A digital microfluidic chip with programmable open system actuation and enhanced optical annealing with near‐infrared light

This work investigates a digital microfluidic chip and presents an advancement to microfluidic optical annealing methods through the application of whispering gallery mode (WGM) with near infrared excitation. Establishing a microfluidic chip with point‐of‐care capabilities, including actuation and annealing, has proven to be important. Unfortunately, poor heat absorption due to the long optical penetration depth of near infrared light creates scaling limitations for applications in optical‐based microfluidics. Through the application of WGM, the interaction length between the droplet and light is increased beyond the droplet diameter to improve heating and optical absorption. This is supported by finite‐difference time‐domain electromagnetic simulations and experimental results showing a greatly improved temperature change. Such a system is implemented in an open system digital microfluidic chip, to facilitate annealing via side illumination of droplets. The open system digital microfluidic chip is programmable for droplet actuation. The fundamental experiment of preprogrammed actuation of microdroplets is demonstrated in a 36 electrode grid. The results of annealing and actuation show potential for implementation in point‐of‐care microfluidic devices.

Initial work on optofluidic annealing has been reported, 26 where infrared (IR) light is used to anneal microdroplets.(This method will be referred to as traditional annealing in this work.)The wavelength of this IR light is 1450 nm, with corresponding optical penetration depth of 330 μm in water. 26,27Therefore, for high absorption (greater than 60%), this technique is limited to reactant volumes of approximately 40 nL.
Past work 1 extended initial reporting 26 by elongating the interaction length between sample droplet and the light beam through total internal reflection, that is, the application of WGM behavior. 28By interacting with the perimeter of the microdroplet, light can be trapped within the droplet due to total internal reflection. 29,30The overall effect is that the interaction path length is elongated (greatly beyond the microdroplet diameter).The past work 1 (hereby referred to as IR WGM annealing) noted an improvement of 125% beyond that of traditional optical annealing.
Although the results achieved using the IR WGM annealing method 1 show enhanced performance, the work was limited to IR optical annealing at 1550 nm wavelength, having an optical penetration depth of approximately 1 cm in water. 27Such IR optical annealing can be integrated with existing erbium-doped fiber lasers [31][32][33] that are available and designed for this telecom wavelength.5][36][37] Previously, optical annealing with NIR light was thought undesirable, due to the (relatively) low absorption coefficient of water for the NIR, corresponding to a very long optical penetration depth (e.g., greater than 10 cm). 27,38t is hypothesized that NIR optical annealing will benefit greatly from the application of WGM optical excitation and total internal reflection, to achieve interaction length at (or beyond) the optical penetration depth.This is a focus of the presented work.This work shows that NIR is a desirable wavelength for optical annealing and heating of microdroplets.
The use of an open digital microfluidic system is presented in this work, as this is required for actuation of droplets that can be side illuminated in a digital microfluidic system for optical annealing with a WGM illumination.Additionally, the digital microfluidic actuation technique should be completely programmable.A microfluidic process requires a great deal of control and preprocessing of fluids, enabled by preprogrammed digital microfluidic actuation.This demonstration of preprogrammed actuation in digital microfluidics is a second focus of the presented work.
In this work, we explore NIR optical excitation and heating of microdroplets with WGM excitation.We use a wavelength of 780 nm to excite microdroplets.(For this manuscript, we will refer to this annealing as WGM NIR annealing.)For a NIR wavelength, we compare our WGM results to that of traditional optical annealing.Our comparison is performed theoretically, with FDTD electromagnetic simulations, and experimentally, with characterization of microdroplet heating as a function of optical excitation position.We develop an open system chip that is programmable with control of the position and actuation of microdroplets.Overall, our results show promise for applications in on-chip analyzes.Such results can enable automated microfluidic analyzes.
A conceptual image of the work is shown in Figure 1. Figure 1A shows the illumination of the droplet for heating.The illumination is off center, thus enabling capturing of the optical power through total internal reflection.Figure 1B shows the motion of the droplet for preprocessing of droplets and for actuation between electrodes used as heat sinks.The motion is around the perimeter of a three by three inner grid of electrodes with the droplet initially placed at location b2 and ultimately returning to this location.Locations b4, d4, and d2 show the intermediate positions and are denoted with a droplet with high transparency.Both annealing and actuation are necessary for lab-on-achip applications.
The main contributions of this work are the following.We considerably improve annealing from the previously reported 125% improvement with IR WGM annealing. 1This is due to the drastic loss of NIR that we are overcoming with WGM annealing.We present an automated and programmable digital microfluidic chip capable of full actuation of the microdroplets as is needed in digital microfluidic devices. 39We experimentally establish cooling time in relation to the actuation time between positions on our digital microfluidic electrode grid.

Preprogrammed actuation in an open system chip
For compatibility with optical annealing, it is important to have a programmable open system chip.This is presented in Figure 2. We choose interdigitated electrodes as they provide enhanced actuation of microdroplets. 40The electrodes are connected to a high voltage source with through-hole via routes.The programmable open system chip can independently activate 36 electrodes arranged in a six by six configuration.The hydrophobic layer is silicone oil, which enables smooth actuation from location to location on the grid of 36 electrodes.The programmable open system chip has three parts: printed circuit board (PCB) housing firmware and electrodes for droplet interaction (JLCPCB, JiaLiChuang Company Limited, China), a voltage power supply (Keithly 2290-5) to supply 300 V DC , and a personal laptop computer to record data.Electrodes can be individually triggered in real-time, or a prerecorded sequence of electrodes can be communicated to the chip.The automated digital microfluidic driving platform is built with a graphical user interface application made through NI LabVIEW.The substrate of the chip is PCB.The electrodes have a center-to-center pitch of 2.75 mm.The bottom plate electrodes are made from copper.The material of the dielectric layer is parafilm with thickness of 135 μm ± 2 μm (mean and standard deviation based on five measurements).The chip is rendered hydrophobic through a coating of silicone oil, spread manually using a gloved finger.
A fundamental experiment is demonstrated and shown in Figure 2, whereby a representative (water) microdroplet is actuated in all directions.The sequence of the electrode activation is preprogrammed into the programmable open system chip.Following the electrode coordinate positions as labeled in (a), the microdroplet begins at c2 and then the electrodes are subsequently activated such that the microdroplet is actuated through a square pattern of (b) b2, (c) b3, (d) b4, (e) c4, (f) d4, (g) d3, (h) d2, and returning to (i) c2.Such an actuation pattern can be used for droplet cooling for microfluidics.With complete directional control, the droplet can be moved north, south, east, and west.Therefore, preprocessing or other droplet activities are achievable.The contact angle is experimentally measured to be 110 • .Given that it is an obtuse angle, there would be minimal effects for the WGM phenomenon, as the droplet is beyond a hemisphere.
The actuation of droplets has dual purpose.First, the foremost purpose is to mix, separate, and provide other such preprocessing of fluids, as is required in on-chip analyzes.Second, the actuation of droplets can move the droplet over many positions, effectively changing to a new heat sink repeatedly and potentially decreasing the cooling time.We establish this cooling time, and also the actuation time between positions on our digital microfluidic electrode grid.In showing that the actuation time between positions on our digital microfluidic electrode grid is significantly shorter than the cooling time, we see the potential benefit of this motion to more and more heat sink locations.For a 500 nL droplet, we have measured the cooling time-domain profile using our experimental setup (shown in Section 2.3), with an associated cooling time constant of 10 s.This time constant is established by mapping a 1/e point on data fitted with a decaying exponential as a function of time.In contrast, we establish that droplets in our digital microfluidic electrode grid can be actuated from one electrode (or heat sink) to the adjacent one in 300 ms, for example, each of the interelectrode motions from locations c2 (start), b2, b3, b4, c4, d4, d3, d2, to c2 (end) in Figure 2.This actuation time is significantly shorter (over 30 times shorter) than the cooling time constant.Thus, the digital microfluidic electrode grid has great potential to move droplets to many heat sink locations within the cooling time constant.Such motion can potentially increase cooling speeds.This is essentially digital microfluidic hotspot cooling performed in reverse, 41 whereby instead of moving droplets to a warm location to accept heat, droplets could be moved to a cool location to lose heat.
This complete control of the microdroplet is required for future use in complicated digital microfluidic operations.The open system configuration allows for optical annealing of droplets.This is to allow the NIR annealing light to interact with the side of the droplet.

Finite difference time-domain simulations
Optical annealing can be modeled by FDTD electromagnetic simulations, to compare traditional annealing 26 to NIR WGM annealing when implemented with an NIR source at 780 nm.These FDTD electromagnetic simulations are presented in Figure 3.A FDTD simulation of electromagnetic propagation is effective when simulating feature sizes on the order of the wavelength (i.e.,  = 780 nm and beyond).The feature of interest in this work is the WGM characteristic wave and the microdroplet.In this experiment, the WGM characteristic wave occurs when the light beam propagates around the perimeter of a microfluidic droplet.This trapping of electromagnetic radiation through a WGM characteristic wave can be modeled effectively using FDTD simulations. 42n the FDTD simulations conducted in this work, the simulation space in Figure 3 represents a cross-section of the z-axis (out-of-the-page) of the physical experiment, where 2-D solutions of Maxwell's equations are implemented to solve for the propagation of an electromagnetic pulse.The electric field (out-of-the-page) is visualized in this simulation space using a color heat map.The maxima of the pulse are represented by red electric field lines, and the minima of the pulse are represented by blue electric field lines.The outlined circle in Figure 3 represents the microdroplet that is used in the physical experimental optical annealing.The exterior and interior of the microdroplet have a different electrical permittivity.This contrast in permittivity is implemented in the discretized spatial grid to replicate the physical experimental conditions that occur between the water and air interfaces.A perfectly matched layer is implemented to absorb the electromagnetic radiation at the boundaries of the simulation space to ensure negligible boundary reflections.
The FDTD analysis begins by modeling traditional annealing in Figure 3A-C.A NIR optical pulse strikes the microdroplet on the optical axis, that is, directly normal to the microdroplet cross-section.In Figure 3A-C, time steps at 0, 125,

F I G U R E 4
The FDTD analysis with a 1300 μm 2 simulation area for WGM optical annealing with an NIR pulse.The NIR pulse can be seen approaching the left interface of the sample droplet in (A), is trapped within the droplet surface in (B) showing WGM behavior, and the continued internal reflection of the pulse is seen in (C), for an elongated interaction length.(A) t = 0 fs; (B) t = 175 fs; (C) t = 250 fs. and 150 fs are observed, respectively.The collimated NIR pulse strikes the left edge of the sample droplet, and travels towards the right edge of the 1300 μm 2 simulation area.Ultimately, the NIR pulse leaves the microdroplet, with minimal absorption over this short interaction length, approximately equal to the microdroplet diameter.
The FDTD analysis continues by modeling WGM NIR annealing in Figure 4A-C.Here, a NIR optical pulse strikes the microdroplet at approximately the radius distance off the optical axis, that is, directed on the perimeter of the microdroplet cross-section for total internal reflection and trapping of the optical pulse.In Figure 4A-C, time steps at 0, 175, and 250 fs are observed, respectively, for the same 1300 μm 2 simulation area.In Figure 4C, we note that the optical pulse is trapped within the microdroplet cross-section, indicating total-internal reflection and increased interaction length.

Experimental annealing results
To experimentally evaluate the NIR optical annealing technique a sample droplet is excited by a NIR beam, and the temperature change caused by the heat transfer through optical absorption is observed.Figure 5 shows the schematics of To characterize multiple striking points on the droplet, a micrometer fixed to a translation stage is used to give precise droplet position measurements.Using a horizontal translation step size of Δx = 50 μm, each temperature change is observed after the thermal camera temperature recording reaches steady state.The envisioned setup will have a digital microfluidic controller that can designate the position of the microdroplet.The modulation of the annealing phases can be controlled through modification of the laser optical emission power by changing the forward current on the laser diode.
The annealing experimental results are presented in Figure 6, where temperature change divided by the maximum of the temperature change, ΔT/ΔT max , of a sample droplet is plotted versus the position, Δx, of the laser with respect to the droplet center.The step size between collected data points is 50 μm and a droplet radius of 400 μm is used.The pseudo collimated NIR beam excites the sample droplet at a vertical offset from the optical axis of Δy = 0.707 × radius = 300 μm, for the 400 μm radius water droplet.The change in temperature is normalized to the point Δx = 300 μm.
Traditional optical annealing is first explored.This is represented by a horizontal offset of Δx = 0.The first inset of Figure 6 shows this case.This temperature change is (relatively) low.This is expected, given the low absorption coefficient of NIR for liquids. 27The WGM case is demonstrated in Figure 6 when Δx = 300 μm, showing maximum temperature change (and associated optical absorption of the NIR collimated beam) due to WGM resonance.At its maximum, the temperature change observed due to WGM annealing is significantly greater than that observed using traditional annealing.This is a significant improvement over both traditional IR annealing results 26 (seen at Δx = 0) and the results of IR WGM annealing 1 (which saw an improvement of 125%).The case where the beam begins to move past the region of optical excitation of the droplet is represented by Δx = 400 μm, where the beam is primarily not interacting with the sample droplet.
The peak temperature change in Figure 6 is ΔT max = 7 • C, corresponding to ΔT/ΔT max = 1.0.For Figure 6, the background temperature ΔT = 0 corresponds to an absolute temperature of room temperature, thus quantifying the complete F I G U R E 6 Experimental results of temperature change divided by the maximum temperature change, ΔT/ΔT max , versus the horizontal position, Δx, of the laser beam with respect to the center of the sample droplet.Results of traditional annealing is seen at Δx = 0 μm, whereas maximum WGM annealing is seen at Δx = 300 μm with a great improvement (ΔT/ΔT max = 1.00 at Δx = 300 μm vs ΔT/ΔT max = 0.16 at Δx = 0 μm).The data points are the mean, with error bars calculated from the mean of the standard deviations, for measurements performed in duplicate.
heating profile of the droplet.This relatively low peak temperature change is because of an experimental power limitation of the Toptica FemtoFiber PRO NIR laser, which has maximum illumination power of 90 mW at the NIR (780 nm) wavelength.The laser spot's power is 81 mW.This is for the beam after focusing (accounting for losses in the lens) and upon entering the droplet (accounting for interface losses entering the droplet).Scaling this peak temperature change to the commonly desired temperature of 95 • C 11 of a 300 nL droplet reveals a desired illumination power of approximately 1000 mW, which is achievable with a titanium sapphire laser at the same NIR (780 nm) wavelength.

HEATING AND COOLING FOR DIFFERENT DROPLET VOLUMES
Heating efficiency and droplet volume 43 will be explored with an analysis of the heating efficiency (i.e., the maximum temperature change reached), and the heating and cooling time constants, for the NIR WGM annealing.The heating efficiency decreases as droplet volume increases, seen as the decrease in WGM temperature change.This is as expected, as larger droplet volumes will be more difficult to heat.The heating and cooling time constants increase with larger volumes.This is undesirable and indicates that smaller droplet volumes should be used.Overall, this analysis of heating, cooling, and droplet volume reveals that the NIR WGM annealing of this work lends itself to use with small droplet sizes, for example, 0.5 μL and below.Although the temperature changes are low (≤7 • C), this should scale to higher temperatures with a combination of smaller droplet volumes and higher optical power, such as is available with a titanium-sapphire laser at the NIR wavelength.A thermal analysis is presented here.The heating droplet temperature temporal profile T can be described with the relationship where T 0 is the initial temperature, t is time, and  heating is the heating time constant.The cooling droplet temperature temporal profile T can be described with the relationship (2)  where  cooling is the cooling time constant.This can be shown for a representative droplet volume of 1.3 μL, corresponding to diameter of 1.1 mm.(Droplet volumes of 0.5 and 1.6 μL, corresponding to droplet diameters of 0.8 and 1.2 mm, respectively, are also studied with results summarized in Table 1.) Figure 7 shows (A) the heating temporal profile and (B) the cooling temporal profile of the representative droplet of volume 1.3 μL, with a formal curve fit tool used to extract parameters.The formal curve fit tool is the MATLAB Curve Fitting Toolbox, and Figure 7 presents one set of this data.For all data analyzed, being droplet volumes of 0.5, 1.3, and 1.6 μL (corresponding to respective droplet diameters of 0.8, 1.1, and 1.2 mm), the square of the correlation coefficient is at or above R 2 = 0.94.The T 0 value is subtracted prior to the curve fit, and thus the data clearly shows the WGM temperature change, ΔT (which is 5.4 • C in the Figure 7 data).The scaling of the droplet heating and cooling time constants is an important point of discussion, and can be seen through further thermal mass transport analysis.The thermal time constant follows the relationship where  is the density, V is the volume, C is the specific heat, h is the heat transfer coefficient, and A is the surface area.This shows a scaling of the time constant with the ratio of the volume over the surface area through Here, V scales with diameter cubed, d 3 , and A scales with diameter squared, d 2 .Thus, the time constant  scales linearly with diameter d.Additionally, the WGM temperature change, ΔT, will scale linearly with d −1 , according to ΔT ∝ d −1 . (5) In the accumulated data of Table 1, we can observe such trends.Specifically, we expect that the WGM temperature change should scale with the reciprocal of diameter, thus the multiplication of ΔT × d should stay relatively consistent through each droplet size.We expect the heating and cooling time constants should scale with d, and thus the ratio  heating /d and  cooling /d should be relatively consistent through each droplet size.From Table 1, we observe that the multiplication of droplet diameter times WGM temperature change is fairly consistent (varying between 5.6 and 5.9 mm • C).We observe that the ratio of droplet diameter over heating time constant, d/ heating , is seen to be fairly consistent (varying between 0.10 and 0.14 mm/s).We see that the ratio of droplet diameter over cooling time constant, d/ cooling , is seen to be fairly consistent (varying between 0.08 and 0.10 mm/s).Thus, the droplet heating and cooling time constant and WGM temperature change is consistent with a thermal mass transport analysis.
The authors believe evaporation will have minimal importance to the ultimate implementation.This is because in digital microfluidic devices evaporation issues have been widely explored and mitigated through techniques demonstrated in the literature, for example, droplet coatings demonstrated. 44Incorporating such a mitigation technique would have minimal effects on our annealing technique because the impedance match is excellent with aqueous solutions, for example, polytetrafluoroethylene has a refractive index of 1.38 45 and water has a refractive index of 1.33, yielding an optical power reflection that is miniscule, that is, 0.03%.

CONCLUSION
This work demonstrated enhanced performance for NIR optical annealing of microdroplets with the application of the WGM method.This hypothesis was supported by FDTD simulations and experimental results that characterized NIR illumination as a function of illumination position on a microfluidic droplet.This work presented a solution to scaling limitations found in microfluidic devices, as the dependence on droplet diameter is removed, and showed that optimal absorption can be achieved with samples of low volume.The work also demonstrated the required programmable droplet actuation in an open system chip.Of particular interest is the application for point-of-care devices.
Conceptual images are shown with (A) microdroplet annealing with WGM NIR annealing and (B) digital microfluidic microdroplet actuation.

F
I G U R E 2 Experimental results of the automated open-system chip are presented here.The open-system chip is preprogrammed to apply voltage on the electrodes systematically to actuate the microdroplet around a square pattern and return to its original position through images (A)-(I).Corresponding time is noted in the counter at the top of each subfigure image.(A) t = 0.00 s; (B) t = 0.13 s; (C) t = 0.33 s; (D) t = 0.63 s; (E) t = 0.90 s; (F) t = 1.10 s; (G) t = 1.40 s; (H) t = 1.63 s; (I) t = 1.96 s.

F I G U R E 3
The FDTD analysis with a 1300 μm 2 simulation area for traditional optical annealing with an NIR pulse.The NIR pulse can be seen approaching the left interface in (A), passing through the center of the sample droplet in (B) where annealing will occur, and exiting through the right interface in (C).(A) t = 0 fs; (B) t = 125 fs; (C) t = 150 fs.

F I G U R E 5
The top views for (A) traditional optical annealing and (B) WGM optical annealing for the NIR source are shown.In (A), the beam strikes the center line of the sample droplet and a large portion of the energy of the beam is seen passing through the sample droplet, unabsorbed.In (B), the beam strikes the outer periphery of the sample droplet and shows the trapping of the beam within the surface due to internal reflection, for enhanced optical absorption.the experimental setup used for testing the traditional and WGM annealing methods.The topviews of the experimental schematics, seen in Figure5A,B, show traditional NIR annealing and NIR WGM annealing, respectively.Here, the NIR source is directed to strike the center and periphery of the sample droplet, respectively.The collimated NIR beam (originating from a Toptica FemtoFiber PRO NIR laser) is pseudo collimated within the microdroplet by a ten times focusing microscope objective, with a focal length of 2.54 cm.The beam strikes the sample droplet at designated points and the change in temperature is observed by a thermal camera.

Note: 7
The multiplication of droplet diameter and WGM temperature change is shown.The ratio of droplet diameter over heating time constant is shown.The ratio of droplet diameter over cooling time constant is shown.The curve fitting of the heating and cooling temporal time constants is shown for a droplet of diameter d = 1.1 mm and volume V = 1.3 μL.The value differential temperature represents the temperature above the background (i.e., T initial ) temperature.The black marker points are the experimental data and the blue curve is the line of best fit to the experimental data.This formal curve fit is accomplished using the MATLAB Curve Fitting Toolbox.
Heating and cooling time constants for different droplet diameters and volumes is shown in this table.
TA B L E 1