Mechanical Properties Based Particle Separation via Traveling

Nov 9, 2016 - Most microfluidics-based sorting methodologies utilize size differences between suspended micro-objects as the defining characteristic b...
2 downloads 9 Views 4MB Size
Download PDF
Article pubs.acs.org/ac

Mechanical Properties Based Particle Separation via Traveling Surface Acoustic Wave Zhichao Ma, David J. Collins, Jinhong Guo, and Ye Ai* Pillar of Engineering Product Development, Singapore University of Technology and Design, Singapore 487372, Singapore S Supporting Information *

ABSTRACT: Most microfluidics-based sorting methodologies utilize size differences between suspended micro-objects as the defining characteristic by which they are sorted. Sorting based on mechanical properties, however, would provide a new avenue for sample preparation, detection and diagnosis for a number of emerging biological and medical analyses. In this study, we demonstrate separation of polystyrene (PS) and poly(methyl methacrylate) (PMMA) microspheres based entirely on their difference in mechanical properties using traveling surface acoustic waves (TSAWs). We theoretically examine the correlation of the applied TSAW frequency, particle density and sound speed with respect to the resultant acoustic radiation force (ARF) that acts to translate particles, and experimentally corroborate these predictions by translating PS and PMMA particles simultaneously in a stationary flow. Even when PS and PMMA particles have the same diameters, they exhibit strongly nonlinear and distinct acoustophoretic responses as a function of their mechanical properties and the applied TSAW frequency. By specifically matching the appropriate acoustic frequency to the desired particle size, each particle population can be selectively translated and sorted. We demonstrate that this mechanical property based sorting can continuously separate these particle populations with at least 95% efficiency in the mixed 10/15 μm diameter PS and PMMA particle solutions tested.

S

difference in the other two intrinsic properties, density and sound speed, vary substantially between target populations in which size is an ineffective sorting parameter. Accordingly, mechanical properties based separation is of great interest for biological research and diagnostics because of their close correlation with cell viability, tumor metastasis, phenotype and differentiation.42,43 Various methods have been applied to separate micron-sized particles based on mechanical properties especially cell deformability, such as microfiltration,44,45 hydrodynamic chromatography,46 and inertial focusing.47 However, the particle size is still a dominant effect in these separation techniques and the ubiquitous particle size variation may substantially compromise the performance of mechanical properties based separation. In the few instances where mechanical properties have been assayed using acoustophoresis, this has occurred in ultrasonic standing waves in which suspended particles preferentially migrate to maximum or minimum time-averaged pressure locations at differential rates.48 Based on the dissimilar acoustophoretic responses arising from the biologically induced difference in cellular mechanical properties, measurements of the cellular mechanical properties have been demonstrated.49,50 With a similar mechanism, continuous particle separation using standing wave acoustic fields have been utilized elsewhere. Petersson

eparation of microscopic particles and cells plays an important role in biological analyses and medical diagnostics.1 For example, the separation of circulating tumor cells (CTCs) from human blood cells2,3 and sorting different stem cell types, such as mesenchymal stem cells (MSCs) and hemopoitetic stem cells (HSCs),4,5 is of great interest for clinical analyses and biological research. Microfluidics, where force gradients on the scale of cells are applied within minute volumes of fluid, have enabled a number of microscale particle manipulation methods including dielectrophoresis,6,7 magnetophoresis,8,9 optical tweezers,10 inertial focusing,11,12 and acoustophoresis.13−18 Acoustophoresis for particle manipulation has gained increasing attention in recent years because of its low power consumption and good biocompatibility, and has been applied in a variety of microfluidic applications such as enrichment,19−21 separation,22−27 patterning,28−34 and continuous focusing.35−37 The acoustic radiation force (ARF) exerted on suspended particles within an acoustic field is a function of their size, density, and sound speed, where the aggregate sound speed and density differences between different micro-object populations can, in theory, be exploited to separate them according to their mechanical properties.38,39 However, the vast majority of acoustophoretic particle and cell separation demonstrations separate particle and cell populations according to the size differences between them. While these volumetric differences can be exploited for separation in a number of important applications, including those dealing with suspensions of microparticles,25 cells,17 bubbles,40 and droplets,41 there are many instances where the © XXXX American Chemical Society

Received: September 10, 2016 Accepted: October 28, 2016

A

DOI: 10.1021/acs.analchem.6b03580 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

several previous publications for fine-grained separation specifically according to particle dimensions.14,16,35,54,57−60 Decades ago, Hasegawa and Yosioka measured the radiation force in a traveling wave field on millimeter-sized silica spheres and, in doing so, confirmed their theoretical predictions of acoustic force as a function of particle size,56,61 though this has yet to be demonstrated or confirmed for a mixed population of different mechanical properties microscale objects in a continuous microfluidic flow, a setup more relevant for clinical applications with potentially confounding features (including acoustic reflections in microfluidic channels) that are not accounted for in the test measurement. In this work, we present a label-free continuous separation of same-sized polystyrene (PS) and poly(methyl methacrylate) (PMMA) particles based on their different mechanical properties (density and sound speed of PS and PMMA) using traveling surface acoustic waves (TSAWs). In contrast to the particle alignment along the pressure or antipressure nodal positions with application of standing acoustic fields, suspended particles exposed to a traveling acoustic wave are displaced in the acoustic wave propagation direction without an inherent limit to their translation distance.26 To assay these forces, an aqueous suspension containing both PS and PMMA particles is injected into a disposable microchannel exposed to TSAW. Their difference in mechanical properties and resultant dissimilar forces imparted gives rise to dissimilar translation velocities. To evaluate the differential response of PS and PMMA particles in traveling wave acoustic fields, these suspended particles are displaced at four different theoretically optimized frequencies generated by one set of chirped interdigital transducers (CIDTs), which permit the generation of SAWs across a wide frequency range.28,62 By tracking and analyzing the motion of individual particles,49,50,63,64 the ratio of the acoustic radiation force on PS particles to that on the PMMA particles is derived, showing good agreement with theoretical predictions.56 Deterministic switching of the two kinds of particles into the desired outlets can be tuned by the dimensionless κ value (e.g., PS particles are preferentially displaced at κ ∼ 1.43, PMMA at κ ∼ 1.78). It is further demonstrated that the TSAW field can exert a larger radiation force on 10 μm PS particles than 15 μm PMMA particles by tuning the κ value such that smaller particles can preferentially displace, despite the larger size of the nondisplaced particle. More generally, this study demonstrates the potential of traveling wave acoustic fields for separating microscopic synthesized particles and biological cells based on their mechanical properties.

et al.51 demonstrated the separation of lipid particles and erythrocytes, where lipid particles with a negative acoustic contrast factor (a parameter measuring the relative densities and the sound speeds of the particle and the suspension media) migrated to the pressure antinodes and erythrocytes with a positive acoustic contrast factor to the pressure nodes. Myeong et al.52 demonstrated a density-dominant separation of similarsized polystyrene and melamine particles in a standing acoustic field since the higher density of the melamine gives rise to a larger acoustic radiation force on them than that on the polystyrene particles. Further, Ding et al.18 demonstrated separation of same-sized polystyrene particles and human leukemia cells based on their different compressibility, though in all of these cases, the incoming particle dimensions should be carefully controlled due to the linear force scaling according to the acoustic contrast factor (F ∼ ϕ) compared to the F ∼ a3 scaling with particle size (both when λ ≫ a),39 where ϕ is the acoustic contrast factor, λ is the wavelength, and a is the particle radius. As a result of the aforementioned force scalings, the acoustic radiation force is predominantly determined by the size of suspended particles rather than the material properties in the use of standing surface acoustic wave (SSAW) fields. In recent and important work, Augustsson et al.53 circumvented this limitation and demonstrated focusing of cells with different acoustic impedance into different equilibrium positions, though this methodology requires the generation and maintenance of an inhomogeneous fluid medium. Moreover, these standing wave separation methods are limited by a maximum displacement of one-quarter of the acoustic wavelength, the distance between neighboring antinodes and nodes. Because of this inherent limitation, the required acoustic wavelength is typically much larger than the particle size to achieve reasonable separation distance between dissimilar particles in sorting applications, though this consideration comes at the expense of the forces that can be generated, which scale with the acoustic force gradients and, thus, the inverse of the wavelength. Additionally, the generation of a standing wave field in the case of SAW field requires the integration of reflectors or an opposing set of the IDTs and the precise alignment of acoustic pressure nodes with channel geometries.54,55 By contrast, the use of traveling acoustic waves simplifies device fabrication and alignment in that their generation requires only a single set of IDTs. Most importantly, the relationship between the imparted force on a particle from a traveling wave and its properties is highly nonlinear when the particle size approaches and exceeds the wavelength in the fluid, potentially permitting more robust and size-insensitive separation than using standing wave fields. A summary table comparing acoustophoresis based on standing waves and traveling waves is shown in the Supporting Information (Table S1). In the case of traveling waves, the theoretical model developed by Hasegawa et al.56 shows that the resonance effect on a particle exposed to a traveling acoustic field can induce a sharply nonlinear relationship between the acoustic radiation force with respect to the particle size relative to the acoustic wavelength, represented by the dimensionless factor κ = 2πaf/cf, where a, f, and cf are the particle radius, the acoustic field frequency, and the sound speed in the fluid, respectively. Further, it is possible for the maximum forces occur at different values of κ for different material properties. Despite these characteristics, continuous separation according to mechanical properties in a traveling wave field remains unexplored, though this nonlinear force scaling has been used to great effect in

■

METHODS Experimental Setup and Methods. Figure 1 shows the schematic working principle and photograph of the acoustophoretic microfluidic system for particle separation based on mechanical properties. The TSAW transducer is comprised of a pair of interdigital transducers (IDTs) on a lithium niobate (LiNbO3, LN) piezoelectric substrate fabricated using a lift-off technique,23 where conductive (Cr/Au) layers were deposited on a 128° rotated Y-cut X-propagating LN substrate using an ebeam evaporator following photolithography to define electrode extents. The individual IDT electrode elements widths are equal to 1/4 of the acoustic wavelength of the plain IDTs. For CIDTs, the wavelengths vary between 60 and 100 μm, which determines the range of usable resonant frequencies (39.6−66.0 MHz) according to f = cLN/λ, where cLN = 3960 m· B

DOI: 10.1021/acs.analchem.6b03580 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry αn = −

βn = −

[Fnjn (κ ) − κjn′ (κ )]2 ⎡F j (κ ) − κj′ (κ )⎤2 + ⎡F y (κ ) − κy′(κ )⎤2 ⎣ nn ⎦ ⎣ nn ⎦ n n

(3)

[Fnjn (κ ) − κjn′ (κ )]·[Fnyn (κ ) − κyn′(κ )] [Fnjn (κ ) − κjn′ (κ )]2 + [Fnyn (κ ) − κyn′(κ )]2

(4)

where jn and yn are the spherical Bessel functions of order n for the first and second kind, respectively. The scattering coefficient Fn is given by Fn =

κ22·ρf 2·ρp

· κ1·jn′ (κ1) κ1·jn′ (κ1) − jn (κ1)

Figure 1. Schematic of the acoustophoretic microfluidic system for particle separation. (a) Three-dimensional perspective illustration. (b) Cross-section of the device. A disposable PDMS channel is placed on a reusable TSAW transducer where the prepatterned IDTs generate the TSAW that couples into the microchannel. Particles flowing through the channel are translated by dissimilar distances according to the difference in their mechanical properties and thus are collected in different outlets. (c) Photograph of the assembled device. The scale bar is 5 mm.

κ1jn′ (κ1) − jn (κ1)

4 k2

−

2n(n + 1)[jn (κ2) − κ2j2′ (κ2)] (n + 2)(n − 1)jn (κ2) + κ22jn″(κ2)

(5)

where ρf, ρp, and σ =

(

c l2 cs2

)(

−2 /

2c l2 cs2

) are the fluid

−2

density, particle density, and the Poisson ratio of the particle material, respectively. The parameters κ1 and κ2 are defined as κ1 = k1·a =

2πfa cl

(6)

κ2 = ks·a =

2πfa cs

(7)

where cl, cs, and f are the longitudinal and shear sound speed of the particle material and the acoustic frequency. The correlation between their ARF factors and the κ value can be obtained by applying the properties of PS and PMMA listed in Table 1 to the theoretical model. As shown in Figure 2, Table 1. Parameters in the YT Calculation properties

PS

PMMA

water

density (kg/m3) longitudinal sound speed (m/s) shear sound speed (m/s) Poisson ratio

1060 2350 1120 0.353

1183 2746 1392 0.327

998 1495

the ARF factor of a PS particle is negligibly small when κ < 1, which is in good agreement with previous experimental studies showing unresponsive particle acoustophoresis when κ < 1.14,16,26,37,65 In the range of κ > 1, the ARF factor of PS particles becomes more pronounced but does not increase monotonically and is comprised of successive peaks (maxima) and dips (minima) corresponding to resonance of free vibration of the sphere.56 PMMA particles have a similar ARF factor versus κ curve with successive peaks and dips when κ > 1. This theoretical model reveals that the relative particle density and sound speed can both shift the location of the first ARF factor peak, demonstrating the ability to separate particle based on density or sound speed or the combination of the two properties. In particular, the shear sound speed contributes more to the shift of the first ARF peak compared to the longitudinal sound speed. More details of the effects of the relative particle density and sound speed on the location of the first peak in YT−κ curve are shown in the Supporting Information (Figure S1). Important for producing differential

(1)

where YT is the ARF factor, ⟨E⟩ is the mean energy density of the incident wave, and a is the radius of the spherical particle. Thus, for the same-sized particles subject to an acoustic field with the same energy intensity, the ARF is proportional to the dimensionless factor YT, which is defined as YT = −

2n(n + 1)jn (κ2) (n + 2)(n − 1)jn (κ2) + κ22jn″(κ2)

κ12[σjn (κ1) / (1 − 2σ ) − jn″(κ1)]

s−1 is the Rayleigh wave sound speed in the x-direction of this cut of LN. The aperture of the IDTs is 8 mm, where this value corresponds to the effective width of the acoustic field. A disposable polydimethylsiloxane (PDMS) channel with three inlets and two outlets is aligned on the TSAW transducer by hand. The channel was fabricated by bonding two PDMS parts with micron-sized structures, as described in a previous study,26 where the upper PDMS layer defines the channel features and the lower layer includes an air cavity for minimizing the wave attenuation along the PDMS/LN interface. As such, the disposable PDMS channel can be easily peeled off and discarded while retaining the TSAW device for reuse. The sample flow is bounded by two sheath flows and is focused into the lower (nonsorted) outlet without the application of TSAW. Once an alternating current (AC) signal is applied to the IDTs, the TSAW generated by the transducer couples into the microchannel in the form of a series of plane waves in the fluid, which subsequently act to translate the desired particle population into different outlets. Theoretical Model. A particle suspended in a fluid and exposed to a traveling wave acoustic field is subject to a timeaveraged acoustic radiation force (ARF, FT), which can be expressed as56 ⟨FT⟩ = YTπa 2⟨E⟩

−

∞

∑ [(n + 1)(αn + αn+ 1 + 2αnαn+ 1 + 2βnβn+ 1)] n=0

(2)

where αn and βn can be derived as C

DOI: 10.1021/acs.analchem.6b03580 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

the suspended particle increases linearly as κ increases. For two same-sized particles with different mechanical properties exposed in the same standing acoustic field, the ratio of their Ys is thus independent of the κ value. By contrast, the nonlinear correlation between YT and κ in a traveling acoustic field induces successive peaks (maxima) and dips (minima) locating at different κ values for different materials, resulting in a significant difference between the ARF on them. Taking samesized PS and PMMA particles exposed in the same standing acoustic field as example, with the parameters shown in Table 1, the ratio of their Fs is calculated to be constant as Rs = PMMA PS FPMMA /FPS /Ys = 1.35. The force ratio in a standing s s = Ys field is much smaller than that for the same situation in a traveling acoustic field at the YT peaks of one particle type, for PMMA PMMA example RT = FPS = YPS = 21.72 when κ = 1.43. T /FT T /YT It is obvious to conclude that the difference in ARF on samesized particles resulting from the variation of their mechanical property is substantially augmented in a traveling field compared to in a standing field. Therefore, the use of TSAW fields is more advantageous for mechanical properties based microparticle separation than the use of SSAW fields.

Figure 2. ARF factors of PS and PMMA particles as a function of κ calculated by the theoretical model. The peaks of the YT−κ curve lie in 1.4−1.5, 2.1−2.2, and 2.7−2.8 for the PS particles, while 1.7−1.8 and 2.5−2.6 for the PMMA particles. The inset graph shows the zoomed in curves between κ = 0 and 1.2.

■

RESULTS AND DISCUSSION Comparison of ARF Factors for PS and PMMA Particles. To experimentally evaluate the different acoustophoretic responses of the same-sized PS and PMMA particles, a set of chirped interdigital transducers (CIDTs) was used to generate frequency tunable acoustic fields, as depicted in Figure 3a. An aqueous suspension containing a mixture of 15 μm PS

particle forces, the successive peaks and dips for these two particle materials are offset with respect to each other. Specifically, the ARF factor peaks for PS particle between κ = 1 and 3 lies in three narrow ranges: 1.4−1.5, 2.1−2.2, and 2.7− 2.8, whereas those for PMMA particles occur at 1.7−1.8 and 2.5−2.6. Therefore, even when the same-sized PS and PMMA particles are exposed to the same acoustic field, it is possible to generate distinct acoustophoretic responses for the two different particle types. Tuning the κ value should enable selective switching from strong acoustophoresis to weak acoustophoresis for one of the particle populations and vice versa. Comparison of ARF in Traveling and Standing Acoustic Fields. Not only in a traveling acoustic field, the suspended particle is also subject to acoustic radiation force, FS, when exposed in a standing acoustic field generated by the interference of two oppositely traveling waves. Standing acoustic fields have been widely demonstrated for size-based microparticle separation in previous studies.15,18,23,66 The theoretical model developed by Yosioka et al.39 is most commonly used to describe the acoustic radiation force in a standing acoustic field as ⟨Fs⟩ = Ysπa 2⟨E⟩sin 2kx Ys =

4 ⎛ 5β − 2 1 ⎞ κ⎜ − ⎟ 3 ⎝ 2β + 1 βγ 2 ⎠

(8)

(9)

where Ys is the ARF factor in a standing acoustic field, ⟨E⟩ is the mean energy density of the standing field, a is the radius of the spherical particle, k = 2π/λ is the wavenumber with λ being the acoustic wavelength, and x is the distance from the pressure node. β and γ is the density ratio and the sound speed ratio of the particle to the suspension medium, respectively. Similar as ARF in traveling acoustic field FT, Fs correlates to the particle density, sound speed, and κ value, which results in a difference in Fs for particles with different mechanical properties, which has been demonstrated for density-based microparticles separation.52 However, eq 9 shows that the ARF acting on

Figure 3. Experimental measurement of ARF factors of PS and PMMA particles. (a) Disposable PDMS channel containing suspended 15 μm PS and PMMA particles was placed on the SAW transducers, where prepatterned chirped IDTs generate acoustic waves at 43.04/45.52/ 50.77/56.57 MHz. The power applied is 126 mW in all cases. (b) Tracked y-position for the two particles and their linear fit during the lateral translation. The ratio of the ARF factors is derived by the average velocities, which are represented as the slope of the fitting lines. (c) One each of PS (fluorescent) and PMMA (plain) particles are translated in a 56.57 MHz TSAW field in this image, obtained by overlapping 18 frames recorded every 0.2 s. D

DOI: 10.1021/acs.analchem.6b03580 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry and 15 μm PMMA particles was introduced into the disposable PDMS channel that was sealed at all the opening inlets/outlets to eliminate effects of Stokes drag resulting from fluid flow and was then aligned on the TSAW transducer. The PS and PMMA particle mixture was laterally translated across the channel in the direction of acoustic propagation with the application of an AC signal to the CIDTs. The average velocities of the moving particles over successive 0.05 s time periods were derived from the trajectories tracked by an academically published particle tracking code.67 The acoustic radiation force is balanced by Stokes drag force FD = 6πμav, where μ is the dynamic viscosity, a is the particle radius, and v is the particle velocity relative to the flow. Thus, the ratio of the ARF force on the PS and PMMA particles, RT, can be derived by the relative average velocities of their translations. Figure 3b,c shows the trajectories of a PS particle and a PMMA particle exposed to a 56.57 MHz TSAW field (κ = 1.78). The experimental observation shows that the 15 μm PMMA particle moves faster than the 15 μm PS particle, which is in qualitative agreement with the theoretical prediction shown in Figure 2. To quantify the acoustophoretic responses at varying κ values, four frequencies 43.04/45.52/ 50.77/56.57 MHz were selected that correspond to κ values of 1.35/1.43/1.60/1.78, respectively. For each frequency, 200 pairs of the PS and PMMA particles were tracked for velocity analysis. Figure 4 shows the comparison between the

experiment, as shown in Figure 4, however, it can be concluded that the acoustic absorption within the particles can considerably dampen the radiation force and thus induce smaller ARF factor peaks in the YT−κ curves, as shown in Figure 3. Moreover, the drag of rolling particles against the channel roof is difficult to account for and thus not explicitly modeled here. Regardless of these effects, theoretical predictions and experimental measurements confirm that κ values at κ = 1.43 and 1.78 generate substantially different ARF factors between PS and PMMA particles, which can potentially be utilized to achieve continuous separation of these populations by tuning the TSAW frequency to match these κ values. Particle Separation Based on Mechanical Properties. Having established the divergent forces imparted on different particle populations according to their mechanical properties, we now demonstrate the separation of PS and PMMA particles by selectively laterally displacing them within a continuous flow using TSAWs. Micrographs of these particles used in the separation experiments are shown in the Supporting Information, Figure S2. In this study, all the PS particles are fluorescent (white) while the PMMA particles are not (gray). The sample flow (1.0 μL/min) containing both PS and PMMA particles is hydrodynamically sandwiched by two asymmetric sheath flows (the upper sheath flow: 4.0 μL/min and the lower sheath flow: 1.5 μL/min). Without the application of the acoustic field, the hydrodynamically confined sample stream flows into outlet B (closer to the IDTs), as shown in the Supporting Information, Figure S3. With the application of a TSAW field at a particular frequency with optimized input power applied to the IDTs, we achieve the selective separation of PS and PMMA particles into outlet A (farther from the IDTs) under the experimental conditions summarized in the Table 2. Table 2. Parameters for the Experimental Mechanical Properties Based Separation

PMMA Figure 4. Comparison of RT = YPS between the theoretical T /YT model and our experimental measurements. The inset graph shows the peak of RT at κ = 1.3−1.55. Error bars represent one standard deviation of the RT measurements. Each of the points represents 200 PS or PMMA particles deflection and tracking events at the four applied frequencies.

frequency (MHz)

particle diameter PS/PMMA (μm/μm)

κ value PS/ PMMA

power (mW)

figure

45.52 56.57 68.28 68.28

15/15 15/15 10/10 10/15

1.43/1.43 1.78/1.78 1.43/1.43 1.43/2.15

50 48 72 65

5a 5b 5c 5d

To demonstrate the effective separation according to the particle’s mechanical properties, we utilize our acoustophoretic microfluidic system to separate 15 μm PS and 15 μm PMMA particles in a 45.52 MHz TSAW field, corresponding to κ = 1.43, at which the ARF imparted on PS particles is maximized relative to that on PMMA ones. With an applied power of 50 mW, the PS particles were selectively translated to outlet A in the direction of wave propagation (farther from the IDTs), while the PMMA particles still exited the sorting region via outlet B, as shown in Figure 5a (Supporting Information, Video S3). Similarly, the PMMA particles can be selectively translated at κ value conditions that maximize the force imparted on them relative to those of PS. Accordingly, we performed the selective PMMA particle separation in a 56.57 MHz TSAW field, corresponding to κ = 1.78 at an applied power of 48 mW, as shown in Figure 5b (Supporting Information, Video S4). We also demonstrate the separation of 10 μm PS and 10 μm PMMA particles while maintaining optimized κ values by

theoretical curve and the experimental measurement of the RT value, in which RT > 1 at κ = 1.35 and 1.43, while RT < 1 at κ = 1.60 and 1.78. These experimental measurements follow the theoretical prediction with regard to relative imparted ARF, though the magnitude of this difference is somewhat smaller than the theoretical prediction. This discrepancy is principally attributed to the fact that the theoretical model does not consider the wave absorption within the particles; though discussed later by Hasegawa et al. and F. G. Mitri,68,69 the wave attenuation coefficients of the PS and PMMA particles are currently unavailable and thus not considered in the present model. Based on our comparison between theory and E

DOI: 10.1021/acs.analchem.6b03580 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

translated to outlet A, while the 15 μm PS particles were collected in outlet B, as shown in Figure 5d (Supporting Information, Video S6). The force scaling between particles of the same composition still applies, however, the subsequent separation distance between these dissimilarly sized particle types was less than that in the case of Figure 5c. The sorting efficiency in each of these test cases is presented in Figure 6, at least 500 particles were counted at the outlet

Figure 5. Mechanical properties based particle separation. (a) In the 45.52 MHz TSAW field, the 15 μm PS particles were translated into the upper outlet while the 15 μm PMMA particles flowing into the PMMA relationship at κ = 1.43. (b) In lower outlet due to the YPS T > YT contrast to the condition in (a), in the 56.57 MHz TSAW field the 15 μm PMMA particles are translated to the upper outlet when YPS T < at κ = 1.78. (c) Similar to the situation at (a), the 10 μm PS and YPMMA T PMMA particles are separated in the 68.28 MHz TSAW field while maintaining κ = 1.43. (d) The 10 μm PS particles can be separated from 15 μm PMMA particles with a larger radiation force acting on the 10 μm PS particles. These images are obtained by overlapping 40 frames recorded every 0.05 s. The scale bar is 100 μm.

Figure 6. Percentage of the particles flowing into outlets A and B analyzed for the separation experiments in Figure 5. The purities of separated particles are all higher than 95%.

junction in each case to calculate the percentage of the two types of particles exiting the separation zone in each outlet. Even for the most challenging separation case, in Figure 5d, the purity of separated PS and PMMA particles exceeds 95%, demonstrating the utility of TSAW based continuous separation according to the composition of the sorted populations.

altering the applied frequency, that is, via the use of a 68.28 MHz TSAW field at κ = 1.43. As expected, the 10 μm PS particles at this κ value also experienced a larger radiation force compared to the 10 μm PMMA particles and are selectively sorted, as indicated in Figure 5c (supplementary video S5). Though κ = 1.43 corresponds to a fixed ARF factor in the YT−κ curve, as indicated in eq 1, the radiation force scales with a2. Therefore, the decrease in the particle size results in a smaller acoustic radiation force on both 10 μm PS and 10 μm PMMA particles, and thus requires a higher input power (72 mW) to achieve similar separation efficiencies. This technique is also robust to substantial differences in size between particle populations. We further demonstrate the separation of 10 μm PS particles from 15 μm PMMA particles, despite the fact that the smaller particle diameter, frontal area and volume is 67, 44, and 29% of that for the larger PMMA particles. In order to do so, we maintain the application of a 68.28 MHz TSAW field, which corresponds to κ = 1.43 for PS particles and κ = 2.14 for PMMA particles. Because of the F ∼ a2 scaling, the radiation force on a 15 μm particle will be 2.25 larger as that on a 10 μm particle of the same material. However, as shown in Figure 2, the ARF factor of PS particles at κ = 1.43 is at its maximum, while the ARF factor of PMMA particles at κ = 2.14 is at its minimum. Despite this smaller size of the PS particles, we demonstrate the selective displacement and separation of 10 μm PS particles from 15 μm PMMA particles due to the larger radiation force acting on the 10 μm PS particles. At an input power of 65 mW, the smaller 10 μm PS particles were

■

CONCLUSIONS In this work we demonstrate for the first time that particles with the same dimensions can be effectively separated solely by the difference in their mechanical properties with traveling surface acoustic waves generated by a single set of IDTs. This separation relies on the fact that the ARF factor YT scales nonlinearly with particle diameter as this value approaches and exceeds that of the acoustic wavelength in the fluid (κ > 1) and can be represented as a series of maxima and minima as a function of a dimensionless factor κ = 2πaf/cf. A theoretical model reveals that PS and PMMA particles have offset ARF factor peaks between κ = 1 and 3, validated by tracking the particle motions in acoustic fields generated by TSAW at different κ values. Our experimental studies confirm the divergent ARF factor peaks for PS and PMMA particles, and PMMA the experimentally derived force factor ratios of YPS at T to YT κ = 1.35/1.43/1.60/1.78 were comparable with the predictions made from a theoretical model that accounts for acoustic scattering. Based on the distinct acoustophoretic responses arising from the difference in their mechanical properties, samesized PS and PMMA particles (10 and 15 μm) were efficiently F

DOI: 10.1021/acs.analchem.6b03580 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry separated in a continuous flow using TSAW fields at varying resonant frequencies (45.52/56.57/62.28 MHz). Though the radiation force scales with a2, we also demonstrated that it is feasible to exert a larger radiation force on 10 μm PS particles than 15 μm PMMA particles at an optimized κ value and, also, to implement separation at efficiencies greater than 95%. Though further study of acoustic interactions with inhomogeneous and nonspherical particles is required to assess the viability of this technique for specific mixed cell population applications, this study demonstrates for the first time the potential of traveling acoustic wave fields to avoid limitations inherent to selective standing-wave based focusing to separate suspended micro-objects according to their mechanical properties.

■

(8) Zhu, G.-P.; Hejiazan, M.; Huang, X.; Nguyen, N.-T. Lab Chip 2014, 14, 4609−4615. (9) Hejazian, M.; Li, W.; Nguyen, N.-T. Lab Chip 2015, 15, 959− 970. (10) Jung, J. H.; Lee, K. H.; Lee, K. S.; Ha, B. H.; Oh, Y. S.; Sung, H. J. Microfluid. Nanofluid. 2014, 16, 635−644. (11) Kim, J.; Lee, J.; Wu, C.; Nam, S.; Di Carlo, D.; Lee, W. Lab Chip 2016, 16, 992−1001. (12) Bhagat, A. A. S.; Hou, H. W.; Li, L. D.; Lim, C. T.; Han, J. Lab Chip 2011, 11, 1870−1878. (13) Augustsson, P.; Magnusson, C.; Nordin, M.; Lilja, H.; Laurell, T. Anal. Chem. 2012, 84, 7954−7962. (14) Skowronek, V.; Rambach, R. W.; Schmid, L.; Haase, K.; Franke, T. Anal. Chem. 2013, 85, 9955−9959. (15) Ding, X.; Peng, Z.; Lin, S.-C. S.; Geri, M.; Li, S.; Li, P.; Chen, Y.; Dao, M.; Suresh, S.; Huang, T. J. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 12992−12997. (16) Destgeer, G.; Ha, B. H.; Jung, J. H.; Sung, H. J. Lab Chip 2014, 14, 4665−4672. (17) Antfolk, M.; Magnusson, C.; Augustsson, P.; Lilja, H.; Laurell, T. Anal. Chem. 2015, 87, 9322−9328. (18) Li, P.; Mao, Z.; Peng, Z.; Zhou, L.; Chen, Y.; Huang, P.-H.; Truica, C. I.; Drabick, J. J.; El-Deiry, W. S.; Dao, M. Proc. Natl. Acad. Sci. U. S. A. 2015, 112, 4970−4975. (19) Wiklund, M.; Christakou, A. E.; Ohlin, M.; Iranmanesh, I.; Frisk, T.; Vanherberghen, B.; Ö nfelt, B. Micromachines 2014, 5, 27−49. (20) Christakou, A. E.; Ohlin, M.; Ö nfelt, B.; Wiklund, M. Lab Chip 2015, 15, 3222−3231. (21) Collins, D. J.; Ma, Z.; Ai, Y. Anal. Chem. 2016, 88, 5513−5522. (22) Franke, T.; Braunmüller, S.; Schmid, L.; Wixforth, A.; Weitz, D. Lab Chip 2010, 10, 789−794. (23) Ai, Y.; Sanders, C. K.; Marrone, B. L. Anal. Chem. 2013, 85, 9126−9134. (24) Schmid, L.; Weitz, D. A.; Franke, T. Lab Chip 2014, 14, 3710− 3718. (25) Collins, D. J.; Alan, T.; Neild, A. Lab Chip 2014, 14, 1595− 1603. (26) Ma, Z.; Collins, D. J.; Ai, Y. Anal. Chem. 2016, 88, 5316−5323. (27) Collins, D. J.; Neild, A.; Ai, Y. Lab Chip 2016, 16, 471−479. (28) Ding, X.; Lin, S.-C. S.; Kiraly, B.; Yue, H.; Li, S.; Chiang, I.-K.; Shi, J.; Benkovic, S. J.; Huang, T. J. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 11105−11109. (29) Bernassau, A.; MacPherson, P.; Beeley, J.; Drinkwater, B.; Cumming, D. Biomed. Microdevices 2013, 15, 289−297. (30) Shilton, R. J.; Travagliati, M.; Beltram, F.; Cecchini, M. Adv. Mater. 2014, 26, 4941−4946. (31) Ma, Z.; Guo, J.; Liu, Y. J.; Ai, Y. Nanoscale 2015, 7, 14047− 14054. (32) Guo, F.; Zhou, W.; Li, P.; Mao, Z.; Yennawar, N. H.; French, J. B.; Huang, T. J. Small 2015, 11, 2733−2737. (33) Collins, D. J.; Devendran, C.; Ma, Z.; Ng, J. W.; Neild, A.; Ai, Y. Sci. Adv. 2016, 2, e1600089. (34) Collins, D. J.; Morahan, B.; Garcia-Bustos, J.; Doerig, C.; Plebanski, M.; Neild, A. Nat. Commun. 2015, 6, 8686. (35) Collins, D. J.; Alan, T.; Neild, A. Appl. Phys. Lett. 2014, 105, 033509. (36) Destgeer, G.; Cho, H.; Ha, B. H.; Jung, J. H.; Park, J.; Sung, H. J. Lab Chip 2016, 16, 660−667. (37) Destgeer, G.; Ha, B.; Park, J.; Sung, H. J. Anal. Chem. 2016, 88, 3976−3981. (38) Gor’Kov, L. Sov. Phys. Doklady 1962, 773. (39) Yosioka, K.; Kawasima, Y. Acta Acust. Acust. 1955, 5, 167−173. (40) Segers, T.; Versluis, M. Lab Chip 2014, 14, 1705−1714. (41) Leibacher, I.; Reichert, P.; Dual, J. Lab Chip 2015, 15, 2896− 2905. (42) Suresh, S.; Spatz, J.; Mills, J.; Micoulet, A.; Dao, M.; Lim, C.; Beil, M.; Seufferlein, T. Acta Biomater. 2005, 1, 15−30. (43) Suresh, S. Acta Mater. 2007, 55, 3989−4014.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.6b03580. S1: Summary of three types of surface acoustic wave based microparticle separation (Table S1); Correlation of the first peak location in YT−κ curve with respect to relative particle density and relative particle sound speed (Figure S1); Microscopic images of used microparticles (Figure S2); Superimposed trajectories of particles when acoustic wave is off (Figure S3; PDF). S2: Video of the deflection experiments (AVI). S3: Video of separating 15 μm PS and 15 μm PMMA particles in 45.52 MHz traveling acoustic field (AVI). S4: Video of separating 15 μm PMMA and 15 μm PS particles in 56.57 MHz traveling acoustic field (AVI). S5: Video of separating 10 μm PS and 10 μm PMMA particles in 68.28 MHz traveling acoustic field (AVI). S6: Video of separating 10 μm PS and 15 μm PMMA particles in 68.28 MHz traveling acoustic field (AVI).

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: (+65) 6499 4553. Fax: (+65) 6779 5161. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by SUTD Startup Research Grant (SREP13053) and Singapore Ministry of Education Academic Research Fund Tier 2 (T2MOE1603) awarded to YA.

■

REFERENCES

(1) Sajeesh, P.; Sen, A. K. Microfluid. Nanofluid. 2014, 17, 1−52. (2) Maheswaran, S.; Sequist, L. V.; Nagrath, S.; Ulkus, L.; Brannigan, B.; Collura, C. V.; Inserra, E.; Diederichs, S.; Iafrate, A. J.; Bell, D. W. N. Engl. J. Med. 2008, 359, 366−377. (3) Yu, M.; Stott, S.; Toner, M.; Maheswaran, S.; Haber, D. A. J. Cell Biol. 2011, 192, 373−382. (4) Forsberg, E. C.; Bhattacharya, D.; Weissman, I. L. Stem Cell Rev. 2006, 2, 23−30. (5) Abdallah, B.; Kassem, M. Gene Ther. 2008, 15, 109−116. (6) Kang, Y.; Li, D.; Kalams, S. A.; Eid, J. E. Biomed. Microdevices 2008, 10, 243−249. (7) Puttaswamy, S. V.; Xue, P.; Kang, Y.; Ai, Y. Biomed. Microdevices 2015, 17, 1−5. G

DOI: 10.1021/acs.analchem.6b03580 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry (44) Jin, C.; McFaul, S. M.; Duffy, S. P.; Deng, X.; Tavassoli, P.; Black, P. C.; Ma, H. Lab Chip 2014, 14, 32−44. (45) McFaul, S. M.; Lin, B. K.; Ma, H. Lab Chip 2012, 12, 2369− 2376. (46) Holmes, D.; Whyte, G.; Bailey, J.; Vergara-Irigaray, N.; Ekpenyong, A.; Guck, J.; Duke, T. Interface Focus 2014, 4, 20140011. (47) Hur, S. C.; Henderson-MacLennan, N. K.; McCabe, E. R.; Di Carlo, D. Lab Chip 2011, 11, 912−920. (48) Nam, J.; Lim, H.; Kim, C.; Kang, J. Y.; Shin, S. Biomicrofluidics 2012, 6, 024120. (49) Barnkob, R.; Augustsson, P.; Magnusson, C.; Lilja, H.; Laurell, T.; Bruus, H. Proceedings of the 15th MicroTAS 2011, 2−6. (50) Hartono, D.; Liu, Y.; Tan, P. L.; Then, X. Y. S.; Yung, L.-Y. L.; Lim, K.-M. Lab Chip 2011, 11, 4072−4080. (51) Petersson, F.; Nilsson, A.; Holm, C.; Jönsson, H.; Laurell, T. Lab Chip 2005, 5, 20−22. (52) Jo, M. C.; Guldiken, R. Sens. Actuators, A 2012, 187, 22−28. (53) Augustsson, P.; Karlsen, J. T.; Su, H.-W.; Bruus, H.; Voldman, J. Nat. Commun. 2016, 7, 11556. (54) Destgeer, G.; Lee, K. H.; Jung, J. H.; Alazzam, A.; Sung, H. J. Lab Chip 2013, 13, 4210−4216. (55) Leibacher, I.; Schatzer, S.; Dual, J. Lab Chip 2014, 14, 463−470. (56) Hasegawa, T.; Yosioka, K. J. Acoust. Soc. Am. 1969, 46, 1139− 1143. (57) Destgeer, G.; Ha, B. H.; Park, J.; Jung, J. H.; Alazzam, A.; Sung, H. J. Anal. Chem. 2015, 87, 4627−4632. (58) Skowronek, V.; Rambach, R. W.; Franke, T. Microfluid. Nanofluid. 2015, 19, 335−341. (59) Behrens, J.; Langelier, S.; Rezk, A. R.; Lindner, G.; Yeo, L. Y.; Friend, J. R. Lab Chip 2015, 15, 43−46. (60) Skowronek, V.; Rambach, R. W.; Schmid, L.; Haase, K.; Franke, T. Anal. Chem. 2013, 85, 9955−9959. (61) Hasegawa, T.; Yosioka, K. J. Acoust. Soc. Am. 1975, 58, 581−585. (62) Ding, X.; Lin, S.-C. S.; Lapsley, M. I.; Li, S.; Guo, X.; Chan, C. Y.; Chiang, I.-K.; Wang, L.; McCoy, J. P.; Huang, T. J. Lab Chip 2012, 12, 4228−4231. (63) Augustsson, P.; Barnkob, R.; Grenvall, C.; Deierborg, T.; Brundin, P.; Bruus, H.; Laurell, T. Proceedings of 14th MicroTAS 2010, 14, n/a. (64) Barnkob, R.; Augustsson, P.; Laurell, T.; Bruus, H. Lab Chip 2010, 10, 563−570. (65) Destgeer, G.; Sung, H. J. Lab Chip 2015, 15, 2722−2738. (66) Shi, J.; Huang, H.; Stratton, Z.; Huang, Y.; Huang, T. J. Lab Chip 2009, 9, 3354−3359. (67) Basu, A. S. Lab Chip 2013, 13, 1892−1901. (68) Hasegawa, T.; Watanabe, Y. J. Acoust. Soc. Am. 1978, 63, 1733− 1737. (69) Mitri, F. Wave Motion 2005, 43, 12−19.

H

DOI: 10.1021/acs.analchem.6b03580 Anal. Chem. XXXX, XXX, XXX−XXX