Wave field and resulting SAW chip requirements
The acoustic wave field, i.e. the distribution of the out of plane amplitude u3 and the phase (i.e.), determines the physical conditions at the fluid supply position, at the atomization zone and in the interconnecting fluid film. As the basic features of any standing SAW field are comparable, they are briefly discussed here, based on Laser-Doppler vibrometry measurements of the devices used in this study. Fig. 3 shows the measured maximum amplitude distribution of the Rayleigh-type sSAW (surface-normal (û3) component distribution only) excited between the two interdigital transducers (IDTs) on the SAW chip. In supplementary video 1, the momentary vertical displacement is shown.
Here, the nodes and antinodes of the standing SAW field caused by interference of counter-propagating SAWs are clearly identifiable. Furthermore, diffraction, caused by the finite IDT aperture and the anisotropic substrate properties, was found to result in three identifiable regions: Region 1, i.e. the center of the acoustic beam, is characterized by a high SAW displacement amplitude and amplitude variation due to diffraction, leading to an amplitude maximum. Depending on the IDT configuration used, several amplitude maxima across the beam aperture are possible (Rezk et al., 2012b). Region 2, i.e. the boundary of the acoustic beam, is determined by low SAW amplitude and an amplitude increase (positive gradient) in the direction towards region 1. Outside of these regions, a region 3 exists on the chip surface, where no measurable substrate displacement occurs. The precise positions of the boundaries between these regions are subject of ongoing investigation, though are ultimately determined by diffractive effects within each IDT. Regardless, every conventional IDT configuration will excite displacement fields qualitatively similar to that in Fig. 3. The diffraction and, therefore, the dimensions of region 1 and 2 strongly depend on the substrate and IDT material, the IDT aperture, the SAW wavelength, the number of IDT finger pairs, and the orientation of the IDT with respect to the crystal orientation of the substrate (Szabo & Slobodnik, 1973; Holm et al., 1996). Special IDTs including focused, slanted-finger or chirped designs can influence the shape of the wave field; though will produce comparable diffraction and boundary features. When devices with more than one IDT are applied, the wave fields of the individual IDTs are superimposed.
In SAW fluid atomization setups, the fluid source is conventionally placed in the center of the acoustic beam (i.e. in region 1 in Fig. 3 – see e.g. (Chono et al., 2004; Ju et al., 2008; Qi & Yeo, 2008)). This may be attributed to the fact that fluid atomization requires high amplitudes to break up individual droplets from the fluid film. However, secondary effects such as fluid accumulation on the chip surface, Eckart streaming and jetting can arise if a „fluid volume“, i.e. a fluid geometry much larger than the acoustic wavelength, is present in region 1, especially if the fluid flow supply rate is not controlled. Additionally, any object placed in region 1, such as a tissue or a capillary used for fluid supply, can interact with the SAW in an undesired manner. Here, the placement of the object with respect to the wave field can then result in the appearance of wave scattering and interference at the object boundaries, high local mechanical stress and – if the material absorbs larger portions of acoustic energy – high local thermal stress. It is therefore impractical to use certain materials in the center of the acoustic beam, as local heat and pressures can lead to deleterious effects including a change in material properties, movement, and detachment or even melting/disintegration. The practical applicability of systems where the fluid is supplied in the center of the SAW propagation path (region 1), as has been the case with virtually every SAW atomization study to date, is thus limited to cases where the materials and designs can cope with these continued stresses or the atomization is limited in duration.
Summarizing the wave field properties, the underlying physics of acoustowetting known so far and our own observations, we suggest an ideal, continuously driven SAW atomization chip to fulfil several criteria:
An on-chip fluid supply (for high accuracy, reproducibility and mass-manufacturability)
Positioning of the fluid supply at the boundary of the acoustic path (for spatial separation of fluid supply position and atomization position, minimized interaction of the means of fluid supply with the SAW and continuous atomization off a fluid film),
Accuracy of the fluid supply geometry and placement well below the SAW wavelength (for reproducible SAW-fluid interaction),
Tailored design of the IDTs and the resulting SAW field to fulfil the fluid supply and atomization needs (for optimal utilization of the SAW power and minimized secondary effects, including device heating, Eckart streaming and jetting), and
Biological and chemical compatibility of all implemented materials to the used fluid solutions.
As a further development of the capillary slit fluid supply (Kurosawa et al., 1995a; Kurosawa et al., 1995b; Soluch & Wrobel, 2003; Soluch & Wrobel, 2006), a fluid supply via SU-8 microchannels placed at the boundary of the acoustic propagation path (region 2) has recently been demonstrated (Winkler et al., 2015). Due to the use of photolithography for the channel structuring, the position of the fluid meniscus as well as its height and width can be tailored with sub-micron resolution, well below the SAW wavelength. When fluid supply and the SAW field are appropriately positioned, the SAW interacts with the fluid directly at the channel outlet, forming a fluid film by longitudinal wave resonance at this position. Thereby, the boundary conditions of height and width of the channel outlet are expected to contribute to fluid film formation and to determine the stable longitudinal wave resonance condition in the formed fluid film under SAW influence.
Furthermore, the microchannel structures can be designed in a way that ensures a sufficient mechanical stability of the channel walls and at the same time minimizes the effective polymer cross section in the acoustic beam boundary to minimize the SAW-SU-8 interaction and associated acoustic energy uptake. Due to these reasons, we employed an SU-8 microchannel fluid supply in this study.
Aerosol generation and crucial process parameters
The behavior of the compact SAW aerosol generator was investigated for different SU-8 channel placement conditions, model fluids, SAW power levels and fluid flow rates. Additionally, the turn-on/−off behavior was studied. The observations made are summarized here.
Subsequently after SAW excitation and concurrent fluid supply, SAW-fluid interaction and associated acoustowetting lead to the extension of a fluid film from the microchannel outlet in region 2 towards the atomization zone in region 1 (as in Fig. 3), where the SAW has a sufficiently high amplitude to eject droplets out of the fluid film. Effects due to gravity are negligible in describing the behavior of this small amount of fluid on the chip surface. Figure 4a shows the aerosol generator during the atomization of ethanol with a flow rate of approx. 140 μl/min. There is no change in the aerosol plume geometry or the atomization behavior, even when the setup is tilted or turned upside down (supplementary video 2).
Experiments with different fluids (Fig. 4b) show a limitation of the maximum realizable flow rate and the aerosol beam height by the fluid properties for otherwise constant experimental boundary conditions (supplementary video 3). While the exact physics of the aerosol generation remain unclear, the dynamic viscosity, surface tension, density and electric conductivity of the fluid as well as the contact angle to the substrate may be responsible, as they may influence the acoustowetting, the longitudinal wave velocity/propagation angle/attenuation and the droplet ejection mechanism.
In general, the observation of the atomization zone with high resolution renders visible several features of the acoustofluidic interaction (Figs. 5 and 6, supplementary video 4-6): The aerosol plume emanates from the acoustically stabilized fluid film at the atomization zone, i.e. a wave field region with sufficiently high SAW amplitude. Due to the finite camera exposure length, the aerosol and the fluid film, both of which are driven by perturbations with timescales in the nanoseconds-range, can only be recorded with blurry boundaries. In addition, a film region with a topography resembling the local sSAW amplitude distribution is visible, whereby a separation of individual fluidic stripes can be measured corresponding to λSAW/2 or 45 μm for the 90 μm wavelength devices used here (compare to wave field in Fig. 3). This film shaping is driven by a balance between acoustic radiation pressure and capillary stress (Manor et al., 2015; Scortesse et al., 2002; Biwersi et al., 2000) in a region with reduced fluid volume.
In order to evaluate the influence of the channel outlet position on the atomization process, the distance between the outlet and the aperture boundary was varied from approx. 100 to 650 μm (Fig. 5 and supplementary video 4). Atomization was carried out at a constant fluid flow rate of 100 μl/min and a SAW power of 2 × 3 W. When the channel outlet is positioned close to the beam boundary (e.g. position A), constant and highly stable aerosol generation can be achieved. However, when the distance to the aperture center is too small, the high SAW amplitude may lead to damage the microchannel material due to local acousto-viscous heating. With increasing distance, the length of the fluid film interconnecting fluid supply and atomization zone increases, while the location of the atomization zone stays more or less constant. With increasing distance, atomization increasingly turns unstable, as the local wetting conditions and capillary stresses become more important to defining the fluid shape (e.g. position B). If the distance is increased further, atomization turns discontinuously (e.g. position C): The supplied fluid accumulates in form of a fluid volume at the microchannel outlet until its contact line reaches the acoustic beam boundary. Then, in short time, the droplet is thinned by acoustowetting as well as fluid jetting, and atomization occurs until the fluid feed is interrupted by retraction of the parental droplet. This dependence of the atomization behavior on the channel outlet position leads to an optimization problem in the device design. To achieve a stable and reproducible atomization, further investigations on the influence of the SAW power, the flowrate and the turn-on/off behavior were carried out in a setup with a reduced separation of 100 μm from channel outlet to aperture center (position A).
To investigate the influence of the SAW power, the flow rate was maintained constant at 100 μl/min while the power was varied. Optical micrographs of the SAW chip during atomization are shown in Fig. 6 and supplementary video 5. These demonstrate that the fluid film extension significantly depends on the applied SAW power. Interestingly, the fluid-covered area decreases in size with increasing SAW power. This can be explained taking into account the local SAW amplitude in respect to the channel outlet position. Higher SAW power leads to an overall increase in SAW amplitude in the acoustic beam. As a certain amplitude threshold has to be reached in order to start atomization, the position of that threshold amplitude and, thus, the atomization zone moves closer to the channel outlet for increased SAW power. As the fluid film properties depend on the local conditions at the atomization zone, also the aerosol properties including the droplet size are influenced by the SAW power (Collins et al., 2012; Bennes et al., 2005). Additionally, the wave field regions not covered by a fluid film do not support fluid atomization and the excess SAW power may promote parasitic heating.
Comparable behavior of the fluid film is observed, when the fluid flow rate is changed at a given SAW power. For decreasing fluid flow rate, the fluid covers a smaller area and the atomization zone moves closer to the channel outlet. The changed film conditions for different flow rate may be an explanation for the change of droplet size distribution with flow rate as reported previously (Winkler et al., 2015). Macroscopically, an increase of the fluid flow rate leads to a higher optical density of the aerosol. Figure 7 shows the aerosol beam for a water flow rate of 0.1 to 1 ml/min. We highlight that SAW fluid atomization with such a high a flow rate has not been previously reported. The maximum flow rate for a given system, however, will ultimately depend on the precise dimensions of the channel and the fluid supply position.
We observe that fluid flow rate and SAW power can be adjusted separately while maintaining stable atomization. Figure 8 shows the conceptual atomization regimes determined for the device used in this study. We find that limitations arise due to the applied power and flow rate; low flow rates will render the atomization process discontinuous as the atomization rate exceeds the rate of fluid supplied to the chip. For moderate fluid flow rates, the amount of fluid on the chip surface fulfils the generation of a fluid film, and atomization occurs continuously if sufficient power is applied. At drastically increased flowrates for a given power, however, the rate of fluid supplied to the chip surface can exceed the atomization rate, leading to the stop of atomization and fluid accumulation, with Eckart streaming and capillary wave excitation in an expanding drop rather than film formation.
Very high SAW power may lead to local temperature increase and device damage, especially if the atomization is carried out in the discontinuous regime. Fluid coverage along the complete width of the acoustic beam (2 × 1.5 W in Fig. 6) can be seen as optimal, as otherwise portions of the wave field are not involved in the droplet generation and the unused SAW power accounts e.g. for device heating or parasitic wave mode excitation.
Droplet measurements in Fig. 9 were performed for the three representative points (I-III) in Fig. 8 were carried out to quantify the relationship between an increasing atomization rate and its effect on the resultant droplet size distributions. In general, a multimodal distribution with individual peaks following a logarithmic normal distribution was observed, where the number and fraction of the peaks vary with atomization rate. The standard deviation (Var(x)1/2) of an individual peak amounts approx. 40…75% of its arithmetic mean E(x).
In each measurement, a fraction of very small droplets (Ø < 5 μm) was observed, comparable to those shown in previous studies (Winkler et al., 2016a; Winkler et al., 2015; Collins et al., 2012). Although no precise parameters could be extracted due to the limited measurement range of the diffractometer (Ømin = 1 μm), the droplet size in this fraction generally decreases with increasing atomization rate. The origin of this droplet fraction is unclear. However, for an atomization with the device used in this study, the bulk of the atomized volume is contained within droplets with a diameter of 5 to 40 μm with x50,3 values of 8.6 μm, 9.2 μm and 11.9 μm for the example cases I, II and III, respectively. Interestingly, two individual peaks were observed in this size range in some of the experiments, e.g. in case III. Individual peaks in the droplet size distribution can be an indicator of different underlying physical mechanisms. We assume, that this additional peak with E(x) = 16…33 μm ± 36% can arise in cases where a fluid accumulation on the chip surface is possible, e.g. temporarily in the regime of discontinuous atomization or continuously at higher flowrates (≥ 200 μl/min). In general, no droplets larger than 80 μm were observed.
Regarding the turn-on behavior, a two-step process was observed. After the application of SAW atomization starts immediately, i.e. after less than 80 ms, once the fluid reaches the atomization zone. Then, after a short settling time, the fluid film and the aerosol beam shape stabilizes (supplementary video 6). This settling time was found to linearly decrease with increasing SAW power, e.g. from 2.5 s (2 × 2 W) to 0.9 s (2 × 3.5 W). Our observations suggest that the stabilization time is principally defined by the height of the outlet channel. In the case of a channel outlet height larger than the height of the forming fluid film, a (concave) fluid meniscus that bridges these two heights is established inside the microchannel. This partial wetting of the inner channel walls requires time to stabilize and, in turn, this transient effect determines the atomization stability. In our case, the channel height measures approx. 140 μm, while the fluid film height is estimated to be below approx. 30 μm, according to optical micrographs. The meniscus had a length of approx. 100 to 150 μm in the channel. We expect that an adjustment of the outlet height to the fluid film height could minimize the meniscus length and, therefore, the time required to stabilize the atomization.
The turn-off behavior can be described as follows: As soon as the fluid supply to the acoustic beam stops, the atomization stops immediately. However, the self-pumping effect of the SAW (Winkler et al., 2015; Kurosawa et al., 1995b) in combination with the usage of compressible tubing can lead to a slowly decreasing fluid flow rate, even if the pump is suddenly deactivated. Therefore, it can be beneficial to apply an inverse volume flow for a short interval to achieve a defined turn off behavior of the fluid flow. A direct integration of a pump in the device (as in (Ariyakul & Nakamoto, 2014)) and the associated reduction of fluidic path length could also be beneficial.
Significant heating is observed with the application of SAW power in the absence of fluid supply. This principally occurs due to (1) Joule heating at the IDT electrodes and (2) the excitation of bulk wave modes originating from reflection of SAW at the chip edges and finger electrodes, which are transmitted to the chip holder and cause viscous heating in the heat conductive foil. No excess heat production is observed in the case of completely efficient continuous atomization. As heat production is representative of under-utilized mechanical energy, it affects the overall efficiency of the SAW atomizer and its usable lifetime. It can accordingly be optimized via the fluid flow rates and SAW powers that result in fluid coverage across the SAW beam, as has been done here, or potentially the use of phononic structures to constrain displacements to the substrate (Reboud et al., 2012).