Ultrasonic trapping and collection of airborne particulate matter enabled by multiple acoustic streaming vortices

Over the years, processing and analyzing airborne particulate matter have garnered increasing attention from the fields of environmental science and technology, healthcare systems, energy engineering, and so forth [1–3]. Conventionally, various air filters have been mostly utilized for capturing and removing in-air particulate matter [1, 4–8]. Many researchers have been devoting their efforts in developing and integrating advanced functional materials into air filtration techniques for performance improvement [9–12]. Recently, Zhang et al found that through incorporating an ionic liquid-polymer composite onto a sponge network, extremely high-performance air filtration effect can be achieved [1]. Electrostatic precipitator techniques have been regarded as another type of effective methods for processing airborne particulate matter [13]. In this principle, corona dischargers are normally used to charge incoming aerosols, and these charged aerosols can be collected by electrostatic forces. However, it is well known that corona dischargers can produce massive oxides of nitrogen and ozone, which is harmful to the environment and human health. Furthermore, these electrostatic precipitator systems are bulky, heavy, and complex due to the usage of high voltages [14–17]. For in-lab analysis of airborne particulate matter, it is demonstrated that microfluidic techniques can also provide useful tools for treating and analyzing in-air particulate matter [2], while this microfluidics-based device is not user-friendly for those operators who are not microfluidic professionals.

Acoustic methods have also exhibited their decent capabilities in manipulating airborne particulate matter throughout the years [18–25]. Generally, airborne particulate matter can be trapped and agglomerated at the nodal positions of acoustic fields via the driving of acoustic radiation force, and particulate aggregates can be formed for post-processing [18]. However, these acoustic tools commonly have complicated operating systems [21], which is not suitable for in-lab sampling of airborne particulate matter. Moreover, airborne particulate matter can only be trapped at the specific positions of acoustic fields in those devices and hardly be directed to an interface for the convenience of downstream analysis due to the nature of acoustic radiation force [18]. Therefore, simple, compact, and effective acoustic manipulation tools are still scarce for manipulation and collection of airborne particulate matter for laboratory applications.

In this work, we develop a facile ultrasonic trapping and collection method towards airborne particulate matter by leveraging a simple and compact ultrasonic device. The ultrasonic device is composed of a Langevin transducer and a radiation plate bonded on the radiation surface of the transducer. After turning on the device, circular standing flexural waves (CSFWs) can be generated in the radiation plate, and an acoustic field in the air gap between the radiation plate and a sampling plate can be excited. Accompanying the acoustic field, an acoustic streaming field in the air gap can also be induced resulting from the nonlinear effect of acoustic field. Numerical simulations show that multiple acoustic streaming vortices account for the ring-shaped aggregation patterns of particulate matter on the sampling plate. Experiments indicate that the collection effects of airborne particulate matter are influenced by the air gap thickness and radius, sonication time, driving voltage, and the angle between the radiation plate and the sampling plate.

Figures 1(a) and (b) show a principle schematic and a photo of the ultrasonic airborne particulate matter sampler, respectively. The ultrasonic device is constructed by axis-symmetrically bonding an aluminum radiation plate onto the surface of a commercial Langevin transducer (HNC-4AH-2560, Hainertec Co., Ltd) with an epoxy layer (Gleihow New Materials Co., Ltd, China). The diameter and height of the transducer are 30 and 35 mm, respectively. The diameter and thickness of the radiation plate are 130 and 1 mm, respectively. The polyvinyl chloride (PVC) sampling plate is put right below and parallel to the radiation plate. The diameter and thickness of the sampling plate are 130 and 0.5 mm, respectively. In the experiments, the whole device system is put in an enclosed chamber (250 mm × 250 mm × 200 mm), and the movement of the sampling plate is controlled by a manual three-dimensional (3D) moving stage (LD125-LM-2, Shengling Precise Machinery Co., Ltd). The ultrasonic transducer is driven by a set of electrical system containing a signal generator, a power amplifier, and an oscilloscope. The airborne particulate matter is produced by smoldering of joss sticks, which mainly contain sugar, nicotine, starch, amino acid, polyphenols, tar, and protein. It is measured that smoldering of 1.5 g joss stick can generate smog of about 50 mg through weighing the mass difference of a piece of filter cotton for smog absorption before and after smoldering. The mass of collected particulate matter on the sampling plate is measured by an electronic balance (ME204E, METTLERTOLEDO).

Zoom In Zoom Out Reset image size

As shown in figure 1(a), after turning on the ultrasonic device in its resonant mode, CSFWs can be generated in the radiation plate. Under the actuation of the CSFWs in the radiation plate and the wave reflection effect enabled by the sampling plate, a standing acoustic field in the air gap between the radiation plate and the sampling plate can be formed [26]. Meanwhile, due to the non-uniformity of acoustic field, an acoustic streaming field in the air gap can also be induced with the feature of multiple acoustic streaming vortices. Thanks to the axial symmetry of the whole device system, the generated multiple acoustic streaming vortices should also be symmetrical about the central axis [27]. As shown in figure 1(a), among the multiple acoustic streaming vortices, the adjacent downward acoustic streaming flows can simultaneously flush airborne particulate matter down to the sampling plate. With the increment of sonication time, multiple acoustic streaming vortices in the air gap can facilitate the formation of aggregations of particulate matter as multiple ring-shaped patterns distributed on the sampling plate.

Before experimentation, it was hard to directly determine an optimal air gap thickness for achieving the best particulate matter collection effect. First, we set the air gap thickness to the wavelength of the acoustic field λ for obtaining a standard standing wave acoustic field. In this setup, since the resonant frequency of the ultrasonic device is measured to be 61.7 kHz, λ is calculated to be 5.6 mm [26]. Figure 2(a) shows a photo of collected particulate matter on the sampling plate when the air gap thickness, sonication time, and driving voltage are λ, 5 min, and 40 V_pp, respectively. It is seen that the collected particulate matter is not uniformly distributed on the sampling plate, and a series of concentric ring-shaped agglomeration patterns of particulate matter are formed. It is counted that the number of the ring-shaped patterns is 10. Figure 2(b) shows the measured mass of collected particulate matter on the sampling plate versus air gap thickness under different sonication time when the driving voltage is 40 V_pp. It is seen that when the air gap thickness is larger than 1 mm, the collection effect becomes obvious. With the increment of air gap thickness, there are several peak values for the mass of collected particulate matter. When the air gap thicknesses are 0.5λ, λ, and 1.5λ, the mass of collected particulate matter reaches the local maximum values, respectively. This is because when the air gap thickness equals integral multiples of the half wavelength of acoustic field, the acoustic field in the air gap is in resonance, under which the intensity of acoustic field reaches locally strongest [26, 28, 29]. Therefore, the consequent acoustic streaming field resulting from the acoustic field is also locally strongest, and the peak values of mass of collected particulate matter occur at these state points. We define the sampling efficiency as e = M/(S·P·T), where M is the mass of collected particulate matter, S is the area of sampling plate, P is the input power of ultrasonic device, and T is the sonication time. The sampling efficiency e is calculated to be 9.6 mg (m² min W)⁻¹ when the driving voltage, sonication time, and air gap thickness are 40 V_pp, 5 min, and λ, respectively, which is about 26 times that obtained by the natural sedimentation method under the same experimental condition. Figures 2(c) and (d) show photos of collected particulate matter on the sampling plates when the air gap thicknesses are 0.5λ and 1.5λ, respectively; the sonication time and driving voltage are 5 min and 40 V_pp, respectively. It is seen that a series of concentric ring-shaped agglomeration patterns of particulate matter are also formed under these air gap thicknesses, while the numbers of ring-shaped patterns are less than that under the air gap thickness of λ, the reason for which will be analyzed in the following section.

Zoom In Zoom Out Reset image size

To unveil the physical mechanism of the formation of ring-shaped patterns of particulate matter enabled by actuating the ultrasonic device, we simulate acoustic fields and acoustic streaming fields in the air gap using the abstracted physical model of the device system based on the finite element method (FEM). Figure 3(a) shows a computational model of the device system. Since the device system in figure 1(a) is axis-symmetrical, we establish the physical model as a two-dimensional (2D) axis-symmetrical structure. Replacing an entire Langevin transducer, we apply a partial actuation to the upper surface of the radiation plate, and the vibration velocity amplitude of the actuation part is experimentally measured to be 0.295 m s⁻¹ under the driving voltage of 61.7 kHz and 40 V_pp using a Doppler Vibrometer (PSV-300F, Polytec, Germany). Boundary conditions for computing acoustic fields are also shown in figure 3(a), in which a set of perfectly matched layers (PMLs) is adopted to mimic the outer air domains in the model. The whole simulation process comprises two steps. In the first step, the acoustic field is computed in the frequency domain by applying a given vibration velocity amplitude (0.295 m s⁻¹) to the actuation part. The governing equation for computing the acoustic field is:

$$\begin{equation}{\nabla ^2}p\, + \,{\left( {\frac{\omega }{{{c_0}}}} \right)^2}p\, = \,0\end{equation} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

where p is the acoustic pressure, c₀ is the sound speed, and ω is the angular frequency of acoustic field [30]. In the second step, with the simulation result of acoustic field, the steady acoustic streaming field in the air gap is computed on the basis of the Reynolds stress method [31–35]. The governing equations for computing the acoustic streaming field are:

$$\begin{equation}\nabla \cdot {{\boldsymbol{{u}}}_{\mathbf{1}}} = 0\end{equation} \tag{ 2 }$$

$$\begin{equation}{\rho _0}\left( {{{\boldsymbol{{u}}}_{\mathbf{1}}} \cdot \nabla } \right){{\boldsymbol{{u}}}_{\mathbf{1}}} = \nabla \cdot \left\{ { - {p_1}{\boldsymbol{{I}}} + \mu \left[ {\nabla {{\boldsymbol{{u}}}_{\mathbf{1}}} + {{\left( {\nabla {{\boldsymbol{{u}}}_{\mathbf{1}}}} \right)}^{\text{T}}}} \right]} \right\} + {{\boldsymbol{{F}}}_{\mathbf{R}}}\end{equation} \tag{ 3 }$$

$$\begin{equation}{{\boldsymbol{{F}}}_{\mathbf{R}}} = - {\rho _0}\langle \left( {{\boldsymbol{{u}}} \cdot \nabla } \right){\boldsymbol{{u}}} + {\boldsymbol{{u}}}\left( {\nabla \cdot {\boldsymbol{{u}}}} \right)\rangle \end{equation} \tag{ 4 }$$

where u₁ is the acoustic streaming velocity, ρ₀ is the density of air, p₁ is the pressure of air, I is the identity matrix, μ is the dynamic viscosity of air, F_R is the body force generated from the acoustic field (< > denotes the time average over a full oscillation time period), and u is the acoustic velocity of acoustic field. Figure 3(b) shows the boundary conditions for computing acoustic streaming fields.

Figure 3(c) shows a mesh plot for the FEM computational model, in which square elements with the side length of 17.6 μm (twice the acoustic boundary layer thickness and 0.317% of the acoustic field wavelength at 61.7 kHz) generated from the Mapping function are adopted for the air domain (except the PMLs). For the rest domains, we generate ten layers of mesh along the horizontal and vertical directions using the Mapping function, respectively. Unless otherwise specified, the material parameters of air and the solids listed in tables 1 and 2 are used in the simulations. All the simulations are performed by the commercial FEM software COMSOL Multiphysics 5.6 (Burlington, MA, USA).

As is well known that when an object is in an acoustic field, it may experience the acoustic radiation force [37–41] and the acoustic streaming-induced drag force [27, 33, 42]. The acoustic radiation force on a single airborne particulate matter in the acoustic field is:

$$\begin{align}{{\boldsymbol{{F}}}_{{\mathbf{rad}}}} & = - \nabla \left\{ \frac{4}{3}\pi {R_{\text{p}}}^3\left[ \left( { - \frac{{3{\rho _{\text{p}}} - 3{\rho _0}}}{{2{\rho _{\text{p}}} + {\rho _0}}}} \right) \cdot \frac{1}{2}{\rho _0}\langle {{\boldsymbol{{u}}}^2}\rangle \right.\right.\nonumber\\ & \qquad \qquad \left.\left. + \left( {1 - \frac{{{\rho _0}{c_0}^2}}{{{\rho _{\text{p}}}{c_{\text{p}}}^2}}} \right) \cdot \frac{1}{2}\frac{{\langle {p^2}\rangle }}{{{\rho _0}{c_0}^2}} \right] \right\}\end{align} \tag{ 5 }$$

where R_p is the radius of the particulate matter, ρ_p is the density of the particulate matter, and c_p is the sound speed in the particulate matter [37]. The acoustic streaming-induced drag force on the single airborne particulate matter is:

$$\begin{equation}{{\boldsymbol{{F}}}_{{\mathbf{drag}}}} = 6\pi \mu {R_{\text{p}}}\left( {{{\boldsymbol{{u}}}_{\mathbf{1}}} - {{\boldsymbol{{u}}}_{\mathbf{p}}}} \right)\end{equation} \tag{ 6 }$$

where u_p is the motional velocity of the particulate matter.

Figure 4(a) shows the simulated wave field in the radiation plate when the driving frequency is 61.7 kHz, from which it is obviously seen that there exist CSFWs in the radiation plate. Figure 4(b) shows the simulated acoustic fields in the air gaps when the air gap thicknesses are 0.5λ, λ, and 1.5λ, respectively. It is seen that there are one, two, and three nodal planes of acoustic field in the air gaps when the air gap thicknesses are 0.5λ, λ, and 1.5λ, respectively, and the maximum value of acoustic pressure decreases with the increase of air gap thickness. With the simulation result of acoustic field, we can obtain the acoustic radiation force field for a specific kind of particle using equation (5). Here, we assume the particle radius to be 140 nm [36, 43–45] and use particle parameters listed in table 2 for computation. Figure 5 shows the simulated acoustic radiation force fields in the air gaps for 280 nm airborne particulate matter when the air gap thicknesses are 0.5λ, λ, and 1.5λ, respectively. It is seen that the directions of acoustic radiation force point to the nodal planes (figure 4(b)), and the magnitude of acoustic radiation force around nodal planes decreases with the increase of air gap thickness.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 6(a) shows the simulated acoustic streaming field in the air gap when the air gap thickness is λ. It is seen that there are multiple acoustic streaming vortices existing in the air gap. Using equation (6) and the simulated acoustic streaming field (figure 6(a)) and assuming the particle velocity u_p to be zero, we can calculate the acoustic streaming-induced drag force on still particulate matter, and the calculation result along the central axis is shown in figure 6(b). According to this calculation result, it is seen that the average value of acoustic streaming-induced drag force along the central axis is about 2 pN, while the maximum value of acoustic radiation force (figure 5(b)) along the central axis is about 6 fN. Thus the acoustic radiation force is negligible compared with the acoustic streaming-induced drag force, which can also be supported by the comparison results in any other positions in the air gap. Therefore, the effect of acoustic streaming is dominant for manipulating airborne particulate matter in this device. Using the simulation result of acoustic streaming field (figure 6(a)), we predict and generate a schematic for the ultrasonic trapping and collection of airborne particulate matter by multiple acoustic streaming vortices, and the schematic is shown in figure 6(c). As shown in figure 6(c), among the multiple acoustic streaming vortices, the adjacent downward acoustic streaming flows can drag and flush particulate matter down to the sampling plate. With the increase of sonication time, the deposited particulate matter on the sampling plate can accumulate and turn into ring-shaped agglomeration patterns along the radial direction of the plate. Thus, the number of ring-shaped patterns should be determined by the number of the adjacent downward acoustic streaming group. In figure 6(a), the number of the adjacent downward acoustic streaming group is 10, so the number of ring-shaped patterns in figure 2(a) is also 10. To validate our prediction, we simulate the particle trajectories in the simulated acoustic streaming field (figure 6(a)) using the Particle Tracing module in COMSOL Multiphysics software, and the result is shown in figure 6(d). It is apparently seen that the adjacent downward acoustic streaming flows enable the deposition of airborne particulate matter on the sampling plate and further facilitate the ring-pattern formation.

Zoom In Zoom Out Reset image size

We also simulate the acoustic streaming fields in the air gaps when the air gap thicknesses are 0.5λ and 1.5λ, and the results are shown in figures 7(a) and (b), respectively. When the air gap thickness is 0.5λ, the ring-shaped patterns are sparse on the sampling plate (figure 2(c)), which is because the acoustic streaming vortices in the air gap (figure 7(a)) are fewer and less regular than those when the air gap thickness is λ (figure 6(a)). When the air gap thickness is 1.5λ, the number of profiles of ring-shaped patterns on the sampling plate (figure 2(d)) is less than that when the air gap thickness is λ (figure 2(a)), which is because the acoustic streaming vortices in the air gap (figure 7(b)) are also fewer than those when the air gap thickness is λ (figure 6(a)).

Zoom In Zoom Out Reset image size

Since the particle trapping and sampling is realized based on the multiple acoustic streaming vortices in the air gap, more specifically, the adjacent downward acoustic streaming flows, it is desired that the acoustic streaming field could be optimized for enhancing the particle collection capability. Here, we simulate the acoustic streaming fields and obtain the corresponding numbers of adjacent downward acoustic streaming groups under different air gap thicknesses and widths (i.e. varying the radius of radiation plate), and the results are shown in figures 7(c) and (d), respectively. From figure 7(c), it is seen that under the constant radius of radiation plate, the number of adjacent downward acoustic streaming groups reaches the maximum when the air gap thickness is λ. Since the intensity of acoustic streaming under this air gap thickness is also strong, the corresponding mass of collected particulate matter on the sampling plate is maximal (figure 2(b)). From figure 7(d), it is seen that the number of adjacent downward acoustic streaming groups increases with the increase of radiation plate radius overall. This implies that the mass of collected particulate matter on the sampling plate should increase with the increase of radiation plate radius, which will be verified in section 4.

Besides the geometric sizes (e.g. radius and height) of the constructed acoustic field, the shape of acoustic field could also be optimized by varying the topological structures of the radiation and sampling plates for achieving more acoustic streaming vortices [46], which could further improve the sampling efficiency of our device.

For the following results, unless otherwise specified, the experimental conditions are as follow: the air gap thickness is λ; the sonication time is 5 min; the driving frequency is 61.7 kHz; the driving voltage is 40 V_pp.

Figure 8(a) shows the measured mass of collected particulate matter versus sonication time under different air gap thicknesses. It is seen that with the increase of sonication time, the mass of collected particulate matter rises rapidly in the first 10 min. After 30 min, the mass of collected particulate matter changes little, which is because most of the airborne particulate matter in the chamber has been settled on the sampling plate or on the bottom of the chamber after 30 min of sonication. Due to the maximum number of the adjacent downward acoustic streaming groups when the air gap thickness is λ, the corresponding mass of collected particulate matter is always maximal.

Zoom In Zoom Out Reset image size

Figure 8(b) shows the measured mass of collected particulate matter versus driving voltage under different air gap thicknesses when the sonication time is kept as 10 min. For the safety purpose, the maximum driving voltage used in the experiments is 80 V_pp. It is seen that with the increase of driving voltage, the mass of collected particulate matter also rises. This is because when applying higher driving voltage to the ultrasonic device, stronger acoustic field and stronger acoustic streaming field can be achieved simultaneously, under which more airborne particulate matter are prone to be trapped and collected onto the sampling plate. We also measure the power consumption of our device under different driving voltages, and the results are listed in table 3. It is seen that the power consumption of our ultrasonic sampler does not exceed 15 W even at a high driving voltage (80 V_pp), which is quite competitive compared with the electrostatic precipitator methods [13–16].

Figure 8(c) shows the measured mass of collected particulate matter versus air gap thickness under different diameters of sampling plates when the sonication time is kept as 10 min. The experimental diameters of the sampling plates are 60, 130, and 160 mm, respectively. It is seen that the mass of collected particulate matter increases with the increment of the diameter of sampling plate. This is because larger sampling plate can generate acoustic field with larger area and consequently create more acoustic streaming vortices in the air gap for trapping and collecting more particulate matter onto the sampling plate (figure 7(d)) [27]. It is also seen from figure 8(c) that when the diameter of sampling plate is 160 mm, the mass of collected particulate matter under 1.5λ becomes larger than that under λ. This is because when the diameter of sampling plate is too large, the particulate matter per unit volume existing in the air gap may become less. In this scenario, increasing the air gap thickness can increase the particulate matter existing in the air gap, so the collected particulate matter on the sampling plate becomes more. It is also calculated that when the diameter of sampling plate is 130 mm, the sampling efficiency e is the highest.

To exclude the influence of non-uniformity of distribution of airborne particulate matter in the experimental chamber on the sampling effect, a set of contrastive experiments in which joss stick smoldered at different places in the experimental chamber were carried out. In two separate experiments, one joss stick smoldered at the left 100 mm of the chamber center and another smoldered at the right 100 mm of the chamber center with the same height, and the sampling results are shown in figures 9(a) and (b), respectively. It is seen that the distributions of collected particulate matter on the sampling plates are almost the same, which indicates that the non-uniformity of distribution of airborne particulate matter in the experimental chamber has little effect on the sampling performance.

Zoom In Zoom Out Reset image size

To investigate the effect of the angle between the radiation plate and the sampling plate on the sampling performance, another set of contrastive experiments in which the angles between the radiation plate and the sampling plate are 0° and 5° were conducted, respectively. Figures 9(c) and (d) show the distributions of collected particulate matter on the sampling plates when the angles between the radiation plate and the sampling plate are 0° and 5°, respectively. It is seen that the ring-shaped patterns in figure 9(d) are less regularly deposited on the sampling plate than those in figure 9(c). The reason for this phenomenon is that when the sampling plate is not completely parallel to the radiation plate, the symmetry of spatial distribution of multiple acoustic streaming vortices in the air gap is broken, which causes the acoustic streaming-induced ring-shaped patterns on the sampling plate more irregular. To validate this assumption, we simulate the 2D acoustic streaming field in the air gap when the angle between the radiation plate and the sampling plate is 5°, and the result is shown in figure 9(e). It is seen that consistent with our assumption, the acoustic streaming vortices asymmetrically distribute in the air gap, which can further result in the asymmetrical distributions of the ring-shaped patterns on the sampling plate (figure 9(d)).

Compared with other conventional airborne particulate matter filtration and sampling techniques, our ultrasonic sampler has some unique merits. For example, due to the sampling in this setup is conducted in a contactless manner, clogging and secondary pollution issues can be naturally avoided and convenience of post treatment can be ensured since the sampling plate can be used in a disposable manner. Moreover, different from traditional electrostatic methods, airborne particulate matter can experience little damage and modification under the effect of acoustic streaming, which is also beneficial to sample post processing and analysis.

There is still significant room for improvement when attempting to adapt this technique for wider application scenarios. First, the Langevin transducers used in this work inherently make the device bulky. To achieve the device miniaturization, high-performance single piezoelectric elements could be employed for substituting the Langevin transducers to make the device more compact [31, 47]. Furthermore, the driving circuit for the ultrasonic device can be integrated into the whole device system to enable a flexible and portable ultrasonic sampler.

Our ultrasonic sampler can also accommodate other advanced air filtration techniques. For example, through pre-functionalizing the sample plate or pet-setting functional coatings, materials, or structures [1, 4, 11, 12] at specific positions on the sampling plate, the sampling capability of the upgraded ultrasonic sampler could be dramatically improved, which would further make this kind of airborne particulate matter processing tools more powerful.

In this article, we developed and demonstrated a facile method of trapping and collecting airborne particulate matter via multiple acoustic streaming vortices, which are generated by the CSFWs in a compact and simple ultrasonic device. Through the driving of multiple acoustic streaming vortices, airborne particulate matter can be trapped and collected onto the sampling plate to form a series of concentric ring-shaped patterns distributed along the radial direction. By changing the thickness of the air gap formed by the radiation plate and the sampling plate, the sampling effect can be tuned owing to the diversified acoustic streaming fields in the air gap. The experimental results show that the sampling capability of the device can be adjusted by the air gap thickness and width, sonication time, driving voltage and the angle between the radiation plate and the sampling plate. Due to the simplicity, compactness, and low-cost of the device and no need of large-scale and complex equipment, the presented trapping and sampling method aiming at airborne particulate matter in this work could provide an effective and versatile tool for dexterously collecting in-air particulate matter for downstream analysis such as air pollutant filtration and detection.

This work is supported by the following funding organizations in China: the Natural Science Foundation of Jiangsu Province (Grant No. BK20201038), the Scientific Research Fund of High Level Talents in Nanjing Institute of Technology (Grant No. YKJ201949), and PAPD.

All data that support the findings of this study are included within the article (and any supplementary files).