Accepted Manuscript Title: Contactless Sensing of the Conductivity of Aqueous Droplets in Segmented Flow Authors: Brian P. Cahill, Raul Land, Thomas Nacke, Mart Min, Dieter Beckmann PII: DOI: Reference: S0925-4005(11)00636-8 doi:10.1016/j.snb.2011.07.006 SNB 13259 To appear in: Sensors and Actuators B Received date: Revised date: Accepted date: 12-4-2011 1-7-2011 4-7-2011 Please cite this article as: B.P. Cahill, R. Land, T. Nacke, M. Min, D. Beckmann, Contactless Sensing of the Conductivity of Aqueous Droplets in Segmented Flow, Sensors and Actuators B: Chemical (2010), doi:10.1016/j.snb.2011.07.006 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. *Manuscript Contactless Sensing of the Conductivity of Aqueous Droplets in Segmented Flow Brian P. Cahilla,*, Raul Landa,b, Thomas Nackea, Mart Mina,b, Dieter Beckmanna Institute of Bioprocessing and Analytical Measurement Techniques, Heilbad Heiligenstadt, Germany b Chair of Electronics Measurement, Tallinn University of Technology, Tallinn, Estonia * Correspondence: Institute of Bioprocessing and Analytical Measurement Techniques, Rosenhof, 37308 Heilbad Heiligenstadt, Germany brian.cahill@iba-heiligenstadt.de us cr ip t a Abstract an We present an electrode arrangement for the inline measurement of the conductivity of droplets in segmented flow by impedance spectroscopy. We use a thin-walled glass capillary with electrodes contacting the outer M surface, so that the contactless measurement of conductivity of the liquid within the capillary is possible. The surface of the glass capillary is silanized resulting in a single hydrophobic surface across which droplets can ed freely move. We model the impedance of such insulated electrodes and use the model to optimize the electrode system. Measurement of solutions with various salt concentrations allows the performance of the electrode structure to be characterized. Subsequently, the measurement of the impedance response of the aqueous ep t segments in two-phase flow was demonstrated. Measurements were firstly performed with an impedance analyzer and subsequently with a multi-sine measurement setup that is better suited to high-speed measurement Ac c of droplets. Previous electrical measurements of segmented flow sensed the difference in dielectric constant between the aqueous phase and the carrier fluid through measurement of capacitance. This work describes an electrical measurement of the conductivity of droplets in segmented flow, that is, the sensor senses a variable property of the droplet itself. Page 1 of 24 1. Introduction Digital microfluidics is a technique based on the generation, handling, and measurement of droplets. The standard droplet generation hardware generates droplets by injecting an aqueous solution into an immiscible oil at a T-junction. The formation of droplets has emerged as a widespread microfluidic tool in the last ten years [112] and, in particular, the application of the technology to cell cultivation [5-10]. us cr ip t Optical measurement has been the most prominent technique for the analysis of droplet content [12-14]. In many cases optical measurements require the addition of fluorescent markers and other molecules that act as indicators. Electrical measurements can often be performed without any need for modification of the object under test. The first electrical measurements of droplets that have emerged have aimed to measure the presence/absence of a drop or its position [15-19]. Chen et al. [15] and Srivastava and Burns [16] used capacitive sensing of droplet position to automate further operations in a microfluidic system and to measure droplet size. an Niu et al. [17] performed inline measurements of the capacitance differences of droplets flowing through a microfabricated chip and subsequently could direct the flow of droplets through a switching mechanism. Nichols M et al. [18] sensed the position of droplets in an electrowetting chip in order to direct the flow of droplets. The capacitive measurement of droplets senses changes in the dielectric constant of the liquid between the ed measurement electrodes; in short, these measurements sense the presence/absence of water between the measurement electrodes. These measurements do not measure the conductivity of the liquid and are therefore ep t less suited to full characterization of the liquid within the droplet. In this paper, we present measurements of the conductivity of aqueous droplets using a galvanically-decoupled electrode system. In most applications of impedance spectroscopy, the electrodes make galvanic contact to the liquid that is to Ac c be measured [20-25]. Direct contact between electrodes and the fluid medium greatly improves measurement sensitivity because the impedance between the electrode and the fluid is low. Insulation introduces a capacitance which greatly increases the electrode impedance. If we consider the practicalities of a droplet-based microfluidic system, the droplet must be transported across the surface of the electrode without breaking up. For this reason, it is convenient that the surface of the electrode be continuous and hydrophobic. To achieve this aim, we choose to use electrodes attached to the outer surface of thin-walled glass capillaries; the surface of glass can be readily silanized with hydrophobic alkysilanes as has been commonly used in other droplet-based microfluidic systems [6-10,26,27]. The measurement setup presented here relies on electrodes that are electrically insulated from the microfluidic system. The insulating layer prevents galvanic contact between the electrodes and the fluid being measured and although this complicates impedance measurement, it is not without some advantages. As the thickness of the insulation is much greater than the thickness of the electric double layer that forms at the Page 2 of 24 surface, any effects resulting from the presence of the electric double-layer can be neglected. Hofmann et al. [28] presented an inline biosensor that measured the growth of yeast cells through contactless measurement by impedance spectroscopy. Olthuis et al. described the measurement of conductivity using interdigitated electrode structures, at first with uninsulated electrodes [29,30] and subsequently extending their work to electrodes insulated with a thin film of tantalum pentoxide [31]. t Contactless measurement of conductivity has gained widespread application in electrophoresis [32-45]. us cr ip Although the design of the sensor presented in this paper is very similar to the original contactless conductivity sensors for electrophoretic applications [32,33], in electrophoresis the variation of the ionic concentration is measured as a function of time without the absolute value of the conductivity being of any great importance. We present a sensor that measures the conductivity of the liquid in the capillary. As will be presented below, the sensor measures the impedance spectrum over a range of frequencies and not just a single measurement at a an fixed frequency. This facilitates the measurement of the conductivity over a wide range. As the thickness of the capillary walls is much lower than that in capillary electrophoresis, the measurement presented here can be M performed at lower excitation amplitudes. This point is important for reducing the complexity and cost of the electronics of any dedicated electronic circuits for performing such measurements. This work aims to lead to the over a wide range of frequencies. Impedance Model ep t 2. ed measurement of the cell content of droplet-based bioreactors, for which it is advantageous to measure impedance Fig. 1a shows an electrode system based on a thin-walled glass capillary. The system consists of a tube with two electrodes contacting the outside of the tube and with a grounded guard electrode shielding the two Ac c electrodes from each other. The general equivalent circuit describing this electrode system is shown in Fig. 1b. Analysis of the equivalent circuit shows how the sensor can be optimized. Cw the capacitance of the insulating walls is given by Cw = wA/dw, where w is the dielectric constant of the wall, A is the area of the electrode and dw is the thickness of the wall. Cp is the parallel capacitance resulting from parasitic capacitance between the electrodes and associated cabling and from the capacitance associated with the measurement system. The quantity that is to be measured is the impedance of the fluid in the capillary. This is determined by the geometry of the sensor and the conductivity and dielectric constant of the fluid and is described by the medium resistance Rm and medium capacitance Cm. Electrode polarization is a topic of much significance for electrodes with direct contact to the measurement fluid and the comparison between two-electrode and four-electrode measurement has been discussed extensively [20,21,35,36]. Although the insulated electrodes were chosen in order to facilitate improved droplet transit Page 3 of 24 through the measurement system, a useful consequence of insulation is the fact that interference due to electrode polarization is minimized. The series addition of a capacitor corresponding to insulation with µm-scale thickness and the electric double layer with nm-scale thickness allows the formation of the electric double layer to be completely neglected. The total impedance of the equivalent circuit in Fig. 1b is given by: 1  jRm C w  C m  C m C w  C p C w  C p C m 1  jRm C w  C p (1) t 1 j C w  C p  us cr ip Z    where C w = C w 2 . Equation 1 is used to generate the graphs shown in Fig. 2. If the thickness of the capillary wall is relatively thin, then the capacitance C w will be much larger than Cp and Cm, so that equation 1 can be Z    1  jRm C w 1 1 1  j 1  jC w 1  jRm C p  C m  jC w 1  j 2 an simplified to: (2) M where 1 and 2 are relaxation times given by Rm C w and Rm (Cp + Cm) respectively. By evaluating equation 2 in the form of a resistance Rs and capacitance Cs in series, Z = Re(Z) + j Im(Z) = Rs + 1/jCs, we obtain variables ed that are easily calculated from the real and imaginary parts of any measured impedance value and can help our understanding of the general form of the curves in Fig. 2: 1   2 22 1   2 1 2 Ac c Cs    Cw 1   2 22 1 ep t Rs    Rm (3) (4) Furthermore, Rs and Cs define a general time constant s:  s  Rs Cs  1 1   2 1 2 (5) In order to describe the behavior of the system, Fig. 2 models the equivalent circuit in Fig. 1b for the following sample values: Rm = 1 M, Cw = 10 pF, Cp = 0.070 pF and Cm = 0.070 pF. Fig. 2a shows the impedance modulus values and phase difference as a function of frequency f. Fig. 2b shows the Rs/Rm, Cs/ C w and s as a function of frequency. At frequencies less than 1/21, the impedance modulus decreases linearly with increasing frequency and the phase modulus is -90° and the imaginary component of the impedance is approximately that of the capacitance of the capillary wall, 1/j C w . In the mid-frequency range, the impedance Page 4 of 24 modulus has a value close to that of Rm, this means that in the mid-frequency range the resistance of the fluid in the capillary is the dominant element of the overall impedance. In the mid-frequency range, C begins to decrease in value. At frequencies greater than 1/22, the impedance modulus decreases linearly with increasing frequency and the phase modulus is -90° and the imaginary component of the impedance is approximately 1/j(Cp + Cm) and the real component. us cr ip t At frequencies much smaller than 1/2 2, Rs is equal to Rm. In practice if 1/j C w is much greater than Rm, such as at frequencies much lower than 1/21, it is not possible to measure Rm accurately because of bit noise. Rm begins to decrease in value at a frequency of 1/22. At frequencies much smaller than 1/2 1, Cs is equal to C w . At frequencies much larger than 1/22, Cs is equal to Cp + Cm. At the frequencies corresponding to 1 = 1 and 2 = 1, s is approximately 1. This is conditional on C w being much larger than Cp and Cm. This allows these frequencies to be conveniently measured using the general time constant  s. The proportionality of Rm and Rm  K1 f1  1 an the frequency f1 at which 1 = 1 to conductivity is given by:  K 2 (6) (7) ed 2 1 M  where K1 and K2 are cell constants that depend on the geometry and material properties of the electrode system. conductivity. Materials and Methods Ac c 3. ep t By measuring impedance as a function of frequency, we have two measurements that are related to medium 3.1 Design of Flow Sensor In order to construct a galvanically-decoupled through-flow sensor system that can dependably measure impedance, it is necessary to have relatively high capacitance electrodes. This means the thickness of the electrode insulation must be as thin as possible. Other objectives of the sensor design are to construct a measurement system that is relatively inexpensive and robust and enables inline measurements to be made on water droplets in segmented flow. We present the design of a thin-walled glass capillary with electrodes connected to the outside walls. The glass capillary can be connected directly to Teflon tubing so that droplets do not break up at the connection. Glass capillaries with nominal wall thickness of 10 µm with 500 µm inner diameter are commercially available for use in X-ray crystallography (W. Müller, Schönwalde, Germany). The walls of these capillaries Page 5 of 24 serve as the insulation of the electrodes in this measurement system. Electrode rings made out of Elastosil LR 3162 conductive silicone (Wacker AG, Burghausen, Germany) were fabricated that could slide across the surface of the glass capillary making conformal contact to the capillary surface. The conductive silicone has a resistivity of 11 cm, which is nearly twice as conductive as sea water (20 cm). The electrode rings of 4mm length were molded around a metal wire of 0.45-mm outer diameter that was removed once the silicone t polymerized leaving a through-hole that later accommodated the glass capillary. Electrical contact to the us cr ip conductive silicone was achieved by inserting another metal wire into the silicone and then curing the silicone at 165° C for 10 minutes. The end of this wire was wound into a coil to ensure that it could not be easily removed from the solidified conductive silicone. A thin copper sheet was used as shielding between the two electrodes and served to reduce parasitic capacitance in the measurement system. This type of shielding has previously been used in contactless conductivity sensors in electrophoresis [34,44,45]. an The glass capillaries were coated with a monolayer of octadecyltrichlorosilane (ODTS) molecules (ABCR, Karlsruhe, Germany). The surface of the capillary was activated using piranha solution: a freshly prepared 1:1 M mixture of sulfuric acid (p.a., Fluka, Buchs, Switzerland) and hydrogen peroxide (30%, Merck, Darmstadt, Germany). The hot piranha solution is flushed through the capillaries for about 10 minutes and afterwards are ed rinsed with pure water and ethanol (p.a., Baker, Griesheim, Germany). The capillaries were dried with ultraclean compressed air and subsequently heated for at least 2 hours at 80 °C. After allowing the capillaries to cool down to room temperature in a desiccator, they were immersed in an octadecyltrichlorosilane solution (1% ep t ODTS in heptane) and stored in the dark for at least 12 hours. Then, the capillaries are rinsed with ethanol and dried with ultraclean compressed air. After this treatment, the surface of the capillaries is sufficiently hydrophic Ac c to allow droplets to traverse the capillary without breaking up. 3.2 Impedance Measurement Impedance measurements were performed using either an HP 4194A Impedance/Gain-Phase Analyzer or an IMPSPEC system (MEODAT, Ilmenau, Germany). The HP 4194A Impedance/Gain-Phase Analyzer is a conventional impedance measurement device that scans the wavelength spectrally, that is, it sequentially applies a single excitation signal at a particular frequency until the preprogrammed frequency scan is completed. It is a very powerful measurement instrument which is very precise but it is not particularly suited for integration into a process control setup where speed of measurement and cost become important factors. The IMPSPEC system is specifically designed for integration in process control systems [46-47] and consists of a 14-bit Digital to Analog Converter (DAC) for signal generation and 14-bit Analog to Digital Converters (ADC) for signal acquisition. The IMPSPEC system used for these measurements had a bandwidth between DC and 20 MHz. The DAC Page 6 of 24 allows the generation of multi-sine signals so that several excitation frequencies can be applied simultaneously. The two ADCs allow the acquisition of signals. A LabView program was written to control data acquisition from the IMPSPEC system. The IMPSPEC system contains a fully programmable gate array (FPGA) and a digital signal processor (DSP). A DHPCA-100 Variable-Gain High Speed Current amplifier (FEMTO Messtechnik GmbH, Berlin, Germany) was used to amplify the signals measured by the IMPSPEC system. Signals of 1 V t amplitude were applied for measurements with both the HP 4194A Impedance/Gain-Phase Analyzer and the us cr ip IMPSPEC system. The low amplitude necessary for measurement facilitates impedance measurement and compares favourably with the relatively high amplitudes necessary for contactless conductivity measurement in electrophoresis. Saline solutions were prepared from ultrapure water and potassium chloride salt. Fig. 3 shows how a syringe pump was used to generate segmented flow by means of a Teflon T-junction with 1-mm inner diameter an (Upchurch Scientific, Oak Harbor, USA) with two perpendicularly-pumped input streams, tetradecane and saline solution. The Teflon tubing had an inner diameter of 0.5 mm and 1.6 mm outer diameter. The tetradecane input M was pumped at a flow rate five times greater than the saline solution input. Teflon tubing was connected to the output and input of the glass capillary by first attaching a short length of heat-shrink tubing (1.6-mm shrunken ed outer diameter) to each end of the capillary and then using silicone tubing to join the capillary to the Teflon tubing. The system was then stabilized mechanically by encasing it in polydimethylsiloxane (PDMS) Sylgard 4. ep t 184 (Dow Corning, Midland, USA). Results and Discussions Ac c 4.1 Ionic Strength Series Each electrode structure was first characterized by impedance measurement of a series of solutions with ionic strengths between 0.39 mM KCl and 200 mM KCl at frequencies between 100 Hz and 20 MHz. This permits the clear evaluation of the electrodes with regard to their effectiveness as a conductivity sensor. Fig. 4 shows the impedance characterization of an ionic strength series for the thin-walled capillary electrode structure with full and dashed lines represent data measured by the HP Impedance Analyzer and IMPSPEC system respectively. The form of the graphs shows a distinct resemblance to that of Fig. 2b. Fig. 4a shows the variation of Rs as a function of frequency and ionic strength. At low frequencies the impedance of the capillary wall is too high for the Rs to be measured reliably. As frequency increases, Rs attains a relatively constant plateau value that is equal to Rm. At higher frequencies, Rs declines at first slowly and then more rapidly. As the ionic strength increases, the plateau value of Rs decreases and the frequency at which the peak value is observed also increases. Page 7 of 24 Fig. 4b shows the variation of Cs as a function of frequency and ionic strength. At low frequencies, Cs is relatively constant and equal to C w . As frequency increases, C w declines until it reaches a constant value at higher frequencies. The data obtained from the HP 4194A Impedance Analyzer and the ImpSpec system are quite similar in the measurement range corresponding to the Rs plateau region but there is deviation from the model described in us cr ip t Section 2 at higher frequencies that is clear for the ImpSpec system at frequencies above 3 MHz and for the HP Analyzer at frequencies above 10 MHz. These deviations are caused by irregularities in the transfer function of the amplifiers. Falkenhagen et al. [48] further developed the theory of Debye-Hückel-Onsager to predict the molar conductivity of strong electrolytes at relatively high concentrations. This work was slightly adapted and strength c in mol/L:   B2 c  B1 c 1     Λ0  F c  1  Ba c  1  Ba c  where the coefficients B, B1, B2 and F are given by: 50.29  1011 B ed  rT  2 B1   rT  2 3 8.25  10 -4   rT   1 2  Ac c B2  8.20  105 ep t 1 F (8) M   an clarified by Wishaw and Stokes [49] and can be used to calculate the molar conductivity on the basis of the ionic exp 0.2929 Ba c  1 0.2929 Ba c (9) (10) (11) (12) where the molar conductivity at infinite dilution 0 is 0.01499 Sm2/mol for KCl, the ionic diameter a is 0.376 × 10-9 m for KCl, the relative dielectric constant of the solvent r is 78 and the viscosity of the solvent is 8.9 × 10-4 kg/m/s at an absolute temperature T of 298 K. Fig. 5a normalizes the data measured by the HP Impedance Analyzer shown in Fig. 4. The frequency was normalized by division by the conductivity (this axis is equal to K2), Rs is normalized by multiplication by the conductivity (this axis is equal to K1) and s. For large frequency ranges, the data for Rs and s collapse to a single curve solely by normalization with a single variable, the conductivity. This coincidence of the Page 8 of 24 normalization is most close for data close to 1 where s = 1, which also coincides with the plateau value of Rs where it is equal to Rm. This point is marked by dashed lines that also indicate the values of the cell constants at this fixed point. At this fixed point the data depend solely on Rm and C w and other equivalent circuit elements play a negligible role. Cell constants K1 and K2 were obtained from calibration using data for the 12.5 mM sample and are given in us cr ip t Table 1. All other molar conductivities were calculated using these constants using equations 6 and 7. Fig. 5b displays the molar conductivity as a function of ionic strength for all the curves shown in Fig. 4, while the solid line is from equation 8 and depends on the ionic strength and the variables given in equations 9-12. The data at both extremes of ionic strength deviate somewhat from the predictions of equation 8. At the two lowest concentrations the datapoints are higher than the theoretical values. This may be because any contamination of the samples is more significant at lower concentrations especially the absorption of carbon dioxide from the an atmosphere. The results obtained from sweeping a single signal in the frequency domain with the HP Impedance Analyzer are clearly closer to the predictions of equation 8 than using multi-sine excitation with the IMPSPEC frequencies above 3 MHz seen in Fig. 4b. M system. The IMPSPEC system is clearly less accurate at higher conductivities which relates to the deviations at ed The only equivalent circuit elements that the measurement of the conductivity depends on are Rm and C w . These can be estimated with the values from the geometry of the electrode system and compared with the Rm  K1  Cw 1  r  0 AC  2 2 dC Ac c Cw  LR AR ep t experimental values obtained from Figure 4b for Rm and Figure 5a for C w by means of the following equations: (13) (14) where LR is the equivalent length of the resistor, AR is the cross-sectional area of the capillary, r is the relative dielectric constant, 0 is the permittivity of free space, AC is the area of the capacitor and dC is the thickness of the capacitor. These values presume that their geometries correspond to that of a conducting bar and a parallel plate capacitor. This is only roughly true but nevertheless the model values can be compared with experimental values to confirm that the results are of the same order of magnitude. As the value of K1 obtained from Figure 5a is given in Table 1 and AR is a circle of radius 0.25 mm, LR is approximately 7.5 mm. This corresponds to less than the total length of the two electrodes even without taking the electrode separation into account. For dC of 10 µm, r of 2 and AC = 2  r L (r = 0.25 mm, L = 4 mm), C w is approximately 5.5 pF which is comparable to Page 9 of 24 the experimental value acquired from Fig. 4b of approximately 4.5 pF. It is likely that the effective length of the capacitor L is less than the full length of the electrode and that the dielectric constant is higher than 2. These approximations of the details of the equivalent circuit elements are relatively empirical and only serve to correlate the experimental data with some rudimentarily calculated theoretical values. More exact calculation of such details is possible by performing finite element analysis that delivers a more satisfactory relation between, us cr ip electrical behaviour of the system. One of the authors has recently presented such a model [50]. t on the one hand, the material properties and the geometry of the electrode system and, on the other hand, the 4.2 Measurement of Droplets Fig. 6 and 7 show impedance measurements of droplets at transport rates of circa 2 and 22 droplets per minute as measured by the IMPSPEC system. Multi-sine wave excitation is used to apply five frequency values between 1 and 10 MHz simultaneously. Fig. 6 shows the measurement of droplets at the lower flow rate. The an aqueous segments in Fig. 6a have clear plateau values for all phase angle measurements and although the measurements in Fig. 6a show some phase boundary effects at higher frequencies, it is still possible to clearly M estimate the plateau values. In Fig. 7, the pumping speed was set to the maximum rate possible with the syringe pump and it is clear the droplet transit rate has reached the limit of what can be reliably measured using the setup ed described in this paper. Possibilities for extending the system to higher throughput are discussed in the following section. The comparison of Fig. 6 and 7 shows that the measurement of the conductivity of droplets at higher ep t throughput is less stable. Measurement speed is not the sole criterion of system performance as the measurement of droplets that serve as bioreactors is not necessarily a high-throughput application; many cells grow in the Ac c timeframe of days to weeks so that a slower throughput can be accepted. Nevertheless, there is much room left for the optimization of the system with regard to increasing measurement speed. In Figs. 6 and 7, the IMPSPEC System took 0.1 s to measure five frequencies between 1 and 10 MHz. In comparison, the HP Analyzer took almost 2 s for a similar measurement. The use of LabView running on a personal computer to control the acquisition of data was responsible for much loss of measurement speed due to slow data transfer combined with slow processing of data by the personal computer’s operating system. Measurement speed can be improved by implementing a real-time measurement system. This could be avoided by directly programming either the fully programmable gate array (FPGA) or the digital signal processor (DSP) present in the IMPSPEC measurement hardware. Measurement speed can be further improved by implementing one of a number of time-domain signal processing techniques (square waves, multi-sine waves, maximum-length sequences and chirp) [51-52]. Min et al. [51] suggested the use of impedance spectroscopy as a technique for Page 10 of 24 high-throughput measurement of the biological content of droplets and suggested a signal processing scheme for making this high-speed measurement possible. The IMPSPEC system is highly suited for implementing timedomain signal processing techniques and in this case we have utilized multi-sine wave excitation. There is much room to improve the speed of measurement of the ImpSpec system while the HP Analyzer is limited to a relatively slower measurement speed. t Conclusions us cr ip 5. This paper presents the inline impedance measurement of the conductivity of droplets in a segmented-flow system. This system is of significance for measurement of the content of droplets that are used as bioreactors in segmented flow in a microfluidic system. A model of the impedance of the system is presented and the results of this model are used to improve system design. An electrode configuration based on electrodes attached to the outside surface of an extremely thin-walled glass capillary is presented. Glass capillaries were hydrophobized by an silanization in order to prevent droplet breakup in the capillaries. Experiments were performed to characterize the electrode structure by measuring the impedance of salt solutions of various ionic strengths in both continuous M flow and segmented flow. The thinness of the walls (10 µm thickness) made it possible to measure conductivity over a relatively large bandwidth at low applied potentials using cell constants based on resistance and frequency ed measurement. The speed of measurement was affected by the relation of the size of the electrodes and the speed of the ep t measurement to the size of the droplets and the speed of the droplet movement: at higher speeds the droplets traverse the electrodes faster than the measurement can be made. There is a need for faster measurement electronics by making use of advanced signal processing and microcontroller technology. The system presented Ac c here is low cost and robust and well suited to inline measurement of impedance spectroscopy in segmented flow. Page 11 of 24 References 1. J. R. Burns and C. Ramshaw, The intensification of rapid reactions in multiphase systems using slug flow in capillaries, Lab Chip 1 (2001) 10-15 2. T. Thorsen, R. W. Roberts, F. H. Arnold and S. R. Quake, Dynamic Pattern Formation in a VesicleGenerating Microfluidic Device, Phys. Rev. Lett. 86 (2001) 4163-4166 T. Nisisako, T. Torii and T. Higuchi, Droplet formation in a microchannel network, Lab Chip 2 (2002) 24- t 3. 4. us cr ip 26 T. Thorsen, R. W. Roberts, F. H. Arnold and S. R. Quake, Dynamic Pattern Formation in a VesicleGenerating Microfluidic Device, Phys. Rev. Lett. 86 (2001) 4163-4166. 5. L. Shui, J. C. T. Eijkel and A. van den Berg, Multiphase flow in micro-and nanochannels, Sensors Actuators B: Chem. 121 (2007) 263-276 6. A. Grodrian, J. Metze, T. Henkel, M. Roth and J. M. Köhler, Segmented flow generation by chip reactors for highly parallelized cell cultivation, Proc. SPIE 4937 (2002) 174-181 K. Martin, T. Henkel, V. Baier, A. Grodrian, T. Schön, M. Roth, J. M. Köhler and J. Metze, Generation of an 7. larger numbers of separated microbial populations by cultivation in segmented-flow microdevices, Lab Chip 3 (2003) 202-207 A. Grodrian, J. Metze, T. Henkel, K. Martin, M. Roth and J. M. Köhler, Segmented flow generation by chip M 8. reactors for highly parallelized cell cultivation, Biosens. Bioelectron. 19 (2004) 1421-1428 9. J. M. Köhler, T. Henkel, A. Grodrian, T. Kirner, M. Roth, K. Martin and J. Metze, Digital reaction ed technology by micro segmented flow—components, concepts and applications, Chem. Eng. J. 101 (2004) 201-216 10. S. Wiedemeier, F. Ehrhart, E. Mettler, G. Gastrock, T. Forst, M. M. Weber, H. Zimmermann, and J. Metze, (2011) 165-173. ep t Encapsulation of Langerhans' islets: Microtechnological developments for transplantation. Eng. Life Sci. 11 11. J. M. Köhler and T. Kirner, Nanoliter segment formation in micro fluid devices for chemical and biological micro serial flow processes in dependence on flow rate and viscosity, Sensors Actuators A: Phys. 119 Ac c (2005) 19-27 12. A. Funfak, J. Cao, O. Wolfbeis, K. Martin and J. M. Köhler, Monitoring cell cultivation in microfluidic segments by optical pH sensing with a micro flow-through fluorometer using dye-doped polymer particles, Microchim. Acta 164 (2009) 279-286 13. P. M. Günther, F. Moeller, T. Henkel, J. M. Koehler and G. A. Gross, Formation of Monomeric and Novolak Azo Dyes in Nanofluid Segments by Use of a Double Injector Chip Reactor, Chem. Eng. Tech. 28 (2005) 520-527 14. J. Schemberg, A. Grodrian, R. Römer, G. Gastrock and K. Lemke, Online optical detection of food contaminants in microdroplets, Engineering in Life Sciences 9 (2009) 391–397 15. J. Z. Chen, A. A. Darhuber, S. M. Troian and S. Wagner, Capacitive sensing of droplets for microfluidic devices based on thermocapillary actuation, Lab Chip 4 (2004) 473–480 16. N. Srivastava and M. A. Burns, Electronic drop sensing in microfluidic devices: automated operation of a nanoliter viscometer, Lab Chip 6 (2006) 744–751 Page 12 of 24 17. X. Niu, M. Zhang, S. Peng, W. Wen and P. Sheng, Real-time detection, control, and sorting of microfluidic droplets, Biomicrofluidics 1 (2007) 044101-12 18. J. Nichols, A. Ahmadi, M. Hoorfar, H. Najjaran and J. F. Holzman, In situ digital microfluidic conductance sampling, Sensors Actuators A: Phys. 152 (2009) 13-20 19. M. J. Schertzer, R. Ben-Mrad and P. E. Sullivan, Using capacitance measurements in EWOD devices to identify fluid composition and control droplet mixing, Sensors Actuators B: Chem. 145 (2010) 340-347. 20. S. Grimnes, O. G. Martinsen, Bioimpedance and Bioelectricity Basics, Academic, London, 2000. 21. U. Pliquett, D. Frense, M. Schönfeldt, C. Frätzer, Y. Zhang, B. Cahill, M. Metzen, A. Barthel, T. Nacke and us cr ip t D. Beckmann, Testing miniaturized electrodes for impedance measurements within the -dispersion – a practical approach, J. Electr. Bioimp. 1 (2010) 41–55 22. M. C. Hofmann, M. Funke, J. Büchs, W. Mokwa and U. Schnakenberg, Development of a four electrode sensor array for impedance spectroscopy in high content screenings of fermentation processes, Sensors Actuators B: Chem. 147 (2010) 93-99 23. E. E. Krommenhoek, J. G. E. Gardeniers, J. G. Bomer, A. Van den Berg, X. Li, M. Ottens, L. A. M. van der Wielen, G. W. K. van Dedem, M. Van Leeuwen, W. M. van Gulik and J. J. Heijnen, Monitoring of yeast an cell concentration using a micromachined impedance sensor, Sensors Actuators B: Chem. 115 (2006) 384389 24. H. E. Ayliffe, A. B. Frazier and R. D. Rabbitt, Electric impedance spectroscopy using microchannels with M integrated metal electrodes, J. Microelectromech. Syst. 8 (1999) 50-57 25. C. Iliescu, D. P. Poenar, M. Carp and F. C. Loe, A microfluidic device for impedance spectroscopy analysis of biological samples, Sensors Actuators B: Chem. 123 (2007) 168-176 ed 26. B. P. Cahill, A. T. Giannitsis, R. Land, G. Gastrock, U. Pliquett, D. Frense, M. Min and D. Beckmann, Reversible Electrowetting on Silanized Silicon Nitride, Sensors Actuators B: Chem. 144 (2010) 380-386 27. J. T. Schumacher, A. Grodrian, C. Kremin, M. Hoffmann and J. Metze, Hydrophobic coating of (2008) 055019 ep t microfluidic chips structured by SU-8 polymer for segmented flow operation, J. Micromech. Microeng. 18 28. M. C. Hofmann, D. Ellersiek, F. Kensy, J. Büchs, W. Mokwa and U. Schnakenberg, Galvanic decoupled Ac c sensor for monitoring biomass concentration during fermentation processes, Sensors Actuators B: Chem. 111 (2005) 370-375 29. W. Olthuis, A. Volanschi, J. G. Bomer and P. Bergveld, A new probe for measuring electrolytic conductance, Sensors Actuators B: Chem.13 (1993) 230-233 30. W. Olthuis, W. Streekstra and P. Bergveld, Theoretical and experimental determination of cell constants of planar-interdigitated electrolyte conductivity sensors, Sensors Actuators B: Chem. 24 (1995) 252-256 31. W. Olthuis, A. J. Sprenkels, J. G. Bomer and P. Bergveld, Planar interdigitated electrolyte-conductivity sensors on an insulating substrate covered with Ta2O5, Sensors Actuators B: Chem. 43 (1997) 211-216 32. A. J. Zemann, E. Schnell, D. Volgger and G. K. Bonn, Contactless Conductivity Detection for Capillary Electrophoresis, Anal. Chem. 70 (1998) 563-567 33. A. J. Zemann, Conductivity detection in capillary electrophoresis, Trends Anal. Chem. 20 (2001) 346-354 34. R. M. Guijt, C. J. Evenhuis, M. Macka and P. R. Haddad, Conductivity detection for conventional and miniaturised capillary electrophoresis systems, Electrophoresis 25 (2004) 4032-4057 Page 13 of 24 35. F. Laugere, R. M. Guijt, J. Bastemeijer, G. van der Steen, A. Berthold, E. Baltussen, P. Sarro, G. W. K. van Dedem, M. Vellekoop and A. Bossche, On-Chip Contactless Four-Electrode Conductivity Detection for Capillary Electrophoresis Devices, Anal. Chem. 75 (2003) 306-312 36. F. Laugere, G. W. Lubking, J. Bastemeijer and M. J. Vellekoop, Design of an electronic interface for capacitively coupled four-electrode conductivity detection in capillary electrophoresis microchip, Sensors Actuators B: Chem. 83 (2002) 104-108 37. Z. Huang, W. Jiang, X. Zhou, B. Wang, H. Ji and H. Li, A new method of capacitively coupled contactless t conductivity detection based on series resonance, Sensors Actuators B: Chem. 143 (2009) 239-245 us cr ip 38. P. Kubán and P. C. Hauser, Capacitively coupled contactless conductivity detection for microseparation techniques - recent developments, Electrophoresis 32 (2011) 30-42 39. J. Bastemeijer, W. Lubking, F. Laugere and M. Vellekoop, Electronic protection methods for conductivity detectors in micro capillary electrophoresis devices, Sensors Actuators B: Chem. 83 (2002) 98-103 40. R. M. Guijt, E. Baltussen, G. van der Steen, H. Frank, H. Billiet, T. Schalkhammer, F. Laugere, M. Vellekoop, A. Berthold, L. Sarro and G. W. K. van Dedem, Capillary electrophoresis with on-chip fourelectrode capacitively coupled conductivity detection for application in bioanalysis, Electrophoresis 22 an (2001) 2537-2541 41. P. Kubán and P. C. Hauser, A review of the recent achievements in capacitively coupled contactless conductivity detection, Anal. Chim. Acta 607 (2008) 15-29 M 42. J. Tanyanyiwa, E. M. Abad-Villar, M. T. Fernández-Abedul, A. Costa-García, W. Hoffmann, A. E. Guber, D. Herrmann, A. Gerlach, N. Gottschlich and P. C. Hauser, High-voltage contactless conductivity-detection for lab-on-chip devices using external electrodes on the holder, Analyst 128 (2003) 1019-1022 ed 43. W. Pfleging, R. Kohler, P. Schierjott and W. Hoffmann, Laser patterning and packaging of CCD-CE-Chips made of PMMA, Sensors Actuators B: Chem. 138 (2009) 336-343. 44. E. Baltussen, R. M. Guijt, , G. van der Steen, F. Laugere, M. Vellekoop, S. Baltussen and G. W. K. van (2002) 2888-2893 ep t Dedem, Considerations on contactless conductivity detection in capillary electrophoresis, Electrophoresis 23 45. K. Mayrhofer, A. J. Zemann, E. Schnell and G. K. Bonn, Capillary Electrophoresis and Contactless Conductivity Detection of Ions in Narrow Inner Diameter Capillaries, Anal. Chem. 71 (1999) 3828-3833 Ac c 46. A. Barthel, D. Beckmann, J. Friedrich, I. Mock, T. Nacke and P. Peyerl, IMPSPEC – Ein Hard- und Softwarekonzept für die (Bio)-Impedanzspektroskopie, Chem. Ing. Tech. 77 (2005) 1948 47. T. Nacke, A. Barthel, J. Friedrich, M. Helbig, J. Sachs, M. Schäfer, P. Peyerl and U. Pliquett, A new hard and software concept for impedance spectroscopy analysers for broadband process measurements in 13th Int. Conf. on Electrical Bioimpedance and the 8th Conf. on Electrical Impedance Tomography (Eds: H. Scharfetter, R. Merva), Springer-Verlag, Berlin 2007, pp. 194-197. 48. H. Falkenhagen, M. Leist, G. Kelbg, Zur Theorie der Leitfähigkeit starker nicht assoziierender Elektrolyte bei höheren Konzentration, Annal. Phys. 446 (1952) 51-59. 49. B F. Wishaw and R. H. Stokes, The Diffusion Coefficients and Conductances of Some Concentrated Electrolyte Solutions at 25°, J. Am. Chem. Soc. 76 (1954) 2065-2071. 50. B.P. Cahill, Optimization of an impedance sensor for droplet-based microfluidic systems, Proc. SPIE 8066 (2011) 80660F-80668. 51. M. Min, U. Pliquett, T. Nacke, A. Barthel, P. Annus and R. Land, Broadband excitation for short-time impedance spectroscopy, Physiol. Meas. 29 (2008) S185-S192 Page 14 of 24 52. U. Pliquett, E. Gersing and F. Pliquett, Evaluation of fast time-domain based impedance measurements on biological tissue, Biomed. Technik 45 (2000) 6-13 Acknowledgements The authors are indebted to Mrs. Ursel Klingebiel’s assistance in the lab and wish her a very happy retirement. The authors thank Dr. Uwe Pliquett for supplying the conductive silicone and for very many enlightening us cr ip t discussions. Dr. Land and Prof. Min would like to thank the European Community for financially supporting the FP6 Marie Curie ToK project 29857 InFluEMP. Dr. Cahill would like to thank the European Community for financially supporting the Marie Curie ERG project EWETDYNAM under reference number PERG05-GA- M an 2009-247784. Biographies ep t ed Brian P. Cahill is a Marie Curie fellow at the Institute for Bioprocessing and Analytical Measurement Techniques in Heilbad Heiligenstadt, Germany. He received his PhD from the ETH Zurich, Switzerland, in 2004 and subsequently carried out post-doctoral research at the University of Geneva, Switzerland. He studied Mechanical Engineering in Ireland receiving his MEng degree in 1998 from Dublin City University and his BE degree in 1995 from University College Dublin. His research interests focus on microfluidics and electrokinetically-driven fluid flow. Raul Land is a senior researcher at the Tallinn University of Technology, Estonia. He received his PhD in Electronic Engineering from Tallinn University of Technology in 1993 and subsequently carried out postdoctoral work at the Royal Institute of Technology in Stockholm, Sweden. Ac c Thomas Nacke is a senior researcher at the Institute for Bioprocessing and Analytical Measurement Techniques in Heilbad Heiligenstadt, Germany. He received the Dipl.-Ing. degree in Electrical Engineering in 1981. His current research interests are focused on impedance-based biosensors and the development of instrumentation for controlling biogas plants. Mart Min has been Professor and Head of the Chair of Electronic Measurement at Tallinn University of Technology since 1992. He received the PhD degree in measurement science from Kiev Polytechnic Institute, Ukraine, in 1983. He received the Dipl.Eng. degree in electronics from Tallinn University of Technology, Tallinn, in 1969. From 1992 to 1993, he was with the Technical University of Munich, Munich, Germany, and the Universität der Bundeswehr München, Munich, Germany, as a Guest Scientist and Professor. His interests include electronic circuits and systems, measurement, and biomedical electronics. Dieter Beckmann is director of the Institute for Bioprocessing and Analytical Measuring Techniques, Germany, and head of the Department of Analytical Measurement Techniques, and also Professor at the University of Applied Sciences in Jena, Germany. He studied physics in Jena and later worked in the field of optical instruments for biotechnology. Today his research interests cover impedance spectroscopy, biosensors, and, in particular, online measuring systems for the life sciences. Page 15 of 24 Figure Headings Fig. 1. (a) Schematic diagram showing a cross-section of the sensor system based on a glass capillary. (b) Equivalent circuit proposed to model the electrode system. Fig. 2. (a) The dependence of the impedance modulus and impedance phase angle on medium conductivity for t the following circuit element values: Cw = 10 pF, Cp = 0.070 pF, Cm = 0.070 pF and Rm = 100 k, 1 M and us cr ip 10 M. (a) The dependence of the Rs, Cs and s on frequency for the following circuit element values: Cw = 10 pF, Cp = 0.070 pF, Cm = 0.070 pF and Rm = 1 M. Fig. 3. Schematic diagram of the equipment for generating and measuring droplets. Fig. 4. (a) The dependence of Rs on signal frequency is displayed. (b) The dependence of Cs on signal frequency an is displayed. Full lines represent data measured by the HP Impedance Analyzer and dashed lines represent data measured by the IMPSPEC system. The ionic strength of the aqueous solutions is indicated by the legend. M Fig. 5. (a) The data measured by the HP Impedance Analyzer displayed in Fig. 4 are normalized by multiplying the Rs by the conductivity of each solution, by dividing the frequency by the conductivity and by evaluating s. ed Full lines represent Rs and dashed lines represent s. The ionic strength of the aqueous solutions is indicated by the legend. (b) The dependence of molar conductivity on molar concentration is displayed. The data displayed ep t in Fig. 4 was used to calculate a molar conductivity value. The solid line is from Falkenhagen-Wishaw-Stokes. Fig. 6. (a) The variation of the phase angle during the passage of 100-mM saline solution fluid segments between the electrodes. (b) The variation of the impedance modulus during the passage of fluid segments Ac c between the electrodes. The individual frequencies of the applied multi-sine wave are indicated by the legend. Measurements acquired using the IMPSPEC system. Fig. 7. (a) The variation of the phase angle during the passage of 100-mM saline solution fluid segments between the electrodes. (b) The variation of the impedance modulus during the passage of fluid segments between the electrodes. The individual frequencies of the applied multi-sine wave are indicated by the legend. Measurements acquired using the IMPSPEC system. Page 16 of 24 Tables Table 1 ImpSpec System K1 [1/mm] 38.32 39.27 K2 [MHz m] 1.002 0.981 Ac c ep t ed M an us cr ip HP Impedance Analyzer t Cell constants used to calculate the molar conductivities shown in Figure 5b. Page 17 of 24 Figure(s) (a) Conductive Silicone Electrodes Glass Capillary Shielding (b) us cr ip t Cw Rm Cm ce pt ed M an Cp Ac Cw Page 18 of 24 Figure(s) (a) 0 9 10 10 * R m Rm Rm / 10 8 10 -20 7 10 |Z| [W] 5 Phase [°] -40 6 10 -60 10 4 10 t -80 3 2 10 3 10 10 (b) 4 5 10 f [Hz] 10 6 7 10 us cr ip 10 8 10 0.1 an Rs/Rm,Cs/C'w,wts 1 0.01 C/C'w R/Rm wts 1/(2pt2) 1/(2pt1) 6 10 7 8 10 ed 10 f [Hz] pt 5 10 ce 4 Ac 10 M 0.001 Page 19 of 24 Figure(s) Syringe Pumps us cr ip t Oil Water T-Junction Ac ce pt ed M an Sensor Page 20 of 24 Figure(s) (a) 200 mM KCl 100 mM KCl 50 mM KCl 25 mM KCl 12.5 mM KCl 6.25 mM KCl 3.125 mM KCl 1.56 mM KCl 780 µM KCl 390 µM KCl 7 10 6 Rs [W] 10 5 t 10 4 3 10 (b) 10 4 5 10 10 f [Hz] 6 10 us cr ip 10 7 10 -11 7 6 5 4 3 2 10 -12 C s [C] 7 6 5 4 -13 7 6 5 4 3 5 f [Hz] 6 10 7 10 ed 10 pt 4 10 ce 3 10 Ac 10 M 2 an 200 mM KCl 100 mM KCl 50 mM KCl 25 mM KCl 12.5 mM KCl 6.25 mM KCl 3.125 mM KCl 1.56 mM KCl 780 µM KCl 390 µM KCl 3 Page 21 of 24 Figure(s) (a) 5 4 3 1 2 4 9 8 7 0.1 wts R ss [1/m] 10 6 200 mM KCl 100 mM KCl 50 mM KCl 25 mM KCl 12.5 mM KCl 6.25 mM KCl 3.125 mM KCl 1.56 mM KCl 780 µM KCl 390 µM KCl 4 3 2 3 0.001 10 5 6 10 7 us cr ip 10 0.01 t 5 8 10 f/s [Wm/s] 10 (b) an 10 5 Falkenhagen-Wishaw-Stokes HP Resistance HP Frequency ImpSpec Resistance Impspec Frequency 2 3 4 5 6 7 2 3 4 5 6 7 2 0.1 ed 0.01 Ionic Strength [M] pt 0.001 ce 4 5 67 M 0 Ac Molar Conductivity [S/m/M] 15 Page 22 of 24 Figure(s) (a) 0 1.00 MHz 1.72 MHz 3.08 MHz 5.64 MHz 10.12 MHz -20 Phase [°] -40 -60 us cr ip t -80 -100 900 (b) 910 920 930 t [s] 940 950 960 940 950 960 5 4 3 2 10 6 7 6 5 an |Z| [W] 4 3 2 7 6 5 M 5 1.00 MHz 1.72 MHz 3.08 MHz 5.64 MHz 10.12 MHz 4 3 920 930 t [s] pt 910 ce 900 ed 2 Ac 10 Page 23 of 24 Figure(s) (a) 0 1.00 MHz 1.72 MHz 3.08 MHz 5.64 MHz 10.12 MHz -20 Phase [°] -40 -60 us cr ip t -80 -100 36 38 40 42 44 42 44 t [s] (b) 5 4 3 2 10 6 an 7 6 5 |Z| [W] 4 3 2 7 6 5 M 5 1.00 MHz 1.72 MHz 3.08 MHz 5.64 MHz 10.12 MHz 4 3 38 40 pt t [s] ce 36 ed 2 Ac 10 Page 24 of 24