6510
IEEE SENSORS JOURNAL, VOL. 17, NO. 20, OCTOBER 15, 2017
Highly Sensitive Microfluidic Chip Sensor for Biochemical Detection Varun Lingaiah Kopparthy and Eric J. Guilbeau
Abstract— Chip calorimetry offers a power tool for fast and high throughput analysis of biochemical process. However, it is challenging to realize an inexpensive, easy to fabricate microfluidic chip-based calorimeter with high sensitivity. This paper describes the design of a novel, highly sensitive, and continuous flow microfluidic chip sensor with an integrated antimony (Sb)–bismuth (Bi) thin-film thermopile heat detection element. The geometry and the design of the microfluidic device facilitate hydrodynamic flow focusing, and the integration and design of the thermopile sensor into the microfluidic device eliminates the need for reference temperature control. The device contains a single flow channel that is 120 µm high and 10-mm wide with two fluid inlets and one fluid outlet. An Sb-Bi thin film thermopile is fabricated on the inner surface of the bottom channel wall using thermal evaporation and was passivized with a 3 µm SU-8 photoresist layer. The device has been successfully used to measure the dynamic temperature changes resulting from heat generation following the mixing of glycerol and water. The effect of flow rates on the sensor’s response was measured. The sensor can detect dynamic temperature changes in the order of 10−6 K. The limit of detection of heat power of the device was calculated to be 8.8 pW. With the obtained remarkable sensitivity and heat power detection limit, the microfluidic chip sensor can potentially be used to investigate biochemical processes, such as enzyme-catalysed reactions, and metabolic activity of cells. Index Terms— Thermoelectric, thermopile, glycerol, microfluidic, biochemical sensor, lab-on-a-chip, calorimetry.
I. I NTRODUCTION
M
EASURING biochemical process has showed increasing importance with the development of chip calorimeters [1]. Chip calorimetry is generally achieved by integrating the calorimetric detection using sensors such as thermistors and thermopiles with the micro flow channels for sample handling. Several groups have shown the advantages of chip calorimeters for the analysis of biochemical process [2]–[4]. Researchers reported the application of chip calorimeters for the detection of bioprocesses such as enzymatic reactions [5]–[7], binding event detection-for instance streptavidin-biotin binding [8], DNA hybridization event [9], [10], incorporation of nucleotide into a base DNA strand [11], [12], and monitoring of biofilm activity/cell Manuscript received May 18, 2017; revised August 23, 2017; accepted August 24, 2017. Date of publication August 29, 2017; date of current version September 25, 2017. The associate editor coordinating the review of this paper and approving it for publication was Prof. Sang-Seok Lee. (Corresponding author: Varun Lingaiah Kopparthy.) V. Kopparthy is with the Department of Mechanical Engineering, Cornell University, Ithaca, NY 14850 USA (e-mail:
[email protected]). E. J. Guilbeau is with the Department of Biomedical Engineering, Louisiana Tech University, Ruston, LA 71270 USA. Digital Object Identifier 10.1109/JSEN.2017.2746419
metabolism [13], [14]. To monitor biochemical process such as metabolic activity of cells and binding event measurements of biomolecules, a highly sensitive chip calorimeter is desired, as the heat released from these processes are very low. Heat power sensitivities of the chip calorimeters must be in the range of nW to pW [13]–[15] to detect complex bioprocesses accurately. Detection elements used in calorimeters such as thermopiles need extreme reference temperature control for sensing the bioprocesses with high sensitivity. To achieve the above specifications, often, chip calorimeters are realized using MEMS fabrication resulting in high heat power sensitivities. MEMS techniques are highly complex, time consuming, expensive and require sophisticated equipment. Few attempts were reported to eliminate the need for reference temperature control for thermopile sensors exploring the common-mode rejection of thermal signals and by using internal fluid flows in the microdevices [4], [16], [17]. Rapid prototyping has shown great promise in realizing inexpensive and easy to fabricate fluidic chips with micrometer features enabling lab-on-chip analysis. This study reports the fabrication of a highly sensitive microfluidic chip sensor using a rapid prototyping technique called Xurography [18]. Reference temperature control is eliminated by the geometry of the microfluidic channel that facilitates hydrodynamic flow focusing. Device operation is demonstrated by detecting heat of mixing of glycerol with water. II. E XPERIMENTAL A. Fabrication of Thin Film Thermopile Thin-film thermopile with antimony (Sb) and bismuth (Bi) metals (Sigma Aldrich, USA) was fabricated using thermal evaporation technique. Sb and Bi metals were chosen because of their higher Seebeck coefficient values 47 and −72 μV K−1 [4], respectively. Complementary metal shadow masks (Towne Technologies, USA) were used to fabricate the thermopile sensor. First, a Bi layer of thickness 0.8 μm with a mask pattern is deposited, and a complementary pattern is aligned and then a Sb layer of thickness 1.2 μm is deposited on a glass coverslip. Second, a metal mask with a pattern for the connector leads is also fabricated on the same coverslip by depositing 1.2 μm Sb layer. Thickness of each metal layer detailed above was an optimized value based on fabrication conditions such as thermal evaporator, and shadow mask technique. Thermopile fabrication procedure described in detail in [4]. Thermopiles have measuring and reference
1558-1748 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
KOPPARTHY AND GUILBEAU: HIGHLY SENSITIVE MCS FOR BIOCHEMICAL DETECTION
Fig. 1. Antimony (Sb)-bismuth (Bi) thin-film thermopile sensor fabricated on a glass coverslip. Enlarged view shows the individual metal lines and thermocouple junctions.
6511
Fig. 2. (a) MCS-1 configuration showing the fabrication of microfluidic device with thermopiles outside the fluidic channel. (b) Schematic of the MCS-2 configuration with a SU-8 passivation layer on thermopiles facing inside of the channel. Pictures next to the schematics in A and B are the fabricated MCS configurations. The microfluidic channel dimensions are 60 mm X 12 mm X 0.1 mm and the channel volume is 72 μl.
junction sets (Fig. 1). The fabricated thermopiles typically have a resistance of 20 K and showed a Seebeck coefficient of 7.14 μV (mK)−1. B. Microfluidic Chip Sensor (MCS) A microfluidic chip sensor (MCS) is fabricated by integrating the thermopile on the channel wall of the microfluidic device. A microfluidic device is fabricated by sandwiching a 100 μm thick dual side adhesive tape (kapton.com) between a microscope glass slide (thickness-1.2 mm and thermal conductivity, k-1.05 W m−1 K−1 ) and a microscope glass coverslip (thickness-170 μm and thermal conductivity, k-1.05 W m−1 K−1 ). Before sandwiching, the dual side adhesive tape is cut with a pattern that forms the channel using a Graphtec plotter (Graphtec Inc, USA). Inlet holes (Inlet 1 and Inlet 2) and an outlet hole were pre-drilled onto the microscope glass slide for fluid flow. Two configurations of the microfluidic chip sensors (MCS-1, MCS-2) were fabricated, tested and evaluated. MCS-1 was fabricated with the glass coverslip containing thermopile sensor facing outside of the channel (Fig. 2a). The thermopiles are protected by attaching a 25 μm thick tape to prevent damage due to handling. MCS-2 was fabricated with the coverslip containing thermopile sensor facing the channel (Fig. 2b). Thermopiles were passivized using a 3 μm thick SU-8 (Microchem, USA, thermal conductivity k-0.2 W m−1 K−1 ) photoresist layer to protect from the fluid flow. Thermal resistance (R) for the reaction generated by the flow in the microfluidic channel to the thermopile for both MCS configurations can be calculated by the following equation [19]: L (1) R= k.A where: L- thickness of the layer (m), k-thermal conductivity (W m−1 K−1 ), and A-area of the layer (m2 ). The area for the calculation is the reaction zone area, which is approximately 24 mm2 (the width of the hydrodynamically focused sample is approximately 4 mm and the length of the thermopile measuring junction’s region is 6 mm). The total thermal resistance in the MCS-1 is due to the thermal resistance of the glass coverslip, which is 6.07 K W−1 , and
Fig. 3.
Schematic of the experimental setup.
the total thermal resistance in the MCS-2 only the 3 μm thick SU-8 layer which is 0.833 K W−1 . C. Experimental Setup The experimental setup (Fig. 3) used for this study consists of two syringe pumps, which continuously flow water into the inlets of the MCS. Since the flows are in microscale, mixing happens only at the interface as viscosity dominates compared to inertia. This results in a hydrodynamically focused flow [20] in the center, when fluid is flown in inlet 1 and inlet 2 simultaneously (Fig. 4). Lee et al. [20] derived the relationship for the width of the hydrodynamic flow in rectangular microfluidic channels as the ratio of the inlet flow rates. An injection valve with a 13 μl sample loop is used to inject the sample into the inlet 2 flow for generating a chemical reaction. The inlet 2 flow is hydrodynamically focused into the MCS. The reaction is generated over the measuring junctions of the thermopile, which generates a proportional voltage. This voltage is detected by a nanovoltmeter (Agilent 34420A, Agilent technologies, USA) and is recorded in a computer. To demonstrate the operation of the MCS, the exothermic nature of glycerol water mixing reaction is utilized. The enthalpies of mixing of water with glycerol over the entire composition range at 298.15 K is determined to be exothermic with a maximum enthalpy of −500 J/mol at molar fraction of 0.5 [21], [22]. III. R ESULTS AND D ISCUSSION The fabricated MSC-1 and MCS-2 were tested using the experimental system for varying glycerol and water mixing scenarios. Effect of flowrates on MCS response was also
6512
IEEE SENSORS JOURNAL, VOL. 17, NO. 20, OCTOBER 15, 2017
Fig. 5. (a) The output of MCS-1 for 10% (V/V) glycerol. (b) Peak height comparison of MCS-1 and MCS-2 for 10 % (V/V) glycerol. (c) Effect of flow rate on MCS-2 response for varying concentrations of glycerol. (d) Sensitivity of the MCS-1 and MCS-2 configurations for varying concentrations of glycerol. Error bars represent the standard error when n = 3. Flowrates used in all cases except (c) are 100 μl min-1 and 25 μl min-1 in inlet 1 and inlet 2 respectively. Fig. 4. Hydrodynamic flow focusing in MCS. (a) CoventerWare®simulation of the hydrodynamic flow in the MCS when the flow rates in inlet 1 and inlet 2 are 100 μl min−1 and 50 μl min−1 respectively. (b) Enlarged image section showing the interfacial mixing. Fluid in inlet 1 and inlet 2 are shown by red and blue colors. (c) Schematic (drawn not to scale) of the top and side view of the MCS illustrating the inlet fluid flow and the location of the thermopile.
reported. After the glycerol sample injection into the inlet 2 flow stream, the sample traveled to the microfluidic device was hydrodynamically focused over the measuring junctions of the thermopile. The glycerol sample mixed with the water flowing from inlet 1 at the interface and generated heat. As the sample passed over the measuring junctions of the thermopile, the voltage increased and then returned to its baseline once the sample flowed past the thermopile. A typical response of the thermopile to glycerol water mixing is shown in Fig 5a. The response curve has two characteristics: 1) the magnitude, which represents the total temperature change detected by the thermopile due to the reaction, and 2) the area under the curve (AUC), this represents the total temperature heat detected by the thermopile. MCS-1 (thermopiles are outside of the flow channel) has high thermal resistance between the reaction zone and the thermopile compared to MCS-2 (thermopiles are inside of the flow channel) with only a 3 μm photoresist layer. MCS-1 and MCS-2 responses to glycerol water mixing reaction were compared in Fig. 5b. As the MCS-2 showed improved performance compared to MCS-1, the effect of flowrate on the MCS-2 was investigated (Fig. 5c). Flowrates of 100 μl min−1 and 25 μl min−1 , and 50 μl min−1 and 25 μl min−1 inlet 1 and inlet 2 respectively were applied and the responses were plotted. The ratio of flowrates in inlet 1 and inlet 2 determines the width of the hydrodynamic flow focusing of inlet 2 flow. As the ratio decreases, the width of the hydrodynamically focused flow increases. Since the reaction heat is generated due to interfacial mixing, a wider focused inlet 2 flow will
TABLE I AUC OF THE T HERMOPILE S IGNAL FOR MCS-1 AND MCS-2 FOR S ENSITIVITY C ALCULATIONS . S TANDARD E RROR R EPRESENTS THE VARIATION IN THE MCS R ESPONSE FOR T HREE S AMPLE I NJECTIONS ( N = 3)
move the interface away from the measuring junctions of the thermopile. This results in a decreased MCS response. The sensitivity of both MCS-1 and MCS-2 were compared (Table I) using the experimental setup when the flowrates used are 100 μl min−1 and 25 μl min−1 in inlet 1 and inlet 2 respectively. The responses of both MCS-1 and MCS-2 for the exothermic reaction of glycerol-water mixing are shown in Fig. 5d. The lowest concentration of glycerol injected was 0.1% V/V, for which a voltage of 400 nV response was recorded by MCS-2. Considering the Seebeck coefficient of the thermopile, 400 nV corresponds to a detected temperature of 57.2 μK. MCS-1 measured no voltage for 0.1 % (V/V) glycerol concentration, because of the high thermal resistance for heat transfer to the thermopile. MCS-2 configuration showed 3.5-fold increase in sensitivity compared to MCS-1. The heat power sensitivity of the fabricated thermopile was characterized in our earlier paper as 0.045 V W−1 [4]. Considering the heat power sensitivity of the thermopile, the lowest heat power detected by MCS-2 was calculated to be ∼8.8 pW. Heat power sensitivity achieved is remarkable (see Table II), provided the MCS was fabricated using a simple technique by integrating the thermopile sensor inside the flow channel.
KOPPARTHY AND GUILBEAU: HIGHLY SENSITIVE MCS FOR BIOCHEMICAL DETECTION
6513
TABLE II C OMPARISON OF THE MCS W ITH THE P REVIOUSLY R EPORTED C HIP C ALORIMETERS
Fabricated MCS occasionally suffered from leakages due to inefficient bonding of the adhesive Kapton®tape to the SU-8 layer. However, Xurography is inexpensive and is suitable for fabricating disposable microfluidic chips in applications where a single test per subject is needed. Incorporation of nucleotide into base DNA strand has been explored using the MCS sensor described in this study. Amine modified DNA strand can be covalently bonded to the epoxy ring chemistry on SU-8. DNA based detection will be detailed in a separate paper. Compatible reactants must be used as SU-8 photoresist is compatible with only few chemicals, for instance ethanol etches away SU-8 layer. A similar coating of liquid PMMA or liquid Kapton® (similar thermal conductivities like SU-8) could be used to passivize the thermopiles for MCS-2 configurations. Two major advantages of the MCS developed in this study are 1) the sensing method is label-free: the principle is based on the calorimetric detection of heat released by the biochemical reactions, and 2) MCS operates without power (self-generating): the sensor generates an output when a temperature difference is maintained over its junctions. A voltmeter is only required to measure the output of the sensor. With the obtained heat power sensitives, the MCS can be used for the detection of biochemical, and bioprocesses involving exothermic or endothermic reaction energies. IV. C ONCLUSION A thermoelectric microfluidic chip sensor was developed to measure dynamic temperature changes in the order of 10−6 K. High sensitivity was obtained by an inexpensive and easy to fabricate MCS without any extreme measures and complex procedures to control reference temperatures. Two configurations of the sensors were fabricated and tested MCS-1 and MCS-2. An improved sensor with low thermal resistance (MCS-2) showed a low heat power detection of 8.8 pW. The sensitivity obtained with the simple to fabricate MCS is suitable for monitoring bioprocesses such as cell metabolism, enzymatic reactions, and binding event measurements. ACKNOWLEDGMENTS Special thanks are also extended to Joshna Nimmala for her assistance in experimental system setup and device fabrication. The authors would like to thank staff at the Institute for
Micromanufacturing (IfM) and Center for Biomedical Engineering and Rehabilitation Science (CBERS) at Louisiana Tech University for their support. R EFERENCES [1] W. Lee, J. Lee, and J. Koh, “Development and applications of chip calorimeters as novel biosensors,” Nanobiosensors Disease Diagnosis, vol. 1, pp. 17–29, Apr. 2012. [2] Y. Zhang and S. Tadigadapa, “Calorimetric biosensors with integrated microfluidic channels,” Biosensors Bioelectron., vol. 19, no. 12, pp. 1733–1743, 2004. [3] W. Lee, W. Fon, B. W. Axelrod, and M. L. Roukes, “High-sensitivity microfluidic calorimeters for biological and chemical applications,” Proc. Nat. Acad. Sci. USA, vol. 106, no. 36, pp. 15225–15230, Sep. 2009. [4] V. L. Kopparthy, S. M. Tangutooru, G. G. Nestorova, and E. J. Guilbeau, “Thermoelectric microfluidic sensor for bio-chemical applications,” Sens. Actuators B, Chem., vols. 166–167, pp. 608–615, May 2012. [5] S. M. Tangutooru, V. L. Kopparthy, G. G. Nestorova, and E. J. Guilbeau, “Dynamic thermoelectric glucose sensing with layer-by-layer glucose oxidase immobilization,” Sens. Actuators B, Chem., vols. 166–167, pp. 637–641, May 2012. [6] V. L. Kopparthy, S. M. Tangutooru, and E. J. Guilbeau, “Label free detection of L-glutamate using microfluidic based thermal biosensor,” Bioengineering, vol. 2, no. 1, pp. 2–14, 2015. [7] J. Lerchner et al., “A new micro-fluid chip calorimeter for biochemical applications,” Thermochim. Acta, vol. 445, no. 2, pp. 144–150, Jun. 2006. [8] S.-I. Yoon, S.-C. Park, and Y.-J. Kim, “A micromachined microcalorimeter with split-flow microchannel for biochemical sensing applications,” Sens. Actuators B, Chem., vol. 134, no. 1, pp. 158–165, 2008. [9] L. M. Ahmad, B. Towe, A. Wolf, F. Mertens, and J. Lerchner, “Binding event measurement using a chip calorimeter coupled to magnetic beads,” Sens. Actuators B, Chem., vol. 145, no. 1, pp. 239–245, Mar. 2010. [10] G. G. Nestorova, B. S. Adapa, V. L. Kopparthy, and E. J. Guilbeau, “Lab-on-a-chip thermoelectric DNA biosensor for label-free detection of nucleic acid sequences,” Sens. Actuators B, Chem., vol. 225, pp. 174–180, Mar. 2016. [11] G. G. Nestorova and E. J. Guilbeau, “Thermoelectric method for sequencing DNA,” Lab Chip, vol. 11, no. 10, pp. 1761–1769, 2011. [12] H. Esfandyarpour and R. W. Davis, “An integrated differential nanocalimeter with on-chip microfluidic multiplexing for high throughput genomics and proteomics,” in Proc. 14th Int. Conf. Miniaturized Syst. Chem. Life Sci., Groningen, The Netherlands, 2010, pp. 1–3. [13] J. Lerchner et al., “Miniaturized calorimetry—A new method for realtime biofilm activity analysis,” J. Microbiol. Methods, vol. 74, nos. 2–3, pp. 74–81, 2008. [14] K. Verhaegen, K. Baert, J. Simaels, and W. Van Driessche, “A high-throughput silicon microphysiometer,” Sens. Actuators A, Phys., vol. 82, nos. 1–3, pp. 186–190, 2000. [15] E. A. Johannessen, J. M. R. Weaver, P. H. Cobbold, and J. M. Cooper, “A suspended membrane nanocalorimeter for ultralow volume bioanalysis,” IEEE Trans. Nanobiosci., vol. 99, no. 1, pp. 29–36, Mar. 2002. [16] B. S. Kwak, B. S. Kim, H. H. Cho, J. S. Park, and H. I. Jung, “Dual thermopile integrated microfluidic calorimeter for biochemical thermodynamics,” Microfluidics Nanofluidics, vol. 5, no. 2, pp. 255–262, 2008.
6514
[17] S.-I. Yoon, M.-H. Lim, S.-C. Park, J.-S. Shin, and Y.-J. Kim, “Neisseria meningitidis detection based on a microcalorimetric biosensor with a split-flow microchannel,” J. Microelectromech. Syst., vol. 17, no. 3, pp. 590–598, Jun. 2008. [18] D. A. Bartholomeusz, R. W. Boutte, and J. D. Andrade, “Xurography: Rapid prototyping of microstructures using a cutting plotter,” J. Microelectromech. Syst., vol. 14, no. 6, pp. 1364–1374, Dec. 2005. [19] T. L. Bergman, F. P. Incropera, D. P. Dewitt, and A. S. Lavine, Fundamentals of Heat and Mass Transfer. Hoboken, NJ, USA: Wiley, 2011. [20] G.-B. Lee, C.-C. Chang, S.-B. Huang, and R.-J. Yang, “The hydrodynamic focusing effect inside rectangular microchannels,” J. Micromech. Microeng., vol. 16, no. 5, p. 1024, 2006. [21] Y. Marcus, “Some thermodynamic and structural aspects of mixtures of glycerol with water,” Phys. Chem. Chem. Phys., vol. 2, no. 21, pp. 4891–4896, 2000. [22] D. Peetersa and P. Huyskensbp, “Endothermicity or exothermicity of water/alcohol mixtures,” J. Mol. Struct., vol. 300, pp. 539–550, Dec. 1993. [23] V. Baier et al., “Highly sensitive thermopile heat power sensor for microfluid calorimetry of biochemical processes,” Sens. Actuators A, Phys., vols. 123–124, pp. 354–359, Sep. 2005. [24] L. Wang, D. M. Sipe, Y. Xu, and Q. Lin, “A MEMS thermal biosensor for metabolic monitoring applications,” J. Microelectromech. Syst., vol. 17, no. 2, pp. 318–327, 2008.
IEEE SENSORS JOURNAL, VOL. 17, NO. 20, OCTOBER 15, 2017
Varun Lingaiah Kopparthy received the B.Tech. degree in biomedical engineering from Jawaharlal Nehru Technological University, Hyderabad, India, in 2008, and the M.S. and Ph.D. degrees in biomedical engineering from Louisiana Tech University in 2016. He is currently a Post-Doctoral Associate with the Mechanical Engineering Department, Cornell University. His current research interests include global health diagnostics, personalized diagnostics, and biosensors.
Eric J. Guilbeau received the Ph.D. degree in chemical engineering from Louisiana Tech University in 1971. He is currently a Professor Emeritus with the Department of Biomedical Engineering, Louisiana Tech University. His current research interests include applied biotechnology and biosensors.