Biosensors and Bioelectronics 121 (2018) 10–18

Contents lists available at ScienceDirect

Biosensors and Bioelectronics

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

Impedance detection integrated with dielectrophoresis enrichment platform T

for lung circulating tumor cells in a microfluidic channel
⁎
Ngoc-Viet Nguyen, Chun-Ping Jen
Department of Mechanical Engineering, National Chung Cheng University, Chia Yi 621, Taiwan, ROC

A R T I C LE I N FO A B S T R A C T

Keywords: Enrichment and detection of rare cells are essential in many biological and medical applications. In this study,
Rare cells circulating tumor cells were detected by combining dielectrophoretic (DEP) manipulation and impedance
Microfluidics measurement using circular microelectrodes within a single microfluidic device. A prototype of the device was
EIS fabricated through standard soft-lithography. With the proposed microchip, target cells (lung cancer cells, A549)
DEPIM chip
were guided toward the center of the working region due to the action of positive DEP and hydrodynamic drag
CTCs enrichment and detection
forces and were then trapped onto the desired sensing electrodes. Impedance was measured to identify the
presence of cells. Impedance spectroscopy was conducted at different numbers of cells. Linear characteristics
were found using differential analysis technique. Experimental results showed that the impedance sensor can
detect A549 cell line with low cell number at the appropriate frequencies. Limit of detection (LOD) of ap-
proximately 3 cells was achieved at frequencies near 50 kHz. This simple, rapid, label-free, and low-cost ap-
proach may open up new opportunities for developing cell diagnosis systems for future applications.

1. Introduction 2014; Rahman et al., 2017; Shim et al., 2013). A conventional DEP
device has many advantages, including simplicity of device fabrication
Manipulation, isolation, and enrichment of biological cells in mi- and integration, low cost, high robustness and throughput, and good
crofluidic devices are essential for accurate detection of specific cell compatibility with other systems (Çetin and Li, 2011; Rahman et al.,
types, especially rare cancerous cells (Y. Chen et al., 2014; Gossett 2017). Many DEP-based cancer cell separation microfluidic devices
et al., 2010; Yousuff et al., 2017). Rare cells, such as circulating tumor have been designed, tested and achieved the high target cell recovery
cells (CTCs), circulating fetal cells, and stem cells, are those with low rates. Microchips that provide non-uniform electric fields for DEP ma-
abundance of less than 1000 cells per milliliter of the sample nipulation to concentrate cancer cells onto microelectrodes have been
(Dharmasiri et al., 2010). During carcinogenesis, CTCs are generated proposed in many studies (G. H. Chen et al., 2014; Huang et al., 2013;
when the cancer cells detach from the primary tumor into the blood Jen et al., 2012, 2011; Jen and Chang, 2011; Jen and Chen, 2009; Wang
flow; this process can start the early stages of metastasis (Jacob et al., et al., 2017). DEP applications were performed by a low actuation
2007). Isolation and separation of CTCs can enrich cellular concentra- voltage in the range from 5 to 16 Vpp (peak-to-peak) at a frequency of
tions to a measurable quantity, thereby enhancing treatment accuracy 1 MHz. Numerical simulations were also conducted to demonstrate the
or efficacy and providing a minimally invasive method to assess the capability of the design. The positive DEP responses of cells were re-
cancer status of patients (Fan et al., 2015). Based on one or several sponsible for their transport to the center of the experimental region.
distinguishing properties of cell lines such as size, shape, deformability, Various techniques, such as fluorescence, mass spectrometry, elec-
morphology, magnetic properties, electrical properties, and surface tromagnetic spectroscopy, electroosmosis, and electrical impedance
markers, cancer cells can be separated from normal blood cells by mi- spectroscopy (EIS), have been used to detect objects in microfluidic
crofluidic techniques (Chen et al., 2012). A typical technique in min- devices (Daniels and Pourmand, 2007; Hajba and Guttman, 2014; Liu
iaturized microfluidic platforms is dielectrophoresis (DEP), which is a et al., 2016; Spencer et al., 2014; Xu et al., 2016). Among these
useful tool that is extensively developed in a wide range of biological methods, label-free impedance measurement has emerged as a non-in-
particles, such as cells, bacteria, viruses, protein, and DNA (Çetin and vasive approach due to its simple fabrication and measurement setup
Li, 2011; Fernandez et al., 2017; Khoshmanesh et al., 2011; Li et al., and minimization capability. Impedance is not only suitable for

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

https://doi.org/10.1016/j.bios.2018.08.059
Received 14 June 2018; Received in revised form 24 August 2018; Accepted 24 August 2018
Available online 03 September 2018
0956-5663/ © 2018 Elsevier B.V. All rights reserved.
N.-V. Nguyen, C.-P. Jen Biosensors and Bioelectronics 121 (2018) 10–18

analyzing the electrical properties of target cells in contact with elec-

trodes, it can also be monitored to quantify the number of cells in so-
lution (Fernandez et al., 2017). The combination of impedance detec-
tion and DEP enrichment termed DEP-based Impedance Measurement
(DEPIM) method (Hamada et al., 2013; Suehiro et al., 1999) is im-
portant in enhancing the capture efficiency of cells and detecting sen-
sitivity at the electrodes. After DEP application, pearl chains caused
electrical connection of cells parallel to the electrode gap, thereby al-
tering the conductance and the capacitance between the electrodes.
Assuming that the trapped cell possesses higher admittance than the
media, the increase rate of conductance in the sensing electrodes was,
therefore, faster and better compared to that obtained without the DEP
concentration. Besides, the total admittance of the microelectrode in-
creases with increasing number of trapped cells (Wang et al., 2017). In
biological cell identification and analysis, impedance spectroscopy is a
powerful tool for the label-free analysis and characterization of cells
and can provide information on membrane capacitance and resistance
and cytoplasm conductivity and permittivity (Xu et al., 2016). The in-
terdigitated electrode configurations were used to apply DEPIM method
for sensitive identification of a bacteria or cell type. However, isolation,
concentration, and detection of rare cancerous cells from peripheral
blood by DEPIM could be revealed using circular microelectrodes in our
present work.
In this study, a rare tumor cell enrichment and detection system that
uses the focus and sensing circular microelectrodes is presented. CTCs
are collected and consequently focused from outside electrodes into the Fig. 1. Structure and working principle of the microfluidic device for the ma-
central sensing region by DEP manipulation. The accuracy of detection, nipulation and detection of target cells (a) Sketching of the designed microchip.
as well as cell identification, is thus improved. Cells are concentrated on (b) Combination of cells sample injection and DEP application to trap almost
the sensing electrodes, result in changing the electrical impedance on cells at the outermost circular electrodes (towards the left side of the sensing
them. This work develops a differential design using double pairs of electrodes). Target cells that are represented by the yellow dots (like as A549).
symmetrical sensing microelectrodes to increase the sensitivity of this Non-target cells that are represented by the green dots (like as RBC). (c) Target
system. The structure can subtract out the common-mode changes cells are collected into the center of the working chamber due to DEP effect and
caused by the electrode properties or surrounding environmental fac- by size-based difference, and trapped onto the pair of the left sensing elec-
tors. Experiments were performed on A549 human lung circulating trodes. Then, the impedance measurement is determined to identify the pre-
sence of the target cells.
tumor cells, which are among the most common causes of cancer death
worldwide (Wu et al., 2018). A549 cell line is a typical early-stage type
in cancer research. Hence, sensitive detection of A549 cells is a po- molecules such as antibody and aptamer in order to catch the desired
tential requirement for early diagnosis and treatment of lung cancer. target cells (Do et al., 2018). The chip consists of microchannel made of
The single type of cell sample was used in order to enhance the cell polydimethylsiloxane (PDMS) and was placed on the gold electrodes
enrichment factor, and reliably calculate the limit of detection for the area of the substrate. The width and the height of the main channel are
target cells. The performance of the impedance sensor was investigated 2 mm and 25 µm, respectively.
with different cell numbers. A highly linear detection with a low de- The designed operation of the chip is illustrated in Fig. 1(b) and (c).
tection limit of fewer than 10 cells was observed at several frequency This device enables enrichment and subsequent detection of rare tumor
ranges. The combination of impedance spectra measurement and dif- cells in the electrodes region. First, a cell sample is pumped into the
ferential analysis technique was proven as a powerful method for de- channel with a low-speed flow. A DEP excitation that would possibly
tection of cancerous cells at low cell concentration. generate a high-electric-field region is applied at the pair of the outmost
electrodes. All cells are expected to be caught between the two biggest
2. Materials and methods circular electrodes by a combination of positive DEP and hydrodynamic
drag forces. The electric field is subsequently switched to the adjacent
2.1. Design of the microfluidics chip with embedded microelectrodes electrode pairs from the outermost to innermost in the working area
(Jen et al., 2012). An inward stepping electric field is generated, re-
The design and operational procedure of target cell detection and sulting in the movement of cells into the central electrode due to DEP
manipulation within a microchip device is illustrated in Fig. 1. DEPIM response. This methodology also allows for the separation between
was applied in this device. Fig. 1(a) shows the schematic diagram of the normal and CTC cells. In the specified sucrose medium, both CTCs and
proposed device structure, including four main parts: a glass substrate, RBCs suspended exhibit a positive DEP effect at a stimulus signal of
circular electrodes, an insulations layer, and a simple straight micro- high frequency (Huang et al., 2013). However, the corresponding
fluidic channel. Eighteen gold-on-chromium electrodes were included magnitude of DEP force for CTCs is much greater than that on RBCs
in the design. Both the electrode width and the gap between two ad- under the same electric field distribution (G. H. Chen et al., 2014). In
jacent electrodes are 30 µm. The outside circles (blue electrodes) called principle, most CTCs could be separated from normal blood cells and
manipulating electrodes were used for DEP-based cell concentration. collected at the sensing electrodes by our design. Thereafter, the im-
Two pairs of symmetrical electrodes at the central region (red elec- pedance of the sensing electrodes was measured to identify the ap-
trodes) called sensing electrodes were used for pulling, capturing, and pearance of cancerous cells. Analyzing the impedance values with dif-
sensing target cells. An insulation layer was coated on the manipulating ferential measurement technique enables the chip to effectively
electrodes to avoid cell adhesion and consequently enhance the DEP recognize the target cell line.
manipulation performance of the chip. In future research, sensing
electrode surfaces can be functionalized by affinity and selectively

11
N.-V. Nguyen, C.-P. Jen Biosensors and Bioelectronics 121 (2018) 10–18

Fig. 2. Chip fabrication: (a) Process of the fabrication. The manipulating electrodes are coated by an insulation layer to avoid the cell adhesion, thus improving the
performance of DEP concentration. (b) A fabricated microfluidic device, composed of a glass substrate, gold-on-chromium circular electrodes, an insulation layer, and
a straight channel.

2.2. Chip fabrication 2.3. Sample preparation

Chip fabrication is shown in Fig. 2(a) using the standard photo- A549 lung CTC cells were cultured for experiment by using the
lithography techniques. The detail of the processes has been described proposed microdevice. Cells were serially passaged as monolayer cul-
in Supplementary material. The substrate slices were cleaned by im- tures in DMEM, supplemented with 3.7 g of NaHCO3 per liter of
mersing in piranha solution (96% H2SO4: 30% H2O2 with a volume medium, 10% fetal bovine serum, and 1% penicillin/streptomycin. Cell
ratio of 3:1) for 30 min, rinsed with deionized water, and dried with culture dishes were incubated in a humidified atmosphere containing
nitrogen following the photolithography process. Gold-on-chromium 5% CO2 and 95% O2 at 37 °C. The medium was replaced every 1–2 days.
microelectrodes and the insulating layer were generated on the glass For the application of DEP, cells were centrifuged and resuspended in
substrate using S1813 positive photoresist. While PDMS channel was an 8.62 wt% sucrose buffer with a measured conductivity of
fabricated by SU-8 2050 cast molding. The PDMS prepolymer mixture 1.76 × 10−2 S/m. Prior to the experiment, cell concentration and
was made, poured and cured on the mold master to replicate the pat- viability were assessed by trypan blue dye exclusion using a hemocyt-
terned structure. After the PDMS replica had been peeled off, inlet and ometer with two counting grids. In some experiments, tumor cells were
outlet fluidic connections to the channel were created by a puncher, stained using a standard fluorescence assay with calcein AM before
and the replica was bonded with glass substrate via treatment of oxygen pumping into the working channel. Because viable tumor cells were
plasma. Fig. 2(b) shows an image of the fabricated microchip. brightly fluorescent, the viability of tumor cells was verified, and cell
ratios, as well as concentration efficiencies, were also counted. The
previous reports demonstrated that all the cells still recovered their

12
N.-V. Nguyen, C.-P. Jen Biosensors and Bioelectronics 121 (2018) 10–18

viability after DEP concentrating in our microchip. Because the applied inability to move further into the center. This finding also reveals that
electric field strength (approximate 5 × 105 V/m) was lower than the the cells would be firmly attached to the outer electrode circles if no
threshold strength of cell lysis to maintain the viability of cells. insulation layer is present. The efficiencies of cellular concentration
were evaluated by calculating the fluorescence intensity of the whole
2.4. Experimental instruments electrodes region at the beginning and at the sensing electrodes after
finishing the DEP concentration. In this study, just the single type of cell
A conventional system used in experiments is schematically shown sample was used in order to enhance the cell enrichment factor, as well
in Supplementary material. The buffer solution with each fixed volume as conveniently calculate the limit of detection for the target cells.
of the cell sample was injected into the channel of the chip using a Therefore, the enrichment rate on the sensing electrodes was accounted
syringe pump. Cell distribution in the channel was monitored and re- for over 90% after the process of applying the stepping electric field
corded using an inverted fluorescence microscope with a mounted CCD from the outermost to central electrodes. The cell trapping time onto
camera connected to a computer running Olympus DP Controller image the outermost electrodes was lower than 5 min. When the stepping
software. Suitable excitation signals for driving DEP were provided by a electric field was applied, the potential was held for about 20 s for each
function generator or a handheld electric module (Jen et al., 2012). The pair before switching to the next adjacent pair of electrodes. Following
manipulating electrodes were attached to the output signal of the impedance measurement was conducted from 1 to 2 min. Therefore, the
generator through a switchboard. The stepping electric fields for DEP total duration of electrical operation on the chip was approximately
manipulation of cells were generated by operating on the switch board. 10 min.
For the impedance measurement, each pair of sensing microelectrodes
was respectively connected to an impedance analyzer via BNC cables. 3.2. Cell trapping and electrical impedance spectroscopy measurement
Impedance spectroscopy can be performed over a wide range of both
amplitude and frequency by using this equipment. The measured sig- Following the steps of DEP-based cell collection, target cells were
nals were collected and transferred to a computer via a digital interface. trapped onto the sensing electrodes. A defined volume of cell sample
The data were then processed using LabVIEW software to calculate the was pumped into the channel during the experiment, which was re-
impedance parameters of the cell sample. peated at cell sample volumes ranging from 0 to 10 µL with an incre-
ment of 2 µL. At the cell concentration of 5 × 104 cells/mL, the cell
3. Results and discussion number injected into the channel could be estimated from 0 to ap-
proximately 50 cells. Fluid flow was stopped during impedance mea-
3.1. DEP-based cell concentration surements. Two pairs of symmetric sensing electrodes are in the center
of the structure. This design aimed to trap cells into a pair of sensing
Experimental results for concentrating CTCs from blood samples to electrodes (facing to the left) called the cell-trapping electrodes. The
the center electrode were published in our previous work (Huang et al., other pair (facing to the right) that was not used to trap cells is called
2013). The concentrations of RBCs and cancer cells in the micro- un-trapping electrodes. The results showed that A549 cells were
chamber were 3.25 × 106 cells/mL and 2.5 × 105 cells/mL, respec- trapped in the cell-trapping electrodes (upstream), whereas the un-
tively, with a RBCs to CTCs ratio of 13. The applied voltage and fre- trapping electrodes were all kept empty (Fig. S2 in Supplementary
quency were 16 Vpp and 1 MHz, respectively. The time interval of relay material).
switching was 20 s. The experimental results indicated that cancer cells For impedance measurement, the wires connected to the sensing
were selectively concentrated at the center pair of microelectrodes from microelectrodes were attached to the probes of the impedance analyzer.
RBCs with a recovery rate of around 80% (Jen et al., 2012). Experi- A sinusoidal alternating potential of 500 mV peak-to-peak was applied
ments on samples with low cancer cell density of 104 cells/mL (in which for all impedance measurements in the same sucrose buffer medium.
the ratio of RBCs to CTCs is 325) were also performed (Huang et al., Electrical impedance spectroscopy of cells is a common theoretical
2013). The experiment revealed that A549 CTCs can also be selectively model on impedance measurement that could be considered as a
concentrated onto the central microelectrode and isolated from the steady-state analysis technique. Electrical impedance spectroscopy was
blood sample even with only a single cell in the manipulation area. conducted at 60 points per decade within 1 kHz to 1 MHz. Fig. 4 shows
In this work, 5 × 104 cells/mL of A549 cell sample was used in the characteristic impedance responses for each pair of central elec-
experiments to identify the main working principles of the device. For trodes. Data are given in the form of amplitude Z (Fig. 4a), phase angle
DEP manipulation, a sinusoidal excitation voltage with a peak-to-peak θ (Fig. 4b), and Nyquist plots (Fig. 4c) at various cell numbers for A549
magnitude of 10 V at a frequency of 1 MHz was applied to the elec- cells. Both impedance magnitude and phase decreased when the fre-
trodes. A549 cells suspended in a sucrose medium were located within quency increased at each level of the cell number. However, significant
the pDEP range. These cells were frequently attracted onto the edge and differences among the different numbers of cells could be observed for
gap between the pairs of the microelectrodes (G. H. Chen et al., 2014; the cell-trapping electrodes. By contrast, these changes were minimal
Huang et al., 2013; Jen et al., 2012; Wang et al., 2017). Fig. 3 shows the for the un-trapping electrodes. According to the investigation of the
microscopy images of a cell sample in the DEP concentration. First, impedance amplitude of the cell-trapping electrodes, there were three
fluid flow was controlled from left to right by syringe pumps with a flow specific zones in the spectra. With an increase in the number of trapped
rate of 1.2 µL/min. The channel was washed with sucrose buffer for cells, the impedance increased rapidly in the low frequency range
30 min before conducting the experiment to block hydrophobic inter- (1–10 kHz), while decreased in the intermediate frequencies
actions between biological samples and PDMS surface. The cell samples (10–100 kHz), and altered not much in the high frequency range
were pumped into the channel of the chip at different volumes. The (100 kHz to 1 MHz). For the un-trapping electrodes, the impedance
electric field at the peripheral was turned on to capture the cells. All difference appeared less at the low frequency range.
cells pumped into the channel were trapped onto the edges between The phase angle describes the relative contributions of the real and
two outermost electrodes. The cells were subsequently concentrated imaginary parts to the total impedance. A phase angle of 0° results from
from the outermost electrode pair toward the sensing electrodes in the a purely resistive structure, and a phase angle of − 90° is the product of
working region by the pDEP effect. Meanwhile, DEP buffer solution was a system that is purely capacitance. For our measurements, the phase
continuously pumped into the channel with a low speed of 0.2 µL/min. angle was close to − 10° at low frequency range of the impedance
Finally, cells were attracted to the central electrodes, located on one spectrum, whereas it approached − 45° to − 90° in the high frequency
side of the sensing electrodes group. It can be seen, there were few cells range. Therefore, the capacitive element of impedance is dominant at
that still strongly attached to the outermost electrodes, leading to their the high frequency range, whereas the resistance dominates the low

13
N.-V. Nguyen, C.-P. Jen Biosensors and Bioelectronics 121 (2018) 10–18

Fig. 3. Example showing the DEP-based ma-

nipulation process of A549 cells. (a) The ori-
ginal chip without injected cells. (b) Fluidic
flow direction was from left to right, with the
flow speed of 1.2 µL/min. Cells were trapped at
the outermost electrodes by pDEP at the fre-
quency of 1 MHz and the potential of 10 Vpp.
(c) Target cells were isolated using DEP effect
and the stepping electric fields from outermost
to innermost electrodes. (d) Focused cells were
collected at the left sensing electrodes for im-
pedance measurement.

frequency range where high magnitude change can be seen. Impedance ∆Z T = Z0T (f ) − ZCN
T
(f ) , (1)
change is up to hundreds of kilo-ohms in the frequency range of
1–100 kHz, where the angle phase is in the range of − 10° to − 30°. ∆θT = θ0T (f ) − θCN
T
(f ), (2)
For the situation of the electrodes in contact with the electrolyte
where the indicators 0 and CN are the representative of cases without
solution, the Randles equivalent circuit model could be applied to
and with cell numbers, respectively, at each individual frequency (f).
analyze the experimental data of EIS (Lisdat and Schäfer, 2008). The
Similarly, the impedance and phase angle difference spectra be-
model is comprised of the solution resistance Rs, the surface modifica-
tween the two pairs of cell-trapping and un-trapping electrodes, called
tion capacitance Cs, the Warburg impedance ZW, and the charge transfer
the differential impedance (ΔZD), and the differential phase angle (ΔθD)
resistance Rct. The values for Rs, Cs, and Rct can be easily determined
given in Fig. 5(b) are calculated using:
with the Nyquist plots shown in Fig. 4(c). Rs corresponds to the
U T
minimum value of the impedance real part. Rct corresponds to the ∆ZD = ZCN (f ) − ZCN (f ) (3)
semicircle diameter of the Nyquist diagram. Cs can be calculated from
U T
the frequency at the maximum of the semicircle (2πf = 1/RctCs). The ∆θ D = θCN (f ) − θCN (f ) (4)
Warburg impedance results from the impedance of the current due to In this second analysis method, the pair of cell-trapping electrodes is
diffusion from the bulk solution to the interface. In this work, the values used as the working group, while the pair of un-trapping electrodes is
of Rs, Rct and Cs were respectively calculated approximate 0, 500 kΩ, considered as the reference group (Daniels and Pourmand, 2007).
and 6 pF, when the channel was filled full in fresh sucrose medium Ideally, no cell is trapped (both the left and right sensing electrodes are
without cells. For the cell-trapping electrodes, the Nyquist responses empty, or CN = 0), and only the buffer medium is sensed. The im-
changed gradually with the increasing amount of captured A549 cells pedance across the cell-trapping electrodes was the same as that across
from 0 to 50 cells. Meanwhile, the EIS representations of the un-trap- the un-trapping electrodes. The magnitude and phase angle differences
ping electrodes expressed negligible differences. These variations in the were almost zero, resulting in straight differential spectrums across the
impedance properties demonstrated that the cells were trapped at the whole frequency. A trend of impedance alterations was observed upon
desired sensing electrodes. increasing the volume of the injected cell sample. When the cells at-
tached to the sensing electrodes, the impedance difference should be
non-zero. The impedance spectra lines were similar to each other at
3.3. Differential impedance spectra different low cell concentrations. The spectra amplitudes were ex-
panded as cell number increases in both analytical methods. The ten-
The impedance and phase are then converted to the parameters dency of impedance responses is positive in the frequency range of
termed the differences in all the investigated frequency range. Fig. 5 10–100 kHz, in which the impedance difference increased for addi-
shows the differential impedance spectra in two analytical methods at tional trapped cells.
the various volume of the injected cell sample. T and U are expressed
for the pairs of cell-trapping electrodes and un-trapping electrodes,
3.4. Detection sensitivity
respectively. The impedance change within the pair of cell-trapping
electrodes, in terms of the amplitude (ΔZT) and phase angle (∆θT )
Fig. 6 shows the number of trapped cells and the relational curves-
shown in Fig. 5(a) are defined by:
fitting the impedance variation versus the volume of the injected cell

14
N.-V. Nguyen, C.-P. Jen Biosensors and Bioelectronics 121 (2018) 10–18

(a)

(b)

(c)
Fig. 4. Electrical impedance spectroscopy responses for A549 cells concentrated into the sensing region of the microchip by using impedance analyzer in sucrose
buffer solution, with an applied potential of 250 mV, and a frequency range from 1 kHz to 1 MHz. (a) Amplitude Z, (b) Phase angle θ, and (c) Relationship between
real Z′ and imaginary Z′′ parts of total electrical impedance in comparison between two pairs of central electrodes (cell-trapping and un-trapping electrodes); at
different numbers of trapped cells: (C0X) 0, (C1X) 10, (C2X) 20, (C3X) 30, (C4X) 40, and (C5X) 50 cells, respectively.

15
N.-V. Nguyen, C.-P. Jen Biosensors and Bioelectronics 121 (2018) 10–18

Fig. 5. Differential impedance spectra for different trapped cell numbers. (a) Impedance change (ΔZT and ΔθT) of the cell-trapping electrodes, (b) Differential
impedance (ΔZD and ΔθD) between the cell-trapping electrodes and un-trapping electrodes.

sample. The linear regression equations are expressed in the graphs. At standard deviation and slope is found from the linear response range.
each same level of cell number, the impedance value at high frequency Herein, a LOD of approximately 3 cells was achieved frequencies near
changed more than the value at low frequency. The divergences in 50 kHz. The obtained results indicated the feasibility of using DEP-
impedance amplitude (both ΔZT and ΔZD) among the five frequencies based enrichment and impedance measurements to detect human lung
increased with increasing cell number. This finding reveals the linear CTCs.
relationships at some frequencies, in which all phase differences at
different cell numbers were approximately zero (Fig. 5).
4. Conclusions
Experiments were repeated three times for each number of cells, and
the error bar was shown to depict the standard error of the mean. The
In this study, DEPIM was successfully developed for the rapid,
results show that highly linear relationships could be achieved with the
highly sensitive, and label-free detection of A549 human lung circu-
correlation coefficient (R2) up to over 99% for both methods of calcu-
lating tumor cells by using a microfluidic channel with circular elec-
lation. In addition, the sufficient frequency ranges were proposed for
trodes. DEP manipulation was conducted to concentrate A549 cells in
the detection of A549 cells based on impedance measurement. The
the central region with the cell enrichment rate accounted for over
range from 10 kHz to 12.5 kHz is suggested for the method using the
90%. Such cells are captured by cell-trapping electrodes and then
impedance change of the cell-trapping electrodes, whereas 45–55 kHz is
identified by impedance responses. The total duration of electrical op-
suggested for the differential impedance analysis between the cell-
eration on the chip was approximately 10 min. Electrical impedance
trapping and un-trapping electrodes. It also shows clearly that the
spectra of the sensing electrodes were recorded in the frequency range
sensitivity of the differential analysis method is higher than that of the
from 1 kHz to 1 MHz. Two differential analysis methods for detection of
first method. With the same change in the number of cells, both the
cells were applied. Linear relationships were found between impedance
impedance variation and the appropriate frequency range of the second
variation and cell number with the correlation coefficient up to 99%. A
method are larger than those of the first one. In this way, the sensitivity
higher detection accuracy of the sensor was observed at the frequency
of the system could be improved using a four-electrode configuration
of 50 kHz using differential calculations between the cell-trapping
instead of the commonly used two-electrode configuration because the
electrodes and the un-trapping electrodes. These results also indicated
effect of the electrode–cell interface impedance on impedance mea-
that the impedance sensor was capable of detecting A549 cell line with
surements could be reduced (Sarró et al., 2012). Limit of detection
low cell number. The proposed DEPIM microchip could be a potential
(LOD) could be calculated from the formula 3σ/slope, where σ is the
approach to recognize cancer cell types. Verifying the isolation

16
N.-V. Nguyen, C.-P. Jen Biosensors and Bioelectronics 121 (2018) 10–18

Acknowledgements

The authors would like to acknowledge the Ministry of Science and

Technology of the Republic of China (Taiwan) under the Grant
MOST105-2923-E-194-002-MY3.

Appendix A. Supplementary material

Supplementary data associated with this article can be found in the

online version at doi:10.1016/j.bios.2018.08.059.

References

Çetin, B., Li, D., 2011. Dielectrophoresis in microfluidics technology. Electrophoresis 32,
2410–2427.
Chen, G.H., Huang, C. Te, Wu, H.H., Zamay, T.N., Zamay, A.S., Jen, C.P., 2014. Isolating
and concentrating rare cancerous cells in large sample volumes of blood by using
dielectrophoresis and stepping electric fields. Biochip J. 8, 67–74.
Chen, J., Li, J., Sun, Y., 2012. Microfluidic approaches for cancer cell detection, char-
acterization, and separation. Lab Chip 12, 1753–1767.
Chen, Y., Li, P., Huang, P.-H., Xie, Y., Mai, J.D., Wang, L., Nguyen, N.-T., Huang, T.J.,
2014. Rare cell isolation and analysis in microfluidics. Lab Chip 14, 626–645.
Daniels, J.S., Pourmand, N., 2007. Label-free impedance biosensors: opportunities and
challenges. Electroanalysis 19, 1239–1257.
Dharmasiri, U., Witek, M.A., Adams, A.A., Soper, S.A., 2010. Microsystems for the capture
of low-abundance cells. Annu. Rev. Anal. Chem. 3, 409–431.
Do, L.Q., Thuy, H.T.T., Bui, T.T., Dau, V.T., Nguyen, N.-V., Duc, T.C., Jen, C.-P., 2018.
Dielectrophoresis microfluidic enrichment platform with built-in capacitive sensor
for rare tumor cell detection. BioChip J. 12 (2), 114–122.
Fan, X., Jia, C., Yang, J., Li, G., Mao, H., Jin, Q., Zhao, J., 2015. A microfluidic chip
integrated with a high-density PDMS-based microfiltration membrane for rapid iso-
lation and detection of circulating tumor cells. Biosens. Bioelectron. 71, 380–386.
Fernandez, R.E., Rohani, A., Farmehini, V., Swami, N.S., 2017. Review: microbial analysis
in dielectrophoretic microfluidic systems. Anal. Chim. Acta 966, 11–33.
Gossett, D.R., Weaver, W.M., MacH, A.J., Hur, S.C., Tse, H.T.K., Lee, W., Amini, H., Di
Carlo, D., 2010. Label-free cell separation and sorting in microfluidic systems. Anal.
Bioanal. Chem. 397, 3249–3267.
Hajba, L., Guttman, A., 2014. Circulating tumor-cell detection and capture using micro-
fluidic devices. TrAC–Trends Anal. Chem. 59, 9–16.
Hamada, R., Takayama, H., Shonishi, Y., Mao, L., Nakano, M., Suehiro, J., 2013. A rapid
bacteria detection technique utilizing impedance measurement combined with po-
sitive and negative dielectrophoresis. Sens. Actuators B: Chem. 181, 439–445.
Huang, C.T., Amstislavskaya, T.G., Chen, G.H., Chang, H.H., Chen, Y.H., Jen, C.P., 2013.
Selectively concentrating cervical carcinoma cells from red blood cells utilizing
dielectrophoresis with circular ITO electrodes in stepping electric fields. J. Med. Biol.
Eng. 33, 51–58.
Jacob, K., Sollier, C., Jabado, N., 2007. Circulating tumor cells: detection, molecular
profiling and future prospects. Expert Rev. Proteom. 4, 741–756.
Jen, C.P., Chang, H.H., 2011. A handheld preconcentrator for the rapid collection of
cancerous cells using dielectrophoresis generated by circular microelectrodes in
stepping electric fields. Biomicrofluidics 5, 034101.
Jen, C.P., Chang, H.H., Huang, C.T., Chen, K.H., 2012. A microfabricated module for
isolating cervical carcinoma cells from peripheral blood utilizing dielectrophoresis in
stepping electric fields. Microsyst. Technol. 18, 1887–1896.
Jen, C.P., Chen, T.W., 2009. Selective trapping of live and dead mammalian cells using
insulator-based dielectrophoresis within open-top microstructures. Biomed.
Microdevices 11, 597–607.
Jen, C.P., Huang, C.T., Chang, H.H., 2011. A cellular preconcentrator utilizing dielec-
trophoresis generated by curvy electrodes in stepping electric fields. Microelectron.
Eng. 88, 1764–1767.
Khoshmanesh, K., Nahavandi, S., Baratchi, S., Mitchell, A., Kalantar-zadeh, K., 2011.
Dielectrophoretic platforms for bio-microfluidic systems. Biosens. Bioelectron. 26,
1800–1814.
Li, M., Li, W.H., Zhang, J., Alici, G., Wen, W., 2014. A review of microfabrication tech-
Fig. 6. High linear relationships between impedance magnitude variation and niques and dielectrophoretic microdevices for particle manipulation and separation.
number of injected A549 cells: (a) Impedance change (ΔZT) of the cell-trapping J. Phys. D: Appl. Phys. 47, 063001.
Lisdat, F., Schäfer, D., 2008. The use of electrochemical impedance spectroscopy for
electrodes at the frequency range from 8.9 to 14.1 kHz, (b) Differential im-
biosensing. Anal. Bioanal. Chem. 391, 1555–1567.
pedance (ΔZD) between the cell-trapping electrodes and un-trapping electrodes Liu, L., Wu, S., Jing, F., Zhou, H., Jia, C., Li, G., Cong, H., Jin, Q., Zhao, J., 2016. Bead-
at the frequency range from 39.8 to 63.1 kHz. The coefficient of linear regres- based microarray immunoassay for lung cancer biomarkers using quantum dots as
sion R2 is over 0.99 at several frequencies in both cases. labels. Biosens. Bioelectron. 80, 300–306.
Rahman, N.A., Ibrahim, F., Yafouz, B., 2017. Dielectrophoresis for biomedical sciences
applications: a review. Sensors 17, 449.
efficiency cancerous cells from the peripheral blood sample, testing the Sarró, E., Lecina, M., Fontova, A., Solà, C., Gòdia, F., Cairó, J.J., Bragós, R., 2012.
Electrical impedance spectroscopy measurements using a four-electrode configura-
biological issues, developing the handled electronics module for dif-
tion improve on-line monitoring of cell concentration in adherent animal cell cul-
ferential impedance measurement and improving the detection sensi- tures. Biosens. Bioelectron. 31, 257–263.
tivity and specificity of the biosensor could be investigated in the future Shim, S., Stemke-Hale, K., Noshari, J., Becker, F.F., Gascoyne, P.R.C., 2013.
Dielectrophoresis has broad applicability to marker-free isolation of tumor cells from
work.
blood by microfluidic systems. Biomicrofluidics 7, 011808.
Spencer, D., Hollis, V., Morgan, H., Spencer, D., Hollis, V., Morgan, H., 2014. Microfluidic
impedance cytometry of tumour cells in blood. Biomicrofluidics 8, 064124.

17
N.-V. Nguyen, C.-P. Jen Biosensors and Bioelectronics 121 (2018) 10–18

Suehiro, J., Yatsunami, R., Hamada, R., Hara, M., 1999. Quantitative estimation of bio- J., Tang, Y., 2018. Ultrasensitive and high specific detection of non-small-cell lung
logical cell concentration suspended in aqueous medium by using dielectrophoretic cancer cells in human serum and clinical pleural effusion by aptamer-based fluor-
impedance measurement method. J. Phys. D: Appl. Phys. 32, 2814–2820. escence spectroscopy. Talanta 179, 501–506.
Wang, H.C., Nguyen, N.V., Lin, R.Y., Jen, C.P., 2017. Characterizing esophageal can- Xu, Y., Xie, X., Duan, Y., Wang, L., Cheng, Z., Cheng, J., 2016. A review of impedance
cerous cells at different stages using the dielectrophoretic impedance measurement measurements of whole cells. Biosens. Bioelectron. 77, 824–836.
method in a microchip. Sensors 17, 1053. Yousuff, C., Ho, E., Hussain, K., I., Hamid, N., 2017. Microfluidic platform for cell iso-
Wu, Y., Zhang, H., Xiang, J., Mao, Z., Shen, G., Yang, F., Liu, Y., Wang, W., Du, N., Zhang, lation and manipulation based on cell properties. Micromachines 8, 15.

18