Analyst View Article Online
Published on 02 December 2013. Downloaded by University of Texas Libraries on 26/05/2014 15:53:43.
PAPER
Cite this: Analyst, 2014, 139, 431
View Journal | View Issue
Real-time measurement of flow rate in microfluidic devices using a cantilever-based optofluidic sensor Mohammad Sadegh Cheri,ab Hamid Latifi,*ab Jalal Sadeghi,a Mohammadreza Salehi Moghaddam,a Hamidreza Shahrakia and Hasan Hajghassemc Real-time and accurate measurement of flow rate is an important reqirement in lab on a chip (LOC) and micro total analysis system (mTAS) applications. In this paper, we present an experimental and numerical investigation of a cantilever-based optofluidic flow sensor for this purpose. Two sensors with thin and thick
cantilevers
were
fabricated
by
engraving
a
2D
pattern
of
cantilever/base
on
two
polymethylmethacrylate (PMMA) slabs using a CO2 laser system and then casting a 2D pattern with polydimethylsiloxane (PDMS). The basic working principle of the sensor is the fringe shift of the Fabry– Pe´rot (FP) spectrum due to a changing flow rate. A Finite Element Method (FEM) is used to solve the Received 21st August 2013 Accepted 6th November 2013
three dimensional (3D) Navier–Stokes and structural deformation equations to simulate the pressure distribution, velocity and cantilever deflection results of the flow in the channel. The experimental results show that the thin and thick cantilevers have a minimum detectable flow change of 1.3 and 4 (mL min1)
DOI: 10.1039/c3an01588b
respectively. In addition, a comparison of the numerical and experimental deflection of the cantilever has
www.rsc.org/analyst
been done to obtain the effective Young's modulus of the thin and thick PDMS cantilevers.
1
Introduction
During the last two decades, lab-on a chip (LOC) and micro total analysis systems (mTAS) have seen an explosive growth in the chemical, biological, biochemical and medical elds due to properties such as rapid analysis time, reduced reagent consumption, high throughput, portability and also reproducibility.1,2 Real-time measurement and accurate control of ow rate is a vital need in applications such as particle sorting and separation,3 ow cytometry,4 air and liquid droplet-based systems,5 mixing,6 chemical synthesis7 and polymerase chain reaction (PCR).8 Several groups of microuidic researchers have been studying various ow sensors with different working principles that have used micro-electro-mechanical systems (MEMS) in ow sensors. MEMS based ow sensors are appropriate candidates for integration with microuidic devices because of their lower power consumption, higher precision, rapid response, improved portability, and lower manufacturing cost. Thermal ow sensors are the most common sensors based on heat transfer detection.9–14 Although these sensors have a high sensitivity, their integration into a microuidic chip is a
a
Laser & Plasma Research Institute, Shahid Beheshti University, Evin, Tehran 1983963113, Iran. E-mail: lati@sbu.ac.ir
b
Department of Physics, Shahid Beheshti University, Evin, Tehran 1983963113, Iran
c
Department of Electrical and Electronic Engineering, Malek Ashtar University of Technology, Tehran, 19395-7179, Iran
This journal is © The Royal Society of Chemistry 2014
challenging issue due to their complex structure. Non-thermal or mechanical ow sensors incorporate a moving structure such as a cantilever, spring, diaphragm etc.15–19 The sensing principle in the mechanical ow sensors can be obtained by mechanisms such as piezoresistivity, piezoelectricity, variable capacitance and optical detection. In optical detection, Fiber Fabry–P´ erot Interferometery (FFPI) is a suitable sensor to integrate with LOC and mTAS due to advantages such as immunity to electromagnetic interference, high sensitivity, excellent resolution, multiplexing capability, and multifunctional sensing abilities. P. Caldas et al.9 fabricated a ber optic hot-wire ow meter based on a metallic coated hybrid long period grating/ber Bragg grating structure with a minimum ow speed of 0.08 (m s1). V. Lien et al.19 used a tapered ber in order to measure ow rates ranging from 0 to 1500 (mL min1). A laminar ow exerts a drag force to the tapered ber resulting in its displacement and decreases the light intensity transmitted through the ber taper into a multi-mode ber. The authors offered the ow sensor for high throughput applications such as ow cytometry and particle sorting/counting. N. Noeth et al.20 measured ow rate based on deection of a SU-8 cantilever perpendicular to the ow stream. They used a bulk setup including a focused beam of laser on the surface of the SU-8 cantilever and measured the displacement of the reected laser beam on a position sensitive detector (PSD). Sanati et al.21 have presented a multilayer so lithography process for fabrication of a PDMS microcantilever-based ow sensor. In that work various ow rates were measured between 0.2 and 1.3 (mL min1) by monitoring and measuring cantilever deection through the
Analyst, 2014, 139, 431–438 | 431
View Article Online
Published on 02 December 2013. Downloaded by University of Texas Libraries on 26/05/2014 15:53:43.
Analyst
Paper
optical microscope and off-line image processing techniques. They reported ow rates as low as 35 (mL min1). In this paper, we have fabricated, simulated and characterized a cantilever-based optouidics ow sensor. The sensor is integrated with optical bers for real-time and accurate measurement of ow rate in LOC applications. The sensing mechanism is based on the displacement of a cantilever due to uidic drag force, which changes the cavity length of a FP and therefore causes a shi in the interference fringe. We present a method for fabricating the sensor, which includes engraving a 2D pattern of cantilever/base on the PMMA (Plexiglas®) material using a CO2 laser and then casting it with the PDMS. This method does not need a complex and multilayer lithography process of SU-8 (ref. 20 and 21) and therefore can be suitable for mass production. Also, PDMS has a low elasticity modulus, low chemical reactivity and good biocompatibility and therefore the PDMS based optouidic ow sensor has a large dynamic range of operation, is highly resistant to corrosive chemical solutions and is suitable for biological applications such as ow cytometry and particle sorting, which strongly depend on the precise control of ow rates.22 In addition, the sensor has the ability to provide local and real-time feedback in complex lab-on-chip applications for ow rate control.
2 Principle of operation 2.1
The principle of Fabry–P´ erot (FP) operation
2D and 3D schematics of cantilever-based optouidic ow sensor are shown in Fig. 1. The sensor consists of a micro Fabry–P´ erot (FP) cavity which includes a cleaved end ber placed on the PDMS base and a vertical micro cantilever fabricated from PDMS. The dimensions of the sensor and the optical path in the ber and FP cavity are shown in Table 1. The ratio of cross section of cantilever to cross section of channel in the z–y plane is 15%. In Fig. 1, the thick and thin cantilevers are shown with a solid line and a dashed line, respectively. A super luminescent diode, which is coupled in the ber (dash-dot line) is partially reected (R1) at the interface of the ber and FP's cavity (dashed line). The rest of the source light enters the FP's cavity (dotted line) and aer passing through the cavity length reaches the copper face of the PDMS cantilever and is consequently reected back from the copper face of the PDMS cantilever. Then, a portion of it (R2) is coupled to the optical ber (solid line). The output light intensity R can be obtained in terms of incident intensity Ri as follows:18 pffiffiffiffiffiffiffiffiffiffiffi 4pnd (1) R ¼ R1 þ R2 2 R1 R2 cos þ 240 l0 where nd, l0 and f0 are the optical path difference (FP cavity length), the central wavelength of the SLD, and the initial phase difference, respectively. The values of R1 and R2 are the reected light at the ber–cavity interface and the coupled light in the ber as shown in Fig. 1. The maxima of the output light intensity in eqn (1) occurs when an integer number of halfwavelength ts in the FP's cavity length. The FP cavity length can be obtained from adjacent peak points in the reection spectrum by:23
432 | Analyst, 2014, 139, 431–438
Schematics of the 2D and 3D view of the optofluidic sensor.
Fig. 1
nd ¼
l1 l2 2ðl1 l2 Þ
(2)
where, d is cavity length and (l1 > l2) are the wavelengths of two adjacent peak points. The ow impinges on the cantilever causing a deection in the cantilever's tip which in turn changes the FP's cavity length resulting in a shi in the FP's interference spectrum. 2.2
The principle of cantilever–uid interaction
When the uid ows in the channel, pressure and viscous drag of the uid causes cantilever deection. In order to investigate the behavior of the uid and the cantilever deection in the channel, three dimensional Navier–Stokes and structural deformation simulations were carried out by a Finite Element Method (FEM) in a laminar ow conditions. The ow channel is 2 mm high and 1.3 mm wide. The uid in the channel is described by the incompressible Navier–Stokes and continuity equations as follows:24
Table 1
Dimensions of the sensor including thin and thick cantilevers
Dimensions Size (mm)
X1 150
X2 45
X3 500
X4 156
Y1 1000
Y2 125
Y3 800
Z 500
This journal is © The Royal Society of Chemistry 2014
View Article Online
Paper
Analyst
Published on 02 December 2013. Downloaded by University of Texas Libraries on 26/05/2014 15:53:43.
r
vu V$ pI þ h Vu þ ðVuÞT þ rðu$VÞ$u ¼ 0 vt V$u ¼ 0
(3)
In these equations, r, u, p, and h are uid density, velocity eld, pressure and dynamic viscosity, respectively. In addition, I and T denote unit diagonal matrix and transpose operator, respectively. The uid has laminar characteristics at the input with a parabolic velocity prole and has a zero pressure conditions at the output. Steady-state and non-slip boundary conditions are used in the Computational Fluid Dynamics (CFD) simulations. The structural deformation of the cantilever resulting from the moving uid is given by:25 FT ¼ n$(pI + h(Vu + (Vu)T))
(4)
where n and FT are the normal vector to the boundary and the uid loading, respectively. The rst term on the right-hand side of eqn (4) is the pressure gradient extracted from the uid simulation results. The second term is the viscous component of the force depending on the velocity and the dynamic viscosity of the uid. A parametric analysis was done by setting the input ow rate from 1 (mL h1) to 60 (mL h1) for the thin cantilever and 10– 300 (mL h1) for the thick one. The input ow is directed toward the positive values of the x-axis. The most important parameters for the cantilever (PDMS) and the uid (water) have been presented in Table 2. Fig. 2(a)–(d) demonstrates the velocity magnitude distribution and pressure magnitude in the x–y plane for thin and thick cantilevers at a ow rate of 10 (mL h 1). In the laminar ow regime, the velocity in the walls is zero due to the dominant viscous force and the ow has maximum velocity in the middle of the channel. Also, ow velocity is zero in the region close to the cantilever (stagnation points), as shown in Fig. 2. According to the Bernoulli equation, the static pressure is highest when the velocity is zero and hence static pressure is at its maximum value. In Fig. 2(b) and (d), maximum pressure is on the le side of the cantilever. In our geometry of the proposed sensor, the cantilever's tip is placed in the middle of the channel and experiences maximum drag force due to maximum velocity. Therefore, for low ow rates the cantilever's tip is deected (sensitive to the low ow rate). The maximum and minimum pressures are on the le and right of the cantilever's tip (free end) for thin and thick cantilevers. This difference pressure on the le and right sides leads
Table 2 Parameters for simulation of cantilever's behaviour in the channel (standard conditions of 20 C and 1 atm)
Parameters
PDMS (unit)
Water (unit)
Young's modulus Poisson's ratio Dynamic viscosity Density
200–1000 (kPa) 0.5 — 970 (kg m3)
— — 1.002 103 (Pa s) 1000 (kg m3)
This journal is © The Royal Society of Chemistry 2014
Fig. 2 Velocity (m s1) and pressure (Pa) distribution for the sensor with thin (a and b) and thick (c and d) cantilevers (in the x–y plane).
to deection of the cantilever to the positive x-axis. The numerical deection of a common beam can be obtained by the following equation. d ¼ FTL3(1 n2)/(3EI )
(5)
where E, n and I are the Young's modulus, Poisson's ratio and moment inertia, respectively. FT is obtained numerically by solving eqn (3) and (4). The Young's modulus of PDMS based structures depends on the geometry and fabrication process and its value has been reported to be between 200 and 1000 (kPa).21,26 Fig. 3 shows the deection of the cantilever's tip versus the ow rate with various values of Young's modulus of PDMS (E) (between 200 and 1000 (kPa) with 50 interval (kPa) steps) for thin and thick cantilevers. Both gures show that by increasing Analyst, 2014, 139, 431–438 | 433
View Article Online
Published on 02 December 2013. Downloaded by University of Texas Libraries on 26/05/2014 15:53:43.
Analyst
Paper
the ow rate, the deection of the cantilever's tip increases due to the increase in drag force of ow rate. In addition, with decreasing (increasing) Young's modulus of PDMS the deection of the cantilever's tip increases (decreases), respectively. Minimum and maximum deections are 3.4 and 1341 (nm) for the thin cantilever at ow rates of 1 and 60 (mL h1), and 1.9, and 1318 (nm) for the thick cantilever at ow rates of 60 and 300 (mL h1), respectively.
3 The fabrication process of the sensor 3.1
Fabrication of the cantilever/base
Fig. 4(a)–(d) shows a schematic view of the whole process. First, a 2D pattern of the cantilever/base was designed and drawn using AutoCAD soware (Fig. 4a). Using a computer-controlled CO2 laser (SYNRAD 48-series) the 2D pattern of the cantilever/ base was engraved on a PMMA slab. Fig. 4(b) shows the cross section of the cantilever/base on the PMMA slab. A mixture with a ratio of 10 : 1 (base : hardener) of PDMS (Sylgard 184 Dow Corning) was prepared and aer degasication, was poured (casting) on the engraved PMMA slab (Fig. 4c). Then, the cast PDMS was placed in an oven for 30 minutes at 90 C. Aer being baked, the 3D pattern of the cantilever/base was obtained by peeling off the PDMS from the engraved PMMA slab. The height
Fig. 4 Fabrication process (a) top view of CAD pattern, (b) cross section of engraved PMMA slab, (c) casting of PDMS, (d) 3D pattern of cantilever/base by peeling PDMS from the engraved PMMA slab, (e) cross section of a Gaussian-like engraved profile on the two clamped PMMA slabs using a CO2 laser.
of the base is lower than the height of the cantilever since the optical ber should be placed on it in order to be aligned with the cantilever's tip. To enhance the surface reectivity of the cantilever, it was coated with copper (Cu) with a thickness of 80 nm by the Physical Vapour Deposition (PVD) method. In the Cu-coating process a mask was placed on the 3D pattern of the cantilever/base. This mask covered all regions on the 3D pattern except the cantilever. 3.1.1 Fabrication of the thin and thick cantilever. We used two PMMA slabs with 1 mm thickness, these two slabs were in contact and clamped tightly with two clamps. The 2D pattern of the cantilever was engraved using the CO2 laser on two slabs by moving the XY motorized stage (Thorlabs BSC 103). Fig. 4(e) shows the cross-section of the cantilever that was taken using a microscope equipped with a CCD camera. As can be seen, the engraved prole is Gaussian-like in its cross section.27–29 By separating the clamp we now have two slabs with two different Gaussian-like patterns. The upper slab has a larger engraved cross section compared to that of the lower slab. The dimensions of the engraved pattern depend on the power and the spot size of the CO2 laser and also on the speed of the XY motorized stage. As a result, we were able to fabricate an engraved prole with thin and thick cross sections. Aer engraving the cantilever's pattern, the pattern of the base was engraved separately on each slab using the CO2 laser. Aer fabricating the cantilever/base, the optical ber was dipped in the PDMS gel (except for its tip) and then placed horizontally on the base. During the placement of the ber on the base, we observed the interferometric spectrum [Fig. 7a] and using eqn (2), the cavity length was calculated. This calculation was implemented in real time using a Labview program. The required value of the X4 (Fig. 1) is obtained by moving the ber in the X-direction. The movement of the ber is controlled by the motorized stage with 100 nm resolution. Aer the required cavity length (X4) is obtained, the ber and PDMS cantilever/base were placed in the oven until the ber is secured to the base. Fig. 5 shows a lateral photo of the sensor with both cantilevers.
3.2 Fig. 3 Numerical deflection of the cantilever's tip for (a) thin cantilever and (b) thick cantilever for Young's modulus (E ¼ 200–1000).
434 | Analyst, 2014, 139, 431–438
Fabrication of the channel
A slab of PMMA was cut using the CO2 laser with a height, width and length of 2 mm, 1.3 mm and 1 cm, respectively. The slab
This journal is © The Royal Society of Chemistry 2014
View Article Online
Paper
Analyst
Published on 02 December 2013. Downloaded by University of Texas Libraries on 26/05/2014 15:53:43.
was placed on a cleaned SiO2 substrate, then the mixture of PDMS (with a ratio of 10 : 1 aer degasication) was poured on it and aer baking in the oven, the channel was obtained by peeling off the baked PDMS. The uid inlet and outlet were punched on the channel and two glass tubes were connected to the inlet and outlet. 3.3
Final packaging of the optouidic sensor
The 3D pattern of cantilever/base, the channel and a cleaned glass substrate were placed separately in the plasma oxygen in order to activate their surfaces. Then, the backside of the 3D pattern of cantilever/base was bonded to the cleaned SiO2 substrate. In the next stage, using a microscope, the activated surface of the channel was placed and bounded on the front face of cantilever/base. 3.4
Fig. 6
Experimental setup.
Experimental setup
The experimental setup is shown in Fig. 6. The sensor was connected to the SLD (Thorlabs S5FC1005SXL) using a singlemode optical ber and a circulator. The ow entered the inlet of the channel via a syringe pump (New Era Pump NE-4000). By injecting various ow rates, the drag force of the uid deects the cantilever's tip and therefore changes the cavity length. This would cause a shi in the interference fringes. In order to measure and analyze the fringe shi, the light reected from the deection of the cantilever's tip is sent through the circulator to an optical spectrum analyzer (OSA) (Agilent 86143B) with a resolution of 10 pm through the circulator.
Fig. 5 Lateral photograph of the sensor with (a) thin and (b) thick cantilevers.
This journal is © The Royal Society of Chemistry 2014
4 Results 4.1
Sensitivity and minimum detectable ow change
Two sensors including thin and thick cantilevers were characterized at ow rates of 1–60 (mL h1) and 10–300 (mL h1), respectively. At rst the channel was lled with DI water (with zero ow rate) and the interference spectrum was recorded as a reference spectrum. Fig. 7(a) shows the Gaussian shaped SLD spectrum, which is modulated by the interference pattern FP at 100 (mL h1) ow rate. According to eqn (2) the length of the FP cavity at a ow rate of zero was found to be 156 mm with n set to be 1.31511 for water at the wavelength of 1550 nm. The drag force of the input ow deects the free end (cantilever's tip) of the cantilever and therefore shis the fringes. The interference spectra of the thick-cantilever sensor versus ow rate are shown in Fig. 7(b). By increasing the ow rates, the cantilever moves more toward the optical ber so that the cavity length decreases. Consequently, the interference fringes shi to smaller wavelengths according to eqn (2). The four peaks of interference spectrum in Fig. 7(b) are shown for the thick cantilever at ow rates of 100, 120, 140 and 180 (mL h1). The calibration curves of the sensors are shown in Fig. 8. The fringe shi (Dl) is obtained by subtracting the initial wavelength at zero ow rate (the reference spectrum) from the same wavelength at a ow rate of 1–60 and 10–300 (mL h1) for thin and thick cantilevers, respectively. The sensor with a thin cantilever has a linear sensitivity equal to 0.12 [nm mL1 h1]. Unlike the thin cantilever, the thick one has a polynomial tting curve. With a good estimation, the sensitivity of the thick cantilever has two linear regions (I) and (II) with sensitivities equal to 0.01 and 0.04 [nm mL1 h1] at ow rates of 10–120 (region I) and 120–300 (region II) respectively. The thick cantilever has less sensitivity in comparison with the thin cantilever. Since the spectral resolution of the OSA was 10 (pm) the minimum detectable ow change (resolution) is 1.3 (mL min1) for the thin cantilever and 16 (mL min1) (region I) and 4 (mL min1) (region II) for the thick one. This minimum detectable ow change for the thin cantilever is slightly better than that of ref. 19 and 21,
Analyst, 2014, 139, 431–438 | 435
View Article Online
Published on 02 December 2013. Downloaded by University of Texas Libraries on 26/05/2014 15:53:43.
Analyst
Paper
Fig. 7 (a) Interference fringes at 100 (mL h1) flow rate and (b) interference fringe shifts for one peak at input flow rates of 100, 120, 140, 180.
1
1
which were 6.1 (mL min ) and 35 (mL min ), respectively. The sensitivity and resolution of the two optouidic ow sensors are presented in Table 3. Using a commercial soware such as LabVIEW (National Instruments) we are able to monitor the calibration curve of the two sensors in real-time and control the ow rate in our optouidic sensor with a resolution of 1.3 (mL min1) by applying feedback to the syringe pump. This capability is suitable for local and real-time feedback in complex lab-on-chip applications such as ow cytometry, particle sorting, separation, liquid droplet-based system, mixing, and chemical synthesis.
4.2
Effective Young's modulus of two cantilevers
As mentioned in Section 2.2, the Young's modulus (E) of PDMS based structures depends on the geometry and fabrication process. According to our knowledge, this is the rst experimental and numerical study of the mechanical deformation for obtaining the effective Young's modulus of PDMS based cantilevers with a conical shape. We obtained the cantilever deection experimentally and compared it with numerical
436 | Analyst, 2014, 139, 431–438
Fig. 8 The calibration curves of two sensors in (a) thin cantilever and (b) thick cantilever.
calculation of deection. The experimental deection is calculated by subtracting the initial cavity length at zero ow rate from the cavity length at 1–60 and 10–300 ow rates (mL h1) using eqn (2). Fig. 9 shows the comparison of numerical and experimental cantilever deections for thin and thick cantilevers. The experimental and numerical deections have similar behaviour. In Fig. 9(a) the experimental deection of the thin cantilever's tip is located between numerical deections with E ¼ 400 and E ¼ 450. For the thick cantilever, the experimental deection is located between numerical deections with E ¼ 500 and E ¼ 550. As can be seen in both gures, the experimental deection due to its uctuation does not t a specic numerical deection (single curve) and is located between the two Young's modulus values. Table 3
Sensitivity and resolution of the two optofluidic flow sensors
Sensor type
Thin cantilever
Thick cantilever
Flow rate range (mL h1) Sensitivity [nm mL1 h1] Resolution (mL min1)
1–60 0.12 1.3
10–120 0.01 16
120–300 0.04 4
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 02 December 2013. Downloaded by University of Texas Libraries on 26/05/2014 15:53:43.
Paper
Analyst
detectable ow change of 1.3 (mL min1) (thin cantilever) at ow rates 1–60 (mL h1) and 16 and 4 (mL min1) (thick cantilever) for ow rates 10–120 and 120–300 (mL h1), respectively. Using commercial soware such as LabVIEW, we are able to monitor and control the ow rate in our optouidic sensor with a resolution of 1.3 (mL min1). Therefore, the sensor provides local and real-time feedback in complex lab-on-chip applications. The effective Young's modulus of PDMS, which depends on the cantilever's geometry and fabrication process, is obtained by comparison between the numerical and experimental deection of the cantilever's tip. The effective Young's modulus has a value between 400 and 450 for the thin cantilever and 500–550 for the thick cantilever.
Acknowledgements The authors gratefully acknowledge the nancial support provided to this study by the Microelectronic Technology Development Council, Iran (http://micro.isti.ir).
References
Comparison of numerical and experimental cantilever tip deflection in (a) thin and (b) thick cantilevers.
Fig. 9
Conclusion An experimental and numerical investigation of a cantilever-based optouidic ow sensor for real-time and accurate measurement of ow rate in lab on a chip and micro total analysis system has been presented. Two sensors including thin and thick cantilevers were fabricated by engraving a cantilever/base pattern on two PMMA slabs using a CO2 laser and casting the pattern with PDMS. The deection of the cantilever due to the drag force of uid was detected by the shi of Fabry–P´ erot spectra. The ow causes a small deection of the cantilever, which changes the cavity length of the FP, which in turn results in fringe shi. A Finite Element Method (FEM) was used for simulation of the pressure and velocity distribution across the cantilever resulting from ow in the channel. In addition, the deection of the cantilever's tip was calculated experimentally and numerically. The experimental results demonstrate that the sensor with a thin cantilever has 0.12 [nm mL1 h1] linear sensitivity at ow rates of 1–60 (mL h1). The thick cantilever has two linear sensitivity regions equal to 0.01 and 0.04 [nm mL1 h1] at ow rates of 10–120 (mL h1) and 120– 300 (mL h1), respectively. These sensors have a minimum
This journal is © The Royal Society of Chemistry 2014
1 S. Haeberle and R. Zengerle, Lab Chip, 2007, 7, 1094–1110. 2 J. West, M. Becker, S. Tombrink and A. Manz, Anal. Chem., 2008, 80, 4403–4419. 3 G. Guan, L. Wu, A. A. Bhagat, Z. Li, P. C. Y. Chen, S. Chao, C. J. Ong and J. Han, Sci. Rep., 2013, 3, 1475. 4 B. K. McKenna, J. G. Evans, M. C. Cheung and D. J. Ehrlich, Nat. Methods, 2011, 8, 401–403. 5 Y. C. Tan, J. S. Fisher, A. I. Lee, V. Cristini and A. P. Lee, Lab Chip, 2004, 4, 292–298. 6 V. Hessel, H. L¨ owe and F. Sch¨ onfeld, Chem. Eng. Sci., 2005, 60, 2479–2501. 7 A. G¨ unther and K. F. Jensen, Lab Chip, 2006, 6, 1487–1503. 8 N. Crews, C. Wittwer and B. Gale, Biomed. Microdevices, 2008, 10, 187–195. 9 P. Caldas, P. A. Jorge, G. Rego, O. Fraz˜ ao, J. L. Santos, L. A. Ferreira and F. Ara´ ujo, Appl. Opt., 2011, 50, 2738–2743. 10 R. J. Rodrigues and R. Furlan, Microelectron. J., 2033, 34, 709– 711. 11 T. H. Kim and S. J. Kim, J. Micromech. Microeng., 2006, 16, 2502. 12 R. Ahrens and K. Schlote-Holubek, J. Micromech. Microeng., 2009, 19, 074006. 13 M. Dijkstra, M. J. de Boer, J. W. Berenschot, T. S. J. Lammerink, R. J. Wiegerink and M. Elwenspoek, Sens. Actuators, A, 2008, 143, 1–6. 14 S. Kim, T. Nam and S. Park, Sens. Actuators, A, 2004, 114, 312–318. 15 A. Cipullo, G. Gruca, K. Heeck, F. De Filippis, D. Iannuzzi, A. Minardo and L. Zeni, Sens. Actuators, A, 2012, 178, 17–25. 16 R. Attia, D. C. Pregibon, P. S. Doyle, J. L. Viovy and D. Bartolo, Lab Chip, 2009, 9, 1213–1218. 17 W. Song and D. Psaltis, Opt. Lett., 2010, 35, 3604–3606. 18 M. S. Cheri, H. Lati, F. Aghbolagh, O. R. Naeini, M. Taghavi and M. Ghaderi, Appl. Opt., 2013, 52, 3420–3427.
Analyst, 2014, 139, 431–438 | 437
View Article Online
Published on 02 December 2013. Downloaded by University of Texas Libraries on 26/05/2014 15:53:43.
Analyst
19 V. Lien and F. Vollmer, Lab Chip, 2007, 7, 1352– 1356. 20 N. Noeth, S. S. Keller and A. Boisen, J. Micromech. Microeng., 2011, 21, 015007. 21 A. Sanati Nezhad, M. Ghanbari, C. G. Agudelo, M. Packirisamy, R. B. Bhat and A. Geitmann, IEEE Sens. J., 2013, 13, 601–609. 22 S. K. Sia and G. M. Whitesides, Electrophoresis, 2003, 24, 3563–3576. 23 G. C. Hill, R. Melamud, F. E. Declercq, A. A. Davenport, I. H. Chan, P. G. Hartwell and B. L. Pruitt, Sens. Actuators, A, 2007, 138, 52–62.
438 | Analyst, 2014, 139, 431–438
Paper
24 K. R. Stein, R. J. Benney, T. E. Tezduyar, J. W. Leonard and M. L. Accorsi, J. Aircr., 2001, 38, 800–808. 25 S. Basak, A. Raman and S. V. Garimella, J. Appl. Phys., 2006, 99, 114906. 26 M. Liu, J. Sun, Y. Sun, C. Bock and Q. Chen, J. Micromech. Microeng., 2009, 19, 035028. 27 H. Klank, J. P. Kutter and O. Geschke, Lab Chip, 2002, 2, 242– 246. 28 T. F. Hong, W. J. Ju, M. C. Wu, C. H. Tai, C. H. Tsai and L. M. Fu, Microuid. Nanouid., 2010, 9, 1125–1133. 29 M. F. Jensen, M. Noerholm, L. H. Christensen and O. Geschke, Lab Chip, 2003, 3, 302–307.
This journal is © The Royal Society of Chemistry 2014