betaFeedback

WorksPhysics and Astronomy

A transition from boundary- to bulk-driven acoustic streaming due to nonlinear thermoviscous effects at high acoustic energy densities

Abstract

Acoustic streaming is studied in a rectangular microfluidic channel. It is demonstrated theoretically, numerically, and experimentally with good agreement, frictional heating can alter the streaming pattern qualitatively at high acoustic energy densities E_ac above 500 J/m^3. The study shows, how as a function of increasing E_ac at fixed frequency, the traditional boundary-driven four streaming rolls created at a half-wave standing-wave resonance, transition into two large streaming rolls. This nonlinear transition occurs because friction heats up the fluid resulting in a temperature gradient, which spawns an acoustic body force in the bulk that drives thermoacoustic streaming.

A transition from boundary- to bulk-driven acoustic streaming due to nonlinear thermoviscous effects at high acoustic en

Microscale acoustofluidic devices are used to manipulate and control microparticles and cells. In such devices, two main forces act on the suspended particles, the acoustic radiation force and the drag force due to acoustic streaming, which is a time-averaged flow caused by the inherent nonlinearities of fluid dynamics. Recent work has clarified many subtle details pertaining to the radiation force on microparticles, including thermoviscous effects [1] and microstreaming [2]. Concurrently, similar progress has been made in the theory of acoustic streaming, especially regarding thermoviscous effects. The fundamental boundary-driven streaming caused by time-averaged forces in the oscillatory boundary-layer flow [3], and the fundamental bulk-driven streaming generated by the time-averaged dissipation of traveling waves [4], have recently been supplemented by bulk-driven baroclinic [5, 6] and thermoacoustic [7, 8] streaming, caused by an interplay between standing acoustic waves and steady temperature gradients. However, as noted in Refs. [7, 8], the validity of the conventional perturbation approach breaks down at moderately high, but experimentally obtainable acoustic energy densities above 100 J/m3^{3} in combination with moderate thermal gradients above 1 K/mm. This need for an extension of the theory beyond perturbation theory is addressed in this Letter and in the accompanying detailed presentation of the nonperturbative model in Ref. [9].

We introduce a nonperturbative iteration approach to investigate theoretically and numerically, the nonlinear effects appearing in a conventional acoustofluidic channel at high acoustic energy density EacE_{\mathrm{ac}}, and we validate experimentally the model predictions. We take as our generic acoustofluidic model system, the widely used rectangular channel driven at resonance with a transverse half-wave standing acoustic wave, for which the streaming at low EacE_{\mathrm{ac}} is dominated by conventional boundary-driven streaming with four streaming rolls [10, 11, 12, 13]. We show how nonlinear effects in the form of heating by viscous dissipation from the acoustic field inside the boundary layers, set up a steady temperature gradient.

This gradient drives a strong thermoacoustic streaming in the bulk, which changes the streaming flow qualitatively from four to two flow rolls, and which by thermal convection alters the temperature field. Our analysis of this nonlinear phenomenon and its underlying mechanism fills a knowledge gap in nonlinear acoustics, and it provides a guidance for understanding and optimizing acoustofluidic systems running at high EacE_{\mathrm{ac}} such as high-intensity ultrasound focusing [14, 15, 16], acoustic streaming-based micromixers [17, 18, 19, 20], particle manipulation devices [21, 22, 23], and high-throughput acoustophoresis devices [24, 25, 26].

*Physical model.—*We consider a long straight microchannel with a cross-sectional of width Wfl=375W_{\mathrm{fl}} = 375 μm and height Hfl=135H_{\mathrm{fl}} = 135 μm embedded in a silicon base with a glass lid, see Figure 1 and Refs. [9, 10, 11, 12, 13]. To excite the horizontal half-wave resonance mode in the fluid, the system is actuated at frequency f0=1.911f_0 = 1.911 MHz by the bottom-edge displacement sketched in Figure 1. The response to the acoustic actuation is governed by the conservation equation for mass, momentum, and energy in the fluid and solid. The independent fields are the pressure pp, the velocity v\mathbf{v}, and the temperature TT in the fluid, and the displacement u\mathbf{u} and TT in the solid.

Image not extracted

Figure 1. Cross-section of the modeled rectangular channel of width Wfl=375W_{\mathrm{fl}} = 375 μm and height Hfl=135H_{\mathrm{fl}} = 135 μm embedded in a silicon base of width Wsl=3W_{\mathrm{sl}} = 3 mm and height Hsi=0.4H_{\mathrm{si}} = 0.4 mm, and covered by a glass lid of height Hgl=1H_{\mathrm{gl}} = 1 mm. The bottom-edge actuation is the displacement u1d=d0(y/Wsl)eyu_1^d = d_0(y/W_{\mathrm{sl}})\mathbf{e}_y with an adjustable amplitude d0d_0 at frequency f0=1.911f_0 = 1.911 MHz.

We study a fluid characterized by the following material parameters: density ρ\rho, isothermal compressibility κT\kappa_T, thermal conductivity kthk^{\mathrm{th}}, specific heat cpc_p, dynamic and bulk viscosity η\eta and ηb\eta^b, thermal expansion coefficient αp\alpha_p, the ratio of specific heats γ=cp/cv\gamma= c_p/c_v, and the isentropic and isothermal compressibility κs\kappa_s and κT=γκs\kappa_T = \gamma\kappa_s. The temperature dependence of the parameters for water are given by the polynomials derived in Ref. [12]. The elastic solid is characterized by density ρ\rho, longitudinal and transverse sound speed cloc_{\mathrm{lo}} and ctrc_{\mathrm{tr}}, thermal conductivity kthk^{\mathrm{th}}, thermal expansion coefficient αp0\alpha_{p0} and isothermal compressibility κT\kappa_T [9].

We use the iterative thermoviscous model presented in our concurrent paper [9], a model that is nonperturbative unlike the perturbative models traditionally used in acoustofluidics [7, 12]. The model exploits that the acoustic fields varies much faster (∼10−7 s\sim10^{-7}\,\mathrm{s}) than the hydrodynamic and thermal flows (∼10−2 s\sim10^{-2}\,\mathrm{s}), so that the fast and slow dynamics can be solved separately. Here, we study the stationary limit of the slow time scale and describe any given physical field QphysQ_{\mathrm{phys}} as a sum of a stationary field Q0Q_0 and a time-varying acoustic field Re⁡{Q1e−iωt}\operatorname{Re}\{Q_1e^{-i\omega t}\} with a stationary complex-valued amplitude Q1Q_1,

Qphys(t)=Q0+Re⁡{Q1e−iωt}.(1)Q_{\mathrm{phys}}(t)=Q_0+\operatorname{Re}\{Q_1e^{-i\omega t}\}. \tag*{(1)}

A product of two acoustic fields will contain a steady part ⟨a1b1⟩=12Re⁡{a1b1∗}\langle a_1b_1\rangle=\frac{1}{2}\operatorname{Re}\{a_1b_1^*\} where the asterisk denotes complex conjugation. We neglect higher harmonics with angular frequency nωn\omega, n=2,3,…n=2,3,\ldots. In Ref. [9], we use this ansatz to separate the governing equations in a set that controls the acoustic fields, and a set that controls the stationary fields.

Acoustofluidic systems also exhibit dynamics on two different length scales, one set by the wavelength of the acoustic fields and one set by the viscous and thermal boundary layers. The boundary conditions on the velocity field, stress, heat flux, and temperature at the fluid-solid interface results in the appearance of thermal boundary layer of width δt\delta_t in both the fluid and the solid, and in a viscous boundary layer of width δs\delta_s in the fluid. These boundary layers are localized near fluid-solid interfaces, and their dynamically-defined widths (jointly called δ\delta) are small compared to a typical device size or wavelength dd [1], δs=2ν0/ω\delta_s=\sqrt{2\nu_0/\omega} and δt≈2Dth/ω\delta_t\approx\sqrt{2D^{\mathrm{th}}/\omega}. Typically, δt≲δs≲500 nm\delta_t\lesssim\delta_s\lesssim500\ \mathrm{nm}, which is more than two orders of magnitude smaller than d∼100 μmd\sim100\ \mathrm{\mu m}. In our model [9], we use δ≪d\delta\ll d to separate the fields into a bulk field (dd) and boundary layer field (δ\delta), that are connected through the boundary conditions at the fluid-solid interface. The model presented in Ref. [9] uses the separation of time and length scales to setup an iterative model with effective boundary conditions that enables simulations of nonperturbative acoustofluidic systems without numerically resolving the viscous and thermal boundary layers. All dependent fields are given analytically by the independent fields as shown in Ref. [9].

Acoustic fields.—In the thermoviscous model of Ref. [9], the acoustics is fully described by the pressure field p1p_1 in the fluid and by the displacement field u1d\mathbf{u}_1^d in the solid through the Helmholtz and Cauchy equations,

∇2p1=−kc2p1,kc=ωc0(1+iΓcfl)(2)\nabla^2p_1=-k_c^2p_1,\qquad k_c=\frac{\omega}{c_0}(1+i\Gamma_c^{\mathrm{fl}}) \tag*{(2)}
−ρω2u1d=∇⋅σ1d,sl,(3)-\rho\omega^2\mathbf{u}_1^d=\nabla\cdot\boldsymbol{\sigma}_1^{d,\mathrm{sl}}, \tag*{(3)}
σ1d,sl=−αpκTT1I+ρctr2[∇u1+(∇u1)T]+ρ(clo2−2ctr2)(∇⋅u1)I.(4)\boldsymbol{\sigma}_1^{d,\mathrm{sl}}=-\frac{\alpha_p}{\kappa_T}T_1\mathbf{I}+\rho c_{\mathrm{tr}}^2\left[\nabla\mathbf{u}_1+\left(\nabla\mathbf{u}_1\right)^{\mathsf{T}}\right] +\rho\left(c_{\mathrm{lo}}^2-2c_{\mathrm{tr}}^2\right)(\nabla\cdot\mathbf{u}_1)\mathbf{I}. \tag*{(4)}

where σ1\boldsymbol{\sigma}_1 is the stress tensor. p1p_1 and u1d\mathbf{u}_1^d are connected through effective boundary conditions on the fluid-solid interface taking the boundary layers into account analytically: a no-slip condition for the velocity and a continuity condition on the stress, as described in Ref. [9]. The acoustic velocity v1\mathbf{v}_1 and the temperature field T1T_1 are given by p1p_1 and the analytical boundary-layer fields [7].

Stationary fields.—The stationary fields are the pressure p0p_0 and streaming velocity v0d\mathbf{v}_0^d in the fluid, and the temperature T0dT_0^d in the solid and fluid. The acoustic timescale affects T0dT_0^d, p0p_0, and v0d\mathbf{v}_0^d through bulk terms (heating and the acoustic body force facd\mathbf{f}_{\mathrm{ac}}^d) and corrections due to the boundary layers to the boundary conditions. The governing equations for v0d\mathbf{v}_0^d and p0p_0 are [9],

0=∇⋅v0d,(5)0=\nabla\cdot\mathbf{v}_0^d, \tag*{(5)}
0=−∇[p0d−⟨pacd⟩]+η0∇2v0d−∇⋅[ρ0v0v0]+facd.(6)0=-\nabla\left[p_0^d-\langle p_{\mathrm{ac}}^d\rangle\right]+\eta_0\nabla^2\mathbf{v}_0^d-\nabla\cdot\left[\rho_0\mathbf{v}_0\mathbf{v}_0\right]+\mathbf{f}_{\mathrm{ac}}^d. \tag*{(6)}
facd=−14∣v1d,p∣2∇ρ0−14∣p1∣2∇κs+(1−2αp(γ−1)β+1)Γωc02⟨v1d,pp1⟩+2αpη0(γ−1)ωc02⟨iv1d,p⋅∇v1d,p⟩.(7)\mathbf{f}_{\mathrm{ac}}^d=-\frac{1}{4}\left|\mathbf{v}_1^{d,p}\right|^2\nabla\rho_0-\frac{1}{4}\left|p_1\right|^2\nabla\kappa_s +\left(1-\frac{2\alpha_p(\gamma-1)}{\beta+1}\right)\frac{\Gamma\omega}{c_0^2}\langle\mathbf{v}_1^{d,p}p_1\rangle +2\alpha_p\eta_0(\gamma-1)\frac{\omega}{c_0^2}\langle i\mathbf{v}_1^{d,p}\cdot\nabla\mathbf{v}_1^{d,p}\rangle. \tag*{(7)}

The boundary layers generate a slip velocity on the solid-fluid interface, and the resulting effective boundary condition on v0d\mathbf{v}_0^d is [9],

v0d0=(A⋅ex)ex+(A⋅ey)ey+(B⋅ez)ez,(8)\mathbf{v}_0^{d0}=(\mathbf{A}\cdot\mathbf{e}_x)\mathbf{e}_x+(\mathbf{A}\cdot\mathbf{e}_y)\mathbf{e}_y+(\mathbf{B}\cdot\mathbf{e}_z)\mathbf{e}_z, \tag*{(8)}
A=−12ωRe⁡{v1d0∗⋅∇(12v1d0−iV10)−iV10∗⋅∇v1d,p+[2−ι2∇⋅v1d0∗+i(∇⋅V10∗−∂zv1zd∗)]v1d0}+12η0Re⁡{η1d0v1d0∗+11−iδs/δtη1δ0v1d0∗}.(9)\mathbf{A}=-\frac{1}{2\omega}\operatorname{Re}\left\{\mathbf{v}_1^{d0*}\cdot\nabla\left(\frac{1}{2}\mathbf{v}_1^{d0}-i\mathbf{V}_1^{0}\right)-i\mathbf{V}_1^{0*}\cdot\nabla\mathbf{v}_1^{d,p} +\left[\frac{2-\iota}{2}\nabla\cdot\mathbf{v}_1^{d0*}+i\left(\nabla\cdot\mathbf{V}_1^{0*}-\partial_zv_{1z}^{d*}\right)\right]\mathbf{v}_1^{d0}\right\} +\frac{1}{2\eta_0}\operatorname{Re}\left\{\eta_1^{d0}\mathbf{v}_1^{d0*}+\frac{1}{1-i\delta_s/\delta_t}\eta_1^{\delta0}\mathbf{v}_1^{d0*}\right\}. \tag*{(9)}
B=12ωRe⁡{iv1d0∗⋅∇v1d,p}.(10)\mathbf{B}=\frac{1}{2\omega}\operatorname{Re}\left\{i\mathbf{v}_1^{d0*}\cdot\nabla\mathbf{v}_1^{d,p}\right\}. \tag*{(10)}

Here, V10\mathbf{V}_1^0 is the velocity of the fluid-solid interface, ez\mathbf{e}_z is the surface normal vector, and ex\mathbf{e}_x and ey\mathbf{e}_y are perpendicular to ez\mathbf{e}_z. The streaming flow v0d\mathbf{v}_0^d can be driven either by the acoustic body force facd\mathbf{f}_{ac}^d, called bulk-driven streaming, or by the effective boundary condition on v0d\mathbf{v}_0^d, called boundary-driven streaming.

The stationary temperature T0d,flT_0^{d,\mathrm{fl}} in the fluid is governed by the heat equation (energy conservation) [9],

0=−∇⋅[k0th∇T0d,fl]−cpρ0v0⋅∇T0d,fl+Pacd,(11)0=-\nabla\cdot\left[k_0^{\mathrm{th}}\nabla T_0^{d,\mathrm{fl}}\right]-c_p\rho_0\mathbf{v}_0\cdot\nabla T_0^{d,\mathrm{fl}}+P_{ac}^d, \tag*{(11)}
Pacd=−∇⋅[⟨k1th,d∇T1d⟩−⟨ρ1v1d,p⟩+⟨v1d,p⋅τ1d⟩]−cp⟨ρ1v1d,p⟩⋅∇T0d,(12)P_{ac}^d=-\nabla\cdot\left[\left\langle k_1^{\mathrm{th},d}\nabla T_1^d\right\rangle-\left\langle\rho_1\mathbf{v}_1^{d,p}\right\rangle+\left\langle\mathbf{v}_1^{d,p}\cdot\boldsymbol{\tau}_1^d\right\rangle\right]-c_p\left\langle\rho_1\mathbf{v}_1^{d,p}\right\rangle\cdot\nabla T_0^d, \tag*{(12)}

and similarly for T0d,slT_0^{d,\mathrm{sl}} in the solid [9],

0=−∇⋅[k0th∇T0d,sl]+Pacd(13)0=-\nabla\cdot\left[k_0^{\mathrm{th}}\nabla T_0^{d,\mathrm{sl}}\right]+P_{ac}^d \tag*{(13)}
Pacd=−∇⋅⟨k1th,sl∇T1d,sl⟩(14)P_{ac}^d=-\nabla\cdot\left\langle k_1^{\mathrm{th},\mathrm{sl}}\nabla T_1^{d,\mathrm{sl}}\right\rangle \tag*{(14)}

Here, PacdP_{ac}^d is the power density delivered by the acoustic wave through frictional dissipation and energy flux. T0d,flT_0^{d,\mathrm{fl}} and T0d,slT_0^{d,\mathrm{sl}} are connected at the fluid-solid interface by the two effective boundary conditions taking the boundary layers into account analytically: continuity of temperature and of heat flux, applied respectively as a Dirichlet condition on T0d,flT_0^{d,\mathrm{fl}} and a flux condition on n⋅∇T0d,sl\mathbf{n}\cdot\nabla T_0^{d,\mathrm{sl}} [9],

T0fl,d=T0sl,d−T0fl,δ0−12Re⁡{u1⋅∇T1fl,d∗−ktfl(u1⋅n)T1fl,δ0∗},(15)T_0^{\mathrm{fl},d}=T_0^{\mathrm{sl},d}-T_0^{\mathrm{fl},\delta0}-\frac{1}{2}\operatorname{Re}\left\{\mathbf{u}_1\cdot\nabla T_1^{\mathrm{fl},d*}-k_t^{\mathrm{fl}}(\mathbf{u}_1\cdot\mathbf{n})T_1^{\mathrm{fl},\delta0*}\right\}, \tag*{(15)}
k0th,sln⋅∇T0sl,d=k0th,fln⋅∇T0fl,d+k0th,fl∂zT1fl,δ+12Re⁡{ktflk1th,flT1fl,δ∗−2iδt2k0th,fl(u1⋅n)T1fl,δ∗}.(16)k_0^{\mathrm{th},\mathrm{sl}}\mathbf{n}\cdot\nabla T_0^{\mathrm{sl},d}=k_0^{\mathrm{th},\mathrm{fl}}\mathbf{n}\cdot\nabla T_0^{\mathrm{fl},d}+k_0^{\mathrm{th},\mathrm{fl}}\partial_zT_1^{\mathrm{fl},\delta} +\frac{1}{2}\operatorname{Re}\left\{k_t^{\mathrm{fl}}k_1^{\mathrm{th},\mathrm{fl}}T_1^{\mathrm{fl},\delta*}-\frac{2i}{\delta_t^2}k_0^{\mathrm{th},\mathrm{fl}}(\mathbf{u}_1\cdot\mathbf{n})T_1^{\mathrm{fl},\delta*}\right\}. \tag*{(16)}

In summary, the bulk temperature T0dT_0^d is governed by the heat equations (5) and (6) together with the effective boundary conditions (7).

Experimental method.—The experiments were performed using a long straight microchannel of width W=375 μmW=375\ \mu\mathrm{m} and height H=135 μmH=135\ \mu\mathrm{m} in a glass-silicon chip with a piezoelectric transducer glued underneath. The transducer was driven at a frequency of 1.97 MHz1.97\ \mathrm{MHz} at input power Pin=6.1P_{\mathrm{in}}=6.1, 86.886.8, and 182.5 mW182.5\ \mathrm{mW}, resulting in the energy density Eac=27.2±1.1E_{ac}=27.2\pm1.1, 388.7±15.9388.7\pm15.9, and 817.3±33.5 J/m3817.3\pm33.5\ \mathrm{J/m^3}, respectively, as measured from the focusing of 5.05.0-μm\mu\mathrm{m}-diameter particles at 140140 fps using confocal micro-particle image velocimetry (μ\muPIV) at the low PinP_{\mathrm{in}} [27]. At higher PinP_{\mathrm{in}}, EacE_{ac} is estimated using the proportionality Eac∝PinE_{ac}\propto P_{\mathrm{in}}. The confocal μ\muPIV technique only captures the particle motion near the focal plane (channel mid-height), excluding particles near the top and bottom walls influenced by hydrodynamic and acoustic particle-wall interactions, and as a result, EacE_{ac} is measured accurately. The acoustic streaming for each EacE_{ac} was measured at 1010 to 6060 fps by tracking the motion of 0.50.5-μm\mu\mathrm{m}-diameter particles using a defocusing-based 3D particle tracking technique [28, 29, 30]. To avoid the resonance frequency shift due to the temperature rise of the transducer under moderate (86.8 mW86.8\ \mathrm{mW}) and high (182.5 mW182.5\ \mathrm{mW}) PinP_{\mathrm{in}}, each measurement was run for 22 s and repeated 100100 times to improve the statistics, resulting in 78007800-1200012000 recorded frames for each driving condition.

Results and discussion.—The simulation and experimental results shown in Figure 2 reveal the dominant nonlinear behavior of the stationary streaming v0d\mathbf{v}_0^d and temperature T0T_0 in a standard acoustofluidic device. In the linear regime at low Eac<30 J/m3E_{ac}<30\ \mathrm{J/m^3}, v0d\mathbf{v}_0^d is dominated by the boundary-driven streaming entering the model through the slip-velocity condition (4), and the usual four boundary-driven streaming rolls appear, see Figure 2(a). Due to friction in the viscous boundary layers, heat is generated both at the top and bottom of the channel. At the bottom, this heat is removed efficiently because of the high heat conductivity of silicon. At the top, however, the heat is removed less efficiently by the lower heat conductivity of glass, and a steady temperature gradient ∇T0\nabla T_0 is established, which explains the temperature T0dT_0^d seen in Figure 2(f).

Image not extracted

Figure 2. Simulation and experimental results for v0v_0 and T0T_0 in the microchannel. (a)–(d) Vector of v0v_0 and color plot of ∣v0∣\lvert v_0\rvert from 0 (blue) to v0max⁡v_0^{\max} (yellow) at four EacE_{\mathrm{ac}}. For each EacE_{\mathrm{ac}}, the left half are simulations and the right half are experiments. Δv\Delta_v (white bar) is the height where v0yv_{0y} is maximal. (e) Plots of v0max⁡v_0^{\max} and Δv\Delta_v vs. EacE_{\mathrm{ac}} with data from simulations and experiments, showing the transition from boundary- to bulk-driven streaming. The error bars on experimental v0max⁡v_0^{\max} and EacE_{\mathrm{ac}} are within the square markers. The round markers represent the simulations shown in panels (d) and (f)–(i). (f)–(i) Color plot of simulated temperature increase T0T_0 from 0 (black) to T0max⁡T_0^{\max} at four EacE_{\mathrm{ac}}. ΔT\Delta_T (white bar) is the height where T0=12T0max⁡T_0=\frac{1}{2}T_0^{\max}. (j) Plots of simulated T0max⁡T_0^{\max} and ΔT\Delta_T vs. EacE_{\mathrm{ac}} showing a transition from diffusion- to convection-dominated heat transport.

The gradient ∇T0\nabla T_0 created by the acoustic frictional heating results in gradients in ∇ρ0\nabla\rho_0 and ∇k0\nabla k_0, thus inducing a thermoacoustic body force (7) facd\mathbf{f}_{ac}^d [7, 8],

facd≈−14∣p1∣2∇ks,0−14∣v1∣2∇ρ0=−14∣p1∣2(∂ks∂T)T0∇T0−14∣v1∣2(∂ρ∂T)T0∇T0.(17)\begin{aligned} \mathbf{f}_{ac}^d\approx-\frac{1}{4}|p_1|^2\nabla k_{s,0}-\frac{1}{4}|\mathbf{v}_1|^2\nabla\rho_0 \\ &=-\frac{1}{4}|p_1|^2\left(\frac{\partial k_s}{\partial T}\right)_{T_0}\nabla T_0-\frac{1}{4}|\mathbf{v}_1|^2\left(\frac{\partial\rho}{\partial T}\right)_{T_0}\nabla T_0. \tag*{(17)} \end{aligned}

Since ∣p1∣2∝∣v1∣2∝Eac|p_1|^2\propto|\mathbf{v}_1|^2\propto E_{ac} and ∣∇T0∣∝Pacd∝Eac|\nabla T_0|\propto P_{ac}^d\propto E_{ac}, we have ∣fac∣∝Eac2|\mathbf{f}_{ac}|\propto E_{ac}^2, and facd\mathbf{f}_{ac}^d will become important in the bulk at high EacE_{ac} and cause qualitative nonlinear changes of the streaming pattern. According to Eq. (17), facd\mathbf{f}_{ac}^d is pointing toward high temperature at the top and is strongest at the pressure antinodes at the sides [7, 8]. Consequently, facd\mathbf{f}_{ac}^d pushes liquid from the sides up toward the top center, with a back-flow down along vertical center axis, thus creating a streaming pattern that consists of two streaming rolls in each side of the channel. This pattern is seen at the high Eac=5300 J/m3E_{ac}=5300\ \mathrm{J/m^3} in Figure 2(d), where the streaming is completely dominated by the thermoacoustic streaming. The transition from boundary-driven streaming at low EacE_{ac} to bulk-driven streaming at high EacE_{ac} is studied qualitatively in Figure 2(a)–(d) and quantitatively in Figure 2(e). During the transition, the two bottom streaming rolls expand and the two top rolls shrink, see Figure 2(b)–(c) at Eac=380E_{ac}=380 and 800 J/m3800\ \mathrm{J/m^3}, respectively. The bottom rolls expand because they rotate the same way as the two thermoacoustic streaming rolls.

This transition is studied quantitatively in Figure 2(e) by plotting the maximum streaming velocity v0max⁡v_0^{\max} and the vertical distance Δv\Delta_v (thick white line) from the bottom of the channel to the position of the maximum horizontal streaming velocity max⁡(v0y)\max(v_{0y}) toward the center occurs. In the log-log plot (dark blue), the perturbative result v0max⁡∝Eacv_0^{\max} \propto E_{\mathrm{ac}} is valid up to Eac≈1000 J/m3E_{\mathrm{ac}} \approx1000\ \mathrm{J/m^3}, but at higher values v0max⁡v_0^{\max} increases faster. A stronger signal is seen in the log-lin plot (dark red), where the perturbative result Δv∝Eac\Delta_v \propto E_{\mathrm{ac}} only holds for Eac≲30 J/m3E_{\mathrm{ac}} \lesssim30\ \mathrm{J/m^3}, after which point Δv\Delta_v increases with increasing EacE_{\mathrm{ac}}.

As the streaming velocity increases, convection becomes increasingly important for the heat transport [5] and strongly affects the temperature field, see Figure 2(f–Figure 2i) for Eac=380E_{\mathrm{ac}} = 380, 800800, 53005300, and 12,600 J/m312{,}600\ \mathrm{J/m^3}. Convection becomes important at a Péclet number Pe=∣v0∣Hfl/Dth∼1\mathrm{Pe}=\lvert v_0\rvert H_{\mathrm{fl}}/D^{\mathrm{th}}\sim1 corresponding to ∣v0∣∼1 mm/s\lvert v_0\rvert\sim1\ \mathrm{mm/s}, consistent with Figure 2(f–Figure 2j). Qualitatively, we see that for Eac≳800 J/m3E_{\mathrm{ac}}\gtrsim800\ \mathrm{J/m^3}, the two flow rolls pull the temperature profile down along the vertical center axis.

We quantify this effect by the maximum temperature T0max⁡T_0^{\max} and the vertical distance ΔT\Delta_T along the center axis from the bottom edge to the point where T0=12T0max⁡T_0=\frac{1}{2}T_0^{\max}. The thermoacoustic streaming increases the heat transport from the fluid-glass interface to the silicon wafer, thus T0max⁡T_0^{\max} increases less steeply than the perturbative result, T0max⁡∝EacT_0^{\max}\propto E_{\mathrm{ac}}, as seen in the log-log plot (blue) of T0max⁡T_0^{\max} vs. EacE_{\mathrm{ac}} for Eac≳5000 J/m3E_{\mathrm{ac}}\gtrsim5000\ \mathrm{J/m^3} in Figure 2(j). A stronger signal is seen in the log-lin plot (dark red), where the perturbative result ΔT∝Eac0\Delta_T\propto E_{\mathrm{ac}}^0 only holds for Eac≲500 J/m3E_{\mathrm{ac}}\lesssim500\ \mathrm{J/m^3}, after which point ΔT\Delta_T decreases with increasing EacE_{\mathrm{ac}}.

**Conclusion.—**In this Letter we have shown numerically and experimentally that the acoustic streaming in a standard microscale acoustofluidic device is changed qualitatively for moderately high acoustic energy densities Eac≳500 J/m3E_{\mathrm{ac}}\gtrsim500\ \mathrm{J/m^3}. We have explained this effect by a nonperturbative model [9], in which a transition from boundary- to bulk-driven acoustic streaming occurs, as the acoustic body force fac\mathbf{f}_{\mathrm{ac}} begins to dominate the streaming at increased EacE_{\mathrm{ac}} due to the internal heating generated in the viscous boundary layers. We have shown good qualitative and quantitative agreement between our model predictions and experimental data.

Eac≳500 J/m3E_{\mathrm{ac}} \gtrsim500\ \mathrm{J}/\mathrm{m}^{3} can easily be obtained in standard acoustofluidic devices, where Eac≈10−50 J/m3×[Upp/(1 V)]2E_{\mathrm{ac}} \approx10 - 50\ \mathrm{J}/\mathrm{m}^{3} \times[U_{\mathrm{pp}}/(1\ \mathrm{V})]^2 has been reported in the literature, UppU_{\mathrm{pp}} being the applied voltage on the piezoelectric transducer [11, 31, 32, 33]. The physical understanding of how such acoustofluidic devices behave at high EacE_{\mathrm{ac}} is important for the continued development of high throughput devices in particular for biotech applications.

Acknowledgments

We thank R. Barnkob and M. Rossi for providing the software DefocusTracker, defocustracking.com/. WQ was supported by MSCA EF Seal of Excellence IF-2018 from Vinnova, Sweden’s Innovation Agency (Grant No. 2019-04856). HB and JHJ was supported by Independent Research Fund Denmark, Natural Sciences (Grant No. 8021-00310B).

References

References

  1. [1]J. T. Karskin and H. Bruus, Forces acting on a small particle in an acoustical field in a thermoviscous fluid, Phys. Rev. E 92, 043010 (2015).
  2. [2]T. Baasch, A. Pavlic, and J. Dual, Acoustic radiation force acting on a heavy particle in a standing wave can be dominated by the acoustic microstreaming, Phys. Rev. E 100, 061102 (2019).
  3. [3]Lord Rayleigh, On the circulation of air observed in Kundt’s tubes, and on some allied acoustical problems, Philos. Trans. R. Soc. London 175, 1 (1884).
  4. [4]C. Eckart, Vortices and streams caused by sound waves, Phys. Rev. 73, 68 (1948).DOI
  5. [5]G. Chini, Z. M. Malaïche, and T. Dreeben, Large-amplitude acoustic streaming, J. Fluid Mech. 744, 329 (2014).
  6. [6]G. Michel and C. Gissinger, Cooling by baroclinic acoustic streaming, Phys. Rev. Applied 16, L051003 (2021).DOI
  7. [7]J. H. Joergensen and H. Bruus, Theory of pressure acoustics with thermoviscous boundary layers and streaming in elastic cavities, J. Acoust. Soc. Am. 149, 3599 (2021).
  8. [8]W. Qiu, J. H. Joergensen, E. Corato, H. Bruus, and P. Augustsson, Fast microscale acoustic streaming driven by a temperature-gradient-induced nondissipative acoustic body force, Phys. Rev. Lett. 127, 064501 1 (2021).DOI
  9. [9]J. H. Joergensen and H. Bruus, Theory and modeling of nonperturbative effects at high acoustic energy densities in thermoviscous acoustofluids, Phys. Rev. E submitted, 15 pages (2021), https://arxiv.org/abs/2112.10737.
  10. [10]R. Barnkob, P. Augustsson, T. Laurell, and H. Bruus, Acoustic radiation- and streaming-induced microparticle velocities determined by microparticle image velocimetry in an ultrasound symmetry plane, Phys. Rev. E 86, 056307 (2012).pubmed.ncbi.nlm.nih.gov/23214876
  11. [11]P. B. Muller, M. Rossi, A. G. Marin, R. Barnkob, P. Augustsson, T. Laurell, C. J. Kähler, and H. Bruus, Ultrasound-induced acoustopherotic motion of microparticles in three dimensions, Phys. Rev. E 88, 023006 (2013).DOI
  12. [12]P. B. Muller and H. Bruus, Numerical study of thermoviscous effects in ultrasound-induced acoustic streaming in microchannels, Phys. Rev. E 90, 043016 (2014).pubmed.ncbi.nlm.nih.gov/25375602
  13. [13]J. S. Bach and H. Bruus, Theory of pressure acoustics with viscous boundary layers and streaming in curved elastic cavities, J. Acoust. Soc. Am. 144, 766 (2018).DOI
  14. [14]M. A. Solovchuk, T. W. Sheu, M. Thiriet, and W.-L. Lin, On a computational study for investigating acoustic streaming and heating during focused ultrasound ablation of liver tumor, Appl. Therm. Eng. 56, 62 (2013).DOI
  15. [15]M. A. Solovchuk, M. Thiriet, and T. W. Sheu, Computational study of acoustic streaming and heating during acoustic hemostasis, Appl. Therm. Eng. 124, 1112 (2017).DOI
  16. [16]A. Sedaghatkish, M. Rezaeian, H. Heydari, A. M. Ranjbar, and M. Soltani, Acoustic streaming and thermosensitive liposomes for drug delivery into hepatocellular carcinoma tumor adjacent to major hepatic veins; an acoustics–thermal–fluid–mass transport coupling model, Int. J. Therm. Sci. 158, 106540 (2020).DOI
  17. [17]A. Oczelik, D. Ahmed, Y. Xie, N. Nama, Z. Qu, A. A. Nawaz, and T. J. Huang, An acoustofluidic micromixer via bubble inception and cavitation from microchannel sidewalls, Anal. Chem. 86, 5083 (2014).
  18. [18]M. V. Patel, I. A. Nanayakkara, M. G. Simon, and A. P. Lee, Cavity-induced microstreaming for simultaneous on-chip pumping and size-based separation of cells and particles, Lab Chip 14, 3860 (2014).DOI
  19. [19]H. Bachman, C. Chen, J. Rufo, S. Zhao, S. Yang, Z. Tian, N. Nama, P.-H. Huang, and T. J. Huang, An acoustofluidic device for efficient mixing over a wide range of flow rates, Lab Chip 20, 1238 (2020).DOI
  20. [20]C. Zhang, P. Brunet, L. Royon, and X. Guo, Mixing intensification using sound-driven micromixer with sharp edges, Chem. Eng. J. 410, 128252 (2021).DOI
  21. [21]M. Wiklund, R. Green, and M. Ohlin, Acoustofluidics 14: Applications of acoustic streaming in microfluidic devices, Lab Chip 12, 2438 (2012).DOI
  22. [22]D. J. Collins, Z. Ma, and Y. Ai, Highly localized acoustic streaming and size-selective submicrometer particle concentration using high frequency microscale focused acoustic fields, Anal. Chem. 88, 5513 (2016).DOI
  23. [23]D. J. Collins, Z. Ma, J. Han, and Y. Ai, Continuous micro-vortex-based nanoparticle manipulation via focused surface acoustic waves, Lab Chip 17, 91 (2017).pubmed.ncbi.nlm.nih.gov/27883136
  24. [24]J. D. Adams, C. L. Ebbesen, R. Barnkob, A. H. J. Yang, H. T. Soh, and H. Bruus, High-throughput, temperature-controlled microchannel acoustophoresis device made with rapid prototyping, J Micromech Microeng 22, 075017 (2012).DOI
  25. [25]M. Antfolk, C. Magnusson, P. Augustsson, H. Lilja, and T. Laurell, Acoustofluidic, label-free separation and simultaneous concentration of rare tumor cells from white blood cells, Anal. Chem. 87, 9322 (2015).DOI
  26. [26]A. Urbansky, F. Olm, S. Scherding, T. Laurell, and A. Lenshof, Label-free separation of leukocyte subpopulations using high throughput multiplex acoustophoresis, Lab Chip 19, 1406 (2019).
  27. [27]W. Qiu, J. T. Karlsen, H. Bruus, and P. Augustsson, Experimental characterization of acoustic streaming in gradients of density and compressibility, Phys. Rev. Appl. 11, 024018 (2019).DOI
  28. [28]R. Barnkob, C. J. Kähler, and M. Rossi, General defocusing particle tracking, Lab Chip 15, 3556 (2015).
  29. [29]R. Barnkob and M. Rossi, General defocusing particle tracking: fundamentals and uncertainty assessment, Exp. Fluids 61, 110 (2020).
  30. [30]M. R. Rossi and R. Barnkob, A fast and robust algorithm for general defocusing particle tracking, Meas. Sci. Technol. 32, 014001 (2020).DOI
  31. [31]R. Barnkob, P. Augustsson, T. Laurell, and H. Bruus, Measuring the local pressure amplitude in microchannel acoustophoresis, Lab Chip 10, 563 (2010).DOI
  32. [32]P. Augustsson, R. Barnkob, S. T. Wereley, H. Bruus, and T. Laurell, Automated and temperature-controlled micro-PIV measurements enabling long-term-stable microchannel acoustophoresis characterization, Lab Chip 11, 1452 (2011).DOI
  33. [33]R. Barnkob, I. Iranmanesh, M. Wiklund, and H. Bruus, Measuring acoustic energy density in microchannel acoustophoresis using a simple and rapid light-intensity method, Lab Chip 12, 2337 (2012).DOI

Paper details

vlnc:W_z2fuzbjb
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 8

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents