Lab on a Chip

Publ

ishe

d on

02

Dec

embe

r 20

13. D

ownl

oade

d by

Tex

as A

& M

Uni

vers

ity o

n 04

/02/

2014

17:

02:1

7.

PAPER View Article OnlineView Journal | View Issue

Lab CThis journal is © The Royal Society of Chemistry 2014

aDepartment of Electrical and Computer Engineering, Texas A&M University,

College Station, TX, 77843, USA. E-mail: arum.han@ece.tamu.edu;

Fax: +1 979 845 6259; Tel: +1 979 845 9686bDepartment of Mechanical Engineering, Texas A&M University, College Station,

TX, 77843, USAc Department of Biomedical Engineering, Texas A&M University, College Station,

TX, 77843, USAdDepartment of Mechanical System Design Engineering, Seoul National University

of Science and Technology, Seoul, Republic of KoreaeMSDE (Manufacturing Systems and Design Engineering) Program, Seoul National

University of Science and Technology, Seoul, Republic of Korea

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c3lc51032h

Cite this: Lab Chip, 2014, 14, 947

Received 9th September 2013,Accepted 2nd December 2013

DOI: 10.1039/c3lc51032h

www.rsc.org/loc

Microfluidic acoustophoretic force basedlow-concentration oil separation and detectionfrom the environment†

Han Wang,a Zhongzheng Liu,b Sungman Kim,a Chiwan Koo,c Younghak Cho,d

Dong-Young Jang,e Yong-Joe Kimb and Arum Han*ac

Detecting and quantifying extremely low concentrations of oil from the environment have broad

applications in oil spill monitoring in ocean and coastal areas as well as in oil leakage monitoring on land.

Currently available methods for low-concentration oil detection are bulky or costly with limited sensitivities.

Thus they are difficult to be used as portable and field-deployable detectors in the case of oil spills or for

monitoring the long-term effects of dispersed oil on marine and coastal ecosystems. Here, we present a

low-concentration oil droplet trapping and detection microfluidic system based on the acoustophoresis

phenomenon where oil droplets in water having a negative acoustic contrast factor move towards acoustic

pressure anti-nodes. By trapping oil droplets from water samples flowing through a microfluidic channel,

even very low concentrations of oil droplets can be concentrated to a detectable level for further analyses,

which is a significant improvement over currently available oil detection systems. Oil droplets in water were

successfully trapped and accumulated in a circular acoustophoretic trapping chamber of the microfluidic

device and detected using a custom-built compact fluorescent detector based on the natural fluorescence

of the trapped crude oil droplets. After the on-line detection, crude oil droplets released from the trapping

chamber were successfully separated into a collection outlet by acoustophoretic force for further off-chip

analyses. The developed microfluidic system provides a new way of trapping, detecting, and separating

low-concentration crude oil from environmental water samples and holds promise as a low-cost

field-deployable oil detector with extremely high sensitivity. The microfluidic system and operation principle

are expected to be utilized in a wide range of applications where separating, concentrating, and detecting

small particles having a negative acoustic contrast factor are required.

Introduction

Crude oil spills pose severe environmental and economicthreats.1,2 Thick oil slicks are typically treated with chemicaldispersants to be broken down into small oil dropletsin the sub-100 μm range for rapid dispersion into theocean.3 Oil leaked from spills and other chronic dischargesalso experiences a series of naturally-occurring physical and

biochemical processes, such as spreading, biodegradation,and oil–mineral aggregate (OMA) formation, which arereferred to as “weathering”.4,5 The weathered oil eventuallybreaks down into small droplets in the aqua environment.Therefore, in regions away from an initial spill site and alsoin weathered spill sites, the concentration of oil droplets islower than 1 part-per-million (ppm).6,7 Large oil spills areusually monitored by microwave or optical sensors carried onsatellites, aircraft, or ships that track thick oil slicks on thesurfaces of water.8,9 However, detecting and quantifying dis-persed oil contents are difficult due to their extremely lowconcentration, especially in coastal areas where environmentalimpacts are the greatest.10 As reported by surveys conductedspanning from May to June of 2010 in the Deepwater Horizonspill in the Gulf of Mexico, after the application of a dispersant,COREXIT 9500 (Nalco Energy Services, TX), the dispersed oilconcentration was in the range of 8.1–14.5 mg L−1 at thewellhead measured using fluorescence spectrophotometersand quickly dropped to less than 1% to the range of parts-per-billion (ppb) at locations 1 km away from the plum.11-13

hip, 2014, 14, 947–956 | 947

Lab on a ChipPaper

Publ

ishe

d on

02

Dec

embe

r 20

13. D

ownl

oade

d by

Tex

as A

& M

Uni

vers

ity o

n 04

/02/

2014

17:

02:1

7.

View Article Online

Most commercial portable ultraviolet (UV) fluorometers for oildetection utilize the natural fluorescence fingerprints of crudeoil excited by UV light but have limited sensitivities (typicallyat 0.05–1 ppm).14-16 To detect oil droplets below the ppmrange, complex flow chambers combined with expensiveoptical instruments are required and thus not suitable forfield deployment.17 Therefore a low-cost, portable, and highlysensitive system that can be used to quantify low concen-trations of oil droplets in a marine environment on site aswell as to sample them for further analyses in laboratory set-tings with mass spectrometers could have broad implications.

We propose the use of acoustophoretic force to first con-centrate oil droplets to improve the detection sensitivity andthen to separate them from water for further downstreamanalyses. Particles suspended in a liquid solution experiencethe acoustophoretic force when exposed to a resonant acous-tic standing wave (given that the particles are much smallerthan the excitation wavelength), where such force dependson the vibro-acoustic properties of the particles and the sur-rounding solution.18-20 The acoustic standing wave formsacoustic pressure nodes (i.e. minimum acoustic pressurepoints) and anti-nodes (i.e. maximum acoustic pressurepoints). The particles can then move either to the acousticpressure nodes or to the anti-nodes by the acoustophoreticforce. The direction of this movement depends on the vibro-acoustic properties of the particles and the surrounding solu-tion and can be represented as a constant referred to as theacoustic contrast factor. This force has been utilized inmanipulating and separating particles and biological samplessuch as cells.21-26 Conventional layered ultrasound resonatorsare bulky and require a tedious acoustic impedance matchingprocess to achieve accurate matching of the layer thicknessesto form an acoustic standing wave.27,28 The recent develop-ment of transversal type resonators integrated with microfluidicsystems enabled simple implementation of acoustophoreticforce based particle manipulation systems.26 Reported first byNilsson et al., this transversal resonator utilized a piezoelectrictransducer attached to the bottom of a silicon/glass microchipto generate planar acoustic standing waves inside the micro-channel and allowed continuous particle separation.22

The system presented here utilizes the fact that oil dropletsin water have negative acoustic contrast factors and thus willmove towards the acoustic pressure anti-nodes under acousticexcitations. Petersson et al. developed an acoustophoreticforce based system for continuous lipid separation fromerythrocytes in blood.24 Due to their positive acoustic con-trast factor, erythrocytes in blood moved towards the acousticpressure nodes in an acoustic standing wave field, while thelipid particles moved towards the pressure anti-nodes due totheir negative contrast factor, and thus could be separatedfrom the erythrocytes.

To concentrate oil droplets from continuously flowing watersamples as needed in our application, an acoustophoretictrapping structure was designed to have acoustic pressureanti-nodes away from the main flow so that oil droplets couldbe moved and trapped at the pressure anti-nodes. This

948 | Lab Chip, 2014, 14, 947–956

concept is similar in principle to a few acoustophoretic parti-cle trapping schemes that utilized localized acousticresonance in trapping regions.29-32 Manneberg et al.30 andSvennebring et al.31 demonstrated that the retention andtrapping of particles and cells could be achieved in a circular-shaped resonance cavity. However, the acoustic standingwave was created by using stand-alone wedge transducers,which are bulky and cannot be easily integrated into compactand portable acoustic actuation systems.33 Evander et al.29

and Hammarstrom et al.32 developed acoustic particle andcell trapping sites in a fluidic channel by using miniaturetransducers mounted on a printed circuit board (PCB) to cre-ate localized acoustic fields. However, this method could notbe easily integrated with other microfluidic operations andrequired labour-intensive manual assembly for the miniaturetransducers. Most importantly, none of these acoustophoretictrapping schemes has been applied to selectively trap parti-cles having negative acoustic contrast factors such as crudeoil droplets as presented in this paper.

Manipulation of droplets using acoustophoretic force canalso be achieved by using both Rayleigh surface acoustic wave(SAW) and standing surface acoustic wave (SSAW).34-36

Franke et al. demonstrated droplet and particle sorting viaRayleigh SAW generated using an interdigital transducer(IDT).34 In this work, acoustic streaming in the bulk fluid sur-rounding the droplet or particle generated by an applied highfrequency acoustic wave deflected the droplet or particle tra-jectory to achieve successful separation. Li et al. utilizedchirped IDT to generate SSAW with excellent tunability andcontrollability for droplet sorting.35 The acoustic radiationforce directly applied to droplets allowed fast sorting in awide range with multiple outlets. Even though SAW-baseddroplet manipulation provides more flexibility, such systemsrequire specialized substrates that are typically much morecostly than utilizing a piezoelectric transducer and bulkacoustic wave. Thus, in cases where only simple separationor trapping of droplets is required, bulk acoustic wave basedsystems can provide more advantages.

Here a round-shape acoustic resonance cavity in a micro-fluidic channel operating in the transversal mode was devel-oped for oil droplet trapping. The presented microfluidicsystem is the first low-concentration oil droplet accu-mulation, detection, and separation system based on theacoustophoretic force. Oil droplets in water samples could betrapped and concentrated, and detected using a custom-builtcompact fluorescent detector. The trapped oil droplets couldthereby be sampled for further off-chip analyses such asconducting fluorescence spectrum analyses or gas chroma-tography to identify the collected crude oil compositions andtheir origins that are important in oil spill detection.

Principles

A particle suspended in a liquid solution where an acousticstanding wave is formed experiences an acoustic radiationforce, which is also referred to as the acoustophoretic force

This journal is © The Royal Society of Chemistry 2014

Lab on a Chip Paper

Publ

ishe

d on

02

Dec

embe

r 20

13. D

ownl

oade

d by

Tex

as A

& M

Uni

vers

ity o

n 04

/02/

2014

17:

02:1

7.

View Article Online

given in eqn (1) in the case of a one-dimensional (1-D) acous-tic standing wave.18-20 The acoustophoretic force acts alongthe y-axis perpendicular to the flow direction (x-axis), i.e.,

F p V kyas

2

0

22

, sin (1)

where p is the acoustic pressure amplitude, V0 is the volumeof the particle, βs is the compressibility of the solution, λ isthe wavelength of the acoustic field, k is the wavenumberdefined as 2π/λ, and Φ(β, ρ) is the acoustic contrast factordefined by:

,

5 220

0

0s

s s

(2)

where ρ0 and ρs are the densities of the particle and thesolution, respectively, and β0 and βs are the compressibilitiesof the particle and the solution, respectively. Thus, dependingon the properties of the particle and the surrounding solu-tion, the acoustic contrast factor can have either a positive ora negative value. The polarity of this acoustic contrast factordetermines whether the particle moves to acoustic pressurenodes (for a positive contrast factor) or anti-nodes (for a nega-tive contrast factor). Since oil droplets in water have negativecontrast factors, they move towards the pressure anti-nodesof an acoustic standing wave (Fig. 1A).

During the movement of the particle to the acoustic pres-sure nodes or anti-nodes, there is a relative velocity differencebetween the particle and the solution, inducing a viscousdrag force in the opposite direction to the particle motion.

This journal is © The Royal Society of Chemistry 2014

Fig. 1 Illustration of the acoustophoretic oil droplet trapping, detection, anoil droplets experience an acoustophoretic force towards the acoustic phalf-wavelength resonator at the first acoustic resonant frequency. (B)sidewalls, where the oil droplets are trapped. (C) Schematic of the microfluididetection module. The oil droplets trapped in the circular chamber were exwas collected by a photomultiplier tube (PMT). The concentrated oil dropletthe collection outlet by using another piezoelectric transducer (PZT 2) for fur

In a flow condition with a low Reynolds number (less than 1),the drag force can be expressed by eqn (3), which is propor-tional to the particle's radius and the relative velocity.37

Fv = − 6πηavr (3)

where η is the dynamic viscosity of the solution, a is theradius of the particle, and vr is the relative velocity betweenthe particle and the solution. Here, acoustic streaming velocityinduced by the resonant excitation is assumed to be includedin the solution flow velocity. Then, the particle motion canbe determined by using both the acoustophoretic force andthe viscous drag force when the gravitation and buoyancyforces are applied perpendicular to the direction of theacoustophoretic force.

In order to generate the high-level resonant acoustic fieldsin the transversal mode, a piezoelectric transducer is usuallyattached to the bottom of a microchip containing a micro-fluidic channel and excited at the first resonance frequencyof the microchannel. In a straight microchannel, one or twoof the parallel sidewalls of the microchannel can be excitedat the first resonance frequency whose one half-wavelength isequal to the channel width to generate a 1-D acoustic fieldas in eqn (1). As for a circular chamber, it is similar to the1-D straight channel case except that the circumferentialsidewalls of the chamber serve as the excitation surfaces.Although the excitation surface is not a plane surface, it cangenerate acoustic fields resembling the “1-D” acoustic fieldsdue to the rigid boundary conditions on the sidewalls. In thiscase, the acoustic pressure anti-nodes are formed at thesidewalls while the node is at the centre of the chamber.

Lab Chip, 2014, 14, 947–956 | 949

d separation scheme. (A) Due to their negative acoustic contrast factor,ressure anti-nodes that are formed near the sidewalls of the straight,In a circular chamber, the acoustic pressure anti-nodes form at thec oil droplet trapping/separation chip aligned to the compact fluorescentcited by a light emitting diode (LED) and the emitted fluorescence lights could then be released from the trapping chamber and separated intother off-chip analyses.

Lab on a ChipPaper

Publ

ishe

d on

02

Dec

embe

r 20

13. D

ownl

oade

d by

Tex

as A

& M

Uni

vers

ity o

n 04

/02/

2014

17:

02:1

7.

View Article Online

The oil droplets having a negative acoustic contrast factorand flowing into this circular chamber thus move towardsthe circumferential sidewalls of the circular chamber (i.e.acoustic pressure anti-nodes) and are trapped there due tothe acoustophoretic force (Fig. 1B). An additional benefit ofthis configuration is that the oil droplets trapped near thesidewalls can avoid the high flow rate region of the chambercentre, and thus effective trapping can be achieved by reduc-ing the fluidic drag force.

Materials and methodsSystem design

As shown in Fig. 1C, the developed acoustic oil droplet quan-tification system consists of a circular chamber for oil droplettrapping and detection, microfluidic inlet and outlet chan-nels for the circular chamber, and a downstream micro-channel coupled with a trifurcation outlet for separating theconcentrated oil droplets into the collection outlet uponrelease from the trapping chamber. Two piezoelectric trans-ducers are attached underneath the trapping chamber andthe separation channel areas, respectively, to generate theacoustic standing waves in the corresponding areas of themicrochip at two different frequencies.

The circular chamber was designed to be 750 μm in diam-eter and the corresponding first resonance frequency wastuned to 1.00 MHz. The microfluidic channels connectingthe circular chamber were designed to be 125 μm in width,and the downstream separation channel was designed to be350 μm wide for separating the concentrated oil droplets atthe first half-wavelength frequency of 2.14 MHz. The down-stream channel width was selected to avoid crosstalk withthe circular acoustic trapping chamber by deviating suffi-ciently away from a channel width that corresponds to higherorder harmonic resonance frequencies of the circular trap-ping chamber.

The circular trapping region also serves as the optical oildroplet detection zone. Crude oil has natural fluorescence,and this has been utilized for optical detection of the trappedand accumulated oil droplets. A custom-made compact fluo-rescent detector is aligned to the circular trapping chamberand focused on the fluidic channel plane for the detectionand quantification of the accumulated oil droplets.

Microchip fabrication

The microchannel was fabricated in silicon by using deepreactive-ion etching (DRIE) to a depth of 100 μm and coveredwith a borofloat cover glass by using anodic bonding. Thefabricated trapping chamber had a diameter of 745 μm andthe separation microfluidic channel in the downstream ofthe trapping chamber was 345 μm wide. Fluidic access holeswere drilled in the cover glass using a bench drill press(DP101, Ryobi Ltd, SC) and connected via flat-bottom ferrules(P-200N, Upchurch Scientific, WA) to Tygon® tubing (Saint-Gobain Performance Plastics, OH). The two piezoelectric trans-ducers (PZ26, Ferroperm Piezoceramics A/S, Denmark) were

950 | Lab Chip, 2014, 14, 947–956

attached to the bottom of the microchip at desired positionsby using wax (CrystalbondTM 509, SPI Supplies/StructureProbe Inc., PA) instead of permanent glue so that the micro-chip could be recycled.

The T-junction droplet generators used to produce oil-in-water emulsions as a pseudo sample of oil spills were fabri-cated in polydimethylsiloxane (PDMS, Sylgard® 184, DowCorning Corp., MI) using soft lithography. The master moldswere fabricated in a negative photoresist (SU-8, MicroChemCorp., MA) on a silicon substrate using standard photolithog-raphy. PDMS was then poured onto the master molds andcured at 80 °C for 2 hours. The cured PDMS replicas werethen released from the master molds and access holes werepunched before bonding to a glass substrate using oxygenplasma. The fabricated PDMS droplet generators were rinsedwith DI water and coated with surfactant (Pluronic F-108, 2%w/v in DI water, Sigma-Aldrich Inc., MO) in a 37 °C incubatorfor 1 hour prior to test.

Oil droplet generation

Crude oil sample from Midland, Texas, was used to generatethe oil-in-water droplets using the microfabricated T-junctiondroplet generators. The oil droplets with diameters of 14, 32,and 62 μm were generated and used as pseudo referencesamples. Samples taken from a weathered oil spill site typi-cally contain oil droplets and oil–mineral aggregates mainlyin the range of 10–50 μm in diameter, which fits well withinthe range of the pseudo samples used here.

A series of the T-junction droplet generators with differentchannel widths and heights were fabricated to generate theoil droplets of different sizes. The flow rates of crude oil andbuffer solution (DI water with 2% v/v Tween® 20 surfactant)(Sigma-Aldrich Inc., MO) were adjusted to obtain the desireddroplet generation frequencies and droplet sizes. The oildroplets 14 μm in diameter were generated by flowing buffersolution at 100 μl h−1 and crude oil at 1 μl h−1 in a T-junctiondevice with a channel width and height of 15 μm. The oildroplets 32 μm in diameter were generated by flowing buffersolution at 200 μl h−1 and crude oil at 3 μl h−1 in a T-junctiondevice with a channel width and height of 30 μm. The oildroplets 62 μm in diameter were generated by flowingbuffer solution at 300 μl h−1 and crude oil at 5 μl h−1 usinga T-junction device with a channel width and height of 60 μm.

The droplet generator chip and the acoustic oil detectionchip were connected with a 0.25 mm inner diameter tube todirectly inject the generated oil droplets into the detectionchip. To characterize the relationship of required power foracoustic trapping and flow rate, a syringe pump (Pico Plus,Harvard Apparatus, MA) was used to inject the generated oildroplets at various flow rates. Two different droplet genera-tion frequencies were utilized. For the single oil droplettrapping and detection experiments, approximately 1 dropletper second, which represents 1 droplet passing through thetrapping region of the acoustophoretic chip per second, wasgenerated. For the oil droplet accumulation experiments,approximately 5–10 droplets per second were generated.

This journal is © The Royal Society of Chemistry 2014

Lab on a Chip Paper

Publ

ishe

d on

02

Dec

embe

r 20

13. D

ownl

oade

d by

Tex

as A

& M

Uni

vers

ity o

n 04

/02/

2014

17:

02:1

7.

View Article Online

To better resemble real oil spill samples, oil droplet mixtureshaving various sizes of oil droplets were generated by stirringcrude oil in water using a magnetic bar. Crude oil (500 μl) wasmixed with DI water (25 ml) that was pre-mixed with 2% v/vTween® 20 surfactant and stirred for 15 min at 300 rpm. Thegenerated oil droplet mixtures were injected into the acousticoil detection chip using a syringe at a flow rate of 300 μl h−1.

Acoustophoretic oil droplet trapping and separation

Sinusoidal waves from a function generator (DG4102, RigolTechnologies Inc., OH) were supplied to the piezoelectrictransducers via a 40 dB power amplifier (406L, ElectronicNavigation Industries Inc., NY) to generate the acoustic standingwave inside the microfluidic device. For oil droplet trapping,sinusoidal waves with an amplitude of approximately100–400 mV were applied from the function generator, whilefor oil droplet separation, about 20–50 mV was applied. Forthe fabricated circular trapping chamber with a diameter of745 μm, a 1.01 MHz sinusoidal wave was applied to obtainthe first resonant acoustic standing wave.

Once all the oil droplets were collected in the trappingchamber and optically quantified, the piezoelectric trans-ducer under the trapping chamber was turned off and the oildroplets were released and carried downstream with thebuffer solution. The second transducer in the downstreamchannel was then turned on and the concentrated oil drop-lets were separated into the collection outlet at the trifurca-tion outlet. The separation fluidic channel width was 345 μm,resulting in the first resonance frequency of 2.18 MHz. Thiswas distinctive from the first and higher order resonancefrequencies of the trapping chamber to avoid any potentialcross-talk.

Optical detection of accumulated oil droplets

The compact fluorescent detection module was aligned to thetrapping chamber for recording the fluorescence intensityincrease as the oil droplets were trapped inside the chamber.This allowed quantification of the oil droplet concentrations inthe sample. The fluorescent detection module was composedof a blue LED (excitation peak at 470 nm, NSPB310B, Nichia,Japan) for excitation, an excitation filter (ET470/40x, ChromaTechnologies, VT) to pass only blue spectrum to excite the oildroplets, a dichroic mirror (495DCLP, Chroma Technologies,VT) to direct the excitation light on the trapping chamber, anemission filter (ET535/50m, Chroma Technologies, VT) totransmit only the emitted light from the oil droplets, and acompact photomultiplier tube (PMT, H10721, Hamamatsu,Japan) for fluorescence emission collection. An NI data acquisi-tion module (NI9219 DAQ, National Instruments Inc., TX) wasused for recording the fluorescence intensity from the PMTand to export the data to a custom LabVIEW™ (NationalInstruments Inc., TX) program. The entire detection modulewas securely enclosed in a plastic housing fabricated by a 3Dmaterial printer (ULTRA®, Envisiontec Inc., MI, USA). Tomonitor the oil droplet trapping and separation characteristics,

This journal is © The Royal Society of Chemistry 2014

an upright fluorescent microscope (LV100, Nikon InstrumentsInc., NY) with a digital camera was used.

Acoustic pressure field simulation in the trapping chamber

The acoustic pressure field distribution in the circular trappingchamber was simulated by developing a computer program inMATLAB® (MathWorks, MA). Since an analytical solution forthe 2-D acoustic standing wave in the circular chamber doesnot exist, a numerical modelling procedure was implementedto simulate the acoustic fields in the chamber and analyze themotion of the incident oil droplets. For this simulation, amapping function was first defined to map the discretizedcircular chamber into a regularly-meshed computationaldomain. This mapping allowed a second-order finite differ-ence method (FDM) to be applied to the computational domain.Here, the Mass and Momentum Conservation Equations andthe State Equation were decomposed into zeroth-, first- andsecond-order governing equations by using a perturbationmethod. By applying the second-order FDM approximationand appropriate boundary conditions, zeroth-order fluidmedium velocities, the first- and second-order acoustic pres-sures and particle velocities were numerically calculated.38

The second-order acoustic fields can be further decomposedinto both the time-dependent and time-independent ones.The second-order time-independent acoustic particle velocityis referred to as the acoustic streaming velocity. By applyingGorkov's equation20 to the first-order acoustic pressure andparticle velocity fields, the acoustophoretic force was calcu-lated. The zeroth-, first-, and second-order acoustic velocitieswere used to calculate the viscous drag force as described ineqn (3). Finally, the fourth-order Runge–Kutta method38 wasapplied to Newton's Second Law of Motion along with theacoustophoretic and viscous drag forces to predict themotion of the oil droplet. The gravitation and buoyancyforces were also considered when the vertical droplet motionwas of interest. The acoustic streaming effect was alsoincluded in this model. The overall numerical modelling pro-cedure is summarized in Fig. 2A. In this simulation, the size,density and compressibility of the oil droplet were set to 32 μm,900 kg m−3 and 6.5 × 10−10 Pa−1, respectively. In order to eval-uate the motion of multiple oil droplets inside the trappingchamber, the trajectories of three oil droplets at three differ-ent entrance positions (one close to the top channel wall, oneclose to the channel centre, and one in-between) were pre-dicted at the inlet fluid medium velocity of 10 mm s−1.

ResultsSimulated acoustic field and oil droplet trajectory

The simulated first-order acoustic pressure amplitude insidethe circular trapping chamber shows that the top and bottom“poles” of the circular chamber had the highest acousticpressure regions and represented the acoustic pressure anti-nodes (Fig. 2B). The acoustic pressure node formed along thecentral axis of the chamber, which was similar to the case ofthe 1-D acoustic standing wave.

Lab Chip, 2014, 14, 947–956 | 951

Fig. 2 Numerical modelling of acoustic pressure field and oil droplet motion. (A) The numerical modelling procedure to simulate acoustic fieldsinside a circular trapping chamber. (B) Simulated acoustic pressure amplitude for first-order acoustic pressure field in the circular trapping chamberand trajectories of three oil droplets having three different incident locations from the channel centre. Droplets flow in from the left side.

Lab on a ChipPaper

Publ

ishe

d on

02

Dec

embe

r 20

13. D

ownl

oade

d by

Tex

as A

& M

Uni

vers

ity o

n 04

/02/

2014

17:

02:1

7.

View Article Online

The pressure distribution in this circular chamber led theoil droplets to move to the two poles of the chamber and betrapped there. This could be observed from the simulated oildroplet trajectories overlaid on top of the pressure field(Fig. 2B). All of the three incident oil droplets had the samephysical properties (diameter = 32 μm, density = 900 kg m−3,compressibility = 6.6 × 10−10 Pa−1) at the same inlet velocityof 10 mm s−1 but differed in their entrance positions alongthe y-axis (15, 30, and 45 μm from channel centre, Fig. 2B).In the range of the generated acoustic field, acoustic stream-ing was insignificant and less than an order of magnitude ascompared to the acoustic radiation force.

Trapping of single crude oil droplets

To characterize crude oil droplet movement and trapping, the62 μm diameter oil droplets were flown through the circulartrapping chamber at a flow rate of 200 μl h−1 (i.e., the inlet

952 | Lab Chip, 2014, 14, 947–956

Fig. 3 Demonstration of acoustophoretic oil droplet trapping inside the c(diameter: 62 μm) passing through the circular chamber when the acousticchamber when the acoustic transducer was turned on (bottom row).

velocity of 4.44 mm s−1) and at a frequency of approximately1 droplet per second. This relatively slow flow speed was usedto observe single oil droplet movement in the trappingchamber with the piezoelectric transducer turned on and off.The top row images in Fig. 3 show a single oil droplet travellingthrough the trapping chamber when the piezoelectric trans-ducer was off. The bottom row images show a single oil dropletbeing successfully trapped on the sidewall of the chamber withthe transducer on. Videos S1 and S2† show single oil dropletsof two different sizes (62 μm and 14 μm) being trapped in thecircular chamber under the acoustophoretic force.

Oil droplet accumulation and detection

Small oil droplets (14 μm in diameter) were used to characterizethe accumulation and quantification of oil content in watersamples since they represent the typical size of oil droplets oroil–mineral aggregates in natural spill samples. Larger oil

This journal is © The Royal Society of Chemistry 2014

ircular trapping chamber. Time-lapse images show a single oil droplettransducer was turned off (top row) and being trapped in the circular

Fig. 4 Optical detection of oil droplets based on their natural fluorescence. (A) Time-lapse bright-field and fluorescent images showing oil droplets(diameter: 14 μm) being trapped and accumulated over time in the circular trapping chamber. (B) Fluorescence intensity increased over time as oildroplets accumulated inside the circular trapping chamber and decreased when the trapped oil droplets were released from the trapping chamber.

Fig. 5 Threshold input voltage needed to trap oil droplets of varioussizes in the circular trapping chamber with 100% efficiency. For the oildroplets with the diameters of 14 μm, 32 μm, and 62 μm, the thresholdinput voltage required to drive the piezoelectric transducer via the 40 dBpower amplifier was characterized as a function of the flow rate in therange of 0.25–2 ml h−1. Data represent average ± STD (n = 3). Error barsare not visible in this graph since they are smaller than the marker size.

Lab on a Chip Paper

Publ

ishe

d on

02

Dec

embe

r 20

13. D

ownl

oade

d by

Tex

as A

& M

Uni

vers

ity o

n 04

/02/

2014

17:

02:1

7.

View Article Online

droplets will require less power to be trapped since theacoustophoretic force is proportional to the volume of thedroplets at the same input power as shown in eqn (1).Bright-field and fluorescent time-lapse microscopic images showoil droplets being accumulated inside the circular trappingchamber when excited at the first resonance frequency (Fig. 4A,Video S3†). Consistent with the simulated trajectories (Fig. 2B),the oil droplets were successfully trapped and accumulated atthe top and bottom poles of the circular trapping chamberwhere the pressure anti-nodes were formed and graduallyexpanded into a semi-circular pattern.

Then, the fluorescence intensity from the trapped oildroplets was measured by placing the microfluidic chip onthe custom-built compact fluorescent detector module. Whenthe system was operating in the trapping mode, the graduallyaccumulated crude oil droplets (Fig. 4A) contributed to thegradual increase in the measured fluorescence intensity. Oncethe system was turned off, the oil droplets were released asrepresented by the sharp decrease in the measured fluorescenceintensity (Fig. 4B).

The current setup based on the highly sensitive PMT hasthe capability of detecting even a single 14 μm diameter crudeoil droplet without trapping it. However, the acoustic trappingscheme was still of central importance in the developed oildroplet detection system. First, for portable oil spill detectionsystems, the capability of collecting samples for off-chip analysesis crucial. Second, the ultimate goal is to develop a low-costsystem where the use of a low-cost photodiode is desirableinstead of a PMT. Thus, trapping oil droplets to a detectablelevel is necessary. Third, at the higher flow rate that is requiredto achieve higher sample processing throughput, the dataacquisition module may not properly resolve a single dropletfor quantification. Thus the oil droplet trapping approach wasimplemented in our design.

Threshold voltage for oil droplet trapping anddetection sensitivity

Based on eqn (1), the acoustophoretic force is proportional tothe acoustic pressure amplitude and the particle's volume,density, and compressibility. Therefore, oil droplets of different

This journal is © The Royal Society of Chemistry 2014

sizes experience different acoustophoretic force amplitudes ina resonant acoustic standing wave, requiring different levelsof acoustic excitation for successful trapping. The input powerrequirement was characterized with the oil droplets of threedifferent sizes (14, 32, and 62 μm in diameter) at varying flowrates (0.25 to 2 ml h−1) (Fig. 5). The threshold voltage wasdefined as the input voltage into the RF power amplifier thatprovided 100% oil droplet trapping efficiency for a given con-dition. As expected, smaller oil droplets required largerthreshold voltages at the same flow rates. As the flow rateincreased, the required threshold voltage also increased in aquasi-linear fashion, revealing the need for a largeracoustophoretic force as the fluidic drag force increased.

This continuous-flow oil droplet detection scheme withintegrated trapping pattern facilitated highly sensitive detec-tion of oil droplets. The current maximum flow rate of 4 ml h−1

and the capability of resolving a single 14 μm diameter oildroplet represent a detection sensitivity of 0.36 ppb oil con-tent in water samples when running the system for 1 minute.This detection sensitivity can be further improved by utilizing

Lab Chip, 2014, 14, 947–956 | 953

Fig. 6 Trapping and optical detection of a pseudo oil spill sample containing mixtures of oil droplets having different sizes. (A) Bright-field imagesshowing oil droplets of varying sizes passing through when the transducer was turned off (left) and then being trapped by the acoustophoreticforce when the transducer was turned on (right). (B) Detected fluorescence intensity increased as oil droplets were trapped and accumulated anddecreased as oil droplets were released by turning off the transducer.

Lab on a ChipPaper

Publ

ishe

d on

02

Dec

embe

r 20

13. D

ownl

oade

d by

Tex

as A

& M

Uni

vers

ity o

n 04

/02/

2014

17:

02:1

7.

View Article Online

for example multiple chambers in parallel to process moresample volumes for a given time.

Accumulated oil droplet release and separation

Once the oil droplets from the water samples were trappedand quantified, they were released by turning off the firstpiezoelectric transducer in the circular trapping region andseparated in the downstream separation channel by turningon the second piezoelectric transducer (driven at 2.18 MHz)to collect the accumulated oil droplets (Video S4†). Thereleased oil droplets moved towards the acoustic pressureanti-nodes near the sidewalls of the separation channel andflowed into the collection outlet. Collecting concentrated oildroplets from water samples can make the downstream off-chip physicochemical analyses easier since no extra pre-concentration step is needed. For example, the chemicalcompositions and spectra of the detected oil droplets can beanalysed using gas chromatography and spectrometry.13

Trapping and detection of poly-dispersed oil droplets aspseudo oil spill sample

Pseudo oil spill samples of poly-dispersed oil droplet mixtureswere successfully trapped and detected by the acoustic oildetection chip (Fig. 6, Video S5†). Oil droplets with signifi-cantly different sizes, ranging from 5 to 70 μm in diameter,were generated (Fig. 6A, left). When the acoustophoretic forcewas applied, oil droplets of different sizes were all trappedand accumulated in the circular chamber (Fig. 6A, right). Thiswas further validated by fluorescence detection using thecompact fluorescent detector. When the transducer wasturned on, oil droplets started to accumulate and thusresulted in rapid increase in fluorescence intensity (Fig. 6B).When the piezoelectric transducer was turned off, the intensitydropped instantly as the trapped oil droplets were released.

Discussion

The developed acoustophoretic oil droplet detection systemdemonstrated the successful trapping, detection, and

954 | Lab Chip, 2014, 14, 947–956

separation of crude oil droplets generated by a T-junctiondroplet generator. The oil-in-water droplets closely resemblethe natural spill samples in terms of their sizes and concen-trations, which are the key factors affecting oil trapping anddetection from water samples. Depending on the origins ofoil spills and weathering processes, oil droplets possess dif-ferent chemical compositions and show different fluores-cence spectra.5,39 Therefore, even for the same amount, oildroplets from different types of crude oil would exhibit differ-ent levels of fluorescence intensity. Thus without advanceknowledge of the fluorescence properties of oil droplets, acalibration process to determine the relationship between theoil concentration and fluorescence intensity is typically neededprior to field test. The microfluidic system presented here pro-vides the capability for accumulating and concentrating suffi-cient amount of oil samples for off-chip analyses that can beused for such crude oil identification based on their fluores-cent signatures. The system also provides the potential for on-chip optical spectroscopy since the accumulated oil dropletswill provide sufficient sensitivity for even a low-cost spectros-copy instrument.

Real seawater may contain a lot of debris and particles.Larger particles can be easily pre-filtered using simple filtrationsystems (e.g., mesh filter); however, smaller sub-100 μm dustparticles can still enter the acoustic device. Dust particles inseawater are expected to mostly possess a positive acoustic con-trast factor and thus would be trapped in the chamber centre.However even if particles are accumulated in the centre part ofthe chamber, because of their relative small size compared tothe overall chamber size, we do not expect to see any signifi-cant interference to the flow or generated acoustic standingwave. Moreover, the detection of oil content depends on theautofluorescence of crude oil at a particular wavelength range,which in most cases does not exist in other particles. Overall,the trapping and detection of oil droplets are most likely notaffected by other particles that may exist in the seawater.

The developed acoustic oil droplet separation and detectionsystem demonstrated the potential for a new type of low-costportable oil content detector. In the current form, a bench-toppower amplifier was used, but in the future, it can be replaced

This journal is © The Royal Society of Chemistry 2014

Lab on a Chip Paper

Publ

ishe

d on

02

Dec

embe

r 20

13. D

ownl

oade

d by

Tex

as A

& M

Uni

vers

ity o

n 04

/02/

2014

17:

02:1

7.

View Article Online

by a compact amplifier circuit on a printed circuit board (PCB).This PCB-based power amplifier currently under developmentin our lab would enable fabricating a completely portableacoustic oil detection system for field deployment. By usingthe acoustic trapping strategy, a low-cost photodiode canreplace the currently used PMT that is both significantly pric-ier and heavier than a photodiode. Changing the excitationLED and the corresponding filter sets to have a peak excita-tion at approximately 300 nm can give even stronger fluores-cence emission from the trapped crude oils, thus increasingthe detection sensitivity.16

Overall, the developed microfluidic device is a broadlyapplicable platform for trapping, concentrating, detecting,and separating not only crude oils but also other samples ofinterest that have negative acoustic contrast factors.

Conclusions

An acoustophoretic force based oil droplet trapping, detec-tion, and separation microsystem has been developed.Successful trapping of crude oil droplets of three differentsizes (14, 32, and 62 μm in diameter) was demonstrated byusing the circular trapping chamber design that generatedthe resonant acoustic standing wave and formed the acousticpressure anti-nodes at the sidewalls of the chamber. Thecustom-built compact fluorescent detection module was usedto successfully detect the increasing fluorescence intensity asthe oil droplets were trapped and accumulated in the circularchamber, and the amount of oil content could be quantifiedby using the measured fluorescence intensity. The trappedoil droplets after the detection were released and separatedinto the collection outlet in the downstream acoustophoreticseparation channel for further off-chip analyses of thesampled oil. A pseudo spill sample containing a mixture ofdroplets ranging from 5 to 70 μm in diameter was also suc-cessfully trapped and detected. This acoustophoretic micro-system holds promise as a low-cost portable system to detectextremely low concentration oil spills in the environment.

Acknowledgements

This work was supported by the U.S. Army Corps of EngineersEngineer Research and Development Centre (ERDC) grant# W9132T-12-2-0022. The project was also partially supportedby the Korean Ministry of Knowledge Economy grantno.: 10039890 to AH. The authors would like to thank theTexas A&M University Department of Petroleum Engineeringfor providing the crude oil samples.

Notes and references

1 C. H. Peterson, S. D. Rice, J. W. Short, D. Esler, J. L. Bodkin,

B. E. Ballachey and D. B. Irons, Science, 2003, 302,2082–2086.

2 D. A. Gill, J. S. Picou and L. A. Ritchie, Am. Behav. Sci.,

2012, 56, 3–23.

This journal is © The Royal Society of Chemistry 2014

3 R. R. Lessard and G. DeMarco, Spill Sci. Technol. Bull.,

2000, 6, 59–68.

4 P. Nounou, Ambio, 1980, 9, 297–302.

5 R. K. Nelson, B. M. Kile, D. L. Plata, D. P. Sylva, L. Xu,

C. M. Reddy, R. B. Gaines, G. S. Frysinger andS. E. Reichenbach, Environ. Forensics, 2006, 7, 33–44.

6 G. D. Marty, D. E. Hinton, J. W. Short, R. A. Heintz,

S. D. Rice, D. M. Dambach, N. H. Willits and J. J. Stegeman,Can. J. Zool., 1997, 75, 989–1007.

7 R. A. Heintz, S. D. Rice, A. C. Wertheimer, R. F. Bradshaw,

F. P. Thrower, J. E. Joyce and J. W. Short, Mar. Ecol.:Prog. Ser., 2000, 208, 205–216.

8 M. F. Fingas and C. E. Brown, Spill Sci. Technol. Bull.,

1997, 4, 199–208.

9 M. Jha, J. Levy and Y. Gao, Sensors, 2008, 8, 236–255.

10 P. D. Keizer and D. C. Gordon Jr, J. Fish. Res. Board Can.,

1973, 30, 1039–1046.11 R. Camilli, C. M. Reddy, D. R. Yoerger, B. A. S. Van Mooy,

M. V. Jakuba, J. C. Kinsey, C. P. McIntyre, S. P. Sylva andJ. V. Maloney, Science, 2010, 330, 201–204.

12 A.-R. Diercks, R. C. Highsmith, V. L. Asper, D. Joung,

Z. Zhou, L. Guo, A. M. Shiller, S. B. Joye, A. P. Teske,N. Guinasso, T. L. Wade and S. E. Lohrenz, Geophys. Res.Lett., 2010, 37, L20602.

13 W. G. Mendoza, D. D. Riemer and R. G. Zika, Environ. Sci.:

Processes Impacts, 2013, 15, 1017–1030.

14 P. Lambert, J. Hazard. Mater., 2003, 102, 39–55.

15 K. Trudel, R. Belore, M. VanHaverbeke and J. Mullin, in

Proceedings of the Thirty-second Arctic and Marine OilspillProgram (AMOP) Technical Seminar, Environment Canada,Ottawa, Canada, 2009, pp. 397–410.

16 M. Kim, U. H. Yim, S. H. Hong, J.-H. Jung, H.-W. Choi, J. An,

J. Won and W. J. Shim, Mar. Pollut. Bull., 2010, 60, 383–389.

17 A. M. Castro, J. Delgado and G. Orellana, J. Mater. Chem.,

2005, 15, 2952–2958.

18 L. V. King, Proc. R. Soc. London, Ser. A, 1934, 147, 212–240.

19 K. Yosioka and Y. Kawasima, Acustica, 1955, 5, 167–173. 20 L. P. Gorkov, Phys.-Dokl., 1962, 6, 773. 21 M. Wiklund, P. Spégel, S. Nilsson and H. M. Hertz,

Ultrasonics, 2003, 41, 329–333.22 A. Nilsson, F. Petersson, H. Jonsson and T. Laurell,

Lab Chip, 2004, 4, 131–135.23 T. Lilliehorn, U. Simu, M. Nilsson, M. Almqvist, T. Stepinski,

T. Laurell, J. Nilsson and S. Johansson, Ultrasonics, 2005, 43,293–303.

24 F. Petersson, A. Nilsson, C. Holm, H. Jonsson and T. Laurell,

Lab Chip, 2005, 5, 20–22.

25 F. Petersson, L. Åberg, A.-M. Swärd-Nilsson and T. Laurell,

Anal. Chem., 2007, 79, 5117–5123.

26 T. Laurell, F. Petersson and A. Nilsson, Chem. Soc. Rev.,

2007, 36, 492–506.

27 O. Doblhoffdier, T. Gaida, H. Katinger, W. Burger,

M. Groschl and E. Benes, Biotechnol. Prog., 1994, 10,428–432.

28 K. Yasuda, S. S. Haupt, S. Umemura, T. Yagi, M. Nishida and

Y. Shibata, J. Acoust. Soc. Am., 1997, 102, 642–645.

Lab Chip, 2014, 14, 947–956 | 955

Lab on a ChipPaper

Publ

ishe

d on

02

Dec

embe

r 20

13. D

ownl

oade

d by

Tex

as A

& M

Uni

vers

ity o

n 04

/02/

2014

17:

02:1

7.

View Article Online

29 M. Evander, L. Johansson, T. Lilliehorn, J. Piskur,

M. Lindvall, S. Johansson, M. Almqvist, T. Laurell andJ. Nilsson, Anal. Chem., 2007, 79, 2984–2991.

30 O. Manneberg, B. Vanherberghen, B. Onfelt and

M. Wiklund, Lab Chip, 2009, 9, 833–837.

31 J. Svennebring, O. Manneberg, P. Skafte-Pedersen, H. Bruus

and M. Wiklund, Biotechnol. Bioeng., 2009, 103, 323–328.

32 B. Hammarstrom, M. Evander, H. Barbeau, M. Bruzelius,

J. Larsson, T. Laurell and J. Nilsson, Lab Chip, 2010, 10, 2251–2257.

33 O. Manneberg, J. Svennebring, H. M. Hertz and M. Wiklund,

J. Micromech. Microeng., 2008, 18, 095025.

34 T. Franke, A. R. Abate, D. A. Weitz and A. Wixforth,

Lab Chip, 2009, 9, 2625–2627.

956 | Lab Chip, 2014, 14, 947–956

35 S. Li, X. Ding, F. Guo, Y. Chen, M. I. Lapsley, S.-C. S. Lin,

L. Wang, J. P. McCoy, C. E. Cameron and T. J. Huang,Anal. Chem., 2013, 85, 5468–5474.

36 X. Ding, P. Li, S.-C. S. Lin, Z. S. Stratton, N. Nama, F. Guo,

D. Slotcavage, X. Mao, J. Shi, F. Costanzo and T. J. Huang,Lab Chip, 2013, 13, 3626–3649.

37 C. Collinson and T. Roper, Particle Mechanics, Elsevier

Science, 1995.

38 Z. Liu, H. Wang, A. Han and Y.-J. Kim, Numerical modeling

for analyzing microfluidic acoustophoretic motion of cellsand particles with application to identification of vibro-acousticproperties J. Acoust. Soc. Am., 2013, 133, 3279.

39 Z. Wang and M. Fingas, J. Chromatogr., A, 1997, 774, 51–78.

This journal is © The Royal Society of Chemistry 2014