the mems handbook - eetmizsei/mikrorejegy/the mems handbook(comple… · light emitting polymer...

30Microdroplet Generators

30.1 Introduction30.2 Operation Principles of Microdroplet Generators

Pneumatic Actuation • Piezoelectric Actuation •Thermal-Bubble Actuation • Thermal-Buckling Actuation • Acoustic-Wave Actuation • Electrostatic Actuation • Inertial Actuation

30.3 Physical and Design IssuesFrequency Response • Thermal/Hydraulic Cross-Talk and Overfill • Satellite Droplets • Puddle Formation • Material Issues

30.4 Fabrication of Microdroplet GeneratorsMultiple Pieces • Monolithic Fabrication

30.5 Characterization of Droplet GenerationDroplet Trajectory • Ejection Direction • Ejection Sequence/velocity and Droplet Volume • Flow Field Visualization

30.6 ApplicationsInkjet Printing • Biomedical and Chemical Sample Handling • Fuel Injection and Mixing Control • Direct Writing and Packaging • Optical Component Fabrication and Integration • Solid Freeforming • Manufacturing Process • Integrated Circuit Cooling

30.7 Concluding Remarks

30.1 Introduction

Microdroplet generators are becoming an important research area in microelectromechanical systems(MEMS), not only because of the valuable marketing device—inkjet printhead—but also because of themany other emerging applications for precise or micro-amount fluidic control. There has been a longhistory of development of microdroplet generators ever since the initial inception by Sweet (1964, 1971),who used piezo actuation, and by Hewlett-Packard and Cannon [Nielsen et al., 1985] in the late 1970s,who used thermal bubble actuation. Tremendous research activities regarding inkjet applications havebeen devoted to this exciting field. Emerging applications in the biomedical, fuel-injection, chemical,pharmaceutical, electronic fabrication, microoptical device, integrated circuit cooling, and solid freeformfields have fueled these research activities. Thus, many new operation principles, designs, fabricationprocesses and materials related to microdroplet generation have been explored and developed recently,supported by micromachining technology.

In this chapter, microdroplet generators are defined as droplet generators generating microsizeddroplets in a controllable manner; that is, droplet size and number can be accurately controlled andcounted. Thus, atomizer, traditional fuel-injector or similar droplet-generation devices that do not offersuch control are not discussed here.

Fan-Gang Tseng National Tsing Hua University

© 2002 by CRC Press LLC

Microdroplet generators usually employ mechanical actuation to generate high pressure to overcomeliquid surface tension and viscous force for droplet ejection. Depending on the droplet size, the appliedpressure is usually greater than several atmospheres. The operation principles, structure/process designsand materials often play key roles in the performance of droplet generators.

The applications of microdroplet generators, in addition to the well-known application of inkjet printing,cover a wide spectrum of fields, including direct writing, fuel injection, solid freeform, solar cell fabrication,light emitting polymer display (LEPD) fabrication, packaging, microoptical components, particlesorting, microdosage, plasma spraying, drug screening/delivery/dosage, micropropulsion, integrated circuitcooling and chemical deposition. Many of these applications may become key technologies for integratedmicrosystems in the near future.

This chapter provides the reader with an overview of the operation principles, physical properties, designissues, fabrication process and issues, characterization methods and applications of microdroplet generators.

30.2 Operation Principles of Microdroplet Generators

Many attempts have been made to generate controllable microdroplets [Buehner et al., 1977; Twardeck,1977; Carmichael, 1977; Ashley et al., 1977; Bugdayci et al., 1983; Darling et al., 1984; Lee et al., 1984;Myers and Tamulis, 1984; Nielsen, 1985; Bhaskar and Aden, 1985; Allen et al., 1985; Krause et al., 1995;Chen and Wise, 1995; Tseng et al., 1996; Hirata et al., 1996; Zhu et al., 1996]. Most of these methodshave employed the principle of creating pressure differences, either by lowering the outer pressure orincreasing the inner pressure of a nozzle, to push or pull liquid out of the nozzle to form droplets. Typicalexamples are pneumatic, piezoelectric, thermal bubble, thermal buckling, focused acoustic-wave andelectrostatic actuations. The basic principles of those droplet generators are introduced in the followingsections. An ejection method by acceleration is also included in the last section, in which inertial forceis employed for droplet generation.

30.2.1 Pneumatic Actuation

The spray nozzle is one of the most commonly used devices for generating droplets nowadays for airbrushor sprayer applications. Two types of spray nozzles are shown in Figure 30.1. Figure 30.1a shows that theair brush generates lower pressure at the outer edge of the capillary tube by blowing air across the tube

FIGURE 30.1 Operation principle of airbrush and sprayer. (After Tseng, 1998.)

(b) Sprayer

Refilling cycle

(a) Air brush

Liquid inlet Air inlet

Airflow

Air flow

Air flowDroplet

Droplets

Spraying cycle

Liquid


end, which forces liquid to move out of the tube and form droplets. The sprayer shown in Figure 30.1bemploys high pressure to push liquid through a small nozzle to form droplets. Typical sizes of the dropletsgenerated by the spray nozzle range from tens to hundreds of microns in diameter. The device can befabricated in microsize using micromachining technology, however, it is difficult to control each nozzleseparately in an array format.

30.2.2 Piezoelectric Actuation

The droplet ejection by piezoelectric actuation was invented by Sweet in 1964. Based on the piezoelectrictechnology, there are two types of piezoelectric devices. One is called the continuous inkjet [Buehneret al., 1977; Twardeck, 1977; Carmichael, 1977; Ashley et al., 1977]. Figure 30.2a shows the operationprinciple of this type of inkjet. Conductive ink is forced out of the nozzles by pressure. The jet breaksup continuously into droplets with random sizes and spacing. Uniformity of the size and spacing of thedroplets can be controlled by applying an ultrasonic wave with fixed frequency to the ink through a piezo-electric transducer. The continuously generated droplets pass through a charge plate, and only the desiredones are charged by the electric field and deflected to printout, while the nondesired ones are collected in a

FIGURE 30.2 (a) Operation principle of a piezoelectric-actuated continuous droplet generator. (After Buehneret al., 1977.) (b) Operation principle of a piezoelectric-actuated droplet-on-demand droplet generator. (After Lee et al.,1984.)

Pump

NozzleInk

Droplet collector

DeflectionelectrodeCharge

electrode

Printsurface

Deflecteddroplets

Undeflecteddroplets

Piezoactuator

(a)

Ink

Piezoactuator

Membrane

Droplets

Nozzle

Ink supply

(b)


gutter and recycled. One piezoelectric transducer can support multiple nozzles, so the nozzle spacing can beas small as desired for high-resolution arrays, however, the complexity of the droplet charging and collectingsystem is a major obstacle to practical use of this device.

The other device, called droplet-on-demand inkjet, utilizes a piezoelectric tube or disc for droplet ejectiononly when printing a spot is desired [Bugdayci et al., 1983; Darling et al., 1984; Lee et al., 1984]. Figure30.2b shows a typical drop-on-demand drop generator. The operational principle is based on the generationof an acoustic wave in a fluid-filled chamber by a piezoelectric transducer through the application of avoltage pulse. The acoustic wave interacts with the free meniscus surface at the nozzle to eject a single drop.The major advantage of the drop-on-demand method is that a complex system for droplet deflection andcollection is not required. However, the main drawback is that the size of the piezoelectric transducer tubeor disc, on the order of submillimeters to several millimeters, is too large for high-resolution applications.It has been reported that the typical frequency for stable operation of a piezoelectric inkjet would be tensof kilohertz [Chen et al., 1999].

30.2.3 Thermal-Bubble Actuation

The thermal bubble jet was developed by Hewlett-Packard in the U.S. and by Canon in Japan in the early1980s [Nielsen, 1985; Bhaskar and Aden, 1985; Allen et al., 1985]. There are also many other designs reportedin the literature [e.g., Krause et al., 1995; Chen and Wise, 1995; Tseng et al., 1996; Tseng, 1998]. Figure 30.3shows the cross section of a thermal bubble jet. Liquid in the chamber is heated by a pulse current appliedto the heater under the chamber. The temperature of the liquid covering the surface of the heater risesto around the liquid critical point in microseconds, and then a bubble grows on the surface of the heater,which serves as a pump. The bubble pump pushes liquid out of the nozzle to form a droplet. Afterthe droplet is ejected, the heating pulse is turned off and the bubble starts to collapse. Liquid refillsthe chamber by surface tension on the free surface of the meniscus for recovery to the original position.The second pulse starts again to generate another droplet. The energy consumption for ejecting eachdroplet is around 0.04 mJ for the H-P ThinkJet printhead. Because bubbles can deform freely, thechamber size of the thermal bubble jet would be smaller than that of other actuation means, which isimportant for high-resolution applications. The resolution reported in the literature ranges from 150 to

FIGURE 30.3 Operation principle of a thermal-bubble-actuated droplet generator. (After Allen et al., 1985.)

Nozzle plate Spacer

Liquid

Main droplet

Satellite droplet

Droplet column

Vaporbubble

Substrate Heater


600 dpi [Krause et al., 1995] and 1016 dpi [Chen and Wise, 1995]. The typical operational frequency forthe contemporary thermal bubble jets is from several to tens of kilohertz.

30.2.4 Thermal-Buckling Actuation

Hirata et al. (1996) employed a buckling diaphragm for droplet generation. Figure 30.4 shows the basicoperation principle. A composite circular membrane, consisting of silicon dioxide and nickel layers, isfixed on the border with a small gap between it and the substrate. A heater is placed at the center of thecomposite membrane and electrically isolated from it. Pulsed current is sent to the heater and then themembrane is heated for several microseconds. When the thermally induced stress is greater than the criticalstress, the diaphragm buckles up abruptly and ejects a droplet out of the nozzle. The power required togenerate a droplet at a speed of 10 m/s is around 0.1 mJ for a 300-µm-diameter diaphragm. The powerconsumption and the device size of buckling membrane inkjet are much larger than those of the thermalbubble jet. The reported frequency response of membrane buckling jet ranges from 1.8 to 5 kHz,depending on the desired droplet velocity.

30.2.5 Acoustic-Wave Actuation

Figure 30.5 shows a lensless liquid ejector using a constructive interference of acoustic waves to generatedroplets [Zhu et al., 1996]. A PZT thin-film actuator with the help of an on-chip Fresnel lens wasemployed to generate and focus acoustic waves on the air–liquid interface for droplet formation. Theactuation comes from excitation of a piezoelectric film under a burst of radio-frequency (RF) signal. Thedevice does not require a nozzle to define droplets, thus reducing the troublesome clogging problemoccurring in most of the droplet generators employing nozzles. Droplet size can also be controlled byacoustic waves with specific frequencies, however, due to vigorous agitation of the acoustic wave in theliquid, it is difficult to maintain a quiet interface for reliable and repeatable droplet generation. As aresult, a “nozzle area” is still desired to maintain the interface at a stable level. The applied RF rangesfrom 100 to 400 MHz and the burst period is 100 µs. The power consumption for one droplet is around1 mJ, which is high compared to other principles. The droplet size ranges from 20 to 100 µm, depending onthe RF. The reported size of the device is 1 × 1 mm

2, which is much larger than other droplet generators

mentioned in the previous sections.

FIGURE 30.4 Operation principle of a thermal-buckling-actuated droplet generator. (After Hirata et al., 1996.)

Substrate Cooling hole

Nozzle plate

Spacer

Nozzle

Ink supply

Membrane

Electrode

Heater

Liquid

Droplet


30.2.6 Electrostatic Actuation

The electrostatically driven inkjet printhead was first introduced by Seiko–Epson Corp. [Kamisuki et al.,1998, 2000] for commercial printing purposes. As shown in Figure 30.6, the actuation beings by appli-cation of a dc voltage between the electrode plate and pressure plate to deflect the pressure plate for inkfilling. When the voltage turns off, the pressure plate reflects back to push the droplet out of the nozzle.This device was developed for use in electric calculators, as it offers low power consumption of less than0.525 mW/nozzle. The driving voltage for a SEAJet

TM is 26.5 V and the driving frequency can be up to

18 kHz with uniform ink ejection. A device with 128 nozzles/chip with 360-dpi pitch resolution has alsobeen demonstrated to have high printing quality (for bar coding), high-speed printing, low powerconsumption, a long lifetime under heavy-duty usage (lifetime more than 4 billion ejections) and lowacoustic noise. However, the fabrication comprises complex bonding processes among three differentmicromachined pieces. Besides, the pressure plate requires a very precise etching process to control theaccuracy and uniformity of the thickness. Due to the deformation limitation of solid materials andalignment accuracy required in the bonding process, the nozzle pitch may not easily be reduced furtherfor higher resolution applications.

FIGURE 30.5 Operation principle of an acoustic-wave-actuated droplet generator. (After Zhu et al., 1996.)

FIGURE 30.6 Operation principle of an electrostatic-actuated droplet generator. (After Kamisuki et al., 1998.)

Substrate

Droplet

Liquid

Membrane

PZT actuator

Fresnel lens

Pressure plate Membrane

NozzleOrifice

Bottomelectrode

Upperelectrode

Bottom glass substrate

Top glass substrateDroplet

Liquid

Reservoir


30.2.7 Inertial Actuation

Inertial droplet actuators apply high acceleration to the nozzle chip for droplet ejection. Such an apparatusis shown in Figure 30.7 [Gruhler et al., 1999]. The print module consists of large reservoirs on the topplate, which connects to the nozzles on the bottom plate. The print module is mounted on a long cantileverbeam with a piezo–bimorph actuator for acceleration generation. It requires 500 µs to generate 1 nl ofdroplets from 100-µm-diameter nozzles. Twenty-four liquid droplets of different solution types weredemonstrated to be ejected simultaneously from the nozzles in a 500-µm-pitch. The smallest dropletclaimed to be generated is 100 pl from 50-µm-diameter nozzles. This principle provides a gentle ejectionprocess for bioreagent applications. However, the ejection of smaller droplets may encounter strong surfacetension and flow drag forces at the microscale that are much larger than the droplet inertial. Also, thedroplets cannot be selectively and individually ejected from the desired nozzles which limits its applications.

30.3 Physical and Design Issues

The sequence of droplet generation involves wide ranges of physical issues and design concerns, includingmicrofluid flow, heat transfer, wave propagation, surface properties, material properties and structurestrength. The following sections discuss frequency response, thermal cross-talk, hydraulic cross-talk, over-fill, satellite droplets, puddle formation and material issues commonly seen in microdroplet generators.

30.3.1 Frequency Response

The frequency response of droplet generators is an important measure for assessing the performance ofa device. The reported typical frequency response for thermal bubble jets, piezo jets, thermal bucklingjets, acoustic wave jets and electrostatic and inertial jets ranges from a kilohertz to tens of kilohertz. Piezoand acoustic wave jets typically have higher frequency responses than the others. Recently, the novel designby Tseng (1998) improved the frequency response of thermal bubble jets by about three times (35 kHz),which is comparable to the speed of piezo jets. Higher speed devices (in the range of hundreds of kilohertz)are currently under development by major inkjet printer makers and further breakthroughs will follow inthe next few years.

The three important time constants related to the frequency response of microdroplet generators areactuation, droplet ejection and liquid refilling, as shown in Figure 30.8.

The typical time constant from heating to bubble formation in a thermal bubble jet is around 5 to10 µs for a chamber size ranging from 20 to 100 µm [Tseng, 1998]. Thermal buckling and electrostaticmicrodroplet generators have actuation times of tens to hundreds of microseconds (estimated from Hirataet al., 1996), due to the large actuation plate required for sufficient displacement for droplet formation.The inertial type required even longer actuation time (estimated from Gruhler et al., 1999), typicallyhundreds to thousands of microseconds, due to the large cantilever structure necessary for generatinglarge droplets with sufficient inertial force to overcome liquid surface tension and viscosity. The actuation

FIGURE 30.7 Operation principle of an inertia-actuated droplet generator. (After Gruhler et al., 1999.)

NozzlePrint moduleReservoirPiezo bimorphactuator

Droplets


time for piezo or acoustic wave types of generators may be shorter [Darling et al., 1984; Zhu et al., 1996],around one microsecond to tens of microseconds.

After application of actuation pressure to the liquid, the droplet starts to eject. The ejection sequenceusually takes a couple to hundreds of microseconds for a droplet volume of 1 pl to 1 nl [Tseng, 1998],which does not vary much for different operation principles.

The liquid refills back automatically by surface tension force after the droplet ejection. Liquid refillingtime can vary by three orders of magnitude (e.g., from less than 10 µs to over 1 s), depending on thelength and geometry of the refilling path. In most of the commercial inkjet printhead designs, a chamberneck [Nielsen, 1985], elongated chamber channel [Neilsen, 1985] or physical valve [Karz et al., 1994] isused to prevent hydraulic cross-talk and maintain a high pressure in the firing chamber. However, thosedesigns, if not arranged properly, may greatly increase the refilling time, causing a reduction in the frequencyresponse. As a result, how to prevent the hydraulic cross-talk without sacrificing the device speed becomesan important issue in the design of super-high-speed and high-resolution droplet generators. Tseng et al.(1998a; 1998b; 1998c) introduced the concept of a virtual chamber neck to speed up the refilling processof thermal bubble jets and to suppress cross-talk. The virtual neck consists of vapor bubbles that providesealing pressure while the droplet is ejecting and opens up to reduce flow resistance while liquid is refilling,thus increasing the frequency response. The concept is shown in Figure 30.9.

For simulation of droplet actuation and the formation sequence, much work [Curry and Portig, 1977;Lee, 1977; Levanoni, 1977; Pimbley and Lee, 1977; Fromm, 1984; Bogy and Talke, 1984; Asai et al., 1987;1988; Asai, 1989; 1991; 1992; Mirfakhraee, 1989; Chen et al., 1997a; 1997b; 1988a; 1988b; 1999; Rembeet al., 2000] has been conducted on the bubble formation sequence, droplet generation, and aerodynamicsof droplets traveling in the atmosphere for thermal bubble as well as piezo jets. Interested readers canrefer to those works for details.

30.3.2 Thermal/Hydraulic Cross-Talk and Overfill

When nozzle pitch becomes small, two types of cross-talk—hydraulic and thermal (for thermal bubblejet)—become significant in multiple-nozzle droplet generators. Hydraulic cross-talk relates to the transpor-tation of a pressure wave from the firing chamber to the neighboring chambers, as shown in Figure 30.10.The vibration of the meniscus of the neighboring chambers may result in poor droplet volume control

FIGURE 30.8 The sequence of droplet generation.

a) Actuation

b) Droplet formation

c) Droplet ejection

d) Liquid refills Satellite droplets


or, even worse, unexpected droplet ejection. Thermal cross-talk, which only appears in the thermal bubblejet, is the phenomenon of thermal energy transportation from the firing chamber to neighboring cham-bers, resulting in poor droplet volume control. After droplet ejection, the refilling process of liquidsometimes causes meniscus oscillation, posing another issue of overfill. Overfill, similar to cross-talk,increases the waiting time for the next droplet ejection and even causes undesired droplet ejection. Thephenomenon of overfill is illustrated in Figure 30.11.

FIGURE 30.9 Operation principle of virtual chamber neck. (After Tseng et al., 1998c.)

FIGURE 30.10 Cross-talk among neighboring microchambers.

FIGURE 30.11 Overfill phenomenon.

Virtual chamber neck

Bubble collapses& liquid refillsDroplet ejection

Liquidrefills

(d)

(b)

Si

ChamberManifold

liquid(a)

Liquid heating

Narrow heater(manifold side)

Wide heater(chamber side)

Virtual chamberneck formation

(c)

Thermal cross-talk

bubble

Firing chamber

Droplet ejection

Hydraulic cross-talkMeniscusoscillation


These issues stem from the fact that there is not enough flow compliance among nozzles. One of thesolutions developed by IBM [Nielsen, 1985] is to increase the channel length for each chamber. However,the increased serial compliance of the lengthened channel between the reservoir and chamber increasesthe flow resistance and inertia significantly, increasing the liquid refilling time.

HP tried to solve this problem by using either a parallel compliant (reservoir) or a chamber neck.Figure 30.12 shows a slot used as a reservoir beside the nozzles to store energy while the bubble isexploding and to release energy while the bubble is collapsing [Nielsen, 1985]. Figure 30.13 shows thesecond approach by narrowing the inlet of the chamber to form a chamber neck [Buskirk et al., 1988].

The effects from the aforementioned methods were simulated by Buskirk et al. (1988). Figure 30.14shows the simulation results of the meniscus position of the firing chamber and neighboring chamberfor various chamber designs. In the first figure, without any design of flow compliance, the meniscuseson both the firing chamber and the neighboring chamber show huge fluctuations. Doubling the channellength (series compliance) damped down some of the fluctuation, but not completely. Only the chamberneck design almost completely damped down the meniscus fluctuation; however, it has the slowest risingresponse among the three methods.

In contrast to fixed chamber neck design, Xerox [Karz et al., 1994] has used a flexible plate as a valveto address the cross-talk issue. However, this design may suffer from low frequency response and material

FIGURE 30.12 Method of applying parallel compliant (reservoir) to overcome cross-talk.

FIGURE 30.13 The cross-talk overcome by employing a chamber neck.

Fluid flowduring firing

Slot

Nozzle plate

Manifold Spacer Nozzle


reliability problems. Tseng et al. (2000) used a novel design of a virtual chamber neck, employing thebubble as a virtual valve to reduce the cross-talk problem while maintaining the high frequency responseof the droplet generator, as shown in Figure 30.9. Their work demonstrated that the bubble has fasterresponse and more reliable operation performance in the microscale than do solid valves.

In addition to hydraulic cross-talk, thermal cross-talk, in Tseng’s work, was also reduced by placingthe heater on the chamber top of a thin film with low thermal conduction, instead of leaving it on thethick substrate through which heat can be conducted to the neighbors [Tseng et al., 1998c; 2001a]. It isclear that as the nozzle pitch becomes closer, the cross-talk problem becomes more severe in the operationof very-high-resolution and high-speed droplet generators.

30.3.3 Satellite Droplets

Satellite droplets result from the breakdown of the long ejected liquid column by the interaction amongsurface tension, air drag force and inertial force. The velocity mismatch along the liquid column, resultingfrom the variation of actuation velocity, promotes the breakdown. The droplet ejection sequence capturedfrom a commercial inkjet reveals the detailed steps of satellite droplet formation, as shown in Figure 30.15[Tseng et al., 1998c]. When applied to printing, the quality is degraded from the occurrence of satellitedrops, as revealed in Figure 30.16. Satellite droplets also reduce the accuracy for precise liquid dispensingcontrol.

A literature survey reveals that there have been many attempts to predict the droplet formation. Asaiet al. (1987; 1988; Asai, 1989; 1991; 1992) conducted both numerical simulation and experimental mea-surements to obtain the temporal variation of droplet length at a thermal bubble jet. However, theformation process of satellite droplets is not included. In the drop-on-demand inkjet, Fromm (1984)

FIGURE 30.14 Meniscus oscillation in various chamber structures. (Data from Buskirk et al., 1988.)

Original design

timeFiring chamber

Neighboring chambertime

Meniscusposition

Double channel length

Meniscusposition

time

time

time

time

Firing chamber

Neighboring chamber

Meniscusposition Firing chamber

Neighboring chamber

With chamber neck


FIGURE 30.15 Droplet ejection sequence of HP 51626A printhead. (After Tseng et al., 1998c.)

FIGURE 30.16 A printed vertical line smeared by satellite droplets. (After Tseng et al., 1998c.)


and Chen et al. (1997a; 1997b; 1998a; 1998b; 1999) solved the Navier–Stokes equation to predict thedroplet formation, and showed the process of satellite droplet formation. Pimbley and Lee (1977) andChen et al. (1997b) demonstrated the evolution of the droplet as well as satellite droplet formation byflow visualization, however, the detailed method for eliminating the separation of satellite droplet fromthe main droplet was not discussed.

Various efforts have been made to eliminate satellite droplets from the commercial product. In piezodroplet generators, triangular waves were used to eliminate satellite drops [Chen et al., 1999]. For thermalbubble jets, Tseng et al. (1998a; 1998c; 2000) proposed a novel method employing the bubble as a trimmerto cut off the long droplet tail to eliminate satellite droplets, as shown in Figure 30.17. The tail of theliquid column, shortened by the bubble trimmer in Tseng’s work, is drawn back by surface tension intothe main droplet, thus eliminating satellite drops.

30.3.4 Puddle Formation

A liquid puddle forms when liquid is pushed to flow outward and accumulates on the outside surfaceof the nozzle. Puddles impose a great blocking force on droplet ejection, causing distortion or evendisruption of droplet ejection. One of the major causes of puddle formation is the hydrophilic nozzlesurface. When it comes into contact with working fluid from the chamber, the liquid accumulates onthe outer surface of the nozzel. It was observed by (Tseng, 1998) that the puddle appears after severalcontinuous operations; that is, the chamber surface does not acquire a puddle until getting wet afterseveral runs. If the operation of droplet ejection stops, the puddle is drawn back into the chamber bysurface tension. However, once the chamber surface becomes wet, puddle formation always occurs whenoperation starts again. Figure 30.18 shows the formation of a puddle during the running of a microinjector[Tseng, 1998]. Notice that the droplet ejection position is away from the nozzle, due to distortion by theliquid puddle.

One way to eliminate puddle formation is to coat the chamber outer surface with a nonwetting material toprevent the wetting process of the working fluid, the inner surface of the chamber must remain hydrophilic

FIGURE 30.17 Droplet ejection without satellite droplets. (After Tseng et al., 1998a.)


for liquid refill. Even with the coating, however, there is still no guarantee that the puddle will not form.More research is under way to fully understand the mechanism of the puddle formation process.

30.3.5 Material Issues

Material issues, including stress, erosion, durability and compatibility, are very complex issues in thedesign of microdroplet generators. Material compatibility, stress and durability problems are commonlydiscussed from the processing perspective. Material compatibility issues result from the processing tem-perature, processing environment (oxidation, reactive gas, etc.), etching method used and adhesionability; stress issues are usually a result of processing temperature as well as doping conditions; materialdurability issues either are material intrinsic properties or are due to the mechanical forces inducedduring the process (i.e., fluid flow force, surface tension force, vacuum forces or handling force). Lots ofcare needs to be taken in the process flow design to eliminate the material issues, such as compensatingmaterial stress during or after the fabrication process, conducting the high-temperature process earlierthan the low-temperature material, finishing aggressive wet etching before metal film deposition or usinglow-temperature bonding material and processes to protect integrated circuit and microdevices.

From the operation aspect, durability, stress and erosion issues are the major concerns. Due to thecycling nature of the droplet generation process, the materials chosen for actuation face challenges notonly from stress but also from fatigue. HP reported that possible reasons for failure of the heaterpassivation materials are cavitation and thermal stress [Bhaskar and Aden, 1985]. Silicon, low-stresssilicon nitride, silicon carbide, silicon dioxide and some metals are usually used to overcome the afore-mentioned problems. In addition to proper material selection, reducing sharp corners in the design isalso an important key to preventing the material from cracking by eliminating stress concentration points.Moreover, the erosion of structure materials from the working fluid is another serious issue. Lee et al.(1999) reported the erosion of spacer material in a commercial inkjet head when diesel fuel was used asthe working fluid. In contrast, materials, including silicon and silicon nitride, used by Tseng et al. [Tsenget al., 1998c; Tseng, 1998] and Lee et al. (1999b) in the microinjector are free of this problem and can beapplied to a wide variety of fluids, including solvents and chemicals. Selecting materials wisely, arrangingthem correctly in process and properly designing the materials in the microstructures are the threeimportant concepts in reducing material issues.

FIGURE 30.18 Liquid puddle formation outside the micro nozzles. (After Tseng, 1998.)


30.4 Fabrication of Microdroplet Generators

The common structures in most of the microdroplet generators include a manifold for liquid storage,microchannels for liquid transportation, microchambers for holding liquid, nozzles for droplet size anddirection definition and actuation mechanisms for droplet generation. Occasionally, droplet generatorsmay not have nozzles but instead use energy focusing as a means to generate droplets locally, such asacoustic wave droplet generators [Zhu et al., 1996]. Before micromachining processes became popular,most of the fabrication processes of microdroplet generators stemmed from the same concept: Nozzleplates, fluid-handling plates and actuation plates are manufactured separately and then integrated intoone final device. However, as the nozzle resolution becomes finer, bonding processes pose severe alignment,yield, material and integrated-circuit compatibility problems. Also, interconnection lines may not haveenough space to pull out individually from each chamber when nozzle resolution is higher than 600 dpi.As a result, monolithic ways to fabricate high-resolution integrated circuit droplet generators have becomevery important. In the following sections, examples of different fabrication means are introduced.

30.4.1 Multiple Pieces

Figure 30.19 shows the traditional fabrication of microdroplet generators by the bonding of fabricatedstructure pieces [Tseng, 1999]. Actuation plates are fabricated separately from the nozzle plates. In thethermal bubble jet, heaters are usually sputtered or evaporated and then patterned with an integratedcircuit on the bottom plate, while piezo, thermal buckling, electrostatic and inertial actuators consist of

FIGURE 30.19 Conventional fabrication process flow of microdroplet generators.

b3: Spacer pattern

Nozzle plate and actuatorplate bonding

a4: Nozzle plate de-mold

b2: Actuator fabrication

b1: Substrate passivationa1: Metal mold fabrication

a2: Sacrificial film pattern

a3: Nozzle plate plating


more complex structures, such as piezo disc, thin plate structure, or cantilever beam. Parallel to the actuatorfabrication, nozzles are fabricated by electroforming [Ta et al., 1988], molding or laser drilling [Keefe et al.,1997]. The combination of those processed pieces is assembled either by using polymer spacer materialas intermediate layers [Siewell et al., 1985; Askeland et al., 1988; Hirata et al., 1996; Keefe et al., 1997] orby directly adhering several pieces through anodic bonding [Kamisuki et al., 1998; 2000], fusion bonding[Gruhler et al., 1999], eutectic bonding or low-temperature chemical bonding. However, most of thebonding means are chip-level not wafer-level processes and face the similar challenges of alignment,bonding quality and material/process compatibility. As the nozzle resolution becomes higher than 600 dpi,alignment accuracy of 4 µm (10% of the nozzle pitch) is difficult to achieve. Higher alignment accuracysignificantly increases the fabrication cost, especially for the chip-level process. Bonding quality is anotherimportant issue corresponding to the fabrication yield of large-array and high-resolution devices. Thebonding materials (mostly polymers) chosen may not be suitable for the application environments andworking fluids. Also, the bonding process, involving heat, pressure, high voltage or chemical situations,restricts integrated circuit integration with the droplet generators which is essential for large-array andhigh-resolution applications.

30.4.2 Monolithic Fabrication

In order to address the aforementioned issues, monolithic processes utilizing micromachining technologyhave been widely employed since the early 1990s. Two major methods have been introduced: One is thecombination of bulk and surface micromachining and the other is the use of bulk micromachining anddeep-ultraviolet (UV) lithography associated with electroforming (or UV lithography only).

For example, Tseng, (1998) combined surface and bulk micromachining to fabricate a microdropletgenerator array with potential nozzle resolution up to 1200 dpi (printing resolution can be 2400 dpi orhigher). In this design, double bulk-micromachining processes were utilized to fabricate a fluid-handlingsystem that included a manifold, microchannels and microchambers. Surface micromachining was usedfor heater, interconnection line and nozzle fabrication. The whole process was finished on (100)-crystal-orientation silicon wafers. Figures 30.20 and 30.21 show the three-dimensional structure of the microinjec-tors and the monolithic fabrication process, respectively. The ejection of 0.9-pl droplets has also beendemonstrated by Tseng et al. (2001b) for high-resolution microinjectors. The structure materials usedin the microinjector are silicon silicon nitride and silicon oxide, durable in high temperature and suitablefor various liquids (even some harsh chemicals). Integrated circuits can be easily integrated with thisdevice on the same silicon substrate.

The second case can be found in Lee et al.’s (1999a) work. In this design, the multi-exposure and signaldevelopment (MESD) lithography method was used to define microchannel and microchamber structures

FIGURE 30.20 Schematic three-dimensional structure view of microinjectors. (After Tseng et al., 1998c.)

Wideheater

NarrowheaterElectrodeManifold Nozzle

Commonline

Liquid

LiquidEntrance

Chamber


(using a photoresist as the sacrificial layer), and the physical structures were constructed by electroformedmetal. The manifold was manufactured from the wafer backside by electrochemical methods [Lee et al., 1995].This device demonstrated compatibility with the very-high-resolution array and integrated circuit process.Another method, using a photoresist as the sacrificial layer and polyimide as the structure layer, wasintroduced by Chen et al. (1998b) for applications compatible with high-resolution and integrated circuitapplications.

30.5 Characterization of Droplet Generation

Droplet trajectory, volume, ejection direction and ejection sequence/velocity are four important quan-titative measures for assessing the ejection quality of microdroplet generators. The following sectionsbriefly introduce the basic methods for the testing of droplet generation.

30.5.1 Droplet Trajectory

Droplet trajectory visualized by utilizing a flashing light on the ejection stream, as shown in Figure 30.22,was introduced by Tseng et al. (1998a). The white dots in Figure 30.23 indicate the droplet stream. Thevisualized droplet trajectory follows an exponential curve, very different from the parabolic one expectedfor normal-size objects with similar initial horizontal velocity. The droplet trajectory was also estimatedby Tseng et al. (1998a) by solving a set of ordinary differential equations from the force balance, in bothhorizontal and vertical direction, of a single droplet flying through air.

FIGURE 30.21 Fabrication process flow of monolithic micro injectors. (After Tseng et al., 1998c.)

Si

A A’

Cross sectionA-A'

PSG & low-stress-nitride deposition

(a)

Top view

KOH etch manifold, PSG removed

KOH etch enlarged chamber depth

(b)

(c)

(d)

Heater & interconnection formation

(e)

Nozzle formation & pad open


From the analysis, the vertical position Y and horizontal position X of the droplet can be expressed by:

(30.1)

(30.2)

where g is the acceleration due to gravity, t is the time, m is the mass, r0 is the radius of the droplet, µ isthe viscosity of air, is the droplet terminal velocity, and UH0 is the initial horizontal velocity.

FIGURE 30.22 Experimental setup for droplet stream visualization.

FIGURE 30.23 Visualized droplet stream and estimated trajectory of a microdroplet generator. (After Tseng et al.,1998a.)

Microinjector

Droplet trajectory

CCD camera

Flashlight

Y Uv∞ tUv∞

g--------- 1 e

6πµr0–

m------------------ t

–

–=

XUH0m6πµr0

-------------- 1 e

6πµr0–

m------------------ t

–

=

Uv∞mg

6πµr0

---------------=


The trajectory is drawn in Figure 30.23 with the experimental result and fits the visualized trajectorywell except at the end, suggesting an interaction among droplets. From this simple analysis, the maximumflying distance of a droplet with a known diameter can be estimated as:

(30.3)

Here, the maximum distance is proportional to the droplet velocity and droplet radius to the secondpower. For droplets of varying sizes and the same initial velocity, the maximum flying distance is reducedquickly for the smaller ones. To obtain 1-mm flying distance, a droplet with 10-m/s initial velocity musthave a minimum radius of 2.7 µm. From the above estimation, droplet size should be maintained beyonda certain value to ensure enough flying distance for printing. Printing with very fine droplets (diametersmaller than a couple of micrometers) requires either increasing the initial velocity of the droplets orprinting in a special vacuum environment to overcome the resistant force from air drag.

30.5.2 Ejection Direction

Droplet direction can be decided by the visualized trajectory. Many parameters, including nozzle shape,roughness, aspect ratio and wetting property, as well as actuation direction and chamber design, affectdroplet direction. In general, symmetric structure design and accurate alignment can help control dropletdirection.

30.5.3 Ejection Sequence/velocity and Droplet Volume

To characterize the detailed droplet ejection sequence, a visualization system [Chen et al., 1997b; Tsenget al., 1998c], as shown in Figure 30.24, has been widely used. In this system, an LED was placed underthe droplet generator to back-illuminate the droplet stream. Two signals, synchronized with adjustabletime delay, were sent to a microinjector and an LED, respectively. Droplets were ejected from a dropletgenerator continuously, and the droplet images were frozen by the flashlight from the LED at specifiedtime delays, as shown in Figure 30.15. Droplet volume can be determined from the images by assumingthe droplet is axisymmetric or by weighing certain numbers of droplets. Droplet velocity can be estimatedby measuring the flying distance difference of the droplet fronts in two successive images.

FIGURE 30.24 Experimental setup for droplet ejection sequence visualization.

Xmax

UH0m6πµr0

--------------2ρliquid

9µair

--------------- UH0r02( ), when t ~ ∞= =

CCDCamera

VCR

LED

Microscope

Microinjector

Signal A

Signal B

Synchronizer


30.5.4 Flow Field Visualization

To better understand flow properties such as cross-talk, actuation sequence, liquid refill and dropletformation inside microdroplet generators, flowfield visualization is one of the most direct and effectiveways. Flow visualization at a small scale has some difficulties that do not occur at a large scale, such aslimited viewing angles, hard-to-apply light sheet reflection from the particles trapped on the wall, shortresponse time and small spatial scale. Meinhart and Zhang (2000) adopted a micrometer-resolutionparticle image velocimetry system to measure instantaneous velocity fields in an electrostatically actuatedinkjet head. In the setup, 700-nm-diameter fluorescent particles were introduced for flow tracing. Thespatial as well as temporal resolutions of the image velocimetry were found to be 5 to 10 µm and 2 to 5 µs,respectively. The four primary phases of the injection operation, including infusion, inversion, ejectionand relaxation, were clearly captured and quantitatively analyzed.

30.6 Applications

More than a hundred applications have been explored employing microdroplet generators. This sectionprovides a summary of some of the applications.

30.6.1 Inkjet Printing

Inkjet printing involves arranging small droplets on a printing medium to form texts, figures or imagesand is the most well-known application. The smaller and cleaner the droplets are, the sharper the printingis, however, smaller droplets cover a smaller printing area and thus increase the printing time. Therefore,in printing applications, high-speed microdroplet generation with stable and clean microsized dropletsis desired for fast, high-quality printing. The printing media can be paper, textile, skin, cans or othersurfaces that can adsorb or absorb printing solutions. Inkjet printing generated a more than $10 billionworldwide market in 2000 and continues to grow.

30.6.2 Biomedical and Chemical Sample Handling

The application of microdroplet generators to biomedical sample handling is an emerging field and isdrawing much attention. Much research effort has been focused on droplet volume control, droplet sizeminiaturization, compatibility issues, variety of samples and high-throughput parallel methods.

Luginbuhl et al. (1999), Miliotis et al. (2000) and Wang et al. (1999) developed piezo- and pneumatic-type droplet injectors, for mass spectrometry. Figure 30.25 shows the design of the injectors, which areutilized to generate submicron- to micron-sized bioreagent droplets for sample separation and analysisin a mass spectrometer, as shown in Figure 30.26. Lugnbuhl et al. (1999) employed silicon bulk micro-machining to fabricate a silicon nozzle plate and Pyrex® glass actuation plate, while Wang et al. (1999)

FIGURE 30.25 The injector design for mass spectrometry. (After Luginbuhl et al., 1999).

GlassPZT disc transducer

Silicon Pressure chamber Nozzle Liquid path


employed the combination of surface and bulk micromachining to fabricate a droplet generator. Theseinjectors are part of the lab-on-a-chip system for incorporating a microchip with a macro-instrument.

Microdroplet generators were also used by Koide et al. (2000), Nilsson et al. (2000), Goldmann andGonzalez (2000) and Szita et al. (2000), for the accurate dispensing of biological solutions. Piezo- andthermal-type injectors were used in this research for protein, peptide, enzyme or DNA dispensing. Inthose applications, the operation principles of the employed devices are similar to those for inject printing.A single biological droplet can be precisely dispensed and deposited onto a desired medium, and thedispensing of droplet arrays can also be carried out. The arrayed bioreagents can be further bioprocessedfor high-throughput analysis.

Continuous-jet-type droplet generators were reported by Asano et al. (1995) to effectively focus andsort particles by electrostatic force. The experimental setup is shown in Figure 30.27. A syringe pumppressurizes the sample fluid containing the particles, which pass through a nitrogen sheath flow forfocusing. The sample is then ejected from a piezoelectric-transducer-disturbed nozzle to form a droplet.The droplets containing the desired particles are charged at the breakup point and deflected into collectors.The reported separation probability for 5-, 10-, and 15-µm particle can be as high as 99%. However, the

FIGURE 30.26 Operation principle of mass spectrometry using microdroplet generators.

FIGURE 30.27 Particle sorting using droplet generators. (After Asano et al., 1995.)

Droplet generators

Droplets

MSInlet

Laser

Sample

Piezo vibratorPressured N2

PMT tube

Charge plate

Deflection plate

Dropletswith particles

Collectors


inner jet diameter limits the particle size for separation. Other than solid particle separation, this methodcan potentially also be applied to cell sorting for biomedical applications.

In addition to biomedical reagent handling, microdroplet generators have been widely used in chemicalhandling. For example, Shah et al. (1999) used an inkjet to print catalyst patterns for electroless metaldeposition. In their system, a Pt solution was employed and ejected by a commercial inkjet printhead asa seed layer for Cu electroless plating. The lines produced by this method were reported to be 100 µmwide and 0.2 to 2 µm high.

30.6.3 Fuel Injection and Mixing Control

Microdroplet generators can be used for fuel injection to dispense controllable and uniform droplets,important for mixing and combustion applications. Combustion efficiency depends on the mixing rateof the reactants. The reactants in a shear flow are first entrained by large vortical structures (Brown andRoshko, 1974) and then mixed by fine-scale eddies. The entrainment can be greatly enhanced by controllingthe evolution of large-scale vortices either actively (Ho and Huang, 1982) or passively (Ho and Gutmark,1987). The effectiveness of controlling large-scale vortical structures by increasing the combustion effi-ciency has also been experimentally demonstrated (Shadow et al., 1987). Although much work has beendone on improving the mixing efficiency in combustion chambers, a significant challenge in combustionresearch is to improve the small-scale mixing and to reduce the evaporation time of the liquid fuel.

Traditional injectors with a nozzle diameter of tens to hundreds of microns can neither supply uniformmicrodroplets for reducing evaporation time and fine-scale mixing nor eject droplets which can becontrolled individually to modulate vortex structure [Lee et al., 1999b]. To overcome those limitations,Tseng et al. (1996) proposed a microdroplet injector array fabricated using micromachining technologiesfor fuel injection. The droplets ejected from microinjectors are uniform and the size can be one to tensof microns in diameter, which is close to the microscale of low-turbulence eddies. The fine-scale mixing canbe carried out by the reaction of the low-turbulence eddies directly with the microdroplets. The evaporationtime is also greatly reduced by increasing the evaporation surface from the reduced and unformed dropletsize. In addition, appropriate selection of microinjectors distributed around the nozzle of a dumpcombustor (Figure 30.28) provides spatial coherent perturbations to control the large vortices. Two typesof coherent structures (i.e., spanwise and streamwise vortices) can be influenced by imposing subhar-monics of the most unstable instability frequency of the air jet. Control of the spanwise vortices can beaccomplished by applying temporal amplitude modulation on the injection. If the ejecting phases of the

FIGURE 30.28 Control of mixing and fuel injection by microdroplet generators. (After Tseng et al., 1996.)

Fuel dropletinjections

Coherentflow structure

Microinjectors

Air

Acousticdriver

Controller

Fuel


microdroplets along the azimuthal direction are the same, the mode zero instability (Brown and Roshko, 1974)is enhanced. When a certain defined phase lag is imposed on these microinjectors, higher mode instability(Brown and Roshko, 1974) waves are generated which are usually beneficial for mass transfer enhance-ment. Because about a thousand injectors are placed around the nozzle, the spatial modulation in theazimuthal direction can perturb the streamwise vortices. The interaction of streamwise and spanwisevortices by microinjectors brings forth fine-scale mixing.

30.6.4 Direct Writing and Packaging

Microdroplet generators offer an alternative to the lithography process for electronics and optoelectronicsmanufacturing. This approach has the advantages of precise volume control of dispensed materials, data-driven flexibility, low cost, high speed and low environmental impact, as mentioned by Hayes and Cox (1998).Materials used with this process include adhesives for component bonding, filled polymer systems for directresistor writing and oxide deposition and solder for solder-bumping of flip-chip ball grid arrays (BGAs),printed circuit boards (PCBs), and chip-scale packages (CSPs) [Teng et al., 1988; Hays et al., 1999]. In theseprinting applications, the temperature should be elevated to 100 ∼ 200°C, and the viscosity of fluidsshould be around 40 cps; in some cases, an inert process environment, such as nitrogen flow, is requiredto prevent oxidation of the materials.

Direct writing by inkjet printing can eliminate the fabrication difficulty inherent in photolithographyor screen-printing processes for solar-cell metallization and light-emitting polymer (LEP) deposition oflight-emitting polymer displays (LEPDs). In solar-cell metallization, metallo-organic decomposition(MOD) silver ink is used to inkjet print directly onto solar-cell surfaces for avoidance of p–n junctiondegradation under the traditional screenprinting method requiring 600 to 800°C for firing process. Inkjetprinting also allows formation of a uniform line of film on rough solar-cell surfaces [Tang and Vest,1988a; 1988b; Somberg, 1990] which is not easily achieved by traditional photolithography.

Organic light-emitting devices that require deposition of many organic layers to perform full-coloroperation face similar problems. Due to the solubility of those organic layers in many solvents and aqueoussolutions, conventional methods such as photolithography, screen printing and evaporation, whichrequire a wet patterning process, are not suitable [Hebner et al., 1998; Shimoda et al., 1999; Kobayashiet al., 2000]. Thus, direct writing of organic materials by inkjet printing has become a promising solutionto provide a safe, patternable process without a wet etching procedure. However, due to pinholes appearingon the patterned materials, high-quality polymer devices may not be easily inkjet printed. Yang’s groupproposed a hybrid way of combining an inkjet-printed layer with another uniform spin-coated polymerlayer to overcome the problem [Bharathan and Young, 1998]. In such a system, the uniform layer servesas a buffer to seal the pinholes and the inkjet-printed layer contains the desired patterns [Bharathan andYoung, 1998].

30.6.5 Optical Component Fabrication and Integration

Integrated microoptics has become a revolutionized concept in the optics field, as it provides the advan-tages of low cost, miniaturization, improved spatial resolution and time response and reduction of theassembly process on optical systems, which is not possible by traditional means. As a result, fabricatingand integrating miniaturized optical components with performance similar or even better than traditionalcomponents are critical issues in integrated microoptics systems. Standard bulk or surface micromachin-ing provides various ways to fabricate active/passive micromirrors, wave-guides and Fresnel lenses, butfabrication of refractive lenses with curved surfaces is not easy. Compared to photolithography, whichutilizes patterned and melted photoresist columns as lenses, the inkjet printing method provides moreflexibility as far as the process, material choices and system integration. Cox et al. [Cox et al., 1994; Haysand Cox, 1998] employed inkjet printing technology to eject heated polymer material to fabricate amicrolenslet array. The shape of the lens was controlled by the viscosity of the droplets at the impactpoint, the substrate wetting condition and the cooling/curing rate of the droplets [Hays et al., 1998]. A70- to 150-µm-diameter lens with a density greater than 15,000/cm

2 has been successfully fabricated and


has focal lengths between 50 and 150 µm. Besides the lens, wave-guides have also been demonstrated byCox et al. (1994) using inkjet technology.

Because the optical components can be selectively deposited onto the desired region with varyingproperties, integration of those components with fabricated integrated circuit or other devices is possibleand efficient.

30.6.6 Solid Freeforming

Two-dimensional patterns and three-dimensional solid structures can be generated by microdropletgenerators. Orme and Huang (1993) and Marusak (1993) reported the application of molten metal dropsfor solid freeform fabrication. Evans’ group demonstrated the application of continuous and drop-on-demand inkjets for ceramic printing to fabricate three-dimensional structures as well as functionallygraded materials [Mott and Evans, 1999; Blasdell and Evans, 2000; Yamaguchi et al. 2000] used a metaljet to print functional three-dimensional microstructures, and Figure 30.29 shows the operation principle.Employing multijets for structure and sacrificial material deposition to print an overhanging structurewas also proposed by Yamaguchi et al. (2000), while Fuller and Jacobson (2000) used laminatedpoly(methyl methacrylate) (PMMA) film as the supporting material for ejection of metal cantileverbeams. The fabrication principle is shown in Figure 30.30.

30.6.7 Manufacturing Process

Droplet generators also provide novel material processing. For example, submicron ceramic particles canbe plasma sprayed, as introduced by Blazdell and Juroda, for surface coating [Blazdell and Kuroda, 2000].The operation principle is shown in Figure 30.31. A continuous-jet printer was used for droplet formationfrom ceramic solution. The produced ceramic stream was delivered into the hottest part of the plasmajet and then sprayed onto the working piece. The splats produced by the plasma spray are similar inmorphology to those produced using conventional plasma spraying of a coarse powder but are signifi-cantly smaller in size, which may provide unique characteristics such as extension of solid-state solubility,refinement of grain size, formation of metastable phases and a high concentration of point defects[Blazdell and Kuroda, 2000].

30.6.8 Integrated Circuit Cooling

Conventionally, blowing fans and fins are widely used to cool integrated-circuit chips, especially forcentral processing unit (CPUs). Recently, as the heating power has increased greatly with increasing CPUsize, more advanced methods, such as heat pipes, CPL and impinging air jets, have been introduced forquick heat removal. However, no matter how the designs improve, the limitation of heat removal ability

FIGURE 30.29 Operation principle of three-dimensional structure fabrication by microdroplet generator. (AfterYamaguchi et al., 2000.)

Diaphragm Microstructure

Metal jet

NozzlePiezo disc

Metal chamber


for those devices is on the order of tens of watts per square centimeter. In addition, being able to detecthot spots and selectively remove the heat only from hot regions to preserve energy is highly desired butnot easy to perform by traditional ways. As a result, the concept of transporting latent heat through thedroplet evaporation process holds promise. This method can, in principle, remove three to four ordersmore heat than conventional methods. Also, the cooling spot can be selected and monitored through theintegrated micro-temperature sensor and integrated circuit array. The conceptual design by Tseng (2001),as shown in Figure 30.32, used a two-dimensional array of microinjectors to selectively deposit liquiddroplets onto the chip surface. The applied droplet frequency and numbers can be adjusted to maintaina dry chip surface with constant temperature. The estimated maximum heat removed by this device is

FIGURE 30.30 Fabrication of three-dimensional structures by droplet ejection and polymer lamination. (AfterFuller and Jacobson, 2000.)

FIGURE 30.31 Operation principle of plasma spraying by microdroplet generators. (After Blazdell and Kuroda, 2000.)

(a) P MMA deposition

Drawdown bar

PMMA

(b) PMMA lamination

Inkjet head(c) Particle deposition by inkjet

(d) Structure

Gas

Arc discharge

Inkjet

Droplets

Plasma jet Deposit


around 300,000 W/cm2, more than 1000 times greater than by conventional means. Temperature sensors

as well as a control circuit can also be fabricated on the same chip to form a self-contained smart system.

30.7 Concluding Remarks

The droplet generator is an important fluid-handling device for precise liquid-dosing control. MEMStechnology makes microsized droplet generators possible and popular for many applications. Variousmethods of droplet generation, including piezoelectric, thermal bubble, thermal buckling, focused acous-tic wave, electrostatic and inertial actuation, have been employed. Compared to other methods, thethermal bubble approach has greater actuation deformation, simpler design/fabrication and less limita-tion on the chamber volume, but it has the drawbacks of being temperature sensitivity and influencedby liquid properties. The piezo-type jet has the advantage of high frequency response, controllable dropletsize, and no satellite drops, but it has the limitation of finite actuation deformation, thus limitingminiaturization of the chamber volume. The electrostatic and thermal buckling jets have size limitationssimilar to the piezo type. Despite the electrostatic generator having the benefit of low power consumption,both have limited frequency response due to size limitations. The acoustic-wave droplet generator, onthe other hand, is not mature and stable enough for commercial applications, while the inertial actuationmethod offers limited miniaturization due to its operation principle. More types of microdroplet gen-erators are under development and may one day replace the ones we have been using for decades.

Physical properties, design issues and manufacturing aspects are important concerns in the design andfabrication of microdroplet generators. The associated issues, including frequency response, cross-talk,satellite droplets, puddle formation, material selection and integration, require great care at the variousdesign and fabrication levels. MEMS technology provides some of the key solutions to those practicalissues.

Many aspects of the applications have made microdroplet generators important and exciting ever sincetheir inception. Inkjet printing is the traditional application and has generated a significant amount ofrevenue in the printing market. Moreover, hundreds of applications are yet emerging, including bioreagenthandling, fine chemical handling, drug delivery, direct writing, solid freeform, integrated circuit coolingand fuel injection, which show promising results and many potential markets. Many exciting applicationsof microdroplet generators are yet to be discovered.

FIGURE 30.32 Conceptual design of integrated circuit cooling by microdroplet injections.

2 cm

2 cm

Signal and energy transport bus

Microinjectors

Parallel packaging

IC chip with integratedsurface temperature sensors Microdroplet generation chip

Dropletejection for cooling


References

Allen, R.R., Meyer, J.D., and Knight, W.R., (1985) “Thermodynamics and Hydrodynamics of ThermalInk Jets,” Hewlett-Packard J. May, pp. 21–27.

Asai, A. (1989), “Application of the Nucleation Theory to the Design of Bubble Jet Printers,” Jpn. J. Appl.Phys. 28, pp. 909–915.

Asai., A. (1991), “Bubble Dynamics in Boiling Under High Heat Flux Pulse Heating,” ASME J. HeatTransfer 113, pp. 973–979.

Asai, A. (1992), “Three-Dimensional Calculation of Bubble Growth and Drop Ejection in a Bubble JetPrinter,” Trans. ASME 114, pp. 638–641.

Asai, A., Hara, T., and Endo, I. (1987) “One Dimensional Model of Bubble Growth and Liquid Flow inBubble Jet Printers,” Jpn. J. Appl. Phys. 26(10) pp. 1794–1801.

Asai, A., Hirasawa, S., and Endo, I. (1988) “Bubble Generation Mechanism in the Bubble Jet RecordingProcess,” J. Imaging Technics 14, pp. 120–124.

Asano, K., Funayama, Y., Yatsuzuka, K., and Higashiyama, Y. (1995) “Spherical Particle Sorting by UsingDroplet Deflection Technology,” J. Electrostatics 35, pp. 3–12.

Ashley, C.T., Edds, K.E., and Elbert, D.L. (1977) “Development and Characterization of Ink for anElectrostatic Ink Jet Printer,” IBM J. Res. Develop. 21(1), pp. 69–74.

Askeland, R.-A., Childers W.-D., and Sperry, W.-R. (1988) “The Second-Generation Thermal InkJetStructure,” Hewlett-Packard J. Aug., pp. 28–31.

Bharathan, J., and Yang, Y. (1998) “Polymer Electroluminescent Devices Processed by Inkjet Printing.I. Polymer Light-Emitting Logo,” Appl. Phys. Lett. 72(21), pp. 2660–2662.

Bhaskar, E.V., and Aden, J.S. (1985) “Development of the Thin-film Structure for the Thinkjet Printhead,”Hewlett-Packard J. May, pp. 27–33.

Blasdell, P.F., and Evans, J.R.G. (2000) “Application of a Continuous Ink Jet Printer to Solid Freeformingof Ceramics,” J. Mater. Proc. Technol. 99, pp. 94–102.

Blazdell, P., and Kuroda, S. (2000) “Plasma Spraying of Submicron Ceramic Suspensions Using a Con-tinuous Ink Jet Printer,” Surface Coatings Technol. 123, pp. 239–246.

Bogy, D.B., and Talke, F.E. (1984) “Experimental and Theoretical Study of Wave Propagation Phenomenain Drop-on Demand Ink Jet Devices,” IBM J. Res. Develop. 28(3), pp. 314–321.

Brown, G.L., and Roshko, A. (1974) “On Density Effects and Large Structure in Turbulent Mixing Layers,”J. Fluid Mech. 64, pp. 775–816.

Buehner, W.L., Hill, J.D., Williams, T. H., and Woods, J.W. (1977) “Application of Ink Jet Technology toa Word Processing Output Printer,” IBM J. Res. Develop. 21(1), pp. 2–9.

Bugdayci, N., Bogy. D.B., and Talke, F.E. ( 1983) “Axisymmetric Motion of Radially Polarized PiezoelectricCylinders Used in Ink jet Printing,” IBM J. Res. Develop. 27(2), pp. 171–180.

Buskirk, W.-A., Hackleman, D.-E., Hall, S.-T., Hanarek, P.-H., Low, R.-N., Trueba, K.-E., and Van dePoll, R.-R. (1988) “Development of a High-Resolution Thermal Inkjet Printhead,” Hewlett-PackardJ. Oct, pp. 55–61.

Carmichael, J.M. (1977) “Controlling Print Height in an Ink Jet Printer,” IBM J. Res. Develop. 21(1),pp. 52–55.

Chen, J.-K., and Wise, K.D. (1995) “A High-Resolution Silicon Monolithic Nozzle Array for InkjetPrinting,” in Proc. 8th Int. Conf. on Solid-State Sensors and Actuators and Eurosensors IX, June 25–29,Stockholm, Sweden, pp. 321–324.

Chen, J.-K., Juan, W., Kubby, J., and Hseih, B.-C. (1998b) “A Monolithic Polyimide Nozzle Aray for InkjetPrinting,” Technical Digest IEEE Solid-State Sensor and Actuator Workshop, Hilton Head Island, SC.

Chen, P.-H., Chen, W.-C., and Chang, S.-H. (1997a) “Bubble Growth and Ink Ejection Process of aThermal Ink Jet Printhead,” Int. J. Mech. Sci. 39, pp. 687–695.

Chen, P.-H., Chen, W.-C., and Chang, S.-H. (1997b) “Visualization of Drop Ejection Process of a ThermalBubble Ink Jet Printhead,” in Proc. 1st Pacific Symp. Flow Visualization and Image Processing, Honolulu,HI, pp. 132–137.


Chen, P.-H., Chen, W.-C., Ding, P.-P., and Chang, S.-H. (1998a) “Droplet Formation of a ThermalSideshooter Inkjet Printhead,” Int. J. Heat and Fluid Flow, 19, pp. 382–390.

Chen, P.-H., Peng, H.-Y., Liu, H.-Y., Chang, S.-L., Wu, T.-I., and Cheng C.-H. (1999) “Pressure Responseand Droplet Ejection of a Piezoelectric Inkjet Printhead,” Int. J. Mech. Sci. 41, pp. 235–248.

Cox, W.R., Chen, T., Hayes, D.J., MacFarlane, D.L., Narayan, V., and Jatum, J.A., (1994) “MicrojetFabrication of Microlens Arrays,” IEEE Photonics Technol. Lett. 6(9), pp. 1112–1114.

Curry, S.A., and Portig, H. (1977) “Scale Model of an Ink Jet,” IBM J. Res. Develop. 21(1), pp. 10–20.Darling, R.H., Lee, C.-H., and Kuhn, L. (1984) “Multiple-Nozzle Ink Jet Printing Experiment,” IBM J.

Res. Develop. 28(3), pp. 300–306.Fromm, J.E. (1984) “Numerical Calculation of the Fluid Dynamics of Drop-On-Demand Jets,” IBM J.

Res. Develop. 28(3), pp. 322–333.Fuller, S., and Jacobson, J. (2000) “Ink Jet Fabricated Nanoparticle MEMS,” in Proc. IEEE MEMS ’2000,

Miyazaki, Japan, pp. 138–141.Goldmann, T., and Gonzalez, J.S. (2000) “DNA Printing: Utilization of a Standard Inkjet Printer for the

Transfer of Nucleic Acids to Solid Supports,” J. Biochem. Biophys. Methods 42, pp. 105–110.Gruhler, H., Hey, N., Muller, M., Bekesi, S., Freygang, M., Sandmaier, H., and Zengerle, R. (1999),

“Topspot—A New Method for the Fabrication of Biochips,” Proc. IEEE MEMS ’99, Orlando, FL,pp. 7413–417.

Hayes, D.J., and Cox, W.R. (1998) “Micro Jet Printing of Polymers for Electronics Manufacturing,” inAdhesive Joining and Coating Technology in Electronics Manufacturing, Proc. 3rd Int. Conf., pp. 168–173.

Hayes, D.J., Grove, M.E., and Cox, W.R. (1999) “Development and Application by Ink-Jet Printing ofAdvanced Packaging Materials,” in Proc. Int. Symp. on Advanced Packaging Materials, pp. 88–93.

Hebner, T.R., Wu, C.C., Marcy, D., Lu, M.H., and Sturm, J.C. (1998) “Ink-Jet Printing of Doped Polymerfor Organic Light Emitting Devices,” Applied Physics Letters, 72(5), pp. 519–521.

Hirata, S., Ishii, Y., Matoba, H., and Inui, T. (1996) “An Ink-Jet Head Using Diaphragm Microactuator,”in Proc. 9th IEEE Micro Electro Mechanical Systems Workshop, San Diego, CA, pp. 418–423.

Ho, C.M., and Gutmark, E. (1987) “Vortex Induction and Mass Entrainment in a Small Aspect RatioElliptic Jet,” J. Fluid Mech. 179, pp. 383–405.

Ho, C.M., and Huang, L.S. (1982) “Subharmonics and Vortex Merging in Mixing Layers,” J. Fluid Mech.119, pp. 443–473.

Kamisuki, S., Hagata, T., Tezuka, C., Nose, Y., Fuji, M., and Atobe, M. (1998) “A Low Power, Small,Electrostatically Driven Commercial Inkjet Head,” in Proc. IEEE MEMS ’98, Heidelberg, Germany,pp. 63–68.

Kamisuki, S., Fuji, M., Takekoshi, T., Tezuka, C., and Atobe, M. (2000) “A High Resolution, Electrostat-ically Driven Commercial Inkjet Head,” in Proc. IEEE MEMS ’2000, Miyazaki, Japan, pp. 793–798.

Karz, R.S., O’Neill, J.F., and Daneile, J.J. (1994) “Ink Jet Printhead with Ink Flow Direction Valves,” U.S.Patent 5,278,585.

Keefe, B.-J., Ho, M.-F., Courian, K.-J., Steinfield, S.-W., Childers, W.-D., Tappon, E.-R., Trueba, K.-E.,Chapman, T.-I., Knight, W.-R., Moritz, J.-G. (1997) “Inkjet Printhead Architecture for High Speedand High Resolution Printing,” U.S. Patent 5,648,805.

Kobayashi, H. et al. (2000) “A Novel RGB Multicolor Light-Emitting Polymer Display,” Synthetic Metals111–112, pp. 125–128.

Koide, A., Sasaki, Y., Yoshimura, Y., Miyake R., and Terayama, T. (2000) “Micromachined Dispenser withHigh Flow Rate and High Resolution,” in Proc. IEEE MEMS ’2000, Miyazaki, Japan, pp. 424–428.

Krause, P., Obermeier, E., and Wehl, W. (1995) “Backshooter—A New Smart Micromachined Single-ChipInkjet Printhead,” in Proc. 8th Int. Conf. on Solid-State Sensors and Actuators and Eurosensors IX,June 25–29, Stockholm, Sweden, pp. 325–328.

Lee, F.C., Mills, R.N., and Talke, F.E. (1984) “The Application of Drop-on-Demand Ink Jet Technologyto Color Printing,” IBM J. Res. Develop. 28(3), pp. 307–313.

Lee, H.C. (1977) “Boundary Layer Around a Liquid Jet,” IBM J. Res. Develop. 21(1), pp. 48–51.


Lee, J.-D., Lee H.-D., Lee H.-J., Yoon, J.-B., Han, K.-H., Kim, J.-K., Kim, C.-K., and Han, C.-H. (1995)“A Monolithic Thermal Inkjet Printhead Utilizing Electrochemical Etching and Two-Step Electro-plating Techniques,” IEEE IEOM 95–60L, pp. 23.3.1–23.3.4.

Lee, J.-D., Yoon, J.-B., Kim, J.-K., Chung H.-J., Lee C.-S., Lee H.-D., Lee H.-J., Kim, C.-K., and Han, C.-H.(1999a) “A Thermal Inkjet Printhead with a Monolithically Fabricated Nozzle Plate and Self-Aligned Ink Feed Hole,” J. MEMS 8(3) pp. 229–236.

Lee, Y.-K., Yi, U., Tseng, F.-G., Kim, C.J., and Ho, C.-M. (1999b) “Diesel Fuel Injection by a ThermalMicroinjector,” in Proc. MEMS, ASME IMECE ’99, Nashville, TN, pp. 419–425.

Levanoni, M. (1977) “Study of Fluid Flow Through Scaled-Up Ink Jet Nozzles,” IBM J. Res. Develop.21(1), pp. 56–68.

Luginbuhl, P. et al. (1999) “Micromachined Injector for DNA Spectrometry,” in Proc. IEEE Transducers ’99,Sendai, Japan, pp. 1130–1133.

Marusak, R.E. (1993) “Picoliter Solder Droplet Dispensing,” in Proc. Solid Freeform Fabrication Symp.,Austin, TX, pp. 81–87.

Meinhart, C.-D., and Zhang H. (2000) “The Flow Structure Inside a Microfabricated Inkjet Printhead,”J. MEMS 9(1), pp. 67–75.

Miliotis, T., Eficsson, D., Marko-Varga, G., Ekstrom, S., Nilsson, J., and Laurell, T. (2000) “InterfacingProtein and Peptide Separation to Maldi-tof MS Using Microdispensing and On-Target EnrichmentStrategies,” in Proc. µTAS ’2000, Enschede, The Netherlands, pp. 387–391.

Mirfakhraee, A. (1989) “Growth and Collapse of Vapor Bubbles in Ink-Jet Printers,” Ph.D. thesis, Uni-versity of California, Berkeley.

Mott, M., and Evans, J.R.G. (1999) “Zirconia/Alumina Functionally Graded Material Made by CeramicInk Jet Printing,” Mater. Sci. Eng. A271, pp. 344–352.

Myers, R.A., and Tamulis, J.C. (1984) “Introduction to Topical Issue on Non-Impact Printing Technol-ogies,” IBM J. Res. Develop. 28(3), pp. 234–240.

Nielsen, N.J. (1985) “History of ThinkJet Printhead Development,” Hewlett-Packard J. May, pp. 4–10.Nilsson, J., Bergkvist, J., and Laurell, T. (2000) “Optimization of the Droplet Formation in a Piezo-Electric

Flow-Through Microdispenser,” in Proc. µTAS ’2000, Enschede, The Netherlands, pp. 75–78.Orme, M., and Huang, C. (1993) “Thermal Design Parameters Critical to The Development of Solid

Freefrom Fabrication of Structural Materials with Controlled Nanoliter Droplets,” in Proc. SolidFreeform Fabrication Symp., Austin, TX, pp. 88–94.

Pimbley, W.T., and Lee, H.C. (1977) “Satellite Droplet Formation in a Liquid Jet,” IBM J. Res. Develop.21(1), pp. 21–30.

Rembe, C., Siesche, S.A.D., and Hofer, E.P. (2000) “Thermal Ink Jet Dynamics: Modeling, Simulation,and Testing,” Microelectron. Reliability 40, pp. 525–532.

Shadow, K.C., Gutmark, E., Wilson, K.J., Parr, D.M., Mahan, V.A., and Ferrell, G.B. (1987) “Effect ofShear-Flow Dynamics in Combustion Processes,” Combustion Sci. Technol. 54, pp. 103–116.

Shah, P., Kevrekides, Y., and Benziger, J. (1999), “Ink-Jet Printing of Catalyst Patterns for Electroless MetalDeposition,” Langmuir 15, pp. 1584–1587.

Shimoda, T., Kimura M., Seki S., Kobayashi, H., Kanbe, S., and Miyahita, S. (1999) “Technology forActive Matrix Light Emitting Polymer Displays,” in Proc. IEEE IEDM ’99, pp. 107–110.

Siewell, G.-L., Boucher, W.R., and McClelland, P.H. (1985) “The ThinkJet Orifice Plate: A Part with ManyFunctions,” Hewlett-Packard J. May, pp. 33–37.

Somberg, H. (1990) “Inkjet Printing for Metallization on Very Thin Solar Cells,” in Conf. Record of the21st IEEE in Photovoltaic Specialists Conf., pp. 666–667.

Sweet, R.G. (1964) “High Frequency Recording with Electrostatically Deflected Ink Jets,” Stanford Elec-tronics Laboratories Technical Report No. 1722-1, Stanford University, Stanford, CA.

Sweet, R.G. (1971) “Fluid Droplet Recorder,” U.S. Patent 3,576,275.Szita, N., Sutter, R., Dual, J., and Buser, R. (2000) “A Fast and Low-Volume Pipettor with Integrated

Sensors for High Precision,” in Proc. IEEE MEMS ’2000, Miyazaki, Japan, pp. 409–413.


Ta, C.-C., Chan, L.-W., Wield, P.-J., and Nevarez R. (1988). “Mechanical Design of Color Graphics Printer,”Hewlett-Packard J. Aug., pp. 21–27.

Teng, K.F., and Vest, R.W. (1988a) “Metallization of Solar Cells with Ink Jet Printing and Silver Metallo-Organic Inks,” IEEE Trans. Components, Hybrids, and Manuf. Technol. 11(3), pp. 291–297.

Teng, K.F., and Vest R.W. (1988b) “Application of Ink Jet Technology on Photovoltaic Metallization,”IEEE Electron. Device Lett. 9(11).

Teng, K.F., Azadpour, M.A., and Yang, H.Y. (1988) “Rapid Prototyping of Multichip Packages UsingComputer-Controlled, Ink Jet Direct-Write,” in Proc. 38th Electronics Components Conf., pp. 168–173.

Tseng, F.-G. (1998) “A Micro Droplet Generation System,” Ph.D. thesis, University of California, Los Angeles.Tseng, F.-G. (1999) Presentation material. Tseng, F.-G. (2001) “Droplet Impinging Micro Cooling Arrays,” proposal for NSC project.Tseng, F.-G., Linder, C., Kim, C.- J., and Ho, C.-M (1996) “Control of Mixing with Micro Injectors for

Combustion Application,” in Proc. Micro-Electro-Mechanical Systems (MEMS ’96), DSC-Vol.59,Atlanta, GA, pp. 183–187.

Tseng, F.-G., Kim, C.-J., and Ho, C.-M (1998a) “A Microinjector Free of Satellite Drops and Character-ization of the Ejected Droplets,” in Proc. MEMS, ASME IMECE ’98, Anaheim, CA, pp. 89–95.

Tseng, F.-G., Shih, C., Kim, C.-J., and Ho, C.-M (1998b) “Characterization of Droplet Injection Processof a Microinjector,” in Abstract Book of 13th U.S. Congress of Applied Mechanics, University ofFlorida, Gainesville, p. TB3.

Tseng, F.-G., Kim, C.-J., and Ho, C.-M (1998c) “A Novel Microinjector with Virtual Chamber Neck,” inTechnical Digest of the 11th IEEE Int. Workshop on Micro Electro Mechanical Systems, Heidelberg,Germany, pp. 57–62.

Tseng, F.-G., Kim, C.-J., and Ho, C.-M. (2000) “Apparatus and Method for Using Bubble as Virtual Valvein Micro Injector to Eject Fluid,” U.S. and international patent (pending).

Tseng, F.-G., Kim, C.-J., and Ho, C.-M. (2001a) “A Monolithic, High Frequency Response, High-Resolution Microinjector Array Ejecting Sub Picoliter Droplets without Satellite Drops. Part I.Concepts, Designs and Molding,” submitted to J. MEMS.

Tseng, F.-G., Kim, C.-J., and Ho, C.-M. (2001b) “A Monolithic, High Frequency Response, High-Resolution Microinjector array Ejecting Sub Picoliter Droplets without Satellite Drops. Part II.Fabrication, Characterization and Performance Comparison,” submitted to J. MEMS.

Twardeck, T.G. (1977) “Effect of Parameter Variations on Drop Placement in an Electrostatic Ink JetPrinter,” IBM J. Res. Develop. 21(1), pp. 31–36.

Wang, X.Q., Desai, A., Tai, Y.C., Licklider, L., and Lee, T.D. (1999) “Polymer-Based Electrospray Chipsfor Mass Spectrometry,” in Proc. IEEE MEMS ’99, Orlando, FL, pp. 523–528.

Yamaguchi, K., Sakai, K., Ymanaka, T., and Hirayama, T. (2000) “Generation of Three-Dimensional MicroStructure Using Metal Jet,” Precision Eng. 24, pp. 2–8.

Zhu, X., Tran, E., Wang, W., Kim, E.S., and Lee, S.Y. (1996) “Micromachined Acoustic-Wave LiquidEjector,” in Proc. Solid-State Sensor and Actuator Workshop, June 2–6, Hilton Head, SC, pp. 280–282.


the mems handbook - eetmizsei/mikrorejegy/the mems handbook(comple… · light emitting polymer...

Documents