By Topic

IEEE Quick Preview
  • Abstract



Portable point-of-care (POC) clinical diagnostic devices are important for implementing effective treatment of diseases in the developing world and for enabling personalized healthcare [1]. A POC diagnostic device is ideally low cost and small sized, and contains a sensor for detecting analytes. The inclusion of a microfluidics system to prepare the sample and present it to the sensor enables wider use of the system, because integrated sample preparation reduces the laboratory requirements and user education necessary for sensor use [1]. Simple lateral flow assays, e.g., the simple dipstick test, are not able to handle complex immunoassays, such as those involving DNA detection and analysis from whole blood. Thus, a more sophisticated microfluidics platform is necessary for more complex assays. In such systems, a very low-cost disposable component is typically envisioned to handle the clinical samples and perform the sensing [1], [2]. For direct interrogation of a sample, the sensor must be in contact with the sample and, to avoid biofouling issues, is ideally disposable and thus low cost. Optical microresonator sensors are low in material cost, simple to fabricate if they are photoimageable, and are high performance sensors that have been demonstrated for applications such as DNA and protein detection [3], [4], [5]. In this paper, the integration of a polymer microresonator sensor with a digital electrowetting-on-dielectric (EWD) fluidic platform is reported.

The integration of compact low-power-consumption sample preparation fluidics and high-performance sensors is critical for portable POC devices. EWD microfluidics is a droplet-based flow platform in which fluids are electrically actuated in the form of droplets for low power consumption (nanowatts to microwatts), automated fluidic functions, such as sample preparation [6], [7], [8], [9]. The EWD microfluidics platform can perform multiplexed DNA sample preparation from whole blood [6], [7] while using very low volumes of reagents and sample material, as well as minimal power to actuate droplets.

Planar microresonator sensors are a promising type of optical surface sensor, because they are compact, label-free, and can be fabricated with low-cost polymer materials. Previous work on polymer microresonators attached to the top plate of EWD systems demonstrated the functionality of a microresonator sensor in a silicone oil medium [10], [11]. However, the sensor was addressed through a hole in the EWD system top plate, which precluded the motion of droplets off the sensor and required a relatively large volume of solution (many droplets) to make contact with the sensor. In this paper, we report on the development of a new integrated microresonator/EWD structure in which the sensor is embedded in the top plate of the EWD microfluidic system. This approach to sensor integration led to improved EWD functionality by enabling addressing of the sensor with a single droplet and by maintaining the ability to move droplets of fluid off the sensor. This paper also reports on the highest figure of merit [FOM; the product of quality (Q) factor and sensitivity (S)] reported to date for SU-8 polymer microresonator sensors probed at a wavelength of 1550 nm, as well as the design optimization and implementation of the highest FOM resonator reported to date for microresonators integrated into EWD systems.

Optical waveguide sensors are particularly attractive for integration with EWD systems, because optical sensors detect changes in the refractive index of media for a significant distance above the sensor surface, roughly in the range of 100 nm to 1 Formula$\mu\hbox{m}$ [12], which is consistent with beam propagation method (BPM; Rsoft) transverse-electric-like (TE) simulations for these devices, which yielded 300 nm–450 nm for the range of microresonator dimensions studied herein. Therefore, a thin film of immiscible liquid medium coating the surface of the sensor, such as the oil used in EWD devices, does not preclude functionality of the sensor, if the film thickness is less than this distance. To access low-power EWD functionality, there is the need for an immiscible medium to fill the fluidic channels, which performs the functions of preventing evaporation [8], preventing surface fouling [8], and reducing the actuation voltage [13]. While actuation in an air medium is possible, the device fabrication cost/complexity is increased relative to using an immiscible liquid medium, such as silicone oil [14]. Thus, the presence of a liquid immiscible medium is an important factor to be considered for sensor integration.



2.1. Design

2.1.1) Microresonator Design

The microresonator sensors were designed to maximize the FOM [which is inversely proportional to the limit of detection (LOD) [15], [16]] and to minimize propagation loss while enabling integration directly onto the top plate of the EWD system. The microresonators were vertically coupled, with the bus waveguide integrated below the resonator surface, to separate the liquid analyte from the bus waveguide/resonator coupling region. To maximize the sensor sensitivity, channel waveguides with the smallest possible cross sections (width and height), which have the highest sensitivities, were used. For waveguides with minimal cross-sectional area, the overlap of the guided-mode field with the sensing region is maximized, which maximizes the change in the resonant wavelength for a perturbation in the refractive index of the sensed region [17], [18]. The BPM was used to calculate the effective indices and design curves for the sensor sensitivity of both the TE and transverse-magnetic-like (TM) fundamental waveguide modes versus waveguide width and height using the physical parameters in Table 1 and the RSoft BPM commercial simulation tool. The method for estimating sensitivity reported in [18] was utilized with a refractive index perturbation of Formula$1 \ast 10^{-3}$ RIU and a fixed wavelength perturbation of 0.1 nm for calculating the group refractive index. The propagation loss due to the combined effects of bending and material losses were also estimated using BPM. The waveguide width was varied from 700 nm to 2 Formula$\mu\hbox{m}$, and the height was varied from 500 nm to 2 Formula$\mu\hbox{m}$, in 100-nm increments for both dimensions. For the bending loss calculation, the design radius of 250 Formula$\mu\hbox{m}$ was applied to the waveguide. A maximum sensitivity for the TE fundamental mode of 214 nm/RIU was predicted for a width of 0.84 Formula$\mu\hbox{m}$ and a height of 2 Formula$\mu\hbox{m}$ (the practical upper limit of thickness for these devices). For the fundamental TM mode, a maximum sensitivity of 174 nm/RIU was predicted for a width of 0.73 Formula$\mu\hbox{m}$ and a height of 1.91 Formula$\mu\hbox{m}$. The maxima for both polarizations corresponded to waveguide dimensions at the waveguiding mode cutoff for the fundamental TE and TM modes. The simulation results indicate that tall and narrow waveguide geometries for waveguides with this particular refractive index profile provide the highest sensitivities and that the highest sensitivity for this waveguide can be realized with the TE polarization. Thus, the sensitivity for these channel waveguides is maximized not simply for waveguides with the smallest possible cross sections but with particular combinations of width and height that optimally balance the amount of field intensity in the sensing region relative to that in the substrate and the waveguide core. For a refractive index profile in which the substrate has a higher refractive index than water, the tall and narrow geometry is favored, because the field intensity for this geometry tends to extend more into the water surrounding the waveguide, rather than into the substrate.

Table 1

Based on the sensitivity and loss analysis, the target microresonator waveguide dimensions were 1.9 Formula$\mu\hbox{m}$ in height with widths between 1 and 1.5 Formula$\mu\hbox{m}$. The target bus waveguide dimensions were set to 1 Formula$\mu\hbox{m}$ in width and 1 Formula$\mu\hbox{m}$ in height in order to minimize the difference in the phase constants between the bus waveguides and the microresonator. Additionally, these dimensions were set such that the effective index of the fundamental TE mode was far enough from the cutoff condition Formula$(n_{eff} = 1.444)$ that the variability in the waveguide height due to the variability of the bus waveguide fabrication process would rarely cause the fundamental TE mode to be cut off in the fabricated devices.

Sample test sensors were fabricated on standard Formula$\hbox{SiO}_{2}$ (thermal oxide)/Si wafers. To enable maskless adjustments of pattern dimensions to study the effect on the FOM of varying the waveguide dimensions, electron-beam lithography (EBL) was utilized instead of photolithography, although the dimensions utilized were within the range of low-cost photolithography, which could be used once optimized waveguide dimensions were identified. The interlayer dielectric thickness was set in order to balance high Q factor with an acceptable on-resonance transmission at the drop port. Because the top-plate embedded sensors tended to have 5–15 dB greater insertion loss than the test sensors and the power measurement noise increased significantly below about −50 dBm with the measurement system utilized, a target on-resonance transmission of greater than or equal to −35 dBm was set for the test sensors. The Formula$\hbox{SiO}_{2}$ interlayer dielectric thickness optimization was explored experimentally, yielding a thickness of 1.4 Formula$\mu\hbox{m}$. For the EWD top-plate embedded sensors, which were fabricated on a Pyrex substrate, the intent was to replicate the optimized dimensions from the test sensor work. The width and height of the fabricated microresonator waveguides were about 1.2 Formula$\mu\hbox{m}$ and 2.0 Formula$\mu\hbox{m}$, respectively, for the top-plate embedded sensor and about 1.5 Formula$\mu\hbox{m}$ and 1.9 Formula$\mu\hbox{m}$ for the test sensor. Although the design optimization for sensitivity yielded widths of 0.84 Formula$\mu\hbox{m}$ and 0.73 Formula$\mu\hbox{m}$ for TE and TM modes, respectively, those widths optimize sensitivity only, which must be balanced with bending loss to optimize the FOM. The target waveguide width to maximize the FOM and to produce a reading at the drop port was found to be between 1 and 1.5 Formula$\mu\hbox{m}$. The fabricated bus waveguide height was, on average, about 850 nm and varied from device to device within a range of approximately ±150 nm. The fabricated bus waveguide width was, on average, 1.5 Formula$\mu\hbox{m}$ at the base with an 80° sidewall slope.

A photomicrograph of the top-plate embedded microresonator sensor and a diagram of its vertical structure are shown in Fig. 1.

Figure 1
Fig. 1. (a) Photomicrograph of a SU-8 polymer microring resonator and the patterned ITO region as embedded in the EWD system top plate. (b) To-scale diagram of the cross-section at the location indicated in (a).

The EWD microfluidics were designed to dispense droplets from reservoirs, actuate the droplets along an electrode path, present the droplets to the sensor site, and then move the droplets off of the sensor. The demonstration of these functions with an embedded microresonator sensor are reported herein for the first time. These fluidic functions were performed by applying a succession of voltages to a series of metal pads that, through the electrowetting process, dragged the droplet from one metal pad onto the next metal pad. The EWD system is composed of a top and bottom plate, containing a ground plane and the metal pads, respectively. The liquid is confined between the top plates vertically and is confined laterally by a patterned gasket, which defines the reservoirs and the fluidic channels [8], [23]. For the system reported herein, the ratio between the gasket height Formula$(\sim\!60\ \mu\hbox{m})$ and the electrode center to center spacing (pitch) was set to 1 : 10 to minimize the droplet dispensing and splitting voltage [13]. The electrode pitch of 605 Formula$\mu \hbox{m}$ kept the size of the microfluidic channel within the extent of the bus waveguides on the test devices, so that the length of the bus waveguides would not have to be increased to accommodate a longer path to the microresonator. This kept the insertion loss from increasing due to increased total propagation loss in the bus waveguides from input to output. A metal pad pitch of 605 Formula$\mu\hbox{m}$ limited the square electrowetting electrodes to about 600 Formula$\mu \hbox{m}$ on a side. Because single droplets are typically only slightly greater in extent than the electrodes with which they are actuated, the size of the electrodes limits the diameter of the ring-shaped microresonator sensors, which must be smaller in extent than the droplet in order to be completely covered by the droplet. The diameter of the microresonator integrated into the top plate of the EWD system was set to 500 Formula$\mu\hbox{m}$ to fit within the area of one electrowetting electrode with some room to spare for alignment error between the top plate and the bottom plate. The microresonator diameter of 500 Formula$\mu\hbox{m}$ was also designed to minimize bending loss within the constraint of 50 Formula$\mu\hbox{m}$ of alignment buffer space between the outer edge of the ring and the edge of the electrowetting electrode.

2.2. Fabrication

2.2.1) Microresonator

Microresonators were fabricated on Formula$\hbox{SiO}_{2}$ substrates and on the Pyrex top plates of the EWD systems. The top plate was fabricated on one quarter of a 4”-diameter 500-Formula$\mu\hbox{m}$-thick Pyrex wafer. A 50-nm layer of PECVD Formula$\hbox{Si}_{3}\hbox{N}_{4}$ (etch stop) was deposited, followed by a 1-Formula$\mu\hbox{m}$-thick layer of PECVD Formula$\hbox{SiO}_{2}$ and a 100-nm-thick Cr etch mask. ZEP 520A, a positive e-beam resist, was used to pattern the bus waveguides with EBL. The ZEP 520A resist was developed in o-xylene and Cr etchant was used to etch the patterns into the Cr. Next, the resist mask was removed, and reactive ion etching (RIE) with an Formula$\hbox{O}_{2}/\hbox{CHF}_{3}$ gas mixture was used to etch trenches into the PECVD Formula$\hbox{SiO}_{2}$. RIE was followed by an etch in buffered oxide etchant (BOE) to smooth the channel surfaces. After etching the trenches, the Cr mask was removed with Cr etchant. The trenches were filled with SU-8 polymer by spin-coating SU-8 2002 and soft curing. The SU-8 polymer was fully cross-linked by UV flood exposure followed by hard curing. The SU-8 film was then etched down to the trench with RIE using an Formula$\hbox{O}_{2}/\hbox{SF}_{6}$ gas mixture. Then, 1.4 Formula$\mu\hbox{m}$ of interlayer dielectric PECVD Formula$\hbox{SiO}_{2}$ was deposited on top of the channel waveguides. To complete the EWD top-plate contact, a 70-nm layer of indium–tin oxide (ITO; 95%/5% Formula$\hbox{InO}_{2}/\hbox{SnO}_{2}$) was sputter deposited. The ITO was patterned in 5% HCl using a photoresist mask. The glass substrate with the completed bus waveguides was diced to expose the bus waveguide facets, and a hydrophobic layer of 50 nm–70 nm of Cytop was then spin coated onto the structure. The Cytop was hard baked and rendered temporarily hydrophilic [24] by a light surface ashing in Formula$\hbox{O}_{2}$ plasma prior to spin-coating SU-8 2002 on the top surface to form the microresonator. The microring resonator was defined in the SU-8 2002 by exposing the SU-8 using EBL. After exposure, the SU-8 was soft cured, developed in SU-8 developer, and hard cured.

2.2.2) EWD Microfluidics

The EWD microfluidic system consisted of a top plate with an embedded microresonator (described above), a bottom plate, and a vertical gasket between the two plates, similar to the work in [23]. The bottom plate was fabricated on a 2” Si wafer by first coating the wafer with 2 Formula$\mu\hbox{m}$ of Formula$\hbox{SiO}_{2}$ using PECVD. The square Cr electrodes were patterned by liftoff of evaporated Cr with a negative photoresist. Photoimageable SU-8 3035 was spun on, patterned by photolithography, and baked to form the gasket. Approximately 800 nm of Parylene C was deposited by CVD to form the dielectric. A hydrophobic coating consisting of a Formula$\sim$50–70 nm layer of Cytop (5 : 1 diluted Cytop 809A) was then spun on and cured. Finally, the bottom plate/gasket structure was diced to the correct system size for testing (4.27 cm × 3.94 cm).



3.1. Measurement and Data Analysis

The microresonators on Formula$\hbox{SiO}_{2}$ were characterized to validate the optimized design, with subsequent fabrication of the microresonators embedded in the EWD top plates, which were mechanically bonded to the EWD bottom plate/gasket structures. The microresonator measurements were performed using optical measurement equipment controlled by Labview software. The EWD microfluidics system was controlled with computer software via USB. The optical microresonator measurements utilized an HP8164A Lightwave Measurement System with an 81680A tunable laser module, an 81623A Ge photodetector module, and an 81618A optical head interface module. Light from the tunable laser was launched into the system using a Corning SMF 28 single-mode fiber, and light was collected from the output waveguide with a Corning 62.5/125-Formula$\mu\hbox{m}$ multimode graded-index fiber. The tunable laser measurement system supported a minimum wavelength scan step of 0.1 pm and had a noise floor of −80 dBm. For sensing experiments, a step size of 0.6 pm over a small range of wavelengths within the mode-hop free range of the laser, i.e., 1520 nm–1570 nm, was used. The selected wavelength range was that which had the highest transmission at the drop port. The step size of 0.6 pm was chosen because it provided the optimal combination of wavelength and temporal resolution (shortest scan time) for the tunable laser measurement system. For measurements using the maximum number of data points at this resolution (19834), each spectrum measurement took about 12 s. For the unintegrated sensor, a smaller number of data points per spectrum was measured in order to increase the spectral measurement rate to about one every 6 s. For these sensors, spectra were measured under two different sensor conditions: immersion in deionized (DI) water and immersion in 2% (w/v) D-glucose solution. For each condition, about 30 spectra were measured to obtain a good estimate of the resonance peak positions for each condition. The same approach was used for the EWD embedded microresonator, using the EWD system to present the water and analyte to the sensor.

The wavelengths of peaks appearing in the spectra, which corresponded to the resonant wavelengths of the resonator, were estimated and tracked in time to generate sensorgrams. The sensorgrams consist of time series data that show the variation of the resonant wavelengths with time. Resonant wavelengths were estimated by first applying a zero-phase low-pass filter to the measured spectra and then estimating the wavelengths of resonance to be the local maxima in the filtered spectra. Resonant wavelength traces were tracked in time by applying an algorithm that assumes that the peak closest to the peak in the previous frame (a frame is a spectrum measured at a specific time) is the same peak. This algorithm breaks down if the resonant peak moves more than one-half of the free spectral range (FSR), the spacing between the peaks, from one frame to the next. The FSR of the test sensor and of the top-plate embedded sensor was measured to be about 0.962 nm and 0.956 nm, respectively. To ensure that this constraint was met, spectral measurements were taken as frequently as possible in order to track the dynamic shift of the resonant wavelengths in time, and the magnitude of the refractive index changes was controlled by adjusting the glucose concentration in the glucose solution, such that resonant wavelength shifts would not exceed one half of the FSR. For a given experiment, several sensorgrams were generated, one for each of the multiple peaks tracked. A representative sensorgram from the set was selected for further analysis for each experiment.

3.2. Microresonator Sensitivity, Q Factor, and FOM

Microresonator sensitivities were measured using droplet swapping of water to establish the baseline variation due to the swapping operation and with D-glucose as a refractive index standard. All of the results in this section are for sensors that were tested outside of the EWD system. Test results for sensors measured inside of the EWD system are presented in the following section. The measurement was performed by applying a 20- Formula$\mu\hbox{L}$ droplet of DI water to the sensors with a pipette to measure a baseline wavelength. To measure the D-glucose sensitivity, the droplet of DI water was removed using a stream of ultra high purity Formula$\hbox{N}_{2}$ gas, and a 20 Formula$\mu\hbox{L}$ droplet of 2% (w/v) D-glucose solution was pipetted onto the sensors. Next, droplet swaps of DI water with DI water were performed to provide a measure of the baseline variability due to the droplet swapping operation. The standard deviations of the baseline shifts were 1.8 pm for the unintegrated test sensor and 11 pm for the top-plate embedded sensor (tested outside of the EWD system), respectively. Fig. 2 shows the sensorgram for five measurements of 2% (w/v) D-glucose solution and four or five measurements of DI water droplets for both the unintegrated test sensor and the top-plate embedded sensor. The wavelength shifts due to these droplet swapping operations were measured by first subtracting the baseline drift from the sensorgram and then estimating the peak wavelength shift. The average and the standard deviation of the wavelength shifts were 230 pm and 1.1 pm, respectively, for the test sensor and 250 pm and 11 pm, respectively, for the top-plate embedded sensor. Using Formula$1.4 \ast 10^{-3}$ RIU/% (w/v) [25] to estimate the refractive index of the glucose solution, the sensitivities of the test sensor and of the top-plate embedded sensor were estimated to be 82 nm/RIU and 89 nm/RIU, respectively. For the top-plate embedded sensor, the optical polarization used for sensing was determined experimentally to be TM. The theoretically predicted sensitivity was 88 nm/RIU for the TM mode of the top-plate embedded sensor, which is in good agreement with the experimental test. The standard deviation of the water swapping baseline shifts were similar to that of the glucose measurements, which indicates that the glucose measurement error could be attributed primarily to the baseline variability caused by the droplet swapping operation.

Figure 2
Fig. 2. Resonant wavelength trace showing 5 replicates of swapping a DI Formula$\hbox{H}_{2}\hbox{O}$ droplet with a 2% (w/v) glucose droplet, followed by 5 replicates of swapping a DI Formula$\hbox{H}_{2}\hbox{O}$ droplet with a DI Formula$\hbox{H}_{2}\hbox{O}$ droplet. 1-Droplet swapped with a 20 Formula$\mu\hbox{L}$ DI Formula$\hbox{H}_{2}\hbox{O}$ droplet. 2-Droplet swapped with a 20 Formula$\mu\hbox{L}$ 2% (w/v) D-glucose droplet. (a) Test sensor. (b) Top plate embedded sensor.

From the spectra measured from the drop port with the sensors immersed in DI water, the nominal Q factor was determined. Spectra measured at the drop port for each device are shown in Fig. 3. The nominal Q factors and the corresponding nominal FOMs for the test sensor and for the top-plate embedded sensor were 15 000 and Formula$1.2 \times 10^{6}\ \hbox{nm/RIU}$ and 8400 and Formula$0.76 \ast 10^{6}\ \hbox{nm/RIU}$, respectively. The FOM of the test sensor is the highest reported for an SU-8 microresonator probed around 1550 nm and is comparable with that of other polymer microresonators, including those made with polystyrene (Formula$1 \times 10^{6}\ \hbox{nm/RIU}$ at 1550 nm) [4], SU-8 (Formula$2.3 \times 10^{6}\ \hbox{nm/RIU}$ at 1310 nm) [26], and ZPU13-430/LFR-S708U (Formula$2 \times 10^{6}\ \hbox{nm/RIU}$ at 1550 nm) [27].

Figure 3
Fig. 3. Measured spectra at the drop and the throughput port. (a) Test sensor on Formula$\hbox{SiO}_{2}$/Si. (b) Top plate embedded sensor.

3.3. Integrated Sensor/EWD System Sensitivity, Q Factor, and FOM

To characterize the sensitivity, Q factor, and FOM of the sensor integrated with the EWD microfluidics system, the EWD system was assembled. The system assembly required several steps. First, the gasket of the EWD system was filled with 2-cSt silicone oil. The 300-nL DI Formula$\hbox{H}_{2}\hbox{O}$ and 2% (w/v) D-glucose were then loaded into two separate reservoirs in the gasket by pipetting the liquids under the layer of silicone oil. After loading the reservoirs, the top plate was brought into contact with and aligned to the bottom plate, such that the sensor was aligned to the center of an electrowetting electrode and mechanically sealed. The assembly was moved to a separate stage, where the electrical and optical connections were made. A spring-loaded array of electrical pins connected to a ribbon cable, which was used to connect the EWD microfluidic system to the EWD control system, was placed in contact with the array of control pads on the bottom plate of the EWD system and fixed in place with a clamp. A ground wire was attached to the top plate with an alligator clip. For the optical connections to the tunable laser system, a single-mode fiber was aligned to the input waveguide port and a multimode fiber was aligned to the drop waveguide port. Alignment was performed by adjusting the position of the fibers in order to maximize the transmitted power at the drop port. Finally, the entire assembly, which rested on two positioning stages, was aligned to a CCD camera to monitor the droplet movements, and the electrical connections were connected to a computer-controlled array of relays to control the voltages applied to the EWD electrodes. The integrated system is shown in Fig. 4(c).

Figure 4
Fig. 4. Glucose measurement with the sensor integrated with the EWD system. (a) Sequence of photomicrographs showing the droplet merging operation (left-to-right) in the electrowetting system. (b) Resonant wavelength trace for the glucose measurement in the electrowetting system. The black arrow indicates when the droplets were merged. (c) Photograph of the integrated system.

To determine the sensitivity of the EWD system with the sensor immersed in 2-cSt silicone oil, 2% (w/v) D-glucose solution was used as a refractive index standard. To measure the sensor response to changes in glucose concentration, a droplet of DI Formula$\hbox{H}_{2}\hbox{O}$ was first dispensed from the reservoir and moved into contact with the sensor. Droplet actuation was achieved by applying an ac voltage typically between 40 V and 80 V peak to peak with 1-kHz frequency. The voltages were applied in sequence to the electrowetting electrodes to direct the droplet along the path toward the sensor. After measuring spectra for nearly a half an hour with the DI Formula$\hbox{H}_{2}\hbox{O}$ droplet in contact with the sensor, a droplet of 2% (w/v) D-glucose was actuated toward the sensor and merged with the droplet of DI Formula$\hbox{H}_{2}\hbox{O}$ as depicted in Fig. 4(a). Spectra were recorded for over 30 min after the merging operation, in order to ensure that complete mixing of the glucose had occurred.

The spectral data was analyzed to determine the wavelength shift due to a 1% (w/v) change in glucose concentration (making the reasonable assumption of equal droplet volumes) after merging the droplet of glucose solution with the DI Formula$\hbox{H}_{2}\hbox{O}$ droplet. The sensorgram for this experiment is shown in Fig. 4(b). After subtracting the baseline drift from this trace, the wavelength shift was estimated to be 101 pm. Using this measurement, the sensitivity was estimated to be 72 nm/RIU, which was 80% of the test value measured when the microresonators were measured outside of the EWD system. The reduction in sensitivity was likely due to the presence of a thin film of oil entrained between the droplet and the sensor surface, which reduces the overlap of the field intensity of the sensor guided mode with the sensing region. The measured sensitivity compares favorably with the sensitivity of 69 nm/RIU reported by Luan et al. [11] for a microresonator sensor bonded to the top plate of an EWD system, which also utilized 2-cSt silicone oil, and the Q factor is much higher: 8400, nearly twice that of the Q factor of 4400 reported previously. The FOM for the integrated microresonator in this paper was Formula$0.60 \ast 10^{6}\ \hbox{nm/RIU}$, which is twice as large as that for the previously reported integrated microresonator sensor, Formula$0.30 \ast 10^{6}\ \hbox{nm/RIU}$ [11]. Furthermore, the sensor reported herein was addressed with single droplets, and droplets could be moved onto and off of the sensor. These capabilities are improvements over the previously demonstrated system, which required several droplets to be merged to make contact with the sensor through a hole in the top plate and could not actuate droplets off of the sensor [10], [11].



A vertically coupled SU-8 polymer microresonator sensor integrated directly into the top plate of an EWD system has been designed, fabricated, and tested, both external to, and integrated with the EWD system. The test sensor fabricated on a Si/Formula$\hbox{SiO}_{2}$ substrate and tested outside of the EWD system had the highest FOM reported to date for a polymer microresonator probed with a wavelength in the 1520–1570 nm range. This vertically coupled sensor structure has been embedded into the top plate of an EWD microfluidic system to demonstrate the functionality of the sensor in a typical oil medium utilized in a droplet-based flow microfluidics system. The FOM of the EWD system integrated sensor has been improved by a factor of two compared with the previously reported EWD system integrated sensor, and the full functionality of the EWD system has been preserved. These results indicate that polymer microresonator sensors are a viable sensor technology for integration with EWD microfluidics toward the development of portable POC diagnostic devices.


The authors would like to thank R. Evans, B.-N. Hsu, Y.-Y. Lin, and A. Madison for their assistance with the electrowetting system design, for their generous donation of materials, and for assistance with microfluidics testing, and Dr. T. Tyler for his work with the EBL system.


This work was supported NSF GrantCNS-1135853. The work of M. W. Royal was supported in part by the Air Force Office of Scientific Research through a National Defense Science and Engineering Graduate Fellowship. Corresponding author: M. W. Royal (e-mail:

The authors are with the Department of Electrical and Computer Engineering Duke University, Durham, NC 27708 USA.


No Data Available


No Photo Available

Matthew W. Royal

No Bio Available
No Photo Available

Nan M. Jokerst

No Bio Available
No Photo Available

Richard B. Fair

No Bio Available

Cited By

No Data Available





No Data Available
This paper appears in:
No Data Available
Issue Date:
No Data Available
On page(s):
No Data Available
INSPEC Accession Number:
Digital Object Identifier:
Date of Current Version:
No Data Available
Date of Original Publication:
No Data Available

Text Size