micromachines

Article A - Technique for Portable Micropneumatic Oscillators

Vidhya Balaji 1,* ID , Kurt Castro 2 and Albert Folch 2

1 Department of Electrical Engineering, University of Washington, Seattle, WA 98195, USA 2 Department of Bioengineering, University of Washington, Seattle, WA 98195, USA; [email protected] (K.C.); [email protected] (A.F.) * Correspondence: [email protected]; Tel.: +1-206-685-2257

 Received: 13 July 2018; Accepted: 22 August 2018; Published: 24 August 2018 

Abstract: Microfluidic automation technology is at a stage where the complexity and cost of external hardware control often impose severe limitations on the size and functionality of microfluidic systems. Developments in autonomous microfluidics are intended to eliminate off-chip controls to enable scalable systems. Timing is a fundamental component of the digital logic required to manipulate fluidic flow. The authors present a self-driven pneumatic oscillator manufactured by assembling an elastomeric sheet of polydimethylsiloxane (PDMS) between two laser-engraved polymethylmethacrylate (PMMA) layers via surface activation through treatment with 3-aminopropyltriethoxysilane (APTES). The frequency of the fabricated oscillators is in the range of 3–7.5 Hz with a maximum of 14 min constant frequency syringe-powered operation. The control of a fluidic channel with the oscillator stages is demonstrated. The fabrication process represents an improvement in manufacturability compared to previous molding or approaches, and the resulting devices are inexpensive and portable, making the technology potentially applicable for wider use.

Keywords: microfluidic automation; laser engraving; manufacturability; pneumatic oscillator; APTES surface activation

1. Introduction Microfluidic automation can feature large arrays of valves and pumps performing multiple analyses for diagnostic assays and biological applications [1]. The development of complete systems that integrate sample preparation, fluidic manipulation, and detection mechanisms [2] is associated with increasingly complex interconnections and microvalve automation. Microfluidic very large-scale integration (mVLSI) [3] has enabled parallel fluidic manipulations using thousands of microvalves, thereby alleviating some of the burden. However, the associated control hardware requires many external connections, thereby making the management of its operation prone to errors (leaks, etc.) and cumbersome to operators unfamiliar with the technology. Designs that minimize the number of control signals [4] have provided an impetus to parallelization, but they have not eliminated entirely the challenge of external connection complexity. In order to drastically reduce off-chip connections, researchers have explored the integration of microfluidic logic control elements on chip of varying complexity [5–8] with the aim of facilitating the development of simple-to-use, portable devices. These circuits have been fabricated with time and labor-intensive techniques such as soft lithography [9] and etching, limiting the manufacturability and mass production capability. The expansion of microfluidic circuits into the realm of point-of-care testing (POCT) hinges on the simplification of the manufacturing process and the automation of the steps involved [10]. To this end, computer numerical control (CNC) milling [11]

Micromachines 2018, 9, 426; doi:10.3390/mi9090426 www.mdpi.com/journal/micromachines Micromachines 2018, 9, 426 2 of 11 and laser engraving [12] have been explored as techniques for microfluidics fabrication. Micro-milled three-dimensional microfluidic flow cells have demonstrated the potential of milling for medical diagnostics [13]. Laser-engraving has emerged specifically as a very promising technology due to factors such as fastest turnaround time, comparatively low operational costs and ease of design processing. A wide range of can be processed by a laser cutter, including thermoplastics such as polymethylmethacrylate (PMMA). Thermoplastics are extensively used for microfluidics due to their high manufacturability and properties amenable to microfluidic analysis [14,15]. Valves and pumps have been demonstrated by sandwiching polydimethylsiloxane (PDMS) between channel structures created on PMMA [16]. Implementing self-driven microfluidic devices requires an on-chip timing reference or clock generation circuit [17]. Pneumatic oscillators fabricated with glass and PDMS have been characterized by Duncan et al. [7]. Scalability of microfluidic logic circuits using prototyping techniques such as CNC milling has also been shown [18]. In this paper, the authors demonstrate an all- (PMMA/PDMS) ring oscillator chip with features manufactured using laser engraving and driven with a syringe which acts as a simple and portable pneumatic power source. The device is assembled by permanently bonding the three layers using chemical surface activation in contrast to mechanical approaches to keeping the layers together, such as binder clips [19] which cannot form a direct bond between the PDMS and PMMA [20].

2. Materials and Methods The ring configuration, wherein an odd number of NOT gates or inverters are connected in a feedback loop, is one of the standard structures for an oscillator used in semiconductor electronics. A fluidic equivalent is based on the analogy between the normally off n-type metal oxide semiconductor field effect transistor (N-MOSFET) operating with a voltage differential and a normally closed valve operating between two pressures—in this case, psi of vacuum (VAC) and atmosphere (ATM). As shown in Figure1, the normally closed valve consisting of a PDMS membrane sandwiched between a PMMA flow layer and a PMMA control layer is connected to the pressure supply (VAC) through a fluidic resistor to form a pneumatic inverter. A logic 0 at the input pulls the output to a 1 through the resistor, whereas an input of 1 opens the valve, driving the output to 0. Device layers of the three-stage pneumatic ring oscillator are fabricated by laser engraving (CO2 laser VersaLaser, Universal Laser Systems, Scottsdale, AZ, USA; 10.2 µm wavelength; 10–60 W power) in PMMA (Astari-Niagara, Jakarta, Indonesia; 0.125-in thick). Micromachines 2018, 9, 426 3 of 11 Micromachines 2018, 9, x FOR PEER REVIEW 3 of 11

Figure 1. (A) Three-layer valve structure before assembly; (B) N-type metal oxide semiconductor Figure 1. (A) Three-layer valve structure before assembly; (B) N-type metal oxide semiconductor (NMOS) inverter (left) along with micropneumatic equivalent (right). Vdd, voltage supply in volts; (NMOS) inverter (left) along with micropneumatic equivalent (right). Vdd, voltage supply in volts; VAC, vacuum supply in psi; ATM, atmospheric pressure; R, pull-up resistance; (C) Pneumatic ring VAC, vacuum supply in psi; ATM, atmospheric pressure; R, pull-up resistance; (C) Pneumatic ring oscillator with the connection between the output of the third inverter and the input of the first oscillator with the connection between the output of the third inverter and the input of the first inverter inverter being the feedback loop; (D) Laser-cut valve (top-view); (E) Height profile of the flow being the feedback loop; (D) Laser-cut valve (top-view); (E) Height profile of the flow chambers and chambers and seat of the valve along the section denoted by the blue dotted line in (D); (F) Laser-cut seat of the valve along the section denoted by the blue dotted line in (D); (F) Laser-cut ring oscillator ring oscillator with silicone tubes attached as vacuum supply inlet for testing the operation. with silicone tubes attached as vacuum supply inlet for testing the operation.

2.1. Parameters of Design and Assembly Process 2.1. Parameters of Design and Assembly Process Oscillation frequency is dependent on two main design parameters: (1) inverter stage resistance Oscillation frequency is dependent on two main design parameters: (1) inverter stage resistance Rs—the sum of valve, pull-up, and parasitic interconnect resistances, and (2) valve chamber volume Rs—the sum of valve, pull-up, and parasitic interconnect resistances, and (2) valve chamber volume Vc Vc (which directly impacts the pneumatic capacitance of the valve). The cross-section of the (which directly impacts the pneumatic capacitance of the valve). The cross-section of the microfluidic microfluidic channel representing the fluidic resistor is approximately rectangular with a designed channel representing the fluidic resistor is approximately rectangular with a designed dimension of dimension of 250 µm × 50 µm (width × height). The fluidic resistance of a rectangular channel is: 250 µm × 50 µm (width × height). The fluidic resistance of a rectangular channel is: 12µ퐿 푅 = 푓 12uL3 (1) R = 푤ℎ 푓 (1) f wh3 f where h is the height of the channel, w is the width, L is the length, µ is the viscosity, and f is the wheregeometrich is form the height factor ofrelated the channel, to the rectangularw is the width, shape.L Whenis the h length, << w, f~µ1 isdoes the not viscosity, influence and thef is value the geometricof Rf. Pressure form factor supply related (Ps) is to the the operational rectangular parameter shape. When thath << dynamicallyw, f ~1 does influences not influence the the oscillator value offrequency.Rf. Pressure Theoretically, supply (P sthe) is time the operationalperiod is proportional parameter thatto (R dynamicallys × Vc)/Ps, analogous influences to the the expression oscillator frequency.for the time Theoretically, constant of an the RC time circuit. period The is c proportionalorrelation with to (experimentalRs × Vc)/Ps, analogousvalues was to confirmed the expression upon fortesting the time the constant oscillators. of anThe RC Poiseuille’s circuit. The equation correlation for with compressible experimental fluids values provides was confirmed an analytical upon testingrelation the between oscillators. pressure, The Poiseuille’s volumetric equation flow rate for, and compressible resistance fluids through provides a pneumatic an analytical circuit relation with a betweenchannel and pressure, chamber: volumetric flow rate, and resistance through a pneumatic circuit with a channel and chamber: 푃 − 푃 푃 + 푃 P 푣− P 푐 P푣 + P푐 푄=푡 = v c v c (2) Qt 푅푐ℎ 2푃푐 (2) Rch 2Pc where Pv is the vacuum supply, Pc is the chamber pressure, and Rch is the fluidic resistance of the where Pv is the vacuum supply, Pc is the chamber pressure, and Rch is the fluidic resistance of the channelchannel connectingconnecting thethe supplysupply toto thethe chamber.chamber. TheThe fluidicfluidic resistanceresistance ofof thethe channelchannel waswas calculatedcalculated asas perper EquationEquation (1).(1). AA 2D2D schematicschematic ofof thethe oscillatoroscillator (Autodesk(Autodesk InventorInventor ProfessionalProfessional 2017,2017, Autodesk,Autodesk, SanSan Rafael, Rafael, CA, CA, USA) USA was) was drawn drawn with with pull-up pull resistors-up resistors and valve and sizes valve based sizes on based the targeted on the frequency. targeted Thefrequency. line diagram The line (AutoCAD diagram (AutoCAD 2017, Autodesk, 2017, SanAutodesk, Rafael, San CA, Rafael, USA)derived CA, USA by) derived identifying by identifying the center linesthe center of symmetry lines of ofsymmetry the channels of the and channels the valve and sections the valve was sections then used was directly then used as input direct toly the as laser’s input to the laser’s proprietary software. A calibration of the engraved line widths with different speed/power settings for a given schematic line width served as a reference to determine the

Micromachines 2018, 9, 426 4 of 11 proprietary software. A calibration of the engraved line widths with different speed/power settings for a given schematic line width served as a reference to determine the appropriate feature dimensions for the oscillator schematic. For the range of channel widths used in the design, the raster mode of the laser was found to be suitable (laser speed: 20–40% of maximum; power: 18–20% of maximum). The consequent multiple scans of the laser-enhanced surface roughness indicated the profile of a valve flow chamber, as shown in Figure1E. The feature depth considered for resistance calculations was the average depth of the profile. A Dektak profilometer (Bruker, Billerica, MA, USA) measured the dimensions of a single valve as 1.4 mm × 1.2 mm (Length × Width) with an average depth of 65 µm. Two or three rinses with isopropyl alcohol (IPA) followed by soaking in deionized water for approximately 10 min were required to remove leftover debris from the features on the PMMA surface after laser engraving. The assembly of the PMMA layers with PDMS film (Rogers HT-6240, 254 µm, Rogers, Woodstock, CT, USA) was accomplished in three steps: (1) Surface activation of PMMA with 3-aminopropyltriethoxysilane (99% APTES, Sigma-Aldrich, St. Louis, MO, USA). After trials of varying the concentration of APTES (1–5%), 5% solution heated to 85 ◦C provided the optimal outcome for bonding. Soaking the PMMA for 30 min in heated APTES solution was adequate; (2) O2 plasma treatment of the PDMS (Diener Zepto, Ebhausen, Germany, at 660 mTorr O2); and (3) Placing the PDMS on the PMMA in conformal contact for bonding and heating in an oven at 70 ◦C for 2 h. Via holes were created in the bonded PMMA-PDMS piece which was then sandwiched with the remaining PMMA layer after APTES and O2 plasma steps were carried out as described above. The assembly was then clamped between flat metal plates to ensure uniform pressure across the device area before placing it in the oven. Upon removal from the oven after 2 h, silicone tubes (Cole Parmer, Vernon Hills, IL, USA, 1.14 mm inner diameter) bonded to PDMS were attached to the vacuum inlet using a silicon-silicone adhesive tape (S1001-1DC11, Adhesive Applications Inc., Easthampton, MA, USA) to carry out testing. The significance of using the APTES approach was the strong and permanent bond formed between the PMMA and PDMS. Comprehensive studies of the bonding of PMMA and PDMS substrates have shown that reliable and reproduceable results are possible with this technique [21]. The experiments carried out for this study confirmed this observation after the procedure was standardized for surface activation and assembly. Additionally, the bond strength of the fabricated devices remained unchanged within a time duration of a few months, demonstrating the structural stability required for use in portable and hand-held application scenarios.

2.2. Characterization Technique A movie of the oscillating valve was captured with a DSLR camera (Canon EOS 5D Mark II at 30 fps, Canon, Tokyo, Japan) mounted on a microscope (Nikon SMZ1500, Nikon, Tokyo, Japan); the movie was split into frames (VLC 2.1.5 media player, VideoLAN, Paris, France) for analysis. Figure2 (Multimedia view) illustrates the frequency calculation using ImageJ (National Institutes of Health, Bethesda, MD, USA) to analyze the frames. At a fixed location from the vertical edge of the valve seat, the distance between the membrane contour (yellow dashed line) and a horizontal edge of the valve seat (referred to as contour offset) was measured for each frame. The time period of oscillations was the time interval between frames having minimal difference in contour offsets. The consistency of the value over a few cycles has confirmed the technique’s reliability. Micromachines 2018, 9, 426 5 of 11 Micromachines 2018, 9, x FOR PEER REVIEW 5 of 11

Micromachines 2018, 9, x FOR PEER REVIEW 5 of 11

Figure 2. Frequency calculation using ImageJ software. The distance from an edge of the valve seat Figure 2. Frequency calculation using ImageJ software. The distance from an edge of the valve seat and the membrane contour (yellow dashed curve) against the PMMA surface is measured for each and the membrane contour (yellow dashed curve) against the PMMA surface is measured for each movie frame of the oscillation. The time interval between similar distances provides the oscillation movieperiod. frame Frames of the 1 oscillation.and 7 have similar The time distances interval of 134 between µm and similar 131 µm, distances respectively, provides corresponding the oscillation to period.a period Frames of six 1 andframes 7 have or a 0.2 similar s oscillation distances period of (Multimedia 134 µm and view) 131 µ. m, respectively, corresponding to a period of sixFigure frames 2. Frequency or a 0.2 calculation s oscillation using period ImageJ (Multimediasoftware. The distance view). from an edge of the valve seat 3. Results andand D theiscussion membrane contour (yellow dashed curve) against the PMMA surface is measured for each 3. Results and Discussionmovie frame of the oscillation. The time interval between similar distances provides the oscillation 3.1. Frequencyperiod. Measurement Frames 1 and 7 have similar distances of 134 µm and 131 µm, respectively, corresponding to a period of six frames or a 0.2 s oscillation period (Multimedia view). 3.1. FrequencyThe variation Measurement in frequency of oscillations with the magnitude of vacuum applied (via Elveflow 3. Results and Discussion pressureThe variation controller in frequency, Elveflow, of Paris, oscillations France) across with the devices magnitude fabricated of vacuum and assembled applied in (via different Elveflow batches is plotted in Figure 3. Frequency increases were directly proportional to vacuum over a pressure controller,3.1. Frequency Elveflow, Measurement Paris, France) across devices fabricated and assembled in different batches vacuum supply range of 3–6 psi (3.08 Hz at 6 psi being twice the 1.48 Hz oscillations at 3 psi). The is plottedcurve instart FigureedThe to3 variation.level Frequency off in at frequency VAC increases = 8 of psi oscillations were attaining directly with an theaverage proportional magnitude 4.15 Hz,of to vacuum suggestive vacuum applied over of (via the a vacuumElveflow saturation supply rangeobserved of 3–6pressure in psi an (3.08 equivalent controller Hz at, MOSFET’sElveflow, 6 psi being Paris, voltage twice France transfer the) across 1.48 characteristics. devices Hz oscillations fabricated A low and at pressure assembled 3 psi). thresholdThe in different curve of 2.5 started to levelpsi was off atobservedbatches VAC is =below plotted 8 psi which in attaining Figu oscillationsre 3. anFrequency average cease, increases corresponding 4.15 Hz,were suggestivedirectly to t he proportional threshold of the saturation tovoltage vacuum required over observed a to in vacuum supply range of 3–6 psi (3.08 Hz at 6 psi being twice the 1.48 Hz oscillations at 3 psi). The an equivalentturn on a NMOS. MOSFET’s The frequency voltage–vacuum transfer supply characteristics. relationship A observed low pressure correlates threshold to the alpha of 2.5power psi was curve started to level off at VAC = 8 psi attaining an average 4.15 Hz, suggestive of the saturation observedlaw commonly belowobserved which used in an oscillationsto equivalent estimate MOSFET’s delay cease, in CMOS correspondingvoltage circuits transfer withcharacteristics. to theshort threshold channel A low transistorspressure voltage threshold required [22], implying of 2.5 to turn on a NMOS.an analogous Thepsi frequency–vacuumwas “short observed-seat” below effect which in supply the oscillations valve. relationship cease, corresponding observed to correlatesthe threshold tovoltage the alpharequired power to law commonly usedturn on to a estimateNMOS. The delay frequency in CMOS–vacuum circuits supply relationship with short observed channel correlates transistors to the alpha [22], power implying an law commonly used to estimate delay in CMOS circuits with short channel transistors [22], implying analogous “short-seat” effect in the valve. an analogous “short-seat” effect in the valve.

Figure 3. Oscillator performance with different magnitudes of constant vacuum supply. Error bars

denote SD across three devices. The linear trend correlates closely with the frequency–vacuum relationship.Figure 3. Oscillator performance with different magnitudes of constant vacuum supply. Error bars Figure 3. Oscillatordenote SD performance across three with devices. different The linear magnitudes trend correlates of constant closely vacuum with the supply. frequency Error–vacuum bars denote SD across threerelationship. devices. The linear trend correlates closely with the frequency–vacuum relationship. Near the vacuum threshold of oscillation (2.5 psi), a higher variation of frequency across devices was observed,Near indicating the vacuum a sensitivity threshold of to o processscillation variations, (2.5 psi), a higher particularly variation in of the frequency adhesion across of the devices PDMS toNear the PMMA thewas vacuum observed, over the threshold indicating valve seat a of sensitivity area oscillation after to bonding. process (2.5 psi), variations, An aexamination higher particularly variation of invalve the of adhesion operatio frequency ofn thethrough across PDMS the devices was observed,capturedto videos the indicating PMMA at frequencies over a sensitivitythe valve below seat to area 2 processHz after confirm bonding. variations,ed Anthis e xaminationhypothesis. particularly of Asvalve infrequency theoperatio adhesionn increas through ofed the the, the PDMS to thelower PMMA variationcaptured over across videos the valve devicesat frequencies seat could area belowbe afterdue 2 toHz bonding. the confirm adhesioned An this being examinationhypothesis. less of As a dominatingfrequency of valve increas operation factor.ed , Asthe throughthe lower variation across devices could be due to the adhesion being less of a dominating factor. As the the capturedsensitivity videos did not at appear frequencies to be a belowlimitation 2 Hz forconfirmed higher frequency this hypothesis. generation, the As frequencydevices could increased, be suitable sensitivityfor medical did applications not appear to that be arequire limitation rapid for analysis.higher frequency A frequency generation, of 4.15 the Hz devices (at VAC could = 8 be psi), the lower variationsuitable for across medical devices applications could that be require due torapid the analysis. adhesion A frequency being less of 4.15 of Hz a dominating (at VAC = 8 psi), factor. As for example, would be useful for a controlled pump sequence. the sensitivityfor didexample, not appearwould be to useful be a for limitation a controlled for pump higher sequence. frequency generation, the devices could be After observing the oscillation frequency stability over a period of a few hours, the authors made suitable for medicalAfter applicationsobserving the oscill thatation require frequency rapid stability analysis. over a Aperiod frequency of a few hours, of 4.15 the Hz authors (at VACmade = 8 psi), a preliminarya preliminary assessment assessment about about the usefthe usefulnessulness of theof the oscillator oscillator as as aa practicalpractical timingtiming reference. reference. As As for example, would be useful for a controlled pump sequence.

After observing the oscillation frequency stability over a period of a few hours, the authors made a preliminary assessment about the usefulness of the oscillator as a practical timing reference. Micromachines 2018, 9, 426 6 of 11

Micromachines 2018, 9, x FOR PEER REVIEW 6 of 11 As FigureMicromachines4 illustrates, 2018, 9, x FOR the PEER frequency REVIEW increased by 25% after more than two hours of operation6 of 11 at Figure 4 illustrates, the frequency increased by 25% after more than two hours of operation at a a constant vacuum supply and stayed at the new value for another two hours. This drift implied Figureconstant 4 illustrates, vacuum thesupply frequency and stayed increased at the by new 25% value after for more another than twotwo hourshours. ofThis operation drift impli at aed a a loweringconstantlowering of vacuum the of threshold the supply threshold and pressure stayed pressurerequired at requiredthe new to value to open open for the the another valve, valve, two attributable attributable hours. This most most drift probably probably implied to a to the the decreaseloweringdecrease in adhesion of thein adhesion threshold of PDMS of pressurePDMS to the to required PMMAthe PMMA over to openover the the long long valve, duration duration attributable [ 7[7].]. TheThe most continuous probably frequency to frequency the increasedecreaseincrease after in a adhesionafter period a period of of constant PDMS of constant to oscillationsthe PMMAoscillations over suggested suggest the longed an durationan apparent apparent [7]. threshold Thethreshold continuous of of surface surface frequency adhesion adhesion deteriorationincreasedeterioration after for a the period for membrane. the of membrane. constant The oscillations The deterioration deterioration suggest continueded continu an apparented linearly linearly threshold and stabiliz stabilized of surfaceed after adhesion after an an hour hour,, beyonddeteriorationbeyond which which oscillations for the oscillations membrane. became bec Theame constant deterioration constant again. again. continu BeyondBeyonded linearly the the five five and hours hours stabiliz of ofoscillationsed oscillations after an plotted hour plotted, in in FigurebeyondFigure4, thewhich 4, frequencythe oscillations frequency remained remainedbecame constant steady steady afterafteragain. further further Beyond observation observation the five hours suggesting suggesting of oscillations the possibility the plotted possibility of inlong - of long-termFigureterm stable 4, stablethe operation.frequency operation. remained After After the the steady cessation cessation after andfurtherand subsequentsubsequent observation initiation initiation suggesting of of the the the oscillations, oscillations, possibility theof the long starting starting- frequency was unaltered (pre-drift value of 3 Hz as shown in Figure 4), implying that the original frequencyterm stable was unaltered operation. (pre-drift After the valuecessation of 3 and Hz assubsequent shown in initiation Figure4), of implying the oscillations, that the the original starting PDMS frequencyPDMS adhesionwas unaltered was restored.(pre-drift This value behavior of 3 Hz reflect as showned an in initial Figure settling 4), implying time for that the the oscillator original and adhesion was restored. This behavior reflected an initial settling time for the oscillator and supported PDMSsupport adhesioned the was concept restored. of a preThis-operational behavior reflect runninged an period initial to settling ensure timestable for oscillations the oscillator for onand-chip the conceptsupportapplications.ed of athe pre-operational concept of a pre running-operational period running to ensure period stable to ensure oscillations stable oscillations for on-chip for applications.on-chip applications.

Figure 4. Stability of frequency with time for an oscillator operated with the vacuum supply set to 8 Figure 4. Stability of frequency with time for an oscillator operated with the vacuum supply set to 8 psi. Figurepsi. 4. Stability of frequency with time for an oscillator operated with the vacuum supply set to 8 psi. The frequencyThe frequency vs. vacuum vs. vacuum pressure pressure applied applied relationship relationship for for an an oscillator oscillator with a a smaller smaller pull pull-up-up resistanceresistanceThe is depictedfrequency is depicted invs. Figure vacuum in Fig5.ure Halvingpressure 5. Halving applied the resistancethe relationshipresistance did did notfor not an produce produce oscillator a a proportionate withproportionate a smaller decreasepull decrease-up in in oscillationresistanceoscillation period. is depicted period. At VAC inAt Fig =VAC 4ure psi = 5. 4 and Halvingpsi and 6 psi, 6 the psi, the resistance the corresponding corresponding did not produce average average a frequencies frequenciesproportionate were were decrease close close to in tot hat that of theoscillation oscillatorof the oscillatorperiod. with pull-upAt with VAC pull resistance= 4 -psiup and resistance R6 psi,(see the FigureR corresponding(see 1FigC).ure A reason1C). average A reasonfor frequencies the for frequency the frequencywere remaining close remainingto that largely of thelargely oscillator unaffected with pullcould-up be resistance the contribution R (see Fig ofure the 1C). parasitic A reason resistance for the of frequency the interconnect remaining lines unaffected could be the contribution of the parasitic resistance of the interconnect lines (connecting largely(connecting unaffected the valvecould stages)be the and contribution valve resistance of the domi parasiticnating resistance the overall of resistance the interconnect in comparison lines to the valve stages) and valve resistance dominating the overall resistance in comparison to the pull-up (connectingthe pull- upthe resistance.valve stages) Reducing and valve the resistance valve and domi interconnectnating the line overall lengths resistance enabl edin comparisonthe tuning ofto the resistance.the frequencypull Reducing-up resistance. with the thevalve pullReducing-up and resistor interconnectthe valve length and and lineinterconnect simultaneously lengths enabledline lengthsachiev theed enabl tuning areaed optimization. ofthe the tuning frequency of the with the pull-upfrequency resistor with the length pull-up and resistor simultaneously length and simultaneously achieved area optimization.achieved area optimization.

Figure 5. Oscillator frequency response with inverter stages having half the pull-up resistance R/2 Figure(top) 5. ofOscillator the device frequency shown in response Figure 1C. with inverter stages having half the pull-up resistance R/2 Figure 5. Oscillator frequency response with inverter stages having half the pull-up resistance R/2 (top) of the device shown in Figure 1C. (top) of the device shown in Figure1C.

Micromachines 2018, 9, 426 7 of 11 Micromachines 2018, 9, x FOR PEER REVIEW 7 of 11

3.2. Portability3.2. Portability Experiment Experiment TheThe portability portability of theof the device device was was demonstrated demonstrated by by powering powering the the oscillatoroscillator with a syringe. syringe. The The vacuumvacuum levellevel appliedapplied was was measured measured by by connecting connecting a tubea tube at theat the mouth mouth of theof the syringe syringe to an to analog an analog pressurepressure gauge. gauge. Applying Applying a steady a vacuum steady by vacuum clamping by a clamping pulled 60 mLa pulled syringe 60 generated mL syringe oscillations generated thatoscillations could be sustained that could for be as sustained long as 14 for min as at long an averageas 14 min frequency at an average of 7.5 freq Hz,uency as shown of 7.5 in Hz Figure, as shown6 whichin depictsFigure 6 the which three depicts characteristic the three performances characteristic of performances seven oscillators of seven tested. oscillators Although tested. the average Although frequencythe average of five frequency oscillators of was five between oscillators 3 andwas4.2 between Hz, two 3 and pulsated 4.2 Hz, at two approximately pulsated at doubleapproximately the valuedouble (7.5 Hz), the manifestingvalue (7.5 Hz), a property manifesting of the a property laser-engraving of the laser process:-engraving the effect process: of channel the effect roughness of channel on theirroughness average on depth. their Features average made depth. using Features the same made raster using mode the laser same settings raster on mode different laser substrates settings on haddifferent a variability substrates in roughness had a variability depending in on roughness the scanning depending pattern on of the the scanning beam for pattern the particular of the beam run. for Thethe resultant particular variation run. The in average resultant channel variation depth in average factors channel into fluidic depth resistance factors into by anfluidic inverse resistance cube. by A depthan inverse of ~85 cube.µm would A depth reduce of ~85 the µm channel would resistance reduce the by channel almost resistance a third and by the almost oscillation a third time and the periodoscillation by approximately time period 60%. by approximately 60%. AfterAfter the initialthe initial period period of steady of steady oscillations, oscillations, the eventual the eventual frequency frequency decline decline observed observed in Figure in Fig6 ure could6 could be caused be caused by the by syringe the syringe losing losing vacuum. vacuum. During During valve valve closing, closing, air was air pulledwas pulled into into the valvethe valve chamberchamber from from the atmosphere.the atmosphere. When When the valvethe valve opened, open aired,in air the in chamberthe chamber flowed flow outed throughout through the the pull-uppull resistor-up resistor into the into supply the supply line which line was which attached was toattached the syringe. to the The syringe. time interval The timefor frequency interval for drop-offfrequency to zero drop from-off the to initial zero from value the was initial 13–14 value min was across 13– devices,14 min across indicating devices, a uniform indicating rate a of uniform air flowrate into of the air syringe. flow into This the unvarying syringe. This air flowunvarying was due air to flow the dependencywas due to the of thedependency flow on the of pull-upthe flow on resistorthe ratherpull-up than resistor the frequency rather than of the valve frequency operation. of valve operation.

FigureFigure 6. Variation 6. Variation of oscillation of oscillation frequency frequency with with time time using using a syringe a syringe clamped clamped at 47.4 at mL47.4 level mL level as the as the supply.supply. Ring Ring oscillator oscillator powered powered by a by syringe a syringe clamped clamped with with a paper a paper clip (insetclip (inset picture). picture). Error Error bars bars denotedenote SD across SD across three three trials. trials.

3.3. Flow3.3. Flow Control Control in Fluidic in Fluidic Channel Channel A circuitA circuit wherein wherein the individualthe individual ring ring oscillator oscillator inverter inverter outputs outputs are drivingare driving three three valves valves in series in series in a fluidicin a fluidic channel channel is shown is shown in Figure in7 Fig. Theure figure 7. The also figure depicts also the depicts relative the phases relative of the phases three inverterof the three outputsinverter and their outputs dependence and their onthe dependence time delay of on each the inverter time delaystage and of the each interconnecting inverter stage channel and the l xl x yl y x x y y x y delay:interconnecting If τl = RC, τxl = channelRxC, and delayτyl: =IfR yτC=( RRCxC,x ,τR y=C yR, RCC, xand, and τ RC = yRareC negligible),(R C , R C , theRC equivalent, and RC are D l D l xl D l yl timenegligible), delay at the the input equivalent of each time stage delay is T atD the+ τ linput, TD + ofτ eachl + τ xlstage, and isT TD + τ l, T+ τ +y lτ, where+ τ , andTD Tis + the τ + τ , where TD is the time delay of each inverter. The corresponding relative phase shift at each stage is τ τ as follows: 0°, 푥푙 × 360°, 푦푙 × 360°. 푇퐷+τ푙+τ푥푙 푇퐷+τ푙+τ푦푙

Micromachines 2018, 9, 426 8 of 11

◦ timeMicromachines delay of 2018 each, 9, x inverter. FOR PEER The REVIEW corresponding relative phase shift at each stage is as follows:8 0of 11, τ τ xl × 360◦, yl × 360◦. TD+τl +τxl TD+τl +τyl ByBy modifying modifying the the channel channel dimensions dimensions and and consequently consequently the resistancethe resistance between between the output the output of one of stageone stage and the and input the input of another of another (specifically (specificallyRx and Rx andRy), R they), the phase phase difference difference can can be be suitably suitably varied. varied. TheThe circuit circuit serves serves to to demonstrate demonstrate the the possibility possibility of of a a timed timed sequence sequence of of opening opening and and closing closing of of valves, valves, usefuluseful in in an an assay, assay, for for example, example, where where specific specific reagents reagents need need to to be be mixed mixed in in at at different different times. times. Red-coloredRed-colored water water assisted assisted in in the the visualization visualization of of the the phase phase shifts shifts of of the the pulsating pulsating fluidic fluidic valves valves (see(see Figure Figure8, 8, Multimedia Multimedia view). view). The The three three phases phases corresponding corresponding to to the the state state of of the the valve valve in in sequential sequential orderorder are are the the following: following partially: partially open, open open,, open and, and closed. closed The. The time time period period of of fluidic fluidic valve valve operation operation withwith a a vacuum vacuum supply supply of of 8 8 psi psi for for the the oscillator oscillator circuit circuit was was 0.2 0.2 s withs with each each phase phase corresponding corresponding to to ~0.067~0.067 s. s. As As seen seen in in Figure Figure8, 8, the the open-to-close open-to-close transition transition was was sudden, sudden whereas, whereas the the close-to-open close-to-open transitiontransition was was phased-out phased-out with with an an intermediate, intermediate, partially-open partially-open state. state. This This behavior behavior is attributed is attributed to the to elasticitythe elasticity of the of PDMS the PDMS and the and strength the strength of its of adhesion its adhesion to the to PMMA the PMMA valve valve seat surface. seat surface.

Figure 7. (A) Oscillator driving three valves in series out-of-phase in a fluid line. (B) Equivalent circuit Figure 7. (A) Oscillator driving three valves in series out-of-phase in a fluid line. (B) Equivalent circuit for ring oscillator with varying interconnect channel lengths. TD is the time delay of each inverter for ring oscillator with varying interconnect channel lengths. TD is the time delay of each inverter stage; stage; R, C represent the resistance and capacitance, respectively, of the minimum connection line R, C represent the resistance and capacitance, respectively, of the minimum connection line between between stages; Rx, Cx and Ry, Cy represent the resistance and capacitance of the delay line of a specific stages; Rx, Cx and Ry, Cy represent the resistance and capacitance of the delay line of a specific length ‘x’length and ‘y ’,‘x respectively,’ and ‘y’, respectively added to, theadded minimum to the connectionminimum connection line to obtain line the to necessary obtain the phase necessary difference. phase Indifference. the corresponding In the corresponding fluidic circuit, fluidic the interconnects circuit, the interconnects have negligible have capacitances negligible capacitances and the resistance and the resistance primarily contributes to the phase shifts. (C) The phase diagram shows the shifts primarily contributes to the phase shifts. (C) The phase diagram shows the shifts introduced by R, Rx andintroduced how they by affect R, R thex and delay how of they the affect subsequent the delay stage of output.the subsequent stage output.

The closed valve remained in that state due to the PDMS valve seat stiction until a threshold The closed valve remained in that state due to the PDMS valve seat stiction until a threshold vacuum to open was reached. Thicker PDMS membranes were less elastic and hence separated from vacuum to open was reached. Thicker PDMS membranes were less elastic and hence separated from the rest (closed) position at a slower speed, but on the other hand, they snapped back faster to the the rest (closed) position at a slower speed, but on the other hand, they snapped back faster to the closed state than thinner ones. With one full cycle of the valve operation equivalent to 360°, the phase closed state than thinner ones. With one full cycle of the valve operation equivalent to 360◦, the phase difference between partially open and open/closed states was 90°. Actuation patterns depended on difference between partially open and open/closed states was 90◦. Actuation patterns depended on the phase difference between the stages and could be set by changing the channel length between the phase difference between the stages and could be set by changing the channel length between stages. Adding inverter stages to the oscillator and using the output from the appropriate nodes to stages. Adding inverter stages to the oscillator and using the output from the appropriate nodes to drive the fluidic valves is another method to achieve the desired actuation outcome. drive the fluidic valves is another method to achieve the desired actuation outcome.

Micromachines 2018, 9, 426 9 of 11 Micromachines 2018, 9, x FOR PEER REVIEW 9 of 11

Figure 8. Oscillator driving three valves in series out-of-phase in a fluid line. The three phases are Figure 8. Oscillator driving three valves in series out-of-phase in a fluid line. The three phases are open, closed, and partially open. When partially open, a small volume of red-dyed water enters the open, closed, and partially open. When partially open, a small volume of red-dyed water enters the flow chamber.flow chamber. In the In open the open phase, phase, the chamber the chamber is maximally is maximally filled fille withd with the water, the water, imparting imparting it a darker it a darker red hue.red Valve hue. Valveclosing closing evacuates evacuates the chamber the chamber indicated indicated by the minimally by the minimally red color (Multimedia red color (Multimedia view). view). 3.4. Benchmarking of Oscillators 3.4. Benchmarking of Oscillators Comparing the oscillators used in the study to previously demonstrated pneumatic and hydraulic Comparing the oscillators used in the study to previously demonstrated pneumatic and circuit implementations provided a benchmark for assessment. The semi-autonomous fluid handling hydraulic circuit implementations provided a benchmark for assessment. The semi-autonomous by means of on-chip digital logic was initially demonstrated by Hui et al. [23]. Pneumatic oscillators fluid handling by means of on-chip digital logic was initially demonstrated by Hui et al. [23]. fabricated with glass/PDMS and controlling the liquid pumps formed a section of the control circuitry Pneumatic oscillators fabricated with glass/PDMS and controlling the liquid pumps formed a section having five external inputs. Kim et al. [24] developed micro-hydraulic oscillators using a gravity of the control circuitry having five external inputs. Kim et al. [24] developed micro-hydraulic water-head as the power source that had the advantage of low-pressure actuation and enabled oscillators using a gravity water-head as the power source that had the advantage of low-pressure parallelization without pressure drops. However, a device relying on gravity is dependent on actuation and enabled parallelization without pressure drops. However, a device relying on gravity orientation and movement, affecting its suitability for portable applications. Additionally, the lower is dependent on orientation and movement, affecting its suitability for portable applications. pulse frequencies achieved by such circuits (<3 Hz) versus the higher frequencies (up to 50 Hz) Additionally, the lower pulse frequencies achieved by such circuits (<3 Hz) versus the higher attainable via pneumatic oscillators are characteristic of high-resistance liquid-medium devices frequencies (up to 50 Hz) attainable via pneumatic oscillators are characteristic of high-resistance (resistance being approximately two orders of magnitude greater in the case of water compared liquid-medium devices (resistance being approximately two orders of magnitude greater in the case to air). of water compared to air). Rhee and Burns [9] demonstrated digital circuits constructed with three-layer PDMS valves. Their Rhee and Burns [9] demonstrated digital circuits constructed with three-layer PDMS valves. pneumatic clock generator was implemented as a single inverter fed back through a resistance (narrow, Their pneumatic clock generator was implemented as a single inverter fed back through a resistance long channel) and capacitance (chambers in the flow and control layers separated by a membrane) (narrow, long channel) and capacitance (chambers in the flow and control layers separated by a in series. Clock frequencies between 2 and 4 Hz for varying RC values and high stability over a ~2 membrane) in series. Clock frequencies between 2 and 4 Hz for varying RC values and high stability min time period of observation were reported. One reason for the relatively short period may have over a ~2 min time period of observation were reported. One reason for the relatively short period been the use of a one-inverter configuration which is inherently less stable than the three-inverter may have been the use of a one-inverter configuration which is inherently less stable than the three- structure. Loss of power in the device due to air-permeability of the PDMS flow and control layers inverter structure. Loss of power in the device due to air-permeability of the PDMS flow and control likely contributed to the rapid decline in oscillations as well. layers likely contributed to the rapid decline in oscillations as well. Pneumatic oscillators fabricated by sandwiching a PDMS membrane between etched glass flow Pneumatic oscillators fabricated by sandwiching a PDMS membrane between etched glass flow and control chambers have been shown to work at frequencies up to approximately 5 Hz depending and control chambers have been shown to work at frequencies up to approximately 5 Hz depending on the pull-up resistance, demonstrating a logarithmic drop in frequency with vacuum supply [7]. on the pull-up resistance, demonstrating a logarithmic drop in frequency with vacuum supply [7]. Robustness against vacuum fluctuations at small strengths has been claimed based on this behavior. Robustness against vacuum fluctuations at small strengths has been claimed based on this behavior. The oscillators have a frequency drift of <1%/h. The authors achieved the same frequency range with The oscillators have a frequency drift of <1%/h. The authors achieved the same frequency range with a comparatively simplified fabrication process. Their oscillators showed a linear drop in the frequency a comparatively simplified fabrication process. Their oscillators showed a linear drop in the of oscillations with vacuum supply with the rate of decrease becoming smaller below 3 psi vacuum, frequency of oscillations with vacuum supply with the rate of decrease becoming smaller below 3 psi which correlated to robustness in the case of the glass-etched oscillators. Their oscillators displayed vacuum, which correlated to robustness in the case of the glass-etched oscillators. Their oscillators zero frequency drift for operation periods less than 2 h, which made them deployable (as an example) displayed zero frequency drift for operation periods less than 2 h, which made them deployable (as for analysis involving certain yeast cell cycles or measurements in cell density changes. an example) for analysis involving certain yeast cell cycles or measurements in cell density changes.

Micromachines 2018, 9, 426 10 of 11

4. Conclusions The authors fabricated a plastic pneumatic oscillator circuit that showed both long-term stability of frequency and robustness of operation, although the resolution of the laser engraver used did not allow for high-density fabrication. The out-of-phase three-valve fluidic device established oscillator suitability for self-driven control logic for microfluidic devices. A syringe was sufficient to drive the oscillator for a considerable period of time, demonstrating its applicability in portable lab-on-a-chip diagnostics. The fabrication process, based on laser engraving and APTES treatment, was reliable for the fast, efficient prototyping of these circuits during the design and development phase, as it took approximately 10–20 min to complete the laser engraving of PMMA and then four hours for the assembly. Higher-resolution laser engravers should allow for fabricating higher-density, smaller valve devices responding at higher frequencies. No additional tool adjustments or calibration is required as in techniques such as CNC milling or hot embossing. Provided that the resolution can be improved, the uniformity of features and the low cost of the process make these circuits potential candidates for simplifying the control circuitry of lab-on-a-chip devices and for driving all-plastic autonomous robots [25].

Author Contributions: V.B. designed the oscillators, conceptualized the fabrication methodology, characterized (validated) the devices, and wrote the original draft of the paper and further revisions. K.C. carried out the laser engraving of the PMMA pieces, including the calibration process. A.F. offered advice and enabled funding for the project, and reviewed and edited the paper. Acknowledgments: This work was partially supported by the National Institutes of Health (Grant No. 5R01NS064387). The discussions and guidance from Karl F. Bohringer (Electrical Engineering Department, University of Washington) and Hallie R. Holmes (Bioengineering Department, University of Washington) are gratefully acknowledged. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Garcia-Cordero, J.L.; Maerkl, S.J. A 1024-sample serum analyzer chip for cancer diagnostics. Lab Chip 2014, 14, 2642–2650. [CrossRef][PubMed] 2. Toriello, N.M.; Douglas, E.S.; Thaitrong, N.; Hsiao, S.C.; Francis, M.B.; Bertozzi, C.R.; Mathies, R.A. Integrated microfluidic bioprocessor for single-cell gene expression analysis. PNAS 2008, 105, 20173–20178. [CrossRef] [PubMed] 3. Araci, I.E.; Quake, S.R. Microfluidic very large scale integration (mVLSI) with integrated micromechanical valves. Lab Chip 2012, 12, 2803–2806. [CrossRef][PubMed] 4. Lai, H.; Folch, A. Design and dynamic characterization of “single-stroke” peristaltic PDMS micropumps. Lab Chip 2011, 11, 336–342. [CrossRef][PubMed] 5. Grover, W.H.; Skelley, A.M.; Liu, C.N.; Lagally, E.T.; Mathies, R.A. Monolithic membrane valves and diaphragm pumps for practical large-scale integration into glass microfluidic devices. Sens. Actuators B Chem. 2003, 89, 315–323. [CrossRef] 6. Weaver, J.A.; Melin, J.; Stark, D.; Quake, S.R.; Horowitz, M.A. Static control logic for microfluidic devices using pressure- valves. Nat. Phys. 2010, 6, 218–223. [CrossRef] 7. Duncan, P.N.; Nguyen, T.V.; Hui, E.E. Pneumatic oscillator circuits for timing and control of integrated microfluidics. PNAS 2013, 110, 18104–18109. [CrossRef][PubMed] 8. Jensen, E.C.; Grover, W.H.; Mathies, R.A. Micropneumatic digital logic structures for integrated microdevice computation and control. J. Micromech. Syst. 2007, 16, 1378–1385. [CrossRef] 9. Rhee, M.; Burns, M.A. Microfluidic pneumatic logic circuits and digital pneumatic microprocessors for integrated microfluidic systems. Lab Chip 2009, 9, 3131–3143. [CrossRef][PubMed] 10. Jung, W.; Han, J.; Choi, J.; Ahn, C.H. Point-of-care testing (POCT) diagnostic systems using microfluidic lab-on-a-chip technologies. Microelectron. Eng. 2015, 132, 46–57. [CrossRef] 11. Guckenberger, D.J.; de Groot, T.E.; Wan, A.M.D.; Beebe, D.J.; Young, E.W.K. Micromilling: A method for ultra-rapid prototyping of plastic microfluidic devices. Lab Chip 2015, 15, 2364–2378. [CrossRef][PubMed] Micromachines 2018, 9, 426 11 of 11

12. Schilling, E.A.; Kamholz, A.E.; Yager, P. Cell lysis and protein extraction in a microfluidic device with detection by a fluorogenic enzyme assay. Anal. Chem. 2002, 74, 1798–1804. [CrossRef][PubMed] 13. Kummrow, A.; Theisen, J.; Frankowski, M.; Tuchscheerer, A.; Yildirim, H.; Brattke, K.; Schmidt, M.; Neukammer, J. Microfluidic structures for flow cytometric analysis of hydrodynamically focussed blood cells fabricated by ultraprecision micromachining. Lab Chip 2009, 9, 972–981. [CrossRef][PubMed] 14. Tsao, C.-W. microfluidics: Simple, low-cost fabrication process bridging academic lab research to commercialized production. Micromachines 2016, 7, 225. [CrossRef] 15. Roy, E.; Galas, J.-C.; Veres, T. Thermoplastic elastomers for microfluidics: Towards a high-throughput fabrication method of multilayered microfluidic devices. Lab Chip 2011, 11, 3193–3196. [CrossRef][PubMed] 16. Zhang, W.; Lin, S.; Wang, C.; Hu, J.; Li, C.; Zhuang, Z.; Zhou, Y.; Mathies, R.A.; Yang, C.J. PMMA/PDMS valves and pumps for disposable microfluidics. Lab Chip 2009, 9, 3088–3094. [CrossRef][PubMed] 17. Mosadegh, B.; Kuo, C.-H.; Tung, Y.-C.; Torisawa, Y.; Bersano-Begey, T.; Tavana, H.; Takayama, S. Integrated elastomeric components for autonomous regulation of sequential and oscillatory flow switching in microfluidic devices. Nat. Phys. 2010, 6, 433–437. [CrossRef][PubMed] 18. Duncan, P.N.; Ahrar, S.; Hui, E.E. Scaling of pneumatic digital logic circuits. Lab Chip 2015, 15, 1360–1365. [CrossRef][PubMed] 19. Werner, E.M.; Chu, M.; Hui, E.E. Rapid prototyping of microfluidic digital logic. In Proceedings of the 20th International Conference on Miniaturized Systems for Chemistry and Life Sciences, Dublin, Ireland, 9–13 October 2016. 20. Duffy, D.C.; McDonald, J.C.; Schueller, O.J.A.; Whitesides, G.M. Rapid prototyping of microfluidic systems in poly(dimethylsiloxane). Anal. Chem. 1998, 70, 4974–4984. [CrossRef][PubMed] 21. Kim, K.; Park, S.W.; Yang, S.S. The optimization of PDMS-PMMA bonding process using silane primer. J. BioChip 2010, 4, 148–154. [CrossRef] 22. Sakurai, T.; Newton, A.R. Alpha-power law MOSFET model and its applications to CMOS inverter delay and other formulas. J. Solid-State Circuits 1990, 25, 584–594. [CrossRef] 23. Nguyen, T.V.; Duncan, P.N.; Ahrar, S.; Hui, E.E. Semi-autonomous liquid handling via on-chip pneumatic digital logic. Lab Chip 2012, 12, 3991–3994. [CrossRef][PubMed] 24. Kim, S.-J.; Yokokawa, R.; Lesher-Perez, S.C.; Takayama, S. Multiple independent autonomous hydraulic oscillators driven by a common gravity head. Nat. Commun. 2015, 6, 7301. [CrossRef][PubMed] 25. Wehner, M.; Truby, R.L.; Fitzgerald, D.J.; Mosadegh, B.; Whitesides, G.M.; Lewis, J.A.; , R.J. An integrated design and fabrication strategy for entirely soft, autonomous robots. Nature 2016, 536, 451–455. [CrossRef][PubMed]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).