. 5 μl of precipitated RBCs was dispensed in a 5 ml 1X-PBSsolution (dilution of 1:1000). We purchased mononuclear B lympho-blastic CCRF-SB leukemia cell line (ATCC® CCL-120TM) for impedi-metric leukemia cell measurements. The cells were grown in the sus-pension cultures in the RPMI complete medium (88 % RPMI 1640medium, 10 % fetal bovine serum with %1 Pen-Strep and %1 gluta-mate) and sub-cultured every 3–4 days. Confluent cells were cen-trifuged at 5000 rpm for 5 min, then re-suspended in 1X-PBS solution (2× 107 cells/ml). PS bead and cell suspensions were pumped throughthe microchannel using a computer controlled pressure pump (ElveflowOB1).

2.2. Microfluidic chip design and fabrication

The microfabrication steps and a photo of the fabricated μIFC chipare given in Fig. 1. The fabrication consists of two main steps: electrodeand microfluidic channel fabrication. Both processes were carried out atUNAM (National Nanotechnology Research Center, Bilkent University)cleanroom with conventional soft lithography techniques [42]. First,AZ5214E (MicroChem) photoresist was spin coated on a glass slide (1mm thick, 75 × 25 mm2). Then, electrode patterns on a transparencymask was transferred onto the substrate with the following steps of UV

exposure (EVG620 Mask Aligner) and development (AZ400 K Devel-oper). Au and Cr layers were thermally evaporated (Vaksis ThermalEvaporator) to obtain 20 and 50 nm thickness, respectively. Finally, themetal deposited glass substrate was dipped in acetone bath overnightfor lift-off.

The mold fabrication was started with spin coating of 10 μm thickSU-8 2005 photoresist (Microchem) onto a 4- inch silicon wafer, fol-lowed by UV exposure and developing steps. PDMS (DowcorningSylgard 184) was mixed with curing agent at 10:1 ratio, poured ontothe prepared mold and cured on hot plate for 3 h at 100 °C. The curedPDMS was peeled off, giving the female replica of the mold. At the end,the electrode-patterned glass slide and PDMS channel layer are alignedand pressed against each other without any plasma treatment. They aresandwiched in between two laser-cut Plexiglas layers (Epilog Zing CO2

laser) and kept together using bolts as shown in Fig. 1. The top Plexiglaslayer has two circular openings for inlet and outlet access. A throughcut was made at the center of the top Plexiglas layer to decrease thepressure on PDMS, since shallower microchannels are more prone toclogging. The elimination of plasma bonding, thus having reversiblephysical bonding, enables us to reuse the microfluidic chip in case ofany clogging.

2.3. Working principle

Lock-in amplifiers and custom electronic circuits are commonlyused as the detection unit for impedimetric particle detection systems[15,43]. In our previous studies, we used an LCR meter for particle anddroplet detection [2,4,44]. Low sampling rate is the main disadvantageof the LCR meters in comparison to lock-in amplifier systems. In thisstudy, we used an FPGA-based HF2LI lock-in amplifier, with a plug andplay trans-impedance amplifier module HF2TA (Zurich Instruments).The particle detection principle of the μIFC is illustrated in Fig. 2. Thecentral electrode is excited with an AC signal, which induces a currentbetween the other electrodes. Induced current is converted to a voltagesignal through the trans-impedance amplifier (TIA). The voltage dif-ference between the electrodes is amplified in a differential amplifier(DA). At the end, the lock-in amplifier is fed with the amplified signals.When there is no particle between the electrodes, the voltage output iszero, since the current is the same in both sensing areas. Once theparticle enters the first electrode region, the positive voltage signal risesat the output. Similarly, when the particle reaches to the second elec-trode region, the negative voltage output rises due to the polarity of thesignal.

At the output of the DA, voltage signal is demodulated using 4thorder 1 kHz bandwidth low-pass filter at 20 kSa/s sampling rate. Afterthe demodulation, output data (for 1 min measurement duration) isexported to a PC via USB 2.0 port in. txt format to be post-processedusing custom-written python script. The script removes the voltageoffset by subtracting the average value from each data points. Then,voltage signal extremums are detected using a derivative sub-functionto acquire the peak amplitude (peak-to-peak amplitude/2) and transi-tion time (passing time between the voltage max and voltage minpoints).

3. Results and discussion

We initially investigated the effect of different parameters such aschannel width (Section 3.1), channel height (3.2), solution conductivity(Section 3.3), excitation voltage (Section 3.4), and the particle size(Section 3.5). Then, we performed single cell differentiation to differ-entiate human RBCs and malignant leukemia cells. We used an ex-citation frequency of 1.5 MHz for the experiments.

3.1. The effect of the channel width

The coplanar electrodes (10 μm width and gap) were sealed using

I. Bilican, et al. Sensors & Actuators: B. Chemical 307 (2020) 127531

2

straight channels (10 μm height) with three different widths: 30, 50,and 70 μm. 3-μm diameter PS bead solution was pumped through themicrochannel; the resulting voltage signals are given in Fig. 3(a)-(c). Asthe channel width gets larger, the peak-to-peak noise increases, but themean peak amplitude does not change significantly. These results can

be explained via electrical equivalent models representing the im-pedance of the solution and particle. Two states are defined for theparticle outside or inside the electrode region, as shown in Fig. 3(d) and3(e) respectively. Due to the symmetry of electric field lines, a singlepair of coplanar electrodes was analyzed. The effect of channel width

Fig. 1. Microfluidic chip. (a) Fabrication of PDMS microchannels and electrode integration. Elastomer solution is poured (i-ii) onto the mold with 10 μm-thickmicrostructures and cured (iii). The PDMS channel layer (iv) and electrode-patterned glass slide (v) are sandwiched in between two laser-cut Plexiglas layers andtightened using bolts (vi). (b) The optical microscope image (left) and the photograph (right) of the final microfluidic device.

Fig. 2. Schematic of microfluidic chip, lock-in amplifier and the output signal for a particle traveling over the electrodes. The middle electrode is excited with the ACbias voltage which induces current between middle and outer electrodes. The current is converted to voltage signal in TIA. The voltage difference between two outerelectrodes at the output of TIA is amplified in DA. The differential voltage is monitored through the lock-in amplifier. A positive (negative) peak was observed whenthe particle passes over the first (third) and the middle electrode.

I. Bilican, et al. Sensors & Actuators: B. Chemical 307 (2020) 127531

3

on the sensing signal can be seen from the equivalent circuit analysisresults based on experimentally observed values. A detailed explanationof the equivalent electrical circuit approach is given in ESI document.The buffer resistance inside 30 μm channel was represented as a singleimpedance element of 20 kΩ and a single particle inside electrode re-gion was simulated as an additional 1 kΩ impedance. If the system isexcited with 1 V, 50 μA current passes from the 20 kΩ channel (Fig. 3(d1)). If the particle enters the sensing region, the total current drops to47.6 μA for the equivalent impedance of 21 kΩ. It means that 2.4 μAcurrent variation (4.8 % in ratio to 50 μA) is obtained by a singleparticle (Fig. 3(e1)). A 60 μm width channel can be represented withtwo parallel impedances of 20 kΩ. For this case, the total impedance

drops to 10 kΩ, and the total current across the channel rises to 100 μAfor the same 1 V excitation voltage (Fig. 3 (d2)). Presence of the particlebetween the electrodes generates only 243.9 Ω variation in theequivalent impedance, which corresponds to 2.4 μA decrease in current(2.4 % change in ratio, Fig. 3(e2)). While the current variation (2.4 μA)is the same for both widths (30 μm and 60 μm), the ratio of the blockedcurrent to the total current (% change) is half for 60 μm width channel.The average signal levels (peak heights) are the same for both widths;however, peak-to-peak noise increases with the increasing channelwidth.

Fig. 3. Experimental results and the equivalent circuit model comparison of the effect of the channel width on SNR and peak-to-peak noise. The experimental resultsof 3 μm diameter beads for (a) 30 μm, (b) 50 μm, and (c) 70 μm width channels (Peak-to-peak noise level is indicated by red dashed line). The equivalent electricalmodel representations; (d) without the particle in the electrode region for (d1) 30 μm, (d2) 60 μm width channel, (e) with the particle in electrode region for (e1) 30μm, (e2) 60 μm width channel. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

I. Bilican, et al. Sensors & Actuators: B. Chemical 307 (2020) 127531

4

3.2. The effect of the channel height

Non-homogeneous electric field distribution, stemming from co-planar microelectrodes, causes variation in impedance signals when thedielectric particle hinders the electric field lines. Identical particlesflowing at different vertical positions through the microchannel createdifferent impedance signals as they encounter different field intensity incoplanar electrode configuration. Therefore, particle focusing techni-ques are used to reach a single train of particles at a fixed verticalposition and reduce the signal variations. Such techniques are attrac-tive, yet sacrifice the simplicity of the fluidic system. Fortunately, if thechannel dimensions and the size of targeted particles are well-adjusted,it is possible to achieve focusing-free separation of different size par-ticles as shown in this section.

Our design optimization is based on numerical results obtained by afinite element modeling software (COMSOL v 5.3). 3D simulations ofparticle detection was performed using the electric field analysisformed by the coplanar electrodes placed inside the microfluidicchannel. Three coplanar electrodes (10 μm width and gap) were used asshown in Fig. 4(a). The middle electrode was excited with 1 kHz, 1 V

excitation voltage. The channel width was kept constant at 70 μm,whereas the channel height was varied between 10 μm and 50 μm with10 μm intervals. We ran the simulations to show the particle size de-pendent impedance change for channel height variations as illustratedin Fig. 4(b). PS particle is located at the halfway between the first andsecond electrode, and the impedance change is recorded as peak am-plitude (PA) while the particle height is vertically varied as channelbottom (particle edge kept at 0.5 μm distant from the wall), center andtop. The simulations were repeated for two different size particles (3and 6 μm) and five channel heights. PS particles disturb the electricfield and generates current variation [45–49]. When the particle getscloser to the electrodes, it generates a higher impedance peak, since theelectric field lines are more intense at the edge of the electrodes. Im-pedance change is proportional to the size of the particle moving inelectro-active field [50].

The PA vs particle height for each channel height is plotted inFig. 4(b). The simulation results revealed that the particles passingcloser to the electrodes generate higher impedance peaks. The PAvariation, ΔPA= PAmax – PAmin, decreases with decreasing channelheight. Due to the polymeric upper channel wall, decreasing the

Fig. 4. Simulation results illustrating signal variation as a function of channel height (a) Illustration of particle size dependent impedance for channel heightvariations. Red-blue gradient bar represents 6 μm particle; black-yellow gradient bar represents 3 μm particle. Electrode and gap widths are 10 μm. (b) The simulatedpeak amplitude variation results obtained from the simulations. PAmax, PAmin, and ΔPA values are tabulated (c) The simulation results for electric current densityacross the channel cross section for varying channel heights; showing that electric current density homogeneity is improved by reducing the channel height resultingin lower signal variation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

I. Bilican, et al. Sensors & Actuators: B. Chemical 307 (2020) 127531

5

channel height squeezes the electric field lines to a narrower space andcreates more homogenous electric field distribution as shown inFig. 4(c), so that PA value is less dependent to the vertical positon in-side the channel. This effect is a function of the electrode width andgap, which are both chosen as 10 μm in our study.

The results summarized in Fig. 4(b) show that the differentiation ofparticles based on their size is possible for a channel height of 10 μm. Asthe channel height increases, the electric field intensity becomes morenon-uniform and peak amplitude variation range increases sig-nificantly, especially for the 6 μm diameter particle. Hence, as a rule ofthumb, we suggest keeping the channel height similar to the electrodewidth and gap to achieve robust focusing-free impedimetric particledetection results. In Fig. 4(c) the electric current density results weregiven for the channel cross section for all channel heights. As seen fromthese simulations, decreasing the channel height traps the electric fieldlines in the channel region that leads to a more uniform electric fieldintensity distribution throughout the cross section. It should be notedthat these results also depend on the channel and solution electricalproperties, so we recommend performing such 3D simulations to opti-mize the channel height for obtaining successful focusing-free im-pedance detection results.

3.3. The effect of the solution conductivity

Solution conductivity plays an important role for high sensitivitymeasurements. The solution conductivity modifies the total currentflowing through the system, but the ratio of the current variation staysconstant. The noise level is inversely related to the solution con-ductivity, hence higher ionic concentration in the solution is favorablefor higher signal to noise ratio (SNR). However, there is a limitation inthe ionic concentration of the solution in biological applications. Cellviability decreases with the increasing ionic concentration of the solu-tion, which restricts further enhancement of SNR.

Fig. 5(a)–(c) give the experimental results for 6 μm PS beads

suspended in three different conductivity (0.1, 1, 2.5 S/m) buffer so-lutions for 0.5 V excitation voltage. Fig. 5(d) shows the peak amplitudehistogram plot for each conductivity and summarizes the results ofSNR, mean value (Mean), standard deviation (SD), and coefficient ofvariations (% CV) for the output signal. The solution conductivity isdirectly proportional to SNR, mean and SD values. We observed a % CVimprovement from 14.36 % to 10.40 % for higher conductivities.

3.4. The effect of the excitation voltage

For demanding particle detection and differentiation tasks, thefundamental limitation comes from the sensitivity of the measurementunit. For most electronic circuits, as a rule of thumb, the higher theexcitation signal, the higher the SNR. For our system, the excitationvoltage determines the current that will be supplied to lock-in ampli-fier. Since the same size particles block the same ratio of the totalcurrent flowing between the electrodes, the ratio of the blocked currentto the total current is independent of excitation voltage. On the flipside, since the noise level does not linearly increase with the excitationsignal, an improvement in SNR is expected using stronger excitationsignals. We characterized SNR using three levels of excitation voltagesfor 3 μm diameter PS beads. The resultant experimental electrical signalprofiles are shown in Fig. 6. We pumped the suspension through themicrochannel (10 μm height, 30 μm width) at 30 mbar pressure dif-ference. Excitation voltage of 0.5, 1 and 3 V resulted in 1.86, 3.52, and9.63 SNR, respectively. However, increasing the excitation voltage isnot unlimitedly favorable. At higher excitation voltage, metal elec-trodes start to deteriorate (erosion of the metal layer) due to the elec-trochemical effects [51,52]. Also higher excitation voltage would harmthe living cells and organisms.

3.5. The effect of the particle size

In this section, we compared experimental impedance measurement

Fig. 5. Experimental results showing the effect of solution conductivity. Impedance signal obtained by 6 μm diameter PS beads at (a) 0.1 S/m, (b) 1 S/m, (c) 2.5 S/mconductivities, and (d) peak amplitude histogram plots. The solution conductivity improves the sensitivity due to lower impedance of the solution through whichmore current is induced by the excitation voltage.

I. Bilican, et al. Sensors & Actuators: B. Chemical 307 (2020) 127531

6

results for 3 μm and 6 μm diameter PS beads. The solution conductivity,excitation voltage and excitation frequency were 2.5 S/m, 3 V, and 1.5MHz respectively. We defined two parameters to distinguish the dif-ferent size particles. Peak amplitude (PA) is the signal amplitude that isproportional to the volume of the particle size, and transit time (TT) isthe time between the peak and dip points of a signal corresponding to asingle particle. The results are given in Fig. 7. The average PAs for 3 μmand 6 μm PS beads were measured as 0.15 mV and 1.65 mV, respec-tively. There is an order of magnitude difference in PA values, whichcan be used as a reliable metric for differentiation of particles based ontheir size. The average TT of 3 μm and 6 μm PS beads are approximately1.88 ms and 3.08 ms, respectively. At the end of the system char-acterization, we performed a classification experiment using similarconcentrations of 3 μm and 6 μm beads. The PA voltage level set to 0.40mV (three times bigger than the mean PA value of 3 μm PS beads) forthe classification threshold based on the single population experiments.Fig. 7c shows the 30 s of time duration for the signal obtained duringthis particle mixture measurement. The impedance signal was analyzedoff-line and two similar particle populations were obtained.

Although there are numerous studies that demonstrate impedi-metric particle detection, there is a lack of consensus on the perfor-mance metrics or certified reference solutions that can be used forcomparison of systems. Hence, in general it is challenging to benchmarka given system. One of the most commonly used particles for char-acterization of impedimetric detection systems are spherical PS beadsas they are readily available at different sizes with relatively highmonodispersity and also yield orientation independent results due totheir spherical symmetry. Interestingly, 3 μm and 6 μm diameter PSbeads have been used by a study that utilize parallel electric fieldconfiguration using a novel floating electrode [43,53]. As can be seenwe were able to obtain an order of magnitude higher SNR using ourfocusing-free system despite the simplicity of the fabrication. Theseresults demonstrate the potential of focusing-free impedimetric detec-tion using coplanar electrodes and re-sealable microfluidic devices in

comparison to some devices that require elaborate alignment stepsduring microfabrication [20,43,54].

3.6. Impedimetric cell detection using red blood cells and malignantleukemia cells in μIFC

Differential impedance measurements were performed using humanRBCs and leukemia cells, both of which were separately suspended in2.5 S/m conductivity buffer solutions. Cell concentrations in the solu-tion were measured to be 2 × 107 cells/ml by a hemocytometer. Bothsuspensions were pumped to the channel at 30 mbar pressure differenceand 5 μl/h flow rate, yielding a throughput of 2000 cells/min. Theelectrodes were excited using a 3 V, 1.5 MHz signal. The lock-in am-plifier sampling rate was set to 10 kSa/s.

Fig. 8(a) and (b) give read-out signals for 10-s close-up views forsingle particle events and histogram plots of peak amplitude. InFig. 8(c), scatter plot of transit time versus peak amplitude is given forRBC and leukemia cell suspensions for 30-s experimental results. Dif-ferential signal waveform is clearly observed for both cell types in close-up single events. While the peak amplitude of RBCs are accumulatedaround 0.65± 0.3 mV, leukemia cells yield 1.35±0.3 mV. Eventhough the transit time overlaps for two different cell populations, weobserved clear differentiation in terms of peak amplitude as shown inFig. 8(c) without using any particle focusing mechanism. The first im-plication of the chronic lymphocytic leukemia (CLL) is the risingnumber of the lymphocytes in human blood. Although lymphocytes areslightly larger than the RBCs in diameter, we still reached a higherdifferentiating contrast with their volumetric ratio due to the lympho-cyte’s spherical shape, unlike RBC’s discoid shape. The other whiteblood cell types such as monocytes (15−30 μm), neutrophils (10−12μm), eosinophils (10−12 μm), and basophils (12−15 μm) are muchbigger than lymphocytes (7−8 μm), enabling size-based differentiationof the lymphocytes from other types [55]. An extension of this focusing-free blood cell differentiation system would be analysis of blood cells

Fig. 6. Experimental results showing the effect of excitation voltage signal for 3 μm diameter PS beads at 10 μm channel height: (a) 0.5 V, (b) 1 V, and (c) 3 V. Whilethe ratio of the blocked current to the total current remains constant, the total current flowing from the excitation electrode to the lock-in amplifier increases togetherwith both the signal and the noise. However, the noise does not increase as much as the signal.

I. Bilican, et al. Sensors & Actuators: B. Chemical 307 (2020) 127531

7

with simultaneous optical verification. Here, it is important to note thatthe simplistic focusing-free impedimetric detection presented in thiswork is limited to a short range of particle sizes that should be opti-mized for the particles of interest.

4. Conclusions

In this study, RBC and malignant lymphocyte detection was in-vestigated using a focusing-free μIFC. The system was optimized formicrofluidic channel dimensions, conductivity of the carrier solution,excitation voltage, and particle size using analytical, computational andexperimental approaches. Design guidelines were provided for label-free on-chip microfluidic detection of micro-scale particles. Thanks to

the precision and sensitivity of the measurement setup, differentiationof RBCs and lymphocytes, was achieved. With the focusing-free μIFCsystem, we demonstrated a cost-effective, high throughput, and reu-sable technique for cytometry applications without the need of ad-vanced signal-processing techniques or complex electrode designs. Thisstudy may also find applications in diagnosis of some other hematolo-gical diseases like malaria and anemia based on the shape differences ofRBCs.

Declaration of Competing Interest

The authors declare that they have no known competing financialinterests or personal relationships that could have appeared to

Fig. 7. Impedance measurements with histogram plots of Peak Amplitude and Transit Time for (a) 3 μm and (b) 6 μm diameter PS beads. Since bigger particlesoccupy larger volume inside the electroactive region, peak amplitude is higher for 6 μm particles. Transit time also increases with increasing particle size: 1.88± 0.6ms for 3 μm PS beads and 3.08±0.9 ms for 6 μm PS beads. (c) Impedance measurement for the 3 μm and 6 μm diameter PS beads mixture. The PA voltage of 0.4 mVwas used as a classification threshold value. PA: Peak Amplitude, PPA: Peak-to-peak Amplitude, PA = PPA/2, and Transit Time = tPAmin - tPAmax.

I. Bilican, et al. Sensors & Actuators: B. Chemical 307 (2020) 127531

8

influence the work reported in this paper.

Acknowledgments

This study was financially supported by Kirikkale University (BAPProject No: 2016/077). Ismail Bilican was supported by TÜBİTAKScientific Support Department within the scope of 2211-Domestic PhDscholarship program. The authors thank Ziya Isiksacan for proofreadingthe manuscript.

Appendix A. Supplementary data

Supplementary material related to this article can be found, in theonline version, at doi:https://doi.org/10.1016/j.snb.2019.127531.

References

[1] G.L. Damhorst, C.E. Smith, E.M. Salm, M.M. Sobieraj, H. Ni, H. Kong, et al., Aliposome-based ion release impedance sensor for biological detection, Biomed.Microdevices 15 (2013) 895–905.

[2] M.T. Guler, I. Bilican, Capacitive detection of single bacterium from drinking waterwith a detailed investigation of electrical flow cytometry, Sens. Actuators A Phys.269 (2018) 454–463.

[3] H. Schwan, Linear and nonlinear electrode polarization and biological materials,

Fig. 8. The μIFC results of read-out signals for single particle events, and histogram plots of peak amplitude for (a) RBCs, (b) WBCs, and (c) Scatter plot of both celltypes in terms of transit time vs peak amplitude. Peak amplitude of RBCs and leukemia cells were accumulated around 0.65± 0.3 mV, and 1.35± 0.3 mV,respectively.

I. Bilican, et al. Sensors & Actuators: B. Chemical 307 (2020) 127531

9

Ann. Biomed. Eng. 20 (1992) 269–288.[4] M.T. Guler, I. Bilican, S. Agan, C. Elbuken, A simple approach for the fabrication of

3D microelectrodes for impedimetric sensing, J. Micromechanics Microengineering25 (2015) 095019.

[5] O.F. Schanne, P. Ruiz, E. Ceretti, Impedance Measurements in Biological Cells,Wiley, 1977.

[6] P. Shah, X. Zhu, C. Chen, Y. Hu, C.-Z. Li, Lab-on-chip device for single cell trappingand analysis, Biomed. Microdevices 16 (2014) 35–41.

[7] S. Zheng, M.S. Nandra, C.-Y. Shih, W. Li, Y.-C. Tai, Resonance impedance sensing ofhuman blood cells, Sens. Actuators A Phys. 145 (2008) 29–36.

[8] W.H. Coulter, Means for counting particles suspended in a fluid, US Patent, 2(1953)508-65.

[9] O. Saleh, L. Sohn, Quantitative sensing of nanoscale colloids using a microchipcoulter counter, Rev. Sci. Instrum. 72 (2001) 4449–4451.

[10] U. Hassan, R. Bashir, Electrical cell counting process characterization in a micro-fluidic impedance cytometer, Biomed. Microdevices 16 (2014) 697–704.

[11] Y. Wu, J.D. Benson, M. Almasri, Micromachined coulter counter for dynamic im-pedance study of time sensitive cells, Biomed. Microdevices 14 (2012) 739–750.

[12] J. Zhe, A. Jagtiani, P. Dutta, J. Hu, J. Carletta, A micromachined high throughputcoulter counter for bioparticle detection and counting, J. MicromechanicsMicroengineering 17 (2007) 304.

[13] S. Zheng, M. Liu, Y.-C. Tai, Micro coulter counters with platinum black electro-plated electrodes for human blood cell sensing, Biomed. Microdevices 10 (2008)221–231.

[14] K.C. Cheung, M. Di Berardino, G. Schade‐Kampmann, M. Hebeisen, A. Pierzchalski,J. Bocsi, et al., Microfluidic impedance‐based flow cytometry, Cytom. Part A 77(2010) 648–666.

[15] S. Gawad, L. Schild, P. Renaud, Micromachined impedance spectroscopy flow cyt-ometer for cell analysis and particle sizing, Lab Chip 1 (2001) 76–82.

[16] D. Holmes, D. Pettigrew, C.H. Reccius, J.D. Gwyer, C. van Berkel, J. Holloway,et al., Leukocyte analysis and differentiation using high speed microfluidic singlecell impedance cytometry, Lab Chip 9 (2009) 2881–2889.

[17] B.J. Bain, Blood Cells: a Practical Guide, John Wiley & Sons, 2014.[18] L. Che-Hsin, L. Gwo-Bin, F. Lung-Ming, H. Bao-Herng, Vertical focusing device

utilizing dielectrophoretic force and its application on microflow cytometer, J.Microelectromechanical Syst. 13 (2004) 923–932.

[19] R. Rodriguez-Trujillo, C. Mills, J. Samitier, G. Gomila, Low cost micro-coultercounter with hydrodynamic focusing, Microfluid. Nanofluidics 3 (2007) 171–176.

[20] H. Wang, N. Sobahi, A. Han, Impedance spectroscopy-based cell/particle positiondetection in microfluidic systems, Lab Chip 17 (2017) 1264–1269.

[21] N. Watkins, B.M. Venkatesan, M. Toner, W. Rodriguez, R. Bashir, A robust electricalmicrocytometer with 3-dimensional hydrofocusing, Lab Chip 9 (2009) 3177–3184.

[22] W. Tang, D. Tang, Z. Ni, N. Xiang, H. Yi, Microfluidic impedance cytometer withinertial focusing and liquid electrodes for high-throughput cell counting and dis-crimination, Anal. Chem. 89 (2017) 3154–3161.

[23] M. Serhatlioglu, M. Asghari, M. Tahsin Guler, C. Elbuken, Impedance-based vis-coelastic flow cytometry, Electrophoresis 40 (2019) 906–913.

[24] Y. Choongho, J. Vykoukal, D.M. Vykoukal, J.A. Schwartz, S. Li, P.R.C. Gascoyne, Athree-dimensional dielectrophoretic particle focusing channel for microcytometryapplications, J. Microelectromechanical Syst. 14 (2005) 480–487.

[25] M. Shaker, L. Colella, F. Caselli, P. Bisegna, P. Renaud, An impedance-based flowmicrocytometer for single cell morphology discrimination, Lab Chip 14 (2014)2548–2555.

[26] S. Yan, J. Zhang, M. Li, G. Alici, H. Du, R. Sluyter, et al., On-chip high-throughputmanipulation of particles in a dielectrophoresis-active hydrophoretic focuser, Sci.Rep. 4 (2014) 5060.

[27] G. Goddard, J.C. Martin, S.W. Graves, G. Kaduchak, Ultrasonic particle‐concen-tration for sheathless focusing of particles for analysis in a flow cytometer, J. Int.Soc. Anal. Cytol. 69 (2006) 66–74.

[28] C. Grenvall, C. Antfolk, C.Z. Bisgaard, T. Laurell, Two-dimensional acoustic particlefocusing enables sheathless chip Coulter counter with planar electrode configura-tion, Lab Chip 14 (2014) 4629–4637.

[29] J. Shi, S. Yazdi, S.-C.S. Lin, X. Ding, I.-K. Chiang, K. Sharp, et al., Three-dimensionalcontinuous particle focusing in a microfluidic channel via standing surface acousticwaves (SSAW), Lab Chip 11 (2011) 2319–2324.

[30] F. Caselli, A. De Ninno, R. Reale, L. Businaro, P. Bisegna, A novel wiring scheme forstandard chips enabling high-accuracy impedance cytometry, Sens. Actuators BChem. 256 (2018) 580–589.

[31] V. Errico, A. De Ninno, F.R. Bertani, L. Businaro, P. Bisegna, F. Caselli, Mitigatingpositional dependence in coplanar electrode Coulter-type microfluidic devices,Sens. Actuators B Chem. 247 (2017) 580–586.

[32] R. Reale, A. De Ninno, L. Businaro, P. Bisegna, F. Caselli, High-throughput electricalposition detection of single flowing particles/cells with non-spherical shape, LabChip (2019).

[33] M. Solsona, E.Y. Westerbeek, J.G. Bomer, W. Olthuis, A. van den Berg, Gradient inthe electric field for particle position detection in microfluidic channels, Lab Chip19 (2019) 1054–1059.

[34] D. Spencer, F. Caselli, P. Bisegna, H. Morgan, High accuracy particle analysis usingsheathless microfluidic impedance cytometry, Lab Chip 16 (2016) 2467–2473.

[35] H.E. Ayliffe, A.B. Frazier, R. Rabbitt, Electric impedance spectroscopy using mi-crochannels with integrated metal electrodes, J. Microelectromechanical Syst. 8(1999) 50–57.

[36] A.C. Rawstron, K.A. Kreuzer, A. Soosapilla, M. Spacek, O. Stehlikova, P. Gambell,et al., Reproducible diagnosis of chronic lymphocytic leukemia by flow cytometry:an European Research Initiative on CLL (ERIC) & European Society for Clinical CellAnalysis (ESCCA) Harmonisation project, Cytometry B Clin. Cytom. 94 (2018)121–128.

[37] J. Chen, C. Xue, Y. Zhao, D. Chen, M.-H. Wu, J. Wang, Microfluidic impedance flowcytometry enabling high-throughput single-cell electrical property characterization,Int. J. Mol. Sci. 16 (2015) 9804–9830.

[38] H. Morgan, D. Spencer, Microfluidic impedance cytometry for blood cell analysis,Microfluidics Med Appl 36 (2014) 213.

[39] M. Hallek, Chronic lymphocytic leukemia: 2017 update on diagnosis, risk stratifi-cation, and treatment, Am. J. Hematol. 92 (2017) 946–965.

[40] P. Strati, T.D. Shanafelt, Monoclonal B-cell lymphocytosis and early-stage chroniclymphocytic leukemia: diagnosis, natural history, and risk stratification, Blood 126(2015) 454–462.

[41] V. Tseng, A.S. Morgan, C.P. Leith, D.T. Yang, Efficient assessment of peripheralblood lymphocytosis in adults: developing new thresholds for blood smear reviewby pathologists, Clin/ Chem. Lab. Med. 52 (2014) 1763–1770.

[42] I. Bilican, M.T. Guler, N. Gulener, M. Yuksel, S. Agan, Capacitive solvent sensingwith interdigitated microelectrodes, Microsyst. Technol. 22 (2016) 659–668.

[43] K. Cheung, S. Gawad, P. Renaud, Impedance spectroscopy flow cytometry: on‐chiplabel‐free cell differentiation, Cytom. Part A 65 (2005) 124–132.

[44] P.K. Isgor, M. Marcali, M. Keser, C. Elbuken, Microfluidic droplet content detectionusing integrated capacitive sensors, Sens. Actuators B Chem. 210 (2015) 669–675.

[45] C. Iliescu, D.P. Poenar, M. Carp, F.C. Loe, A microfluidic device for impedancespectroscopy analysis of biological samples, Sens. Actuators B Chem. 123 (2007)168–176.

[46] C. Küttel, E. Nascimento, N. Demierre, T. Silva, T. Braschler, P. Renaud, et al.,Label-free detection of Babesia bovis infected red blood cells using impedancespectroscopy on a microfabricated flow cytometer, Acta Trop. 102 (2007) 63–68.

[47] H. Morgan, D. Holmes, N.G. Green, High speed simultaneous single particle im-pedance and fluorescence analysis on a chip, Curr. Appl. Phys. 6 (2006) 367–370.

[48] S. Murali, A.V. Jagtiani, X. Xia, J. Carletta, J. Zhe, A microfluidic Coulter countingdevice for metal wear detection in lubrication oil, Rev. Sci. Instrum. 80 (2009)016105.

[49] D. Wood, S.-H. Oh, S.-H. Lee, H. Soh, A. Cleland, High-bandwidth radio frequencycoulter counter, Appl. Phys. Lett. 87 (2005) 184106.

[50] H. Morgan, N. Green, AC Electrokinetics: Colloids and Nanoparticles, ResearchStudy Press, 2003.

[51] A.M. Johnson, D.R. Sadoway, M.J. Cima, R. Langer, Design and testing of an im-pedance-based sensor for monitoring drug delivery, J. Electrochem. Soc. 152 (2005)H6–H11.

[52] D. Palanker, A. Vankov, P. Huie, S. Baccus, Design of a high-resolution optoelec-tronic retinal prosthesis, J. Neural Eng. 2 (2005) S105.

[53] L.I. Segerink, A.J. Sprenkels, J.G. Bomer, I. Vermes, A. van den Berg, A new floatingelectrode structure for generating homogeneous electrical fields in microfluidicchannels, Lab Chip 11 (2011) 1995–2001.

[54] C. Bernabini, D. Holmes, H. Morgan, Micro-impedance cytometry for detection andanalysis of micron-sized particles and bacteria, Lab Chip 11 (2011) 407–412.

[55] P.R. Wheater, H.G. Burkitt, V.G. Daniels, Functional Histology. A Text and ColourAtlas, Churchill Livingstone, 23 Ravelston Terrace, Edinburgh, EH4 3TL, 1979.

Ismail Bilican is currently a visiting researcher at Bilkent University UNAM and lecturerat Aksaray University. He received his M.S. and Ph.D. degree from Kirikkale UniversityDepartment of Physics. His research interests are in the field of flow cytometry applica-tions, microfluidics components, rapid prototyping and lab-on-a-chip devices.

Mustafa Tahsin Guler received his Ph.D. degree from Kirikkale University Department ofPhysics. After a post-doctoral study at Harvard Medical School, he is currently a visitingresearcher at Bilkent University UNAM and research assistant at Kirikkale University. Hisresearch interest covers development of microfluidics components, rapid prototypingmethods and micro-electrical biosensors.

Murat Serhatlioglu is a Ph.D. candidate at UNAMMaterials Science and NanotechnologyProgram at Bilkent University. He received his M.S. degree from Meliksah University. Hisresearch interests are in the field of microfluidics and optical integration, flow cytometryapplications, single cell characterization, viscoelastic fluids and their microfluidic appli-cations.

Talip Kirindi is a Professor at Kirikkale University, Department of Mathematic andScience Education. His research interests include shape memory, martensitic transfor-mation, solid-state phase transition and material science.

Caglar Elbuken is an assistant professor at Bilkent University, UNAM - NationalNanotechnology Research Center. He obtained his B.Sc. degree in Electrical andElectronics Engineering at Bilkent University and his Ph.D. degree in MechanicalEngineering at University of Waterloo. His research interests include lab-on-a-chip de-vices, droplet microfluidics, point-of-care diagnostics, and viscoelastic fluids. He is therecipient of the Science Academy’s 2019 Young Scientist Award.

I. Bilican, et al. Sensors & Actuators: B. Chemical 307 (2020) 127531

10