Multi-parameter analysis using photovoltaic cell-based optofluidic cytometer.

A multi-parameter optofluidic cytometer based on two low-cost commercial photovoltaic cells and an avalanche photodetector is proposed. The optofluidic cytometer is fabricated on a polydimethylsiloxane (PDMS) substrate and is capable of detecting side scattered (SSC), extinction (EXT) and fluorescence (FL) signals simultaneously using a free-space light transmission technique without the need for on-chip optical waveguides. The feasibility of the proposed device is demonstrated by detecting fluorescent-labeled polystyrene beads with sizes of 3 μm, 5 μm and 10 μm, respectively, and label-free beads with a size of 7.26 μm. The detection experiments are performed using both single-bead population samples and mixed-bead population samples. The detection results obtained using the SSC/EXT, EXT/FL and SSC/FL signals are compared with those obtained using a commercial flow cytometer. It is shown that the optofluidic cytometer achieves a high detection accuracy for both single-bead population samples and mixed-bead population samples. Consequently, the proposed device provides a versatile, straightforward and low-cost solution for a wide variety of point-of-care (PoC) cytometry applications.


Introduction
Flow cytometry is a well-established technique for counting and analyzing single cells or particles. In general, flow cytometers utilize three different types of light for, namely forward scattered light, side scattered light and fluorescent light. Typically, forward scatter measurement is used to determine the cell size, side scatter measurement is used to evaluate the internal properties and structures of the cell, and fluorescence measurement is used to detect fluorescence tag or biomarker, and to count the cells as they pass through the detection zone [1]. However, traditional cytometers are physically bulky and expensive, and hence they are unsuitable for portable or in situ applications. Advances in micro-electro-mechanical systems (MEMS) technology now make possible the realization of microfluidic analysis systems with a low power consumption, a low cost, a small physical size, a small sample and reagent volume, a high sensitivity, and a straightforward operation. Many microfluidic platforms have been presented for point-of-care (PoC) diagnosis [2,3], pathogen sensing [4], and cell biophysical characterization [5]. In addition, many methods for performing the enhanced detection and analysis of single cells have been proposed, including optical image processing [6][7][8][9], and electrical and optical techniques [10][11][12]. Consequently, the potential for integrating microfluidic platforms with enhanced detection techniques to realize microflow cytometers capable of performing the multi-function analysis of single cells has attracted great interest in recent years.
The literature contains many versatile optical imaging techniques based on complementary metal oxide semiconductors (CMOS) and charged-coupled devices (CCDs) for PoC applications [7][8][9]. For example, Seo et al. [7] proposed a lens-free on-chip cytometer for the detection of red blood cells, yeast cells, E. coli and micro-particles, in which holographic images of the cells or particles were captured using a high-resolution CCD and CMOS sensor array and were then matched with a library of existing images for identification purposes. Zhu et al. [9] presented a cell-phone based optofluidic imaging platform capable of detecting red and green fluorescent-labeled particles over a field-of-view of 81 mm 2 with a raw spatial resolution of ~20 µm. The feasibility of the proposed platform was demonstrated by detecting labeled white blood cells, Giardia Lamblia cysts and micro-particles.
Many microfluidic-based cytometers for single cell analysis have been proposed, including acoustic wave microflow cytometers [13] enhanced the detection sensitivity by using a metal-oxide-semiconductor-field-effecttransistor (MOSFET) as an amplifier. Wu et al.
[16] presented a microfluidic resistive pulse sensor (RPS) based on a symmetric mirror channel design and a sensing aperture with a size of 50x16x20 µm 3 . It was shown that the mirror channels improved the signal-to-noise ratio of the detection signal and resulted in a record low volume ratio of the particle to sensing channel of 0.0004%. In a later study [17], the same group proposed an integrated lab-on-chip (LoC) device capable of performing the simultaneous detection and counting of labeled and label-free particles by means of an RPS sensor and a fluorescence detection technique, respectively. Moreover, Zhe et al.
[18] presented a multichannel RPS platform for the highthroughput detection and differentiation of micro-scale particles. The feasibility of the proposed device was demonstrated using both natural and synthetic micro-particles with diameters ranging from 20~40 µm.
Kim et al. [19] proposed an impedance-based microfluidic chip incorporating two polyelectrolytic gel electrodes (PGEs) for detecting the number and size of red blood cells (RBCs) in diluted whole blood by monitoring the number and amplitude of the peaks in the impedance signal between the PGEs. The detection results obtained for over 800-fold diluted samples were shown to be consistent with those obtained using a commercial human hematoanalyzer. Guo et al. [20] presented an alternating current (AC) impedance-based microflow cytometer built on a printed circuit board (PCB) and sealed with polydimethylsiloxane (PDMS) thin film. Similarly, Shi et al.
[21] developed a differential amplifier-based microflow cytometer on an SU-8 coated PCB for the detection and enumeration of biological cells. The performance of the proposed device was evaluated using HeLa cells. The results showed that the device provided an effective and low-cost solution for PoC diagnostic applications.
Various integrated platforms based on impedance flow cytometers and fluorescent microscopes have been presented for the simultaneous detection and sizing of biological cells and particles. Holmes et al. [23] proposed an impedance-based fluorescence flow cytometer for the discrimination and enumeration of human blood. The whole blood and mixed CD antibody-conjugated cells samples were tested on the combining electrode deposited microchip and fluorescent microscope. The performance of the proposed device was evaluated by identifying and enumerating the CD4 + T-lymphocytes subpopulation in human whole blood. Barat et al.
[24] presented a microfluidic cytometer incorporating optical fibers and photomultiplier tubes (PMTs) to measure the side scattered light, signal extinction and fluorescence signals, respectively, and an RPS sensor for sizing purposes. Wang et al. [25] developed a PDMS-based microchip for counting the number and percentage of fluorescentlabeled CD4 + T-lymphocytes in human blood using a MOSFET-enhanced RPS sensor and a fluorescence detection method. It was shown that the device had an accuracy comparable to commercial flow cytometer.
Chen and Wang [28] presented a multi-functional optical flow cytometer for the simultaneous detection, counting and sizing of particles based on forward scattered light measured along the axis of the laser beam and backward fluorescent light measured along the laser excitation fiber. Fu and Wang [30] proposed an optical cytometer for the simultaneous enumeration and sizing of labeled and non-labeled micro-particles. In the proposed device, the size and total number of the particles was determined via the non-scattered light signal reflected from a plane mirror and the number of fluorescent-labeled particles was determined via the back scattered fluorescence signal. Watts et al. [31] proposed an integrated optofluidic device for scattering and fluorescence detection based on an SU-8 photoresist waveguide lens system deposited on a wafer and an external multimode optical fiber waveguide. It was shown that the detection reliability obtained for beads with diameters of 2.5 µm and 6 µm, respectively, was consistent with that obtained using conventional excitation methods.
The present study develops an optofluidic cytometer incorporating two power-free lowcost photovoltaic (PV) cells and an avalanche photodetector (APD) for the simultaneous counting, sizing and discrimination of biological cells and particles based on side scattered (SSC), extinction (EXT) and fluorescence (FL) signals. The proposed optofluidic cytometer utilizes a free-space light transmission technique, and thus not only avoids the need for a precise waveguide alignment procedure, but also reduces the cost and complexity of the fabrication process. The practical feasibility of the proposed device is demonstrated by detecting labeled and label-free beads with sizes of 3 μm, 5 μm, 7.26 μm and 10 μm, respectively.

System design
In commercial bench top flow cytometers, label-free particles are detected by measuring the change in intensity of the extinction light and side scattered light as the particles pass through a detection region illuminated by a laser beam, while labeled particles are detected by capturing the fluorescence signals emitted by the particles under laser excitation using a fluorescence detection system equipped with an appropriate optical filter. In the present study, the same detection principles are applied to enumerate and differentiate labeled and label-free particles using the PDMS-based optofluidic cytometer illustrated in Fig. 1. As shown, the extinction (EXT) and side scattered (SSC) signals are detected by two silicon photovoltaic (PV) cells (70 mm x 35 mm, SC 7035, Taiwan) mounted above and to one side of the microchip, both distances to the microchip are 15 mm respectively. In addition, the fluorescence (FL) signal is detected by an avalanche photodetector (APD, C5460-01, Hamamatsu, Japan) coupled to the optical fiber used to supply the laser excitation light. Notably, the EXT and SSC detection signals are modulated by the passage of both labeled and label-free particles through the detection region, while the FL detection signal is modulated only by the passage of the labeled particles. Furthermore, the amplitudes of the EXT and SSC detection signals vary with the particle size, while the amplitude of the fluorescence signal varies with the fluorescence intensity. Thus, by continuously monitoring the EXT, SSC and FL signals as the sample flows through the optofluidic cytometer, both the size and the type (i.e., labeled or label-free) of the particles can be reliably determined. To enhance the sensitivity of the detection results, the PV cells are connected to amplification circuits in order to boost the output signal. In addition, the relationship between the output voltage of the PV cells and the intensity of the detected laser light is determined using the method described by the present group in [32]. Finally, the APD output voltage signal is recorded by a data acquisition board (NI 9234, National Instruments, Austin, TX) and digitized using self-written LabVIEW® code (National Instruments, Austin, TX). In implementing the proposed optofluidic cytometer, the excitation laser light was provided by a laser diode module with a central wavelength ofλ = 488 nm coupled into a single mode optical fiber with a core diameter of 3.5 μm (Blue Sky Research, CA., optical power output ~25 mW). In addition, the microchip was mounted on an XY translator stage (Thorlabs, Newton, NJ) and a precise alignment between the microchip detection region and the optical fiber was achieved using an XYZ 3-axis travel translation stage (Thorlabs, Newton, NJ) and an optical microscope (SMZ800, Nikon Instruments Inc.). The laser light emitted by the fiber was focused on the detection region of the microchip using the method described previously in [32]. In addition, a multi-mode fiber optic splitter (62.5μm, 7.5% / 92.5%, Yofc, China) was used to guide excitation laser light to the main detection zone in the amount of 7.5% and guide emission fluorescent light to the APD for labeled-particle detection in the amount of 92.5%. To improve the accuracy of the labeled-particle detection results, the excitation laser light reflected by the glass substrate of the PDMS microchip was blocked using a long-pass filter (205876, Chroma, USA).

Materials
The optofluidic cytometer was fabricated on PDMS and glass substrates using a standard soft lithography technique. The microchannel is patterned with SU-8 2025 negative photoresist and the dimensions of detection zone is 20 μm x 25 μm (width x depth). The microfluidic chips were made as described previously [32]. The microchannel geometry of the detection chip was designed by an AutoCAD software and printed with a 12000 dpi high-resolution laser printer on a transparent plastic film used as a mask. A 25 µm thick SU-8 2025 negative photoresist (MicroChem Corp., USA) microchannel pattern on the glass is produced by UV soft lithography processes as PDMS casting mold. The detection experiments were performed using labeled and label-free polystyrene beads of various sizes. Three labeled beads were used, namely (1) fluorescent green 3 μm beads (G0300, 1% solid content, Duke Scientific, USA) with a stock concentration of 6.7 × 10 5 beads/ml; (2) 5 μm beads (G0500, 1% solid content, Duke Scientific, USA) with a stock concentration of 1.4 × 10 5 beads/ml; and (3) 10 μm beads (G1000, 1% solid content, Duke Scientific, USA) with a stock concentration of 1.8 × 10 4 beads/ml. All three beads had a peak excitation wavelength of 468 nm, a peak emission wavelength of 508 nm, and a refractive index of 1.59. Just one label-free sample was used, namely non-fluorescent 7.26 μm beads (PS06N, 9.9% solid content, 0.63 μm standard deviation, Bangs Laboratories, USA,) with a stock concentration of 4.7 × 10 5 beads/ml. The detection experiments were performed using both single-bead population samples (i.e., beads of a single type and size) and mixed-bead population samples (i.e., beads of more than one type and size). For each sample, the beads were diluted and suspended in DI water with a concentration of approximately 2 × 10 4 beads/ml. Furthermore, prior to each test, the beads were re-suspended in the solution by vortexing at 2000 rpm for 10 s.

Results and discussion
For the low cost and low sensitivity photovoltaic cells signals detection, the amplification circuits were designed and applied for the EXT and SSC signals enhancement. Figure 2 indicates the extinction signals of 3 µm particles from the photovoltaic cell and dual stage amplification. The original signals (Fig. 2(a)) from the photovoltaic cell are about 0.1 mV and have negative peaks from a high baseline because the particles in the detection zone block the laser light. After the first stage amplification (Fig. 2(b)), the extinction signal is about 1 mV that enhanced by the amplification circuit. To further enhance the extinction signals, an inverting amplifier circuit was used that causes the extinction signals have positive peaks from baseline and the S/N ratio is improved significantly (Fig. 2(d)). The same amplification circuit is used for SSC signals detection that causes the SSC signals have negative peaks from a high baseline. The detection performance of the optofluidic cytometer was evaluated initially using single-bead population samples. Figure 3(a) shows the side-scattered (SSC), extinction (EXT) and fluorescence (FL) signals acquired over a 10 s observation period for a sample containing labeled 3 μm beads. For each figure, a spike in the baseline signal corresponds to the passage of a bead through the integration region of the microchip. Since the beads in the sample are labeled, spikes are observed not only in the SSC and EXT signals, but also in the FL signal. As expected, the spikes in all three signals are perfectly synchronized. Note that the inverting amplification circuits for signals enhancement cause the SSC signals in negative peaks and EXT signals in positive peaks. From inspection, the spikes in the SSC and EXT signals have intensities of 3~5 mV, and 9~12.5 mV, respectively, while those in the FL signal have an intensity of 75~150 mV. Figure 3(b) shows the SSC, EXT and FL signals acquired over a 10 s observation period for the sample containing label-free 7.26 μm beads. Distinct peaks, corresponding to individual beads, are observed in the SSC and EXT signals. However, as expected, the FL signal lacks such peaks since the beads are label-free. The peaks in the SSC and EXT signals have intensities of approximately 10~30 mV and 30~60 mV, respectively. It is noted that the peaks intensities are higher than those observed for the 3 μm beads. In other words, the intensity of the peaks in the SSC and EXT signals reduces as the bead diameter reduces. The ability of the proposed cytometer to differentiate between beads of a different type (i.e., labeled and label-free) and a different size was evaluated using a mixed-bead population sample containing 3 μm fluorescent beads, 5 μm fluorescent beads, 7.26 μm label-free beads and 10 μm fluorescent beads. Figure 4(a) shows the SSC, EXT and FL signals obtained over a 400 s detection period. As shown, the peaks in the SSC signal have a one-to-one correspondence with those in the EXT signal. However, while the peaks in the FL signal also correspond to those in the SSC and EXT signals, they are fewer in number since no peaks for the label-free 7.26 μm beads appear. Notably, all three signals remain stable over time. In other words, the optofluidic cytometer provides a stable detection performance despite the use of low-cost PV cells. Figure 4(b) shows an enlarged view of the typical peaks in the SSC, EXT and FL signals over a 20 s period. The results show more clearly that the four different sizes and different types of micro-particles are recognized in term of extinction, side scatter, fluorescent and enumeration by the developed optofluidic cytometer. The results confirm that for all three signals, the peak intensity reduces with a reducing bead diameter. Moreover, for each bead diameter, the peak intensity in the EXT signal is greater than that in the SSC signal. Furthermore, the SSC and EXT signals contain peaks corresponding to all four beads, while the FL signal contains peaks only for the three labeled beads. Thus, by inspecting the amplitudes of the peaks in the three signals, and comparing the SSC/EXT signals with the FL signal, it is possible to determine not only the size of each detected bead, but also the type. Table 1 presents detection results for the four single-bead population samples, as determined using the EXT, SSC and FL coefficient of variations (CV), respectively. In performing the detection experiments, the CV values were 32.1%, 37.2% and 24.7%, respectively. The corresponding average intensities of the peaks in the EXT, SSC and FL signals for the 3 μm sample were found to be 11.1 mV, 5.7 mV and 113.3 mV, respectively. For the 5 μm beads, the CV values were equal to 26.9%, 37.1% and 39.2%, respectively, and the corresponding average intensities of the peaks in the EXT, SSC and FL signals were found to be 33.1 mV, 11.0 mV and 615.4 mV. For the label-free 7.26 μm beads, the CV values were 24.3% and 32.5%, respectively, and the corresponding average peak intensities in the EXT and SSC signals were found to be 45.8 mV and 22.1 mV. Finally, for the 10 μm beads, the CV values were 18.5%, 27.7% and 21.5%, and the average peak intensities in the EXT, SSC and FL signals were found to be 102.6 mV, 45.3 mV and 2.44 V, respectively. For comparison purposes, the detection experiments were repeated using a conventional optical flow cytometer (FACScan, Becton Dickinson, USA) equipped with a 488 nm solid-state laser (20 mW, Coherent Sapphire, USA). The corresponding detection results are also shown in Table 1. The CV values for the 3 μm beads were found to 24.2%, 33.4% and 9.0% when using the EXT, SSC and FL signals, respectively. The corresponding CV values for the 5 μm beads were found to be 11.6%, 13.0% and 5.3%, respectively. The CV values for the 7.26 μm label-free particles were 9.6% (EXT signal) and 13.0% (SSC signal). Finally, for the 10 μm beads, the CV values were 6.8% (EXT signal), 10.9% (SSC signal) and 4.5% (FL signal). Figures 5 present histograms of the detection results obtained using the three signals in the proposed optofluidic cytometer, for the 3 μm, 5 μm and 10 μm labeled beads in a mixed-bead population sample. For the proposed optofluidic cytometer, the histogram of the detection results obtained using the SSC signal lacks distinct clusters corresponding to the individual beads since a high CV value is obtained irrespective of the bead size as a result of the microchip and system structure. More specifically, in contrast to the EXT signal, the height of the beads in the microchannel influences the scattered light intensity and light scattering direction. Consequently, the CV value increases. Furthermore, the PV cell used for SSC detection purposes is mounted on the side of the microchip (see Fig. 1), and hence the microchip material blocks the side scattered light and thus causes its intensity to be lower than that of the extinction light. The histogram of the detection results obtained using the EXT signal contains three relatively distinct clusters corresponding to the different bead sizes. The superior discriminating ability of the EXT signal arises because of the thin PDMS microchip optical transmittance structure. Finally, for the histogram obtained using the FL signal, a good discrimination performance among the three beads is observed since the fluorescence intensity varies significantly with the bead size.  Figures 6(a), 6(b) and 6(c) present SSC/EXT, EXT/FL and SSC/FL dot plots of the detection results obtained for a mixed-bead population sample consisting of 3 μm, 5 μm and 10 μm labeled beads and 7.26 μm label-free beads. Note that for each figure, the plot presented on the left corresponds to the detection results obtained using the proposed optofluidic cytometer, while that on the right corresponds to the results obtained using the conventional cytometer. Figure 6(a) shows that for both cytometers, the detection results are grouped into four main clusters, where each cluster corresponds to a particular bead. However, for Figs. 6(b) and 6(c), only three clusters are observed since the FL signal fails to detect the label-free 7.26 μm beads in the sample. The dot plot of four main clusters is very close due to the weak SSC and EXT signals. Figure 6(a) shows that for both cytometers, the detection results are grouped into four main clusters, where each cluster corresponds to a particular bead. The dot plot of four main clusters is very close due to the weak EXT and SSC signals. The amplitudes of the SSC signals range from 5.7 mV to 45.3 mV and the EXT signals range from 11.1 mV to 102.6 mV, respectively. The dynamic range for the SSC and EXT measurements is very small due to the low photoelectric sensitivity of photovoltaic cells. The high photoelectric sensitivity photo-detectors such as APD and PMT can improve the sensitivity and dynamic range of measurement significantly. The experimental results shown in Fig. 6 are summarized in Table 2. As shown, the sample analyzed by the proposed optofluidic cytometer contained a total of approximately 1061 beads, while that analyzed by the conventional cytometer contained around 9525 beads. As shown, the proposed optofluidic cytometer yielded 276, 264, 343 and 178 counts for the 3 μm beads, 5 μm beads, 7.26 μm beads and 10 μm beads, respectively. In other words, the 3 μm, 5 μm, 7.26 μm and 10 μm beads accounted for 26%, 24.9%, 32.3% and 16.8% of the total number of beads in the sample, respectively. For the conventional cytometer, 2714, 2271, 3088 and 1452 counts were obtained for the 3 μm, 5 μm, 7.26 μm and 10 μm beads, respectively, accounting for 28.5%, 23.9%, 32.4% and 15.3% of the total number of beads in the sample. It is seen that the detection performance of the two cytometers is very similar for each bead. Consequently, the practical feasibility of the proposed optofluidic cytometer is confirmed.

Conclusions
This study has presented a optofluidic cytometer consisting of a detection microchip, a laser light source, two power-free photovoltaic (PV) cells and an avalanche photo-detector (APD).
In the proposed device, the two PV cells are placed above and to one side of the microchip, respectively, in order to detect the extinction (EXT) light and side scattered (SSC) light, while the APD is coupled to the optical fiber used to supply the excitation laser light in order to detect the fluorescence (FL) signal. Compared to existing microflow cytometer [26, 27, 33], the proposed device utilizes a free-space light transmission technique, as a result, the need for optical waveguides is removed, and hence the fabrication cost is reduced and the need for a complex waveguide alignment process is removed. The feasibility of the proposed device has been demonstrated using labeled beads with dimensions of 3 μm, 5 μm and 10 μm, respectively, and label-free beads with a size of 7.26 μm. The detection experiments have been performed using both single-bead and mixed-bead population samples. The results have shown that the proposed optofluidic cytometer has the ability to perform simultaneous SSC, EXT and FL detection. Consequently, the proposed cytometer has significant potential for low-cost point-of-care (PoC) diagnostic applications.