piezoactuators for microfluidics

Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 909

Piezoactuators for MicrofluidicsTowards Dynamic Arraying

BY

TOBIAS LILLIEHORN

ACTA UNIVERSITATIS UPSALIENSISUPPSALA 2003

ii

Dissertation presented at Uppsala University to be publicly examined in Siegbahnsalen, Ångströmlaboratoriet, Uppsala, Friday, December 5, 2003 at 13:15 for the degree of Doctor of Philosophy in Engineering Science with Specialisation in Microsystems Technology. The examination will be conducted in Swedish.

Abstract Lilliehorn, T. 2003. Piezoactuators for Microfluidics: Towards Dynamic Arraying. Acta Universitatis Upsaliensis. Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 909. 56 pp. Uppsala. ISBN 91-554-5810-6. Microfluidics can be used to increase performance, reduce reagent consumption and increase throughput in chemical analysis. With the forthcoming development of more advanced microfluidic systems, the integration of actuating elements becomes essential, giving the ability to control and manipulate fluid flow as well as sample or other components. This thesis addresses miniaturisation of piezoceramic actuators, in particular important technological issues when actuators are integrated in microfluidic systems. Thick film multilayer fabrication technology for piezoceramics has been further developed, e.g. by introducing techniques for integration of microfabricated channel structures and via interconnects in multilayer components. New building techniques have been incorporated to allow miniaturisation of devices. A rapid prototyping technique for advanced multilayer actuators based on mechanical machining has also been developed and used in subsequent work.

When interfacing the macro and the micro world in miniaturised chemical analysis systems, non-contact sample dispensing methods such as ink-jet technology are needed. Thus a piezoactuated flow-through microdispenser, suitable for high-speed on-line chemical sample handling has been investigated. A new miniaturised actuator has been developed and integrated in the microdispenser, simplifying assembly and demonstrating an improved performance of the device.

With the prospect of performing automated and highly parallel analysis in reusable microarray devices, a new concept for dynamic arraying is presented. Non-contact trapping of particle or bead clusters in a microfluidic system is demonstrated utilising acoustic radiation forces in standing ultrasonic waves. The integration of piezoceramic microtransducers has been shown to render possible localised and spatially controlled trapping of individually addressable particle clusters in microfluidics. The importance of the acoustic near field in miniaturised devices has been identified and utilised to give strong trapping forces. By making use of disposable chemically activated microbead arrays within a flow-through device, a flexible system is emerging with e.g. applications in proteomics.

Tobias Lilliehorn, Department of Engineering Sciences, The Ångström Laboratory, Uppsala University, Box 534, SE-751 21 Uppsala, Sweden © Tobias Lilliehorn 2003 ISSN 1104-232X ISBN 91-554-5810-6 Printed in Sweden by Fyris-Tryck AB, Uppsala 2003 urn:nbn:se:uu:diva-3784 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-3784)

iii

Till Veronika

iv

Papers included in the thesis

I Rosqvist Tobias, Johansson Stefan. Soft micromolding and lamination of piezoceramic thick films. Sensors and Actuators A: Physical, 97-98: 512-9. (2002).

II Lilliehorn Tobias, Simu Urban, Johansson Stefan. Multilayer

telescopic piezoactuator fabricated by a prototyping process based on milling. Sensors and Actuators A: Physical, 107(2): 159-68. (2003).

III Rosqvist Tobias, Bergkvist Jonas, Nilsson Johan, Laurell Thomas,

Johansson Stefan. Evaluation of a microfabricated piezoceramic tripod actuator in a silicon flow-through dispenser. Presented at MME´00 Micromechanics Europe 2000, Uppsala, Sweden.

IV Bergkvist Jonas, Lilliehorn Tobias, Nilsson Johan, Johansson Stefan,

Laurell Thomas. Miniaturized flow-through micro-dispenser with piezoceramic tripod actuation. Submitted to Journal of Microelectromechanical Systems (November 2003).

V Lilliehorn Tobias, Simu Urban, Nilsson Mikael, Almqvist Monica,

Stepinski Tadeusz, Laurell Thomas, Nilsson Johan, Johansson Stefan. Trapping of microparticles in the near field of an ultrasonic transducer. Submitted to Ultrasonics (November 2003).

VI Lilliehorn Tobias, Nilsson Mikael, Simu Urban, Johansson Stefan,

Almqvist Monica, Nilsson Johan, Laurell Thomas. Dynamic arraying of microbeads for bioassays in microfluidic channels. Submitted to Sensors and Actuators B: Chemical (November 2003).

VII Lilliehorn Tobias, Johansson Stefan. Fabrication of multilayer 2-D

ultrasonic transducer microarrays by green machining. Submitted to Journal of Micromechanics and Microengineering (October 2003).

v

The author’s contributions to the papers

I Major part of planning, all experimental work, major part of evaluation

II Large part of planning, experimental work and evaluation III Large part of planning, experimental work and evaluation IV Part of planning, experimental work and evaluation V Major part of planning, experimental work and evaluation VI Large part of planning, part of experimental work and large part of

evaluation VII Major part of planning, all experimental work, major part of

evaluation

vii

Contents

1 Introduction ........................................................................................1

2 Piezoelectric actuators ........................................................................3 2.1 Piezoelectricity .............................................................................3 2.2 Piezoactuators for microsystems ..................................................6 2.3 Processing.....................................................................................9

2.3.1 Miniature multilayer components ..........................................9 2.3.2 Microstructuring of PZT ......................................................13

3 Microfluidics ....................................................................................17 3.1 The toolbox ................................................................................17 3.2 Microfluidic systems ..................................................................17 3.3 Microdispensing .........................................................................18 3.4 Bead handling in µTAS..............................................................21 3.5 Dynamic arraying.......................................................................22

4 Ultrasonic trapping ...........................................................................25 4.1 Physics of acoustic particle manipulation ..................................25 4.2 Applied acoustic particle manipulation......................................28 4.3 Integrated ultrasonic trapping.....................................................29 4.4 Towards dynamic arraying.........................................................35

5 Concluding remarks and outlook......................................................37

6 Acknowledgements ..........................................................................39

7 References ........................................................................................41

Tobias Lilliehorn ~

1

1 Introduction

The scientific work in this thesis is performed within the area of microsystems technology (MST). A microsystem can be defined as a system with some essential features in the micrometer range, and therefore making use of some typical behaviour on a microscale. It can comprise passive and/or active functions often embracing several domains, such as electrical, mechanical, fluidical, chemical, biological and optical. Sensing and actuating capabilities are often fundamental in microsystems since these allow the transformation of signals from one domain into another. Components performing such transformations are called transducers, and include both sensors and actuators. Sensors are used to convert a stimulus (physical, chemical etc.) into an electrical signal, and actuators to convert an electrical signal into a physical effect. Microsystems are also known under the denomination MEMS, being short for microelectromechanical systems, implying exactly such transformation between electrical and mechanical domains. The common denominator in all of my work is a particular piezoelectric ceramic material. This is used due to its ability to transform electrical stimuli into mechanical work, generating motion or sound, thus working as an actuator.

Piezoelectric materials can e.g. be found in a variety of applications dealing with the emission and detection of acoustic waves, such as medical imaging, non-destructive materials testing and underwater sonar systems. Piezoelectrics has also made its way into some consumer products and you have probably been in direct contact with devices utilising the piezoelectric effect, perhaps without knowing it. On your arm, the pace of time in your wristwatch is set by a small micromachined quartz tuning fork, and in your pocket, surface acoustic wave filters based on piezoelectric materials are integrated in your mobile phone. Some ink-jet printers and auto-focusing motors in cameras utilise piezoelectrics. Piezoelectric fibres have also been integrated in exotic “smart” sportsware, such as vibration damping slalom skis and tennis racquets.

Piezoactuators for Microfluidics ~

2

Figure 1: Ratio of articles in INSPEC containing the abbreviation ”PZT” to the total number of available articles in the database for each year 1973-2003.

The interest in making use of piezoelectric materials in research is still increasing. In Fig. 1 you can see the ratio of published work containing the term “PZT” to all searchable articles in INSPEC* during the last 30 years. PZT is the abbreviation for lead zirconate titanate, being the most widely used piezoceramic material today. As can be seen, the research field has undergone a fast expansion, in particular since the mid 90s. Thus adding to the ongoing development, this thesis introduce new technologies to microstructuring of piezoceramics, with an example shown in Fig 2†, and new applications of piezoceramic actuators within microfluidic systems.

Figure 2: Micromachined piezoceramics as compared to a human hair.

* Bibliographic information service for scientific and technical literature. † What would a thesis in MST be without the comparison of micromachined structures with a strand of human hair?

Tobias Lilliehorn ~

3

2 Piezoelectric actuators

2.1 Piezoelectricity The piezoelectric effect has been known since 1880, when the two brothers Jacques and Pierre‡ Curie discovered that electrical charges appear on the surfaces of some particular crystalline materials, such as tourmaline and quartz, when subjected to mechanical stress. The phenomenon was named “piezoelectricity”, originating from the Greek noun piezein, to press, and is traditionally referred to as the direct effect. One year later the Curies also confirmed the converse effect, i.e. that the same materials respond with a mechanical strain when subjected to an applied electrical field. As the direct effect makes it possible to use the material as a sensor, it is the converse effect that is used within the scope of this thesis. This effect makes the material suitable to be used as an actuator, performing energy conversion from electrical to mechanical.

The first serious application of the phenomenon took place during World War I, with the development of ultrasonic submarine detectors. These detectors were fabricated out of quartz and were designed to emit acoustic signals and to detect reflected echoes from objects situated in front of the them, very much like what can be found in nature used by e.g. dolphins and bats. The ability of piezoelectric devices to emit high frequency acoustic waves is utilised in one part of this thesis to trap microscopic particles in a miniaturised fluidic device.

The piezoelectric effect is originating from the crystalline structure of the material. The atoms in the crystal of a dielectric ceramic material can be considered more or less ionised, i.e. electrically charged. When subjected to an electrical field, the ions are displaced, the positive and negative ions in opposite directions. If the crystal, and the bonds between the ions, is completely symmetric, the displacement of the negative and positive are exactly the same and there is no resulting strain in the material. But if the crystal is non-centrosymmetric (with one exception [1]), the displacement of ‡ Married to Marie Curie together with whom he shared the 1903 Nobel prize in Physics for their joint researches on radioactivity.


4

the negative ions will be larger than the positive or vice versa, and the material will be strained, also on a macroscopic level. Since the utilisation of the piezoelectric properties in a single crystalline material, such as quartz, depend on (knowing) the orientation of the crystal structure; much effort is needed to determine this orientation and to cut the crystal in well-defined directions.

During World War II, researchers from different parts of the world discovered a new class of polar ceramic materials, where an applied electric field could alter the orientation of the polarisation within crystallites by switching the position of the ions. The materials were called ferroelectric in analogy with ferromagnetism. The discovery opened up for the possibility of making polycrystalline ceramics out of powders by conventional sintering§ and then lining up the dipoles within the separate grains. Before alignment the individual dipole moments are randomly oriented, giving no net dipole moment. The lining up of the dipoles by an applied electric field E is called polarisation or “poling” and the process is illustrated in Fig. 3.

Figure 3: Polarisation of ferroelectric ceramics by applying an electric field, E.

The discovered ferroelectric materials were based on the perovskite crystal structure ABX3. This crystal structure has ever since been made to produce a wide array of phases, with properties ranging from high dielectric constant to superconducting [2] and is considered to be one of the most important crystal structures in the history of materials design and engineering. In 1954 the discovery of lead zirconate titanate, or PZT, was presented [3]. This ferroelectric material was based on the perovskite structure, Fig. 4, and demonstrated an unusually large piezoelectric effect. PZT has since then become one of the most commonly used piezoelectric materials, and is the actuator material used in the work of this thesis.

§ Sintering is the process of heating the body to a temperature where the ceramic particles start to grow together. It has been used since about 20000 B.C. and is today utilised e.g. in producing porcelain tableware.

Tobias Lilliehorn ~

5

Figure 4: The perovskite unit cell of lead zirconate titanate, Pb(ZrxTi1-x)O3 (PZT).

The symmetric cubic perovskite shown in Fig. 4 is the phase of PZT when it is heated over a certain temperature where the material has lost its piezoelectric properties. This is called the Curie temperature, where the material exhibits a phase transition between the paraelectric and the ferro-electric phases. The ferroelectric phases are slightly distorted when compared to the cubic perovskite unit cell found at higher temperatures. Shown in Fig. 5 is a phase diagram of PZT, which can be described as a map of the phases in the material by varying the composition and the temperature. Since the material is a solid solution of PbZrO3 and PbTiO3 the composition can be varied in the form Pb(ZrxTi1-x)O3. Thus at room temperature, the material can be made to go from a rombohedral phase for Zr-rich compositions to a tetragonal phase for Ti-rich phases, with a change in phase near a Zr/Ti ratio of 53/47. This is called the morphotropic phase boundary (MPB), and is characterized by a change in phase with composition that is nearly independent on temperature [2]. At the MPB the dielectric and piezoelectric constants of the material are maximised. This is in part explained by a coexistence of different phases surrounding the MPB, increasing the possible directions of polarisation due to transformations between the phases [4].

Figure 5: Phase diagram of Pb(ZrxTi1-x)O3. (After [4]), indicating the distortion of the unit cell in the rombohedral and the tetragonal phases.


6

The properties of PZT can be changed by small modifications of the solid solution composition or with the addition of small amounts of dopants, taking the place of the lead or the zirconium and titanium. This is the origin of so-called “hard” and “soft” PZT materials. These two types of materials have different advantages. Hard materials are mechanically stiffer and show low dielectric losses, making them suitable for e.g. high power ultrasonic applications where losses needs to be minimized to keep temperature down. Soft materials have higher dielectric and piezoelectric coefficients and are easy to polarise, making them suitable for e.g. quasi-static actuator applications where the high piezoelectric coefficients yield large strokes of the component. All PZT materials used in the work within this thesis are soft materials. One of the origins of the higher strains that are generated in soft materials is attributed to domain wall movements. As shown in Fig. 3, different polarization directions can exist in different regions within a single grain. The boundaries between such regions, or domains, are called domain walls. In soft PZT compositions, these walls are moved more easily. This leads to that, when applying an electrical field, the domains with a favourable polarisation grow on the expense of the other. Since the crystal unit cell is elongated in the direction of polarisation, this gives a higher macroscopic strain in the material.

2.2 Piezoactuators for microsystems In the issue of designing actuators to be used in microsystems some important features are identified. For an actuator to be considered to be a microactuator, it should in general occupy less than 1 mm3. The maximum strain in a piezoceramic material is typically 1 ‰ (i.e. 0.1 %) and to reach this, relatively high electric fields, in the order of some MV/m (or V/µm) are needed. For a typical mm-sized component this corresponds to voltages of some thousand volts. Handling such high voltages in microsystems is not trivial, and thus it is important to use actuators that can be activated by using much lower voltages of some tens of volts. This is achieved by supplying the lower voltage to thin layers of the piezoceramic material, and stacking several such layers into a multilayer structure [1], shown in Fig. 6. The integrated electrodes in this structure are also used to polarise the material in the direction of the applied electric field.

Tobias Lilliehorn ~

7

Figure 6: Multilayer actuator with integrated electrode layers (dark).

The lowering of driving voltage is one of the main reasons why all the fabrication processes utilised or developed in the papers aim to making multilayer structures. Another reason is that, when using the components as ultrasonic transducers, the electrical impedance needs to be matched to the drive electronics to be able to couple as much of the electrical energy as possible into the component [5]. The electrical impedance is the ratio of applied voltage to the resulting electrical current in a circuit or a component, and is often a function of frequency. The impedance is usually high for miniaturised piezoceramic components, but is decreased with the intro-duction of the multilayer design. Multilayer transducers are therefore used in the papers on ultrasonic particle manipulation (Papers V-VII). The electrical impedance spectrum of a piezoelectric component can be used to identify mechanical resonances in the structure due to the electromechanical coupling in the material. At resonance, the resistance to mechanical vibration is low, and relatively large electrical currents are generated upon application of an alternating voltage, thus yielding a low electrical impedance.

Since the dimensions of miniaturised piezoelectric components are in the order of 1 mm, a strain of 1 ‰ implies a maximum elongation of about 1 µm. Various smart designed components have been studied, to get larger strokes out of the small intrinsic strain [6-8]. In Paper II, existing designs are reviewed and one alternative design is fabricated and evaluated, the telescopic actuator shown in Fig. 7. In the telescopic actuator, mechanically interconnected concentric tubes with integrated multilayer electrodes are made to expand or contract, giving a displacement magnification of about 5 as compared to an ordinary multilayer actuator of the same height.


8

Figure 7: The telescopic actuator and an illustration of its working principle. The measured stroke of this actuator was 10.8 µm, as shown in Fig. 8, even though one of the tubes was inactive due to a short-circuit. Since piezo-ceramic materials show some hysteresis, the strain in the material is different for increasing and decreasing field strength in an actuation cycle. The hysteresis can be thought of as due to an energy barrier in the material when switching from one polarisation direction to another or when transforming from one phase to another.

Figure 8: Stroke of the telescopic actuator, showing the effect of hysteresis.

Piezoceramic materials can be used to generate large forces even when miniaturised. Ideally, a component with a cross-sectional area of 1 mm2 is able to generate a force of about 50 N, corresponding to a weight of a 5 kg

Tobias Lilliehorn ~

9

mass. In reality, the generated force is however limited by mechanical breakdown due to stress concentrations in the component and to defects in the ceramic material. The telescopic actuator was shown to withstand loads up to 25 N and was upon higher loading broken near the region where simulations indicated the highest tensile stresses. The integration of leverage structures enhance stroke of the component on the expense of generative force. Thus, to compare different actuator designs, or actuator materials, a good measure of actuator performance is the energy density. This corresponds to the work needed to force an actuator from its actuated to its unactuated state per volume of the actuator. An actuator with a high energy density will in general yield a high performance even when miniaturised. The energy density of the telescopic actuator was quite high when compared to other designs with intermediate leverage magnification, but still an order of magnitude smaller than the ideal value corresponding to a multilayer stack due to inactive volume and to deformations in the structure, implying room for improvements.

2.3 Processing One of the main scientific contributions in this thesis regards the processing of miniaturised components made out of piezoceramic material. As mentioned above, layered structures are often advantageous, which is reflected in the process technology being used. PZT can be processed into thin films [9] i.e. less than 1 µm or into thick films [10], typically thicker than 10 µm. Thin films are often deposited atom by atom (or molecule by molecule) using e.g. laser ablation, sputtering or CVD, or by chemical precipitation from precursors, such as sol-gel processes. Thick films are often prepared from micrometer sized ceramic powders, which after processing are sintered to a ceramic body. Also thin film processes often require some kind of thermal step during or after the deposition of materials, to obtain the piezoelectric crystal structure in the film. In order to get good piezoelectric properties for actuators, thick films are preferred over thin film since the piezoelectric properties degrade when the grains in the material get too small [11]. Thick film technology is also suitable for fabrication of multilayer structures. All work performed within this thesis is thus based on thick film processing.

2.3.1 Miniature multilayer components To fabricate actuators the piezoceramic powder has to be prepared in a form that allows processing of the material into multilayer structures. Ceramic


10

slurry, i.e. a suspension of powders in a liquid vehicle, can by quite simple means be cast into layers. The preparation of the ceramic powder suspension is very similar to what is done when preparing paint from pigments [12]. Polymeric binder is introduced in the slurry in order to allow handling and processing of the ceramic layers upon drying. In the dry “green” state, the material is rigid and very suitable for shaping, using e.g. mechanical machining [13]. Green machining can be highly efficient as compared to the demanding and time-consuming machining of sintered ceramics. By shaping the components in the green state one can also avoid the possible damage of inducing cracks in the structures when machining brittle sintered ceramics.

Figure 9: Wet building process.

Most processes used at our laboratory for the fabrication of piezoceramic components [13-15] are based on a technique called wet building. Using wet building, illustrated in Fig. 9, ceramic slurry is cast in layers some tens of micrometers thick using an adjustable device called a doctors blade. On top of the dried green ceramic layer, electrode patterns are screen-printed with the pattern differing between layers to integrate multilayer electrodes as illustrated in Fig. 6 and 9. The casting and printing steps are repeated with the height of the doctor blade adjusted between layers until the full component height is reached.

Tobias Lilliehorn ~

11

Ceramic multilayer components can also be prepared using other similar techniques, the most common being tape casting and lamination [16]. Using the tape casting process, a single layer of green ceramic is cast on a carrier tape using a doctor blade. This can be performed on big rolls of carrier tape, making the process suitable for industrial scale production. Similarly to the wet-building process, electrodes are printed on top of the cast tape. Green ceramic tapes are then aligned, stacked and laminated into a multilayer by applying heat and pressure.

The resulting multilayer slab produced by any of these processes is cut into components while still in the green state, using e.g. a dicing saw. These components are still unusable as actuators, since the ceramic powders are not mechanically interconnected other than via the binder. To get a monolithic ceramic body, the components are subjected to a thermal treatment, where the binder is burned out followed by sintering of the ceramic particles at a high temperature (1280°C). To be compatible with the piezoceramic material at this temperature the electrode paste is made out of the noble metal platinum. Partly due to the cost issues of this material, there is a general trend in trying to lower the sintering temperature of the PZT material, being able to use other electrode material combinations [17].

On the basis of these two processing schemes, we identified some issues of particular importance for further development of thick film laboratory-scale processing of miniaturised actuators. Issue number one was regarding the lateral definition and registry of electrodes in different layers. When miniaturising multilayer piezoactuators, the inactive volumes at the edges in the multilayer design, as shown in Fig. 6, tend to be large when compared with the active volumes, decreasing the performance of the component [13]. This is because the resolution of electrode patterning is poor. The printing resolution of the screen-printing process used at the laboratory was about 100 µm, and considering microelements less than 500 µm squared, almost 50 % of the component would become inactive. Issue number two was regarding the lead-time of prototyping. Prototyping means that you are interested in producing a few components of a certain component design, with the possibility to redesign that component with a short turnaround time. An alternative to large-scale wet building or tape casting processes was therefore wanted. The patterning using screen-printing had to be replaced due to the time consuming manufacturing of screens for each new design. Also the geometric freedom of machining the green components was sought to be improved. Finally, to enable fabrication of some advanced micro-actuator components an electrical via interconnection technology, creating three-dimensional coupling between integrated multilayer electrodes was crucial.


12

To deal with these issues, Paper II presented a prototyping technology using a wet building/lamination hybrid process. In this process electrode patterning and green machining was performed using a computer numerically controlled (CNC) desktop micromilling machine. Small chips containing only a few layers were laminated to a multilayer from which a few, or even one, component was machined, as shown in Fig. 10. A method of forming via interconnects was also introduced. Using this method, 600 µm holes were drilled throughout the component, coated with electrode paste on the interior surface and then machined to allow several electrical phases to be connected using the same, split, via. The resolution of electrode patterning was found at least as good as for screen printing but with a superior flexibility due to the lack of mask changes. The manufacturing time was significantly reduced by using this hybrid process scheme. The telescopic actuator presented in the paper was the first demonstrator of the new prototyping process, which later was utilised in Papers IV-VI. The relatively low yield in the processing of the advanced telescopic actuator was very much improved in the fabrication of the simpler multilayer components utilised in subsequent work.

Figure 10: Lamination of chips and machining of the telescopic actuator.

To further increase the resolution of mechanical machining, a dicing saw was utilised to pattern electrode material in the ultrasonic array transducers presented in Paper VII. Using a dicing saw it is possible to machine structures on a very fine scale, as exemplified in Fig. 11, where a human hair is diced into small cubes. Mechanical machining of electrodes was shown to resolve electrode lines down to 10 µm, which is much better than ordinary screen-printing. In this work a new via interconnection technology was developed for multilayer components. It allowed electrical interconnection between two individual electrode layers with via dimensions of 20 µm, and rendered it possible to produce arrays of miniaturised multilayer transducers. In these array transducers it was desirable to design the acoustical behaviour

Tobias Lilliehorn ~

13

of the array elements. To do this, there was a need for a technique to introduce structured voids in the multilayer structure.

Figure 11: Human hair, cut with a silicon dicing saw (courtesy of Disco Corp.).

2.3.2 Microstructuring of PZT Microstructuring of piezoceramics could be used to e.g. fabricate ultrasonic piezoceramic-polymer composites [4] or acoustic matching layers for ultrasonic transducers, improving acoustic matching to water and human tissue. Integrated channels or voids in a multilayer component could be used to conduct fluid for ejection (inkjet devices), for active or passive cooling (micro heat sinks or micro heat exchangers) or for performing chemical reactions in a microscale (microreactors). Voids could also be integrated, having a mechanical or acoustical function, such as separating vibrating devices, reflecting acoustic waves, or allowing the integration of levered structures for stroke enhancement.

In this thesis, two different approaches have been used to address micro-structuring of piezoceramic components. Paper I, describes the development of a soft micromolding technique, related to soft lithography [18]. This technique resulted in microstructured green tapes that, by making use of a green tape adhesive, also were laminated to allow channel integration. The other approach, treated in Paper VII, deals with sacrificial structures that were integrated in the multilayer and then removed during binder burnout.

Figure 12: Soft micromolding process.


14

(a) (b)

(c) (d)

Figure 13: Scanning electron microscope (SEM) images of process steps in the soft micromolding process; master (a), PDMS mold (b), resulting PZT replica in the green (c) and in the sintered state (d).

Fig. 12 show the process scheme of soft micromolding with some process steps imaged in Fig. 13. A master template (A) was fabricated using lithographic techniques. This master was replicated by casting a silicon rubber, poly(dimethylsiloxane) (PDMS) over the structure (B), which was peeled off after hardening (C). This PDMS “soft micromold” was then used as a mold for piezoceramic slurry (D), which easily was separated after drying (E). The resulting structured green state replica (F) of the master was sintered (G) or laminated before sintering. The same PDMS mold could be reused several times.

(a) (b)

Figure 14: Micromolded pillars of PZT in green (a) and sintered (b) state.

Tobias Lilliehorn ~

15

The soft micromolding technique also allowed fabrication of the labyrinth on the cover page, and the microscopic pillars shown in Fig. 14. Applying a solution that can be referred to as a green tape adhesive made the lamination of micromolded tapes possible. Microchannels such as the ones shown in Fig. 15 were integrated in between ceramic layers by using this solution and applying moderate pressure (less than 1 MPa) in comparison to ordinary lamination (typically 20 MPa). The attained aspect ratio of molded structures, up to 7:1, is comparable to lost mold techniques, which lack this multilayer possibility.

Figure 15: Microchannels in a sintered laminate, perhaps suitable for cooling.

The second technique developed for microstructuring of channels and voids in multilayer ceramics utilises the fact that polymer, and other mainly carbon-based materials, are removed from the ceramic green body during binder burnout. By integrating sacrificial structures of such a material it was possible to get well-defined channels and voids throughout the component. In Paper VII, screen-printed layers of carbon-based paste was structured using a silicon dicing saw before integration in a multilayer.

Figure 16: Integration of voids using sacrificial material that is deposited (a), structured (b), integrated (c) and finally removed during burnout and sintering (d).


16

(a) (b)

Figure 17: Narrow channels (a) and wide voids (b) made using the sacrificial material technique.

The process of integrating sacrificial structures in a ceramic multilayer is shown in Fig. 16. The integrated voids ranged from 10 µm up to some hundred micrometers in width as shown in Fig. 17. As stated earlier the integration of voids was used to design the acoustic behaviour of ultrasonic microtransducer arrays in Paper VII. Due to the difference in acoustic impedance between the ceramic material and air, most of the acoustic energy transported in an acoustic wave is reflected at this type of boundary. The integrated voids, working as acoustic reflectors, therefore allowed the excitation of a thickness vibration mode in the multilayer element supported over the void. The fabrication technology showed the unique ability to producing dense 2-D arrays of individually addressable multilayer micro-transducers (350 x 350 µm2), with integrated acoustic micromachined reflectors. The impedance spectra of two different designs were evaluated and thickness modes were detected above 10 MHz, thus indicating that the reflectors worked as intended.

Tobias Lilliehorn ~

17

3 Microfluidics

3.1 The toolbox Microsystem technology once started off by using tools developed in the field of microelectronics, but has ever since been developed in its own direction [19]. This also means that the materials used in micromachining today range from traditional single crystalline silicon, via glass and ceramics to diamond and polymer materials. I will not go into the processing of silicon, since it is beyond the scope of this thesis, only mention that typical processes used for silicon micromachining [20] are photolithography followed by wet (chemical) or dry (plasma) etching. Some silicon micromachining has been used in Papers III and IV, to form fluidic channels and nozzles in biochemical sample microdispensers. Lithographic techniques were used to fabricate masters for mold fabrication in the soft micromolding process for piezoceramic microstructuring presented in Paper I. The photo-sensitive material used in this process, SU-8 [21] allows fabrication of thick structures with very high aspect ratio, i.e. the ratio of height to width. This is often needed in the fabrication of microchannels in fluidic devices. In Papers V and VI, SU-8 was used to make the fluidic channels in devices for ultrasonic particle manipulation.

3.2 Microfluidic systems By decreasing dimensions in fluidic systems some important parameters for the performance of e.g. chemical analysis or synthesis can be improved. Chemical reactions are faster on a microscale since the diffusion distances decrease. Also the use of expensive reagents can be minimised and by making use of the high surface to volume ratio in microchannels, chemical analysis can be made very sensitive. The small dimensions together with the possibility to perform highly parallel fluidic operations with microfabricated devices is expected to increase throughput substantially. In the early 1990s, the concept of miniaturised total chemical analysis systems (µTAS) was


18

proposed [22]. Since then, pumps and valves have been developed to control fluidic flow in microchannels [23]. Microfluidic technologies are highly amenable for automation, making them suitable for e.g. high throughput screening (HTS) for drug candidates [24]. Automation has shown it possible to make miniaturised laboratories, performing several hundred operations in parallel, integrated in what looks like a common music compact disc (CD) [25]. The issue of interfacing the macro world and the micro world is very important and delicate. In e.g. high throughput screening, substances from vast chemical libraries are tested against biological samples (enzymes, proteins, etc.) to search for candidates to developing new drugs. To be able to inject sample from these chemical libraries to microfluidic analysis systems, techniques to handle small amounts of fluids in this interface are needed.

3.3 Microdispensing When handling fluids in e.g. chemical analysis or combinatorial chemistry, the technique used is depending on the volumes of the dispensed fluid. Pipetting, being the by far most widely spread fluid handling technique, is limited by physical interaction of the fluid and the pipette tip, i.e. capillary forces and wetting (surface tension effects). Also pin or needle techniques, where sample is transferred by dipping a solid or split pin in a fluid reservoir, are limited by surface tension, particularly since the performance depends on the properties of the surface to which sample is transferred. The miniaturisation of chemical analysis systems requires the ability to handle and dispense small quantities of sample fluid using non-contact methods. Therefore ink-jet technologies are advantageous when the amount of dispensed fluid is decreased.

Ink-jet technology [26] is today one of the most important commercial applications of MST. Ink-jetting can be of the continuous type, producing a continuous stream of droplets by Rayleigh instability, or of the drop-on demand type, where single droplets are ejected on demand. The latter has ever since the 1980s been the dominating technique that replaced matrix printers to conquer low-cost market of the rapidly expanding PC industry. Ink-jet printers have become a standard component in both home and small office solutions. Thermally or piezoelectrically actuated drop-on-demand devices dominate the market today. It has been concluded that the main challenging issue of piezoactuated print heads is the miniaturization of the integrated actuator [26]. This again shows the importance of developing processing techniques for miniaturized piezoactuators.

Tobias Lilliehorn ~

19

Thermal printing techniques are not suitable for handling sensitive biosamples since it involves local heating of the fluid up to 300°C, and thus piezoelectric devices are preferred for the dispensing of biofluids. Earlier presented devices [27, 28] have been designed to dispense from an enclosed volume of liquid, i.e. a reservoir to which fluid is supplied by individual fluidic connections or aspired from a sample well. To allow on-line coupling of dispensers to e.g. chromatographic systems, a flow-through micro-dispenser design has been proposed [29]. This allows sample to be ejected from a continuous flow of fluid containing e.g. sample gradients from up-stream separations. It has recently been shown to be possible to dispense several samples separated by washing fluid using such a dispenser with an in-between sample carry over of less than 1 per 10000 [30]. Sample carry over is essential to minimise in flow-through systems since it describes limitations of the devices due to cross contamination between subsequent performed activities. Another important parameter to minimise in flow-through devices for chemical analysis is the dispersion of sample, limiting the resolution of e.g. separations. It has been concluded [31] that the dispersion in the dispenser was decreased by miniaturisation of the device. Further miniaturisation was limited by the actuator and the building technology, and therefore a novel actuator was developed in Paper III and IV. This actuator, named “tripod”, consisted of three electrically separated multilayer actuators, fabricated monolithically with a common backing. The component was designed to allow for mass-fabrication of miniaturised dispensers by simplifying the assembly. The tripod design was shown in the overview of the wet building technique, Fig. 9, and an exploded view of a tripod-actuated flow-through dispenser is shown in Fig. 18.

Figure 18: Flow-through dispenser with fluidic connectors (A), tripod actuator (B), push-bar (C), flow-through channel (D) and nozzle for droplet ejection (E).


20

The tripod actuator was designed to allow push-pull actuation, as illustrated in Fig. 19, but the evaluation was performed with only the middle leg activated. By supplying an electric pulse, droplets were ejected from the nozzle situated along the flow-through channel of the device. Dispensers utilising two generations of tripod actuators with slightly different fabrication techniques were evaluated. The dispensers showed similar performance, with stable droplet generation up to 2-3 kHz, where resonances in the system started to influence operation. Reproducible droplet generation was possible over this frequency, shown up to 8 kHz, demonstrating considerable variations in droplet velocity and volume with frequency. The ejected droplet volumes were about 20-70 pl in the examined voltage range. A maximum volumetric dispensing rate of 7.8 µl/min (33 V pulse amplitude) was reached within the stable dispensing region. Fig. 20 shows a droplet ejection sequence together with a stroboscopic image of dispensing at 5 kHz.

Figure 19: Illustration of push-pull actuation using the tripod actuator.

Figure 20: Flow-through dispenser in operation, the upper image showing droplet formation at 50 Hz, and the lower high speed dispensing at 5 kHz.

Tobias Lilliehorn ~

21

Along with the miniaturisation of the device, a building technique for miniaturised electrical interconnects [32], based on microstructured flexible printed circuits was introduced. This made it possible to miniaturise the device further. The smallest tripod-actuated flow-through dispenser fabricated, 5 mm long, is shown in Fig. 21, where it can be seen to hover above the inlets of a miniaturised CD laboratory. The function of this dispenser was shown, but the performance not fully evaluated.

(a) (b)

Figure 21: Microdispenser with mounted tripod actuator (a), hovering over a miniaturized CD laboratory (Gyros, Uppsala) (b).

Apart from sample injection in analysis systems, dispensers can be used to print e.g. substances onto the solid surface of chemically modified glass slides, being able to fabricate microarrays for DNA [33, 34] or protein [35, 36] analysis. The surface area of the solid phase, and therefore the performance of reactions and the sensitivity of analysis, can be increased by introducing bead-based methods. Beads are small spherically shaped particles made out of e.g. glass or polymer material. Their surface can be chemically modified to allow immobilization of antibodies etc. Performing e.g. studies of protein interactions in a fluidic environment also help in keeping sensitive proteins in natural ambient conditions. Bead handling in µTAS is therefore of particular interest.

3.4 Bead handling in µTAS Particles in microfluidic systems are often considered a problem, since once they have entered the system they can block fluid flow or otherwise have


22

detrimental effects on the system performance. This can be taken care of by integrating filtering in the system to stop particles from entering sensitive parts. However in some cases, the particles are inserted on purpose, to take advantage of some properties of particulate material. This is e.g. the large surface area and thus the high binding capacity of the beads as compared to solid flat surfaces. To be able to control the supply of sample and reagents to the beads and to be able to detect the reactions, techniques to trap beads at well-defined positions in microfluidic systems are needed. Various trapping techniques have previously been examined, such as physical barriers or filters of different designs [37-39]. These types of physical traps are very efficient, letting no beads to pass through if designed correctly. This is often an advantage, but the techniques may be limiting in some cases where there is a need for more flexible non-contact handling.

3.5 Dynamic arraying Microarrays have become indispensable as research tools, but to fulfil requirements of industry applicability, the technology has to be further developed. Automated flow-through systems with reusable devices are identified to fulfil the requirements of the second generation microarrays for bioanalysis [40]. In Paper VI we propose a new concept of dynamic arraying for bioanalytical assays** based on beads in microfluidic systems. Bioassays often incorporate some kind of antibody-antigen interaction to measure the amount, or just detect the presence, of antigen in a sample. This can be used e.g. to detect proteins identified as biomarkers in early disease development. With the new concept the interactions are performed on arrays of disposable chemically activated microbeads. As stated above, bead-based methods can be made flexible and sensitive due to the physical and chemical properties of beads. By trapping chemically activated beads of different antigen specificity in an array pattern (A, B…X) in a flow-through microsystem, one can address each bead array with an individual sample stream (1, 2…Y) as illustrated in Fig. 22. This means testing all interactions (A, B…X)-(1, 2…Y) in parallel. By performing the chemical interactions on microbead clusters one can dispose the solid phase, i.e. the beads after completed analysis. The procedure can be repeated by loading a new set of beads into the array before performing the next assay, thus making the device reusable.

** An assay is the chemical analysis of a substance to determine its content, historically used to describe the quantification of a metal in an ore.

Tobias Lilliehorn ~

23

(a)

(b)

(c)

Figure 22: Illustrations of the concept of dynamic arraying showing insertion of the solid phase of different specificity (a) through inlets (A, B…X), trapping of beads or bead clusters (b) and perfusion of sample (c) through inlets (1, 2…Y) followed by read-out of fluorescent response.

To be able to trap beads in an array configuration in a fluidic chamber a non-contact trapping technique is needed. Optical [41] and magnetic [42, 43] trapping methods are possible, i.e. translucent beads can be manipulated with laser tweezers, and magnets can be used to trap magnetic or paramagnetic particles. Both these techniques however put special demands on the beads, and are often associated with bulky systems not particularly suitable for array trapping. Trapping by acoustic forces is a another, very attractive approach. As discussed earlier, piezoceramic components can be made to emit ultrasonic waves, being suitable for performing acoustic manipulation of beads in microfluidic devices. The tools presented earlier in this thesis make it possible to study miniaturised solutions.


24

Tobias Lilliehorn ~

25

4 Ultrasonic trapping

4.1 Physics of acoustic particle manipulation In the 1870s, Kundt and Lehman reported that dust subjected to an acoustic standing wave was collected in orthogonal lines separated by one half wavelength of sound in the medium. The device used in the experiments is nowadays recognised as Kundt’s tube, and consists of dust enclosed in an air cylinder that set into longitudinal standing wave vibration. Despite that the phenomenon was discovered more than a century ago, acoustic particle manipulation has yet not been widely applied and implemented.

A standing wave is the superposition of two propagating waves of equal amplitude, travelling in opposite directions. Extinguishing for a standing wave is the formation of stationary nodal planes of minimum velocity or pressure amplitudes separated by half a wavelength. Equidistant between the nodal planes there are corresponding anti-nodal planes of maximum velocity or pressure amplitudes. Velocity nodes correspond to pressure antinodes and vice versa.

Figure 23: Particles moving under influence of the primary radiation force in an acoustic standing wave field. The particles gather at velocity anti-nodal planes separated by half a wavelength (λ/2) (After M. Gröschl).


26

The forces acting on particles in a standing wave field are called acoustical radiation forces and can be separated into the primary radiation forces in axial and transverse direction of sound propagation, and the secondary particle-particle interaction forces originating from scattering of the incident wave [44-46]. The effect of the primary radiation force is illustrated in Fig. 23. The forces can be considered to act to minimize the acoustic potential energy φ, i.e.

( )r)r(F vvvφ−∇= (1)

The time averaged acoustic potential energy ⟨φ⟩ of a compressible sphere with a radius much smaller than the wavelength in an arbitrary standing wave field has been derived to be [47]

( ) ( ) ( ) ( ) ⎟⎟⎠

⎞⎜⎜⎝

⎛

ρ+ρ

ρ−ρ−

β

β−β=φ rE

23

rEVr kinfp

fppot

f

pf vvv (2)

with

( ) ( )

( ) ( ) 2in

fkin

2in

fpot

rv2

rE

rp2

rE

vvv

vv

ρ=

β=

(3)

⟨Epot⟩ and ⟨Ekin⟩ are the second order approximations to the time averaged kinetic and potential energy densities from the pressure pin and velocity vin of the incident wave. The volume of the particle is denoted V, β is the compressibility and ρ the density of the fluid medium and the particle (subscripts f and p respectively). These properties of the particle in relation to the surrounding medium, together with the local acoustic field distribution determine the strength and direction of the primary radiation forces acting on the particle. For typical polymer particles in water, the force is directed towards the velocity anti-nodes, as shown in Fig. 23. This corresponds to the circumstances in Paper V and VI. The kinetic and potential energy densities in a plane standing wave are

Tobias Lilliehorn ~

27

( ) ( )( ) ( )

2inf2

ff

2in

s

2skin

2spot

p41

c4p

E

kzsinEzE

kzcosEzE

β=ρ

=

=

=

(4)

The radiation force in the direction of sound propagation can thus be calculated to (combining eq. 1, 2 and 4)

( ) ( )kz2sin2

25kEVzF

fp

fp

f

ps ⎟

⎟⎠

⎞⎜⎜⎝

⎛

ρ+ρ

ρ−ρ−

β

β−= (5)

By introducing the relations

f

p

f

p

cc

=σ

ρ

ρ=δ

(6)

the radiation force can be written as

( ) ( )kz2sin12251kEVzF 2s ⎟⎠

⎞⎜⎝

⎛+δ−δ

−δσ

−= (7)

From (eq. 5 and 7) it can be seen that the radiation force is proportional to the particle volume V (for particle radii much smaller than the wavelength) and to the frequency via the wave number, k. The discussion of radiation forces is valid for general acoustic wave propagation, but since the forces acting on particles are proportional to the frequency it is desirable to work at ultrasonic frequencies. This also makes the technique suitable for implementation in microfluidic devices, where the dimensions of the fluidic channel can be in the order of the wavelength of sound. Short acoustic path lengths are also advantageous from a sound attenuation point of view. Another advantage of increasing the frequency up to the megahertz region is to avoid cavitation [48]. Cavitation is the nucleation of gas bubbles in a fluid due to local pressure drop below the vapour pressure of the fluid, and may be detrimental when handling biological material such as cells [49].

In the general case the primary radiation force depends on acoustic energy gradients both axially and transversally to the sound propagation


28

direction. This means that when the particles have reached their nodal positions, the forces acting on them can be utilised for trapping purposes, both in the axial and the transversal direction. The existence of secondary radiation forces due to scattered waves from surrounding particles was studied by Bjerknes [45], and is therefore called Bjerknes forces. The orientation of the particles determines if these forces are attractive or repulsive. They act to attract particles lined up orthogonal to the sound propagation, but the effect is negligible [44] for particle distances of several particle radii.

4.2 Applied acoustic particle manipulation Applications of acoustic trapping or separation of particles or biological material in fluids are of particular interest to the work performed in this thesis. In streaming fluidic devices, standing waves across the passing fluid have been used to separate particles [50-53] or cells [54] from the fluid flow by using flow splitters, as shown in Fig. 24. This separation technique can be used to clarify suspensions or to increase the concentration of particles in a sample. It has also been showed to enable separation of fat from blood [55], which is useful for blood recirculation during e.g. cardiovascular surgery.

Figure 24: Separation of particles (grey) from a continuous flow by applying an acoustic standing wave at A and using flow splitters.

For cell culture fermentation of mammalian cells, where the bio-product is of high value, ultrasonic cell filters have been developed to boost the cell production rate by allowing continuous perfusion of nutrient medium [56, 57]. Primary radiation forces in the transverse direction trapped cells, creating cell aggregates that were returned to the fermentation reactor by gravitational sedimentation. By manipulation of standing acoustic waves, particle transport has been evaluated in one [58] and two dimensions [59]. Two-dimensional trapping of swimming microorganisms has also been used

Tobias Lilliehorn ~

29

to evaluate their locomotive force [60, 61]. Even large biomolecules, such as DNA, have been concentrated by local precipitation in a standing ultrasonic field [62], implying applications of the technology beyond particle manipulation.

Some bioanalytical applications making use of ultrasonics and particle trapping have been presented. Acoustic particle concentration has been used to enhance agglutination of antibody-coated latex microparticles for the detection of biomolecules [63]. A combination of ultrasonic trapping and capillary electrophoresis has been developed for the detection of trace amounts of proteins [46, 64]. In this set-up, ultrasonic evanescent waves were coupled into a fused silica capillary using an external focused ultrasonic transducer, allowing for size-selective particle separation [65]. The method is thought to allow for separation of bead-based immuno-complexes [66] of low concentration in the sample.

So far, none of the presented applications allows for individual manipulation of beads in arrays in microfluidic systems suitable for realisation of dynamic arraying as discussed earlier. Acoustic trapping by integrated miniature ultrasonic transducers in a microfluidic chamber is therefore investigated in Papers V and VI.

4.3 Integrated ultrasonic trapping Various resonator designs can be used to produce standing waves in a fluidic chamber. The resonator design we have implemented for acoustic particle manipulation is shown in Fig. 25. In this design, individually controlled ultrasonic microtransducers (c) were integrated in the wall (d) of a microchannel, with a particle-conducting fluid layer (b) and an acoustic reflector (a) on top.

Figure 25: Schematic design of investigated trapping device with glass reflector (a), particle-conducting fluid layer (c) and microtransducers (c) mounted in a baffle (d) with holes for air backing (e). The device was designed to trap particle clusters (f) in the middle of the fluid channel.


30

The fabrication of the device is presented in detail in Paper V, and the schematic assembly with an array of three ultrasonic transducers positioned at crossings in a microfluidic channel network is illustrated in Fig. 26. This shows the modular construction with a separate transducer board, a fluidic/reflector plate and a lid for clamping them together. An assembled device can be seen in Fig. 27, with a close-up on the transducer in Fig. 28.

Figure 26: Schematic assembly of a microresonator array for particle trapping showing transducer board (a), fluidic/reflector plate (b) and lid (c) for clamping. The position of one of the transducer elements is indicated in (d).

The acoustic behaviour of layered acoustic resonators for particle manipulation has been modelled [52, 67-69] based on a layered resonator model [70] limited to sound propagation in one dimension. Starting our work with miniaturised systems, we found the trapping behaviour of particles to be strongly influenced by near field effects, which earlier have been identified in some macro scale devices [71]. In analogy with diffraction of light passing a rectangular opening, the finite size of a transducer element gives rise to a complex acoustic field distribution in the direct vicinity of the transducer. These effects are particularly strong when using transducers with dimensions approaching the acoustic wavelength in the medium. The acoustic wavelength in water at 10 MHz is 150 µm and the transducers used were about 800 µm squared. To describe the trapping of particles, calculations of the acoustic field in three dimensions were therefore needed. The acoustic field in front of a transducer element, radiating into a fluid chamber without and with a reflecting plane is shown in Fig. 29. The introduction of a reflector half a wavelength from the transducer shows that pressure nodes are created in the centre of the channel, but also the existence of periodical lateral gradients in the pressure distribution.

Tobias Lilliehorn ~

31

Figure 27: Assembled device showing the transducer array through the glass examination window, electrical connections on the side and tubular fluidic connections on the backside.

Figure 28: Mounted multilayer transducer element (a) with external electrodes (b) connected to the printed circuit board (c) with conductive epoxy (d). The board was covered with epoxy (e).

(a) (b)

Figure 29: Calculated acoustic field in front of a transducer without (a) and with (b) the introduction of a reflector along the dotted line in (a).


32

(a) (b)

Figure 30: Particle suspension forming clusters above transducer element at 7.7 MHz (a) and 10.7 MHz (b).

Experiments on particle trapping were performed by injecting a suspension of 16 vol. % polyamide particles (5 µm). When subjected to the acoustic field in front of the transducer, the particles formed clusters with a structure resembling the near field acoustic pressure distribution, Fig. 30. The agreement can be seen in Fig. 31, where the periodicity in calculated pressure distribution (X) and in trapped particle clusters is compared.

Figure 31: Periodicity the calculated pressure distribution (X) compared with the periodicity in the trapped particle clusters (O).

The acoustic forces acting on particles in the vicinity of the microtransducer allowed for highly localised trapping of particles, with the lateral extension of the trapping site essentially determined by the corresponding transducer dimensions. This is the first evidence that the technique using integrated microtransducers allows for trapping of particles in arrays in a microchannel.

Tobias Lilliehorn ~

33

It was shown that, by tuning the frequency, it was possible to trap particle clusters that were kept away from the transducer and reflector surfaces within this trapping site. An image series of such particle trapping is shown in Fig. 32, where the clusters were trapped at a fluid flow of 4 µl/min (approximately 1 mm/s) by the acoustic forces, and released upon inactivation of the field (b). Poor tuning of the driving frequency resulted in an undesirable condition with particles lingering within the trapping site after inactivation of the acoustic field.

The near field effects described earlier helped in providing strong lateral trapping of the particles by the creation of periodic lateral pressure gradients within the fluid layer. These near field effects can be seen to confine the particles into laterally distributed particle clusters, Fig. 32. The response time of trapping was estimated to be 200 ms, gained by the short distances for particles to reach nodal positions in the near acoustic field.

(a) (b)

Figure 32: Trapping of particle clusters over fluid flow of 4 µl/min to the left (a), and releasing of the clusters by inactivating the transducer (b).


34

In Paper VI, the performance of bead trapping was further investigated and quantified, using monitoring of fluorescence marked beads. The position of the beads, being mainly in the centre of the channel, was confirmed by confocal imaging. The stability of the acoustic trap was evaluated by imaging the bead loss under increasing fluid flow, shown in Fig. 33. As can be seen in Fig. 34, it was found that more than 50 % of the trapped beads were still trapped at a flow of 10 µl/min, and that the flow could be increased up to about 20 µl/min before all beads were pulled from the trap. This corresponds approximately to a linear flow velocity of 4.7 mm/s within the channel. From the experiments it was concluded that the integrated miniature transducers allowed trapping of microparticles, such as beads, for bioassays in microfluidic systems.

(a) (b) (c)

Figure 33: Bead trapping at increasing fluid flow (a) 3 µl/min, (b) 11 µl/min and (c) 17 µl/min.

Figure 34: Relative bead area of trapped beads versus flow, given with standard deviation, showing approximate linear dependence.

Tobias Lilliehorn ~

35

4.4 Towards dynamic arraying A model bioassay, consisting of biotin-coated microbeads and fluorescent marked avidin was used to prove the applicability of the particle-trapping device in Paper VI. Avidin, and its bacterial counterpart, streptavidin, has since the 1970s due to its high affinity for biotin become a standard reagent for diverse detection schemes. Avidin is an egg white derived glycoprotein and biotin a vitamin essential for growth and well being in animals and some microorganisms.

Biotin-coated microbeads were injected and trapped at one of the three trapping sites in the device. A micromolar concentration (µM) solution of fluorescent-marked avidin was perfused over the trapped beads via the crossing fluidic channel, and the fluorescent response was recorded. After washing away unbound avidin solution from the trapping site, the binding of avidin to the trapped beads was seen as a significant change in relative fluorescent intensity ∆, see Fig. 35. The proof of principle experiments thus proved the successful detection of avidin, binding to the acoustically trapped bead clusters in the flow-through device.

Figure 35: Fluorescent response during the bioassay, showing a change ∆ associated with binding of avidin to the biotin on trapped beads.

Finally, the possibility of moving trapped beads between the three trapping sites in the device is shown in Fig. 36. Beads were injected and conducted to the first trapping site, where they were trapped. The bead conducting flow was maintained throughout the experiment, so when the drive signal was switched to activating the second transducer instead of the first, the beads


36

were transported and trapped at the second trapping site. The sequence of moving beads between the different trapping sites together with the performed bioassay shows the prospect of the technology and are considered to be the first steps towards the envisioned dynamic arraying.

Figure 36: Moving trapped beads from one trapping site to another, by activating the transducers in a sequential fashion. Due to long exposure time the beads appear to be smeared out when moving.

Tobias Lilliehorn ~

37

5 Concluding remarks and outlook

Fabrication techniques for miniaturised advanced multilayer piezoactuators by thick film processing have been developed, with the presentation of a new rapid prototyping process utilising high resolution electrode patterning, electrical via formation and component shaping by means of computer controlled mechanical machining. As a first prototype a multilayer telescopic actuator, not previously realised, was presented. The telescopic actuator demonstrated relatively high energy density and a stroke magnification of 5 when compared to a multilayer stack of the same height. Soft micromolding of PZT was presented for the first time, showing replication of microstructures with aspect ratio up to 7:1 with the possibility of green post processing, thus allowing integration of channel structures by adhesive assisted lamination. Well-defined voids have also been fabricated by the integration of sacrificial structures in multilayers, and these voids have been utilised as integrated acoustic reflectors in 2-D ultrasonic microtransducer arrays. The fabrication of arrays with individually addressable multilayer microelements (350 x 350 µm2) was made possible by the introduction of a new electrical via interconnection technology with 20 µm wide via trenches.

A novel multilayer actuator was designed and successfully utilised in a flow-through microdispenser for chemical sample handling. The evaluation showed it possible to dispense droplets 20-70 pl in the examined voltage range, with a maximum volumetric dispensing of 7.8 µl/min within the stable region. The actuator was, together with new building techniques, designed for mass-fabrication of dispensers and found to allow for further miniaturisation of the device. Possibilities to fabricate arrays of flow-through dispensers, suitable for parallel sample handling, are also emerging.

With the new concept of dynamical arraying, arrays of chemically functionalised microbead clusters are trapped in reusable flow-through microfluidic devices. Two important steps towards dynamic arraying have been presented. First, a new technology for acoustic particle trapping was developed, utilising piezoceramic microtransducers integrated in a microfluidic device. The importance of near field effects on particle trapping in miniaturised systems was identified, and near field pressure gradients were utilised to get strong lateral particle trapping. The technique was shown to allow for bead handling in a three element 1-D array, and the applicability


38

was shown by performing a model bioassay on trapped beads followed by fluorescent read-out. Possibilities to increasing the speed of detection are identified, and further work will also address the issue of the detection limit. The bead-based method is believed detect much lower concentrations of biochemical compounds than the ones used in these proof of principle experiments. Second, the above described fabrication technology enabled the fabrication of 2-D microtransducer arrays suitable for integration in a microfluidic array trapping device.

+ =

Figure 37: Proposed route to 2-D microbead handling for dynamic arraying.

The outlook of the work on ultrasonic trapping can thus be summarised with the “simple” mathematics proposal in Fig. 37. This illustrates the route pointed out in Papers V-VII, bringing the results from the papers together to realise 2-D microbead handling suitable for dynamic arraying. To reach this, more work is needed on the acoustic characterisation of microtransducer arrays and on building techniques for the integration of such arrays in microfluidic systems.

Tobias Lilliehorn ~

39

6 Acknowledgements

Först och främst skulle jag vilja tacka alla i PZT-gruppen vid avdelningen för materialvetenskap på Ångströmlaboratoriet, Uppsala. Min handledare Stefan Johansson för för att du trodde på ett oprövat kort och för allt du lärt mig genom åren. Urban Simu, Rickard Gustafsson, Niklas Snis och tidigare gruppmedlemmar, Lena Klintberg och Staffan Karlsson, för all hjälp, allt stöd och ert sällskap under den tid jag har varit på Ångström. Samtidigt vill jag lämna över en stafettpinne till nytillkomna Linda Johansson. Jag vill även sända ett tack alla andra trevliga typer på avdelningen som jag haft nöjet att fika och äta lunch med under åren.

Tack till avdelningens överhuvud, Jan-Åke Schwietz, till Carin Palm, Ann-Sofie Lindberg och Anette Edén samt till Jan-Åke Gustafsson och Rein Kalm. Tack till Mikael Österberg som hjälpte till att gräva ut mina dokument ur en dator som givit upp.

Mina medförfattare vid Elmät, Lunds Tekniska Högskola, vill jag tacka för ett gott samarbete under alla år; Johan Nilsson, Thomas Laurell, Monica Almqvist, Mikael Nilsson och Jonas Bergkvist. Tack till alla inblandade i mikrodispenserprojektet KOFUMA 3, vilka förutom Lundagänget var Jesper Brandt vid Svenska Keraminstitutet, Gérald Jesson och Rolf Ehrnström på Gyros Microlabs, Lars Rosengren (industrihandledare), Mårten Stjernström och andra inblandade vid f.d. Amersham Biosciences samt György Marko-Varga vid AstraZeneca. Tack också till Tadeusz Stepinski vid avdelningen för signaler och system, Uppsala Universitet för alla givande diskussioner kring ultraljud och resonatordesign. Tack till alla andra som borde varit med men som jag glömt nämna i denna sena timme. För finansiellt stöd vill jag tacka NUTEK och Stiftelsen för Strategisk Forskning (SSF).

Sist men inte minst vill jag rikta ett stort tack till hela min familj och till mina vänner (ni vet vilka ni är). Men mest av allt vill jag tacka Veronika, min fru, vars kärlek alltid har gett kraft och stöd. Du är bäst!


40

Tobias Lilliehorn ~

41

7 References

1. Uchino, K., Ferroelectric Devices. 2000, New York: Marcel

Dekker, Inc. 2. Bhalla, A.S., R. Guo, and R. Roy, The perovskite structure – a

review of its role in ceramic science and technology. Materials Research Innovations, 2000. 4(1): p. 3-26.

3. Jaffe, B., R. Roth, S., and S. Marzullo, Piezoelectric properties of lead zirconate-lead titanate solid-solution ceramics. Journal of Applied Physics, 1954. 25: p. 809-810.

4. Cross, L.E., Ferroelectric ceramics: Tailoring properties for specific applications, in Ferroelectric ceramics. Tutorial reviews, theory, processing and applications, N. Setter and E.L. Colla, Editors. 1993, Birkhäuser Verlag: Basel. p. 1-85.

5. Goldberg, R.L. and S.W. Smith, Multilayer piezoelectric ceramics for two-dimensional array transducers. IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, 1994. 41(5): p. 761-71.

6. Dogan, A., K. Uchino, and R.E. Newnham, Composite piezoelectric transducer with truncated conical endcaps "cymbal". IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, 1997. 44(3): p. 597-605.

7. Mohammadi, F., et al., High-displacement spiral piezoelectric actuators. Applied Physics Letters, 1999. 75(16): p. 2488-90.

8. Pearce, D.H., A. Hooley, and T.W. Button, On piezoelectric super-helix actuators. Sensors and Actuators A (Physical), 2002. 100(2-3): p. 281-6.

9. Muralt, P., PZT thin films for microsensors and actuators: Where do we stand? IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, 2000. 47(4): p. 903-15.

10. Bell, A.J., Multilayer ceramic processing, in Ferroelectric ceramics. Tutorial reviews, theory, processing and applications, N. Setter and E.L. Colla, Editors. 1993, Birkhäuser Verlag: Basel. p. 241-71.


42

11. Randall, C.A., et al., Intrinsic and extrinsic size effects in fine-grained morphotropic-phase-boundary lead zirconate titanate ceramics. Journal of the American Ceramic Society, 1998. 81(3): p. 677-88.

12. Shanefield, D.J., Organic additives and ceramic processing, second edition: With applications in powder metallurgy, ink and paint. 1996, Boston: Kluwer Academic Publishers.

13. Simu, U., Piezoactuators for Miniature Robots. Doctoral Thesis, Department of Materials Science, Uppsala University. 2002.

14. Bexell, M., Microfabrication and Evaluation of Piezoceramic Actuators. Doctoral Thesis, Department of Materials Science, Uppsala University. 1998.

15. Klintberg, L., Miniature Phase-Transition Actuators. Doctoral Thesis, Department of Materials Science, Uppsala University. 2002.

16. Mistler, R.E. and E.R. Twiname, Tape Casting: Theory and Practice. 2000, Westerville, Ohio: The American Ceramic Society.

17. Lubitz, K., et al., Material Aspects for Reliability and Life Time of PZT Multilayer Actuators, in Piezoelectric Materials in Devices, N. Setter, Editor. 2002, Setter, N.: Lausanne. p. 183-94.

18. Xia, Y. and G.M. Whitesides, Soft lithography. Angew. Chem. Int. Ed., 1998. 37: p. 550-75.

19. Thornell, G. and S. Johansson, Microprocessing at the fingertips. Journal of Micromechanics and Microengineering, 1998. 8: p. 251-62.

20. Madou, M., Fundamentals of Microfabrication. 1997, Boca Ranton: CRC Press.

21. Lorenz, H., et al., High-aspect-ratio, ultrathick, negative-tone near-UV photoresist and its applications for MEMS. Sensors & Actuators: A. Physical, 1998. 64(1): p. 33-9.

22. Manz, A., N. Graber, and H.M. Widmer, Miniaturized Total Chemical Analysis Systems: a Novel Concept for Chemical Sensing. Sensors & Actuators: B. Chemical, 1990. 1: p. 244-8.

23. Shoji, S., Fluids for Sensor Systems, in Microsystem Technology in Chemistry and Life Science, A. Manz and H. Becker, Editors. 1998, Springer-Verlag: Berlin.

24. Mitchell, P., Microfluidics — downsizing large-scale biology. Nature Biotechnology, 2001. 19: p. 717-21.

25. Ekstrand, G., et al. Microfluidics in a rotating CD, in microTAS 2000. 2000. Enschede, The Netherlands: Kluwer Academic Publishers.

Tobias Lilliehorn ~

43

26. Le, H.P., Progress and Trends in Ink-jet Printing Technology. Journal of Imaging Science and Technology, 1998. 42(1): p. 49-61.

27. Sziele, D., et al., Adaptation of a Microdrop Injector to Sampling in Capillary Electrophoresis. Journal of Chromatography A, 1994. 669(1-2): p. 254-258.

28. Fiehn, H., S. Howitz, and T. Wegener, New technology for the precision dosage of liquids in the range of microlitres and submicrolitres. Pharmazeutische Industrie, 1997. 59(9): p. 814-817.

29. Laurell, T., L. Wallman, and J. Nilsson, Design and development of a silicon microfabricated flow-through dispenser for on-line picolitre sample handling. Journal of Micromechanics and Microengineering, 1999. 9(4): p. 369-376.

30. Jesson, G., G. Kylberg, and P. Andersson. A versatile macro-to-micro dispensing system, in microTAS 2003. 2003. Squaw Valley, California USA.

31. Miliotis, T., et al., Capillary liquid chromatography interfaced to matrix-assisted laser desorption/ionization time-of-flight mass spectrometry using an on-line coupled piezoelectric flow-through microdispenser. Journal of Mass Spectrometry, 2000. 35(3): p. 369-377.

32. Karlsson, S., S. Johansson, and K. Boustedt. A new flexible building technique for microsystems, in IMAPS Nordic 36th Annual Conference. 1999. Helsinki, Finland: p. 133-141.

33. Schena, M. and R.W. Davis, Genes, Genomes and Chips, in DNA Microarrays: A Practical Approach, M. Schena, Editor. 1999, Oxford University Press: New York.

34. Heller, M.J., DNA Microarray Technology: Devices, Systems and Applications. Annual Review of Biomedical Engineering, 2002. 4: p. 129-53.

35. Cutler, P., Protein arrays: The current state-of-the-art. Proteomics, 2003. 3: p. 3-18.

36. Zhu, H. and M. Snyder, Protein chip technology. Current Opinion in Chemical Biology, 2003. 7: p. 55-63.

37. Oleschuk, R.D., et al., Trapping of bead-based reagents within microfluidic systems: On-chip solid phase extraction and electrochromatography. Analytical Chemistry, 2000. 72(3): p. 585-90.

38. Andersson, H., et al., Micromachined flow-through filter-chamber for chemical reactions on beads. Sensors and Actuators B (Chemical), 2000. B67(1-2): p. 203-8.


44

39. Sato, K., et al., Determination of Carcinoembryonic Antigen in Human Sera by Integrated Bead-Bed Immunoassay in a Microchip for Cancer Diagnosis. Analytical Chemistry, 2001. 73: p. 1213-8.

40. Blohm, D.H. and A. Guiseppi-Elie, New developments in microarray technology. Current Opinion in Biotechnology, 2001. 12: p. 41-7.

41. Ashkin, A., History of Optical Trapping and Manipulation of Small-Neutral Particle, Atoms, and Molecules. IEEE Journal on selected topics in Quantum Electronics, 2000. 6(6): p. 841-56.

42. Gosse, C. and V. Croquette, Magnetic Tweezers: Micromanipulation and Force Measurement at the Molecular Level. Biophysical Journal, 2002. 82: p. 3314-29.

43. Deng, T., M. Prentiss, and G.M. Whitesides, Fabrication of magnetic microfiltration systems using soft lithography. Applied Physics Letters, 2002. 80(3): p. 461-3.

44. Woodside, S.M., B.D. Bowen, and J.M. Piret, Measurement of Ultrasonic Forces for Particle-Liquid Separations. AIChE Journal, 1997. 43(7): p. 1727-36.

45. Gröschl, M., Ultrasonic separation of suspended particles. I. Fundamentals. Acustica - Acta Acustica, 1998. 84: p. 432-47.

46. Wiklund, M., Acoustic Trapping for Biomedical Analysis. Licentiate Thesis, Department of Physics, Royal Institute of Technology. 2001, Stockholm.

47. Gor'kov, L.P., On the Forces Acting on a Small Particle in an Acoustical Field in an Ideal Fluid. Soviet Physics - Doklady, 1962. 6(9): p. 773-5.

48. Gould, R.K., W.T. Coakley, and M.A. Grundy, Upper sound pressure limits on particle concentration in fields of ultrasonic standing-wave at megahertz frequencies. Ultrasonics, 1992. 30(4): p. 239-44.

49. Yasuda, K., Non-destructive, non-contact handling method for biomaterials in micro-chamber by ultrasound. Sensors and Actuators B (Chemical), 2000. B64(1-3): p. 128-35.

50. Yasuda, K., S.-I. Umemura, and K. Takeda, Concentration and fractionation of small particles in liquid by ultrasound. Japanese Journal of Applied Physics, Part 1 (Regular Papers & Short Notes), 1995. 34(5B): p. 2715-20.

51. Hawkes, J.J. and W.T. Coakley, Force field particle filter, combining ultrasound standing waves and laminar flow. Sensors & Actuators: B. Chemical, 2001. 75(3): p. 213-22.

Tobias Lilliehorn ~

45

52. Hawkes, J.J., et al., Single half-wavelength ultrasonic particle filter: predictions of the transfer matrix multilayer resonator model and experimental filtration results. Journal of the Acoustical Society of America, 2002. 111(3): p. 1259-66.

53. Nilsson, A., et al. Manipulation of Suspended Particles in a Laminar Flow, in microTAS 2002. 2002. Nara, Japan: Kluwer Academic Publishers. Vol. 2: p. 751-3.

54. Yasuda, K., et al., Using acoustic radiation force as a concentration method for erythrocytes. Journal of the Acoustical Society of America, 1997. 102(1): p. 642-5.

55. Nilsson, A., et al. Autologous blood recovery and wash in microfluidic channel arrays utilizing ultrasonic standing waves, in microTAS 2002. 2002. Nara, Japan: Kluwer Academic Publishers. Vol. 2: p. 625-6.

56. Trampler, F., et al., Acoustic cell filter for high density perfusion culture of hybridoma cells. Bio/Technology, 1994(12): p. 281-4.

57. Gröschl, M., et al., Ultrasonic separation of suspended particles. III. Application in biotechnology. Acustica - Acta Acustica, 1998. 84: p. 815-22.

58. Whitworth, G., M.A. Grundy, and W.T. Coakley, Transport and harvesting of suspended particles using modulated ultrasound. Ultrasonics, 1991. 29: p. 439-44.

59. Kozuka, T., et al., Control of a standing wave field using a line-focused transducer for two-dimensional manipulation of particles. Jpn. J. Appl. Phys, 1998. 37(5B): p. 2974-8.

60. Saito, M., S.-y. Izumida, and J. Hirota, Ultrasonic trapping of paramecia and estimation of their locomotive force. Applied Physics Letters, 1997. 71(14): p. 1909-11.

61. Saito, M., N. Kitamura, and M. Terauchi, Ultrasonic manipulation of locomotive microorganisms and evaluation of their activity. Journal of Applied Physics, 2002. 92(12): p. 7581-6.

62. Yasuda, K., M. Kiyama, and S. Umemura, DNA concentration using acoustic radiation force. Journal of the Acoustical Society of America, 1996. 99(2): p. 1248-51.

63. Sobanski, M.A., et al., Sub-micron particle manipulation in an ultrasonic standing wave: Applications in detection of clinically important biomolecules. Bioseparation, 2001. 9: p. 351-7.

64. Wiklund, M., et al., Ultrasonic-trap-enhanced selectivity in capillary electrophoresis. Ultrasonics, 2003. 41(4): p. 329-33.

65. Wiklund, M., S. Nilsson, and H.M. Hertz, Ultrasonic trapping in capillaries for trace-amount biomedical analysis. Journal of Applied Physics, 2001. 90(1): p. 421-6.


46

66. Nilsson, H., et al., Microparticles for selective protein determination in capillary electrophoresis. Electrophoresis, 2001. 22: p. 2384-90.

67. Gröschl, M., Ultrasonic separation of suspended particles. II. Design and operation of separation devices. Acustica - Acta Acustica, 1998. 84: p. 632-42.

68. Hawkes, J., et al. Positioning particles within liquids using ultrasound force fields, in Proc. Forum Acusticum. 2002. Sevilla, Spain.

69. Hill, M., Y. Shen, and J.J. Hawkes, Modelling of layered resonators for ultrasonic separation. Ultrasonics, 2002. 40(1-8): p. 385-92.

70. Nowotny, H. and E. Benes, General one-dimensional treatment of the layered piezoelectric resonator with two electrodes. Journal of the Acoustical Society of America, 1987. 82(2): p. 513-21.

71. Whymark, R.R., Acoustic field positioning for containerless processing. Ultrasonics, 1975(November): p. 251-61.

Acta Universitatis UpsaliensisComprehensive Summaries of Uppsala Dissertations

from the Faculty of Science and TechnologyEditor: The Dean of the Faculty of Science and Technology

Distribution:Uppsala University Library

Box 510, SE-751 20 Uppsala, Swedenwww.uu.se, [email protected]

ISSN 1104-232XISBN 91-554-5810-6

A doctoral dissertation from the Faculty of Science and Technology, UppsalaUniversity, is usually a summary of a number of papers. A few copies of thecomplete dissertation are kept at major Swedish research libraries, while thesummary alone is distributed internationally through the series ComprehensiveSummaries of Uppsala Dissertations from the Faculty of Science and Technology.(Prior to October, 1993, the series was published under the title “ComprehensiveSummaries of Uppsala Dissertations from the Faculty of Science”.)

piezoactuators for microfluidics

Documents