Theory and modeling of thin-film actuation in microscale acoustofluidics

Acoustofluidics is the field that studies sound in fluids which has been a steadily progressing research field stemming back over 200 years. From back with the study of motion of small particles driven by air currents above Chladni plates, vibrated by dragging a violin bow on the edge of metal plates.

Till today with systems manufactured in microelectromechanical systems (MEMS) compatible clean rooms, where they are designing and using devices with micrometer precision in liquid filed channels the widths of a few hairs. Especially the interplay between acoustofluidics and microfluidics has as of late shown many interesting applications, utilizing in part the acoustic radiation force and the acoustic streaming to move particles submersed in fluids, in what is generally called acoustophoresis.

These acoustofluidic devices have in the literature seen uses from, but not limited to, inkjet printers to cancer cell sorting, continuous flow separation of blood cells from blood plasma, label-free trapping and sorting, microfluidic mixers, acoustic tweezers, and much more. The field is a rapidly expanding and evolving field and there are still room for innovations and improvements.

In particular are there two problems that the field is facing that relates to this thesis. One of the common way to produce acoustfluidic devices to this day is to fabricate it in ways that make them difficult to reproduce results from device to device, as well as make them difficult to mass produce, by for example gluing lead zirconate titanate (PZT) piezoelectric transducers to the devices.

This is done in ways that make each device different from each other, since small changes in the glue makes large difference end results. The acoustofluidic device being based on the use of PZT piezoelectric transducers also means their systems contains lead, which for example the European Union has legislated in order to out phase, as lead is an environmentally hazardous material.

However lead is a difficult material avoid in piezoelectric transducers, as it works much better than most other lead free alternatives. This thesis has investigated ways to implement thin-film piezoelectric (PZE) transducer into acoustofluidic devices. One thin-film PZE that is commonly used in clean room fabrication and which is MEMS compatible is aluminum nitride (AlN), and its newer and more exotic variant, aluminum scandium nitride (AlScN) which is a stronger PZE material, but more difficult to manufacture.

How these thin-films can be used to actuate acoustofluidic devices, understanding the underlying mechanism, as well as comparing our model to experiments and using that model to make predictions on thin-film acoustofludic devices have been the main goal of the thesis. This has been achieved in this thesis and the results have been done in two published papers and one in which is still in preparation.

In the first article an ensemble of millimeter-sized glass blocks with AlN thin-film on top and different top electrode patterns were investigated. In the Paper I it is described how they were made, characterized, simulated, and how the Young’s modulus and Poisson’s ratio of the glass was fitted by minimizing the difference between measured impedance spectra and the calculated impedance spectra from the simulations.

These device showed high reproducibillity, that glass bulk devices can be actuated by a thin-films, and that good agreement with between simulations and measurements could be made. It was found that after having fitted Young’s modulus and Poisson’s ratio of the glass the relative deviation between simulated and measured resonance peaks of (−0.5±0.1)%, which was a order of magnitude lower than the parameter values from the manufacturer, and therefore showed that this could be a method for in situ characterization of parameters.

These parameter values are important, as they play a central role in the accuracy of our predictive simulations. The second paper is about the acoustofludic applications that lies in using PZE thin-films for driven bulk acoustofluidic devices. Building upon the confirmation of the solid mechanical part of the model in the first paper, we have simulated glass devices with microfluidic channels, of a few hundreds micrometers in width and height, actuated with thin-film transducers of the type AlN, AlScN, and PZT thin-films all showing acoustofluidic behavior.

In order to compare how the acoustofluidics, such as acoustic streaming and radiation force, compares to a traditional device we simulate a conventional device driven by a bulk PZT that constitutes 57% of the total volume of the device whereas the thin-film transducer only constitutes less than 0.1% of the volume.

The pressure, acoustic streaming, and acoustic radiation force fields were then compared qualitatively between the two types of devices and it was found that the devices preformed comparably, and this was seen again when comparing them in more quantitative terms, such as the average acoustical energy in the fluid, and the average time it took in each device to focus particle in the center of the channel.

Different aspects of the mechanism were investigated, such as the underlying physical mechanism that governs the principle of thin-film actuation, how the mechanical quality factor and thickness of the film does not effect the system significantly. Also shown was how electrode patterning can enhance the acoustic modes and that the system also works for the standing full-wave solution in the channel, that the modes look very clean and are easy to identify, and that the system is robust to channel off-set.

The last significant result in the thesis is the use of 1-µm-thick AlScN thin-film actuator on a circular 10-µm-thick membrane of silicon. The membrane investigated then has the electrode on the thin-film patterned in order to increase the mechanical actuation of the membrane which allows for the effective actuation of higher order membrane modes.

These higher order membrane modes then allows the pressure nodes to have a focus above the center of the membrane. The patterning increases the pressure hot spot for the fourth order mode many fold and the radiation force that follows, as a second order field, scales squared with this increase. The mechanism is investigated and shown that the system is a combination of standing and traveling waves, which stems from waves traveling out from the anti-nodes in the membrane.

This amplification of the membrane mode from effective electrode patterns is even more pronounced when going to higher order modes, and for such a device it is shown that the acoustic hot spot above the center of the membrane gives rise to acoustic trapping of cancer cells of the type MCF-7, when tuning the medium such that the acoustic contrast factor for the cells are negative, by adding Iodixanol which is a biologically inert fluid that is more compressible than water.

It is shown that even with acoustic streaming there is a hot spot, and the that the barrier that needs to be overcome when entering the trap can be minimized by playing with parameter values of the liquid that the cells are submersed in. The final thoughts on the thesis are given in the conclusion and outlook where the results that were shown this thesis summarized and it is discussed how they might be used in the future, as well what would be investigated had the PhD project continued.