Integrating Ultrasound with Microfluidics
Going “micro” with ultrasound is not straightforward, neither is interpreting the results of experiments or processes, yet its potential for different chemical uses has already been identified [77]. Microfluidics enable the manipulation of chemical reactions using very small amounts of fluid, and ultrasound offers a good “non-invasive” alternative for several processes. Additionally, the small quantities of reagents, solvents, and waste, a precise control of reaction conditions, as well as the integration of functionality for process intensification, have all been highlighted as greener, safer, and often faster protocols [30]. On the other hand, answering the perennial question on how to process or produce larger volumes of liquids with microfluidics, several numbering-up and scaling-up strategies, as well as manifolding have been explored with varied success [78–81].
In designing the experimental setups that researchers have used to perform microfluidics and sonochemistry we can identify two main directions based on the final objective of the researchers and availability of resources of each group:
-
Use of commercial existing equipment, where a capillary or microfluidic device is placed inside ultrasonic baths or in close contact with ultrasonic horns (see Fig. 1) [82–84].
-
Tailor made ultrasonic setups, where a transducer or set of them are glued or clamped against the microfluidic device (see Fig. 4 Bottom) [85–90].
While using commercial devices will lead to fast implementation in the lab, bespoke setups allow an additional degree of customization and optimization in terms of energy efficiency, but also reaction yield. One key design parameter is the coupling between the ultrasound transducer and the microfluidic reactor [91], and such designed systems can then be applied to microfluidic liquid–liquid extraction [89], degradation of contaminants [92–94], and particle synthesis [95, 96].
The large number of techniques at hand and disparate existing knowledge on its effects, has detrimentally influenced the modest utilisation and adoption of ultrasonic cavitation and microfluidics in industry. Furthermore, the differences in experiments designed by chemists and physicists are as varied as the ways a lab researcher and a plant engineer interpret their results. Because of the complex interrelation of physicochemical phenomena resulting from the collapse of a bubble or a cluster of bubbles, replicating results and the choice of quantification techniques (calorimetry, chemical dosimetry, acoustics, optics, etc.) has been troublesome [97–103]. In the particular case of sonochemistry, it has reached the point of being labeled a “black art” [104]. Overall, there is an increasing interest in exploring the potential positive results brought up by combining these two techniques (ultrasound and microfluidics), as observed in specialized conferences and other media. It is still in its early phases, and the reason why the progress is modest will be made more clear in the subsequent sections.
Entangled Effects of Cavitation
Acoustic cavitation can produce phenomena difficult to explain since it has interconnected variables with non-linear dependencies. The acoustic frequency, pressure amplitude, and other complex physicochemical parameters dictate the creation (or nucleation) and dynamic interaction of collapsing bubbles. The collapse of a bubble can be stable or transient, bubbles inside a cluster can be shielded by those outside; interaction forces between the acoustic field and bubbles (Bjerkness forces) are affected by the liquid properties (gas content, surface tension) and the geometry and materials of the reactors, to name a few [105, 106]. All of the above is further complicated “going down” in scale by confinement effects due to small scales, heat, and mass transfer phenomena. For example, when comparing 1D, 2D and 3D equivalent sonochemical reactors [107], an apparent increase in the reaction rate over the volume change was suggested to be due to the relaxation of space confinement when changing from 1D to 2D geometry. The reaction rate increased by 10 times while the volume increased by 57 times from 2D to 3D. A logical explanation was given by the fact that the total volume of a 3D reactor is not used as efficiently as the thinner layered channel, where nodes and antinode planes are not present. The power input dependencies also exhibit behaviours difficult to explain without a proper understanding of the underlying physicochemical mechanisms of cavitation (e.g. adiabatic compression of gas content dependence on the maximum and minimum radius, number, and spacial distribution of bubbles, shielding effects, etc.). In the same study, a decrease in the production rate of hydro-terephthalic acid (HTA) at higher input power density was found for 2D and 3D, but not in the 1D channel. In later studies, other puzzling correlations of power and an unexpected drop in radical production efficiency [85, 108] were explained by the change in the sphericity of bubble collapses. More details will be provided in the following Sect. 3.3.
From the chemical engineering and practical point of view, the most difficult hurdle for the wide acceptance of ultrasound and sonochemistry as a useful tool has been the measly energy efficiency values. The acoustic transducers transform electrical power into mechanical energy which is transmitted to the liquid. Part of the energy generates cavitation and another heats the whole system, hence not all of the energy produces the desired chemical and physical effects, making it difficult to establish a robust energy balance. For simplification purposes and using a relation reported in the literature [82, 109], we can define the sonochemical efficiency (or yield) as \(X_\mathrm{US}=\) measured effect/input power; which in our particular case we define it as:
$$\begin{aligned} X_\mathrm{US}=\frac{\Delta H (\Delta N_\mathrm{rad}/\Delta t)}{P_\mathrm{US}} \end{aligned}$$
(3)
where \(\Delta H\) is the energy required for the formation of OH· radicals, which is equal to the enthalpy of formation of the chemical reaction with a value of 5.1 eV per molecule [110]:
$$\begin{aligned} H_2O \mathop {\rightleftharpoons }\limits ^{\Delta H= 5.1\,\mathrm{eV}}OH\cdot + H\cdot \end{aligned}$$
(4)
\(P_\mathrm{US}\) is the electric power absorbed by the transducer which can be obtained from the measured voltage, current and their phase difference. This is clearly an underestimated value that can help in practical as well as in academic comparisons. Other “measured effects” can be used depending on the specific case, e.g. color dye degradation, mixing efficiency, calorimetric measurements. Depending on the specific study, on average the values reported for OH· radicals are in the order of \(X_\mathrm{US}\sim \mathcal {O}(10^{-6})\) [4, 85, 100, 111].
Ultrasonic cavitation is known to be difficult to reproduce since bubbles are normally created from impurities randomly distributed inside the reactor. Impurities such as defects on the walls (crevices) or dissolved solid particles are efficient traps for gas nuclei, and the acoustic nucleation threshold for bubbles trapped in cavities has theoretically and empirically been predicted [29, 112, 113]. The event of nucleation of a bubble from a crevice serves as a seed for subsequent cavitation. Depending on the acoustic conditions characteristic clouds of bubbles, also known as streamers, can persist for long periods [106, 114–117]. Effects such as inertia, surface tension, and viscous forces in a liquid influence the generation of micro-bubbles and have been studied with the help of non-dimensional numbers such as the ultrasound Weber number (We) and the ultrasound Womersley number (Wo).
$$\begin{aligned} \mathrm{We}=\rho f^{2}d_\mathrm{in}^{3}/\sigma \end{aligned}$$
(5)
$$\begin{aligned} \mathrm{Wo}=d_\mathrm{in}(f/\nu )^{1/2} \end{aligned}$$
(6)
where \(\rho \) is the density of the liquid, f is the ultrasound frequency, \(d_\mathrm{in}\) is the diameter of the pinned bubble, \(\sigma \) stands for the interfacial tension and \(\nu \) is the kinematic viscosity. The ratio of inertial and surface tension forces is given by the Weber number, We, and the Womersley number, Wo, represents the ratio of pulsatile to viscous forces. According to a particular study [114], a uniform diameter of bubbles is obtained when \(8.16< \mathrm {We} < 300\) and \(2< \mathrm{Wo} < 5\). For \(\mathrm{Wo} > 5\), the inertial effect dominate the viscous effect, resulting in bubbles of various sizes being released from the gas-liquid interface. When \(\mathrm{Wo} < \,2\), the interface was not distorted sufficiently to release bubbles. When \(\mathrm{We} > 300\), the inertia dominates surface tension effects producing bubbles of various sizes. For small We, the interface oscillations are stable without strong distortion, and no bubbles are produced due to dominating surface tension effects.
In general, it has been possible to study to a good level of detail physicochemical exotic phenomena such as plasma formation inside the bubbles, the emission of light (sonoluminescence), radical production, shockwaves, streaming and jetting (see Fig. 4) [11, 98, 106, 111, 118–120]. The interaction of individual bubbles or clusters as they collapse among themselves and against nearby surfaces (see Fig. 2) has also been studied given its importance for applications such as erosion prevention, cleaning, surface modifications, biology, and several chemical processes [13, 121–126].
In the particular cases that acoustic cavitation takes place close to a surface (see Fig. 5), interaction forces among the bubbles and the acoustic field alter the otherwise “ideal” spherical collapse. Oscillating bubbles, and bubble clusters oscillating close to a surface are attracted to their “image” on the virtual mirrored space. Jets and shockwaves emitted at different instants during acoustic cavitation are the main mechanisms responsible for the erosion of surfaces (see Fig. 4) [6, 13, 122, 127].
Controlling Cavitation with Microscopic Crevices
For a newcomer to the sonochemistry realm, it is important to understand that the rationale behind all the effects of cavitation begins with considering each bubble as a reactor in itself. Then, a working ultrasonic bath or horn will produce an undetermined number of bubbles of different sizes, and consequently each collapse will result in different temperatures and pressures inside of the bubble. It should be no surprise that as a result, the chemical or physical effects have a broad distribution of values, simply because all bubble-reactors behave different.
Despite the careful experimental precautions, the number of bubbles can be underestimated since some large bubbles overlap, and many smaller bubbles are not counted due to the optical resolution [108]. Nevertheless the qualitative and quantitative measurements of their number, coupled to the radical formation values can give relevant information to compare with other reports. Dividing the radical production per cycle by the average number of bubbles per cycle, gave bubble radical productions in the same order of magnitude measured by Didenko and Suslick [111] who reported data for a single bubble of maximum radius of 28.9 \(\upmu \)m driven at 52 kHz producing OH· radical generation at a rate of 6.6 \(\times \) 10\(^5\) per cycle. These values for radical production can be taken as a reference of the order of magnitude that can be produced by a generic bubble in an ultrasonic reactor [128].
The quest to trap bubbles for the study of nucleation and collapse was initially driven by fundamental research questions. Different approaches have been used, from acoustic trapping to geometrical confinement of conical bubbles or micromachined crevices [86, 111, 114, 129, 130]. Focusing on aspects relevant for the chemical engineering community, we will elaborate on a series of recent results motivated by two practical questions:
-
1.
How can the energy efficiency of sonochemical reactors be improved?
-
2.
How can bubble generation be controlled reproducibly?
The challenge was, not surprisingly, conquered as a result of an interdisciplinary collaboration between microfabrication experts, physicists, and chemists. A microfluidic sonochemical reactor was designed, modeled and tested under laboratory conditions [85, 108, 131, 132], and more recently the same concept has been patented, scaled up, numbered up, and commercialised [133–135]. We consider this case a good example of the positive outcome of the synergy of microfluidics, ultrasound and process intensification concepts.
Batch Micro-Sono-Reactor
The structure of a batch ultrasonic reactor was modified by micromachining artificial crevices of ca. 30 \(\upmu m\) diameter and 10 \(\upmu m\) depth, unto the surface of silicon substrates. This passive modification was used as a trap for a determined number of bubbles (see Fig. 4). Upon sonication, the bubbles served as a seeding gas volume that ensured continuous formation of smaller bubbles (streamers) containing a mixture of gas and water vapor. This continued formation of bubbles can be sustained by virtue of a phenomenon termed “rectified diffusion” [27, 136]. This phenomenon occurs as a bubble expands, and gas-solvent molecules diffuse–evaporate into the bubble; conversely, when the compression part of the sound field arrives, gas-solvent molecules diffuse–condense out of the bubble. Because of an unbalance in the area in the expansion (larger) and compression phase (smaller), there is a net gain inside the bubble which can compensate any mass loss as the smaller bubbles are created from the crevice.
In the first report of a series of studies based on the same device [85], the streamers were demonstrated to produce hydroxyl radicals (OH·) by imaging its reaction with luminol. The reaction rates were measured as a function of electrical power with terephthalic dosimetry (see Fig. 6).
As introduced in Sect. 3.2, though the efficiency when crevices were present was higher at the three powers tested (about one order of magnitude), at the highest power there was an unexpected drop. The efficiency values calculated with Eq. (4), are shown in Fig. 7. From a speculative analysis it was concluded that at high power, when the bubble pattern changes (see Fig. 6) a different radical generation distribution over the reactor volume could take place. Looking at each bubble as individual reactors, the energy efficiency of each collapse is strongly determined by the sphericity of the collapse, and the maximum size reached before collapsing. The smaller bubbles were thought to be stiffer due to a surface tension contribution and were not expected to grow considerably large during expansion. This leads to a consequently weaker compression and lower maximum temperatures reached after each collapse. Conversely, larger bubbles do not collapse spherically especially when close to a solid surface, hence limiting its maximum compression potential.
To clarify these assumptions, two complementary studies were devised. In the first one, the light emission from the collapse of bubbles (sonoluminescence) and as the result of the reaction of OH· with luminol (sonochemiluminescence) was measured with a photomultiplier [131]. Transient cavitation conditions were verified by measuring the SL intensity in propanol solutions. The different light intensities of SL and SCL helped to establish a difference in the bubble population able to emit light and those chemically active. This type of photosensitive study is important when is impossible to capture the fast dynamics of bubble collapse with the available equipment. The second study was directed at taking short exposure images and fast-imaging movies for the determination of the number and sizes of bubbles [108]. The quantification of radicals was correlated with the number of bubbles and radius-time evolution in an acoustic cycle. The main conclusions from this study were that at higher powers, hence larger pressure fields in which bubbles expand and collapse, the shape of each bubble deviates from the ideal sphericity which corresponds to a maximum conversion of potential energy into sonochemical effects (shockwaves, liquid jets, sonoluminescence, and radical formation, see Fig. 8). Those bubbles that collapse towards each other, or in the proximity of an interface, deform and result in jet formation. As a result, not all the energy that could heat the bubble content is available for the rupture of chemical bonds, such as that of water molecule which is measured by the formation of OH· radicals.
From the reactor design perspective, the fact that bubbles remain close to the surface from which they were created can have negative consequences. With this microreactor that allows control over cavitation, the erosion caused by bubbles collapsing in the vicinity of different silicon substrates could be studied right from the initial incubation period and through more advanced stages [127]. The same effect of various sources of damage formation such as jetting, shock waves, direct bubble impact, and surface stress corrosion that can cause the damage observed for three crystallographic silicon surfaces studied, was later used for another useful application: the rapid removal from substrates of deposited organic and inorganic materials, such as biofilms [137]. The practical relevance of these results will be continued in the following subsection.
Contrary to what happens with most commercial ultrasonic equipment, tailor-made devices can in principle be operated at different frequencies. The studies of this micro-sono-reactor were conducted at one frequency (200 kHz) where it was more active. Based on what was discussed in Sect. 3.2, for the case of water and air on a 30 \(\upmu \mathrm{m}\) crevice we have \(\mathrm{We}=15\) and \(\mathrm{Wo}=13.4\), indicating that relatively small interface oscillations of the bubble in the crevice (small We) are unbalanced by inertia dominated effect resulting in bubbles of different sizes (large Wo). Snapshots of videos from experiments of this system indeed show significant bubble deformations of the bubbles ejected from the bubble stabilised on the crevice (see Fig. 8) [108, 117]. We have not found in the literature other studies reporting on such large values for both numbers, hence a parametric study in this direction would be highly valued for future applications. Similar studies addressing the effect of different frequencies in the same reactor will shine more light into the challenging “black” field of sonochemistry.
Scaled-Up Non-Conventional Batch Reactor
Scaling-up or numbering-up are the most frequent strategies used in microfluidics whenever there is an interest in practical applications beyond the lab-scale. Small volumes ca. 250 \(\upmu \ell \), such as the ones used in the studies mentioned in the previous sections are of limited relevance for practical or industrial uses. Making larger crevices or higher amplitudes to obtain larger bubbles before collapse does not correlate linearly with better results, as explained in Sect. 3.3.1. Additionally, despite the unique control over the location of bubble clouds, it is difficult to envisage any commercial appeal that the presence of a few crevices etched on the bottom of a batch reactor can produce. Even when the numbering-up of crevices might be possible by drilling crevices on the surfaces of ultrasonic baths or the tips of horns, it seems like an improbable option for all manufacturers of baths and horns.
In trying to find a universal solution that could be used in most ultrasonic equipment, the new concept of the Bubble Bag was devised [134]. The inner surface of a bag that serves as a container can be indented with crevices for gas entrapment (see Fig. 9). It has been described as a cavitation intensifying bag since it enables a production of bubbles that does not takes place on a simple bag. The Bubble Bag can be interpreted as sophisticated beaker, initially made of plastic, but it can be manufactured in any other material since the principle for creating bubbles from crevices has already been demonstrated in silicon and glass. The first and most versatile advantage is that it does not depend on a particular frequency for cavitation to happen as demonstrated with several pieces of ultrasonic equipment; see in Fig. 9 the results of two ultrasonic baths with different frequencies with and without crevices. The second advantage schematically represented in Fig. 10 is that numbering-up of such a beaker is straightforward.
Designed originally for cleaning arbitrary objects [135], the current dimensions and potential use for radical generation allows us to classify it as a milli-reactor [138]. Using the same configuration, the bags can also be used for other applications such as emulsification, and presumably for any physicochemical processes where ultrasound is currently used [139, 140]. The most relevant scientific advantage is that reproducibility of results and energy efficiency are considerably better than when compared with a bag not having crevices. The standard deviation of radicals detected was reduced by 22 %, accompanied by an increase of 45.1 % in efficiency (see Fig. 9). The number of pits and bags can be in principle as large as the ultrasonic equipment allows for. New optimisation and characterisation studies will be required to find the best position or configuration of transducer-bag positioning. So far, the Bubble Bag has been operated in batch mode; we expect to provide more results as a flowing reactor in the near future in a multiple-stage assembly. We believe this concept will lead to new applications and improved versions to be reported in the specialised literature.
To conclude with this section, in Fig. 10 we illustrate the trends identified up to now and where we believe the future developments will head to: intensified continuous flow and industrial scale sonochemical reactors.
Acoustic Streaming
A phenomena increasingly exploited in microstructured devices is acoustic streaming. In general terms, acoustic streaming is the generation of a convective motion (i.e. fluid flow) through the presence of an acoustic field [141]. Real fluids exhibit a viscous attenuation towards an acoustic wave traveling through them, which in the case of an oscillatory acoustic field results in a time-averaged displacement of individual fluid elements. This local displacement can develop into a steady fluid flow, termed acoustic streaming [142]. Depending on the location and length-scale of the fluid motion several sub-categories of acoustic streaming can be defined, such as boundary layer driven streaming (Schlichting and Rayleigh streaming), Eckart streaming in the bulk fluid, and cavitation microstreaming [141, 143]. In the following we will discuss each streaming phenomena and provide application examples in microfluidics.
Boundary layer driven streaming Because of the no-slip condition at a solid interface, the fluid flow in the boundary layer is characterized by a steep velocity gradient. In turn, this steep velocity gradient is also responsible for a larger viscous dissipation of acoustic energy compared to bulk flow [144]. Applying a standing acoustic wave parallel to the solid interface results in spatially fixed pressure nodes and antinodes, which leads to a steady vortical fluid motion in the boundary layer (termed Schlichting streaming). These vortices in the inner boundary layer will excite counter-rotating vortices in the outer boundary layer, and this induced streaming motion is termed Rayleigh streaming [145].
Boundary layer driven streaming in microfluidics is often employed for particle trapping, e.g. to either control local particle concentration [146], particle aggregation [147], or to separate particles based on their size (in combination with the acoustic radiation force acting on the particles) [54, 143, 148]. However, it can also be used to overcome one of the disadvantages of microstructured devices, i.e. that the only means of mixing is by diffusion due to the predominant laminar flow. Rayleigh streaming has been shown to improve micromixing [149].
Eckart streaming When the dissipation of the acoustic energy takes place in the bulk of the fluid, Eckart streaming is observed. However, as microstructured devices are characterized by a large surface-to-volume ratio, their hydrodynamics is governed by boundary layers, and bulk flow is rarely encountered [141]. In general, Eckart streaming will only occur when high frequency ultrasound is applied, and when the characteristic dimension (defined as the propagation direction of the acoustic wave) is in the order of millimeters.
Consequently, the applications of Eckart streaming are limited in microfluidics. However, based on the experimental observation that the resulting convective motion is characterized by high velocity jets [150], it has been used to design a valveless ultrasonic pump [151].
Cavitation microstreaming Cavitation microstreaming is generated by the acoustically driven oscillations of microbubbles in a liquid. These oscillations are transferred via the boundary layer surrounding the bubble and generate vorticity and convective motion in the fluid [152]. In addition to the streaming motion, oscillating bubbles will also create stress fields in their surrounding, which largely depend on their mode of oscillation [153].
These stress fields coupled with the streaming effect enable the therapeutic use of microbubble-mediated ultrasound. An increased permeability of cell membranes has been observed, which allows e.g. targeted drug delivery. In addition, the combination of microbubbles and ultrasound have been shown to accelerate the breakdown of blood clots (thrombolysis) [154]. Furthermore, cavitation microstreaming is also applied to enhance mixing in microfluidic devices [155].
Clogging Prevention
In this section we will review the handling of solids in microfluidic systems. The solid material can either be comprised of an unwanted and insoluble by-product of a reaction, or the target compound (e.g. nanoparticle synthesis or crystallization of organic molecules). In general, managing solid particles in flow represents a major challenge for the upstream, continuous processing of fine chemicals in microreactors [56]. Many synthetic organic reactions either involve the use or the generation of insoluble compounds [57]. In the following we will discuss the ultrasound application strategies to control the particle formation within the microchannel.
The impact of channel clogging on continuous manufacturing is best illustrated with a study conducted at Lonza [156]: In a screen of 86 different reactions it was found that 59 % would benefit from a continuous process, however, this number reduced to 19 % due to the presence of solids. Consequently, to assist the transition from conventional batch to continuous manufacturing processes exploiting microreaction technology, reliable solids handling needs to be established.
Figure 11 depicts the main interactions governing particle behavior in microfluidic channels, namely particle-fluid, particle-particle, and particle-surface interactions [58].
The relative importance of these interactions differs on a case by case basis, which leads to several phenomena discussed below: Deposition of particles (Fig. 11a) is initiated by particle-fluid interactions transporting the solid to the microchannel wall where it finally sticks due to a dominating particle-surface interaction. Increasing the particle-fluid interaction by e.g. increasing the fluid velocity will lead to resuspension (Fig. 11b). The particles will agglomerate in the bulk of the fluid by particle-particle interactions (Fig. 11c); however, agglomerate break-up can again occur when the particle-fluid interactions overcome the inter-particle interactions. As we will discuss later, this is one of the main avenues where ultrasound comes into play, as the induced cavitation will give rise to hydrodynamic forces on the agglomerates and thus decrease their size. The clogging phenomena itself (Fig. 11d) is governed by all three interactions, and usually occurs via bridging of a constricted microchannel cross-section [56, 57].
The most used passive means to prevent microchannel clogging are:
-
Application of two-phase flow, allowing the introduction of a secondary phase which dissolves the solids. However, this also introduces additional mass transport limitations.
-
Increasing the wall shear stress by increasing the fluid velocity.
-
Reactor surface modification, mostly based on fluoropolymers, to achieve non-sticking reactor walls.
The passive techniques are reviewed elsewhere [56–58], here we primarily focus on active clogging prevention using ultrasound.
The integration of acoustic actuators with microstructures is a new and emerging area, where the acoustic energy is mostly supplied using transducers or piezoelectric microdevices with different sizes and geometries [77, 157–161]. One application of ultrasound integration in microfluidics is the excitation of a standing acoustic wave in the fluidic channel [53]. Particles in an acoustic standing wave field will experience an acoustic radiation force [54], which can be exploited for the manipulation of their trajectories [162]. This concept can even be extended to controlling the trajectory of single bubbles [163].
At increased power, acoustic irradiation has been shown to be successful in reducing agglomerate particle size, which is essential to prevent clogging [158, 159, 164]. A well studied reaction system and also a challenge under flow conditions due to clogging is that of Pd-catalyzed CN cross-couplings [158]. Under typical reaction conditions inorganic by-products precipitate immediately in the non-polar solvents needed for this transformation. Furthermore, the recent developments of highly active palladium catalysts which allow for extremely fast reactions, and thus also a fast generation of these inorganic salts.
One approach to prevent clogging is to immerse Teflon tubing in an ultrasonic bath for irradiation, as shown in Figs. 10 and 12a [159] and discussed in Sect. 3.1.
However, when using an ultrasonic bath, one has to be aware of the fact that not a single frequency is excited, but the resulting waveform can be quite complex [158]. Furthermore, the emitted ultrasonic waves first need to couple with the media in the bath before transferring to the microreactor. As such, integrating a piezoelectric actuator directly into the microfluidic assembly to directly transmit the acoustic waveform to the reactor is energetically more efficient, as mentioned in Sect. 3.1. An example of such a layered microreactor system is shown in Fig. 12b, which was also successfully applied to the aforementioned Pd-catalyzed CN cross-couplings and allowed for long term operation [160]. High-speed imaging revealed the formation of gas bubbles upon ultrasonic irradiation via the mechanism of stable cavitation, and the pressure forces associated with this formation lead to the breakup of the particle agglomerates [165–167]. This phenomena is clearly depicted in Fig. 13, which shows the fragmentation of a calcite crystal due to the collapse of a bubble.
Furthermore, using piezoelectric actuators allows for a precise control of the operating frequency, which is important to control the resulting size of the agglomerates. Figure 12c depicts the particle size distribution of inorganic precipitates subject to the applied ultrasound frequency, and for the particular setup the identified optimum frequency corresponded to 50 kHz.
Microfluidics As a Tool for Particle Synthesis
Microfluidic systems have also seen an increase in use for cases where the particulate matter is the desired product. Continuous manufacturing enables the formation of particles with narrow size distribution, which is an important property for their final application. In general, particles of similar properties will be formed when they experience similar conditions in the reactor, which is difficult to achieve in batch systems due to mixing limitations associated with their large size, but feasible on the micro-scale using continuous flow.
However, focusing on particle formation processes, these devices need not only overcome clogging of the microchannels, but also ensure kinetic control of nucleation, growth, and agglomeration. The same Teflon reactor design with integrated piezoelectric actuator as outlined above was applied to the crystallization of hydroxyapatite (HAp) in continuous flow [168, 169]. Compared to the batch process (stirred tank reactor), the HAp particles formed were more crystalline and less carbonate contaminated. The sonication strategy also lead to a reduction in particle aggregation and primary particle size; however, the resulting particle size distribution was poly-modal, which strongly suggested that the nucleation, growth and agglomeration processes were not precisely enough controlled in this system. A decoupling of nucleation and growth of particles in the reactor will increase this kinetic control, and one example to achieve this is the use of an ultrasonic horn next to the microchannel [95]. This will create a spatially localized zone within the reactor for the generation of crystal nuclei. Using a similar setup, Rossi et al. [96] studied the effect of supersaturation and ultrasound power on nucleation in more detail, and it was concluded that the transient cavitation of bubbles is a significant mechanism for enhancing crystal nucleation.