Investigation of Thermally-Actuated Pumping During Pool Boiling of a Dielectric Liquid on an Asymmetric Microstructured Silicon Heat Sink by Naveenan Thiagarajan A dissertation submitted to the Graduate Faculty of Auburn University in partial ful llment of the requirements for the Degree of Doctor of Philosophy Auburn, Alabama December 14, 2013 Keywords: reentrant cavities, pool boiling, FC72, microgravity, anemometry Copyright 2013 by Naveenan Thiagarajan Approved by Sushil H. Bhavnani, Chair, Professor of Mechanical Engineering Roy W. Knight, Assistant Professor of Mechanical Engineering Jeyhoon Khodadadi, Professor of Mechanical Engineering Daniel Mackowski, Associate Professor of Mechanical Engineering Narendra Govil, Professor of Mathematics and Statistics Abstract Developments in the eld of electronics fabrication have led to signi cant miniaturization of devices. Such a reduction in foot-print of the electronic devices coupled with increased capabilities, such as computing power, have o ered numerous advantages to human comfort in the form of connectivity and portability. But these developments have also resulted in a bottle-neck which is the demand to dissipate the resulting high heat densities in the electronic packages. With heat dissipation demands exceeding more than 1000 W/cm2, traditional air cooling techniques and even evaporative liquid cooling techniques like heat pipes have been pushed to their limits of operation. Future high powered, micro-electronics, thermal management could potentially migrate to liquid cooling involving phase change which is considered as one of the potential solutions for high heat dissipation demands. While a technique such as ow in microchannels with phase-change has proven to meet the demands, it comes with the cost of power required to pump the uid through narrow passages. With modern electronics getting leaner in power consumption, an ideal cooling technique would be one that while dissipating a large volume of heat, also self-propels the uid to enhance the heat transfer characteristics. Such a cooling system with a pump-less ow loop will be power-free, compact and self-regulating. The system proposed will also be applicable to thermal management of space electronics, where power is a precious commodity. The study conducted by the author in collaboration with a heat transfer research group from Oregon State University, investigates the liquid self-propulsion e ects during pool boiling of a dielectric liquid on asymmetric surfaces. The study describes a novel silicon heat sink with an asymmetric saw-tooth cross- sectioned surface structure, which has the potential to be translated into a liquid propulsion system while dissipating heat e ciently. The heat sink was fabricated using a combination of ii gray-scale lithography, deep reactive ion etching and wet etching techniques. The novelty of the heat sink lies in the ability to e ect lateral motion of bubbles due to nucleation from re- entrant cavities fabricated on the shallow slope of the saw-toothed surface. The asymmetric nucleation, growth and departure of bubbles leads to an angular momentum imparted to the liquid, thereby resulting in a net lateral ow. The study investigates the ability of surface structure to propel the liquid in its immediate vicinity under a variety of test conditions. The tested conditions include heat ux in the range of 0-4 W/cm2, liquid subcooling ranging from 0-20 C, and gravity ranging from 0-1.8g. Due to the unique pro le of the surface, the bubble characteristics are very di erent from those reported in the literature for common surfaces and uids. One of the primary objectives of the study is the characterization of bubble dynamics from such a surface. In the experiments conducted, bubble growth and departure from re-entrant cavities on the asymmetric structures were studied using high speed photography and image processing techniques. The asymmetry in shape of the ratchet and location of re-entrant cavities re- sulted in nucleation only from the shallow slope of the ratchets. Interestingly, the bubbles were \light-bulb" shaped which otherwise would be more circular for a highly wetting uid such as FC-72. It was observed that the bubble growth and departure were normal to the shallow slope of the ratchet surface structure. Bubble dynamics such as growth rate, bubble departure diameter and frequency were studied as a function of heat ux and subcooling. Asymptotic growth relationships were expressed as D = At and D = tm for the inertia and heat transfer controlled regimes respectively. The values of A, varying between 48-181, increased with increasing heat ux and decreasing subcooling. Similarly, in the heat transfer controlled growth regime, the value of , termed as the growth constant, was observed to increase between 0.25-0.3 with increasing heat ux and decreasing subcooling. Subcooling or heat ux did not a ect the value of m signi cantly which varied between 0.20 - 0.25, compared to the value of 0.5 that has been widely reported in the literature. The bubble de- parture frequency was estimated to be increasing between 0 - 60 with increasing heat ux iii and decreasing subcooling. However, under the tested conditions bubble departure diameter was not found to be a ected signi cantly with heat ux or diameter. A similar saw-toothed surface was tested at microgravity to understand the e ects of gravity on bubble dynamics and self-propelled bubble motion across the surface. Pool boiling experiments were conducted aboard NASA?s reduced gravity ight. In FC-72, vapor bubbles six times larger in diameter compared to 1g were observed, due to lack of buoyancy. Interestingly, bubbles were observed to be sliding across the asymmetric surface at velocities as high as 27.4 mm/s. This motion was observed at all tested conditions. Pool boiling on a plain surface would result in stagnant, surface residing vapor bubbles that would a ect the heat transfer characteristics adversely causing burn-out of the chip being cooled. The ability to move the bubble along the surface at high velocities prevented any heat transfer deterioration, and actually leads to an enhancement. The sliding motion was attributed to pressure di erences in the thin liquid lm existing between the saw-toothed surface and the vapor bubble. A model has been proposed based for the sliding velocity of bubbles, which proves that the force due to pressure di erences in the liquid lm is a potential driving force for the bubbles among other forces. iv Acknowledgments The author would like to extend his sincere thanks to his advisory committee, chaired by Prof. Sushil H.Bhavnani, and comprising Prof. Jeyhoon Khodadadi, Prof. Roy W. Knight, Prof. Daniel Mackowski and Prof. Narendra Govil for their contributions and time. Special thanks are extended to Prof. Roy Hart eld for serving as the external reader. The author gratefully acknowledges the support of sponsors, NSF and NASA, for the opportunity to work in this eld of research. Words cannot express the joy and pride that this research has brought along, especially, in the form of microgravity testing. The author would like to sincerely thank his adviser and committee chair Prof. Sushil H. Bhavnani for being a constant source of inspiration and motivation through a lot of di cult times, and for all the opportunities provided. Sincere thanks and appreciation are extended to Prof. Vinod Narayanan and his graduate students, Florian Kapsenberg and Logan Strid, for the guidance and ideas provided. The author is also thankful to AMSTC, especially Mr. Charles Ellis, for providing the facilities and knowledge for the fabrication of test devices. The author would like to acknowledge the guidance and assistance provided by Dr. Shakib Morshed, and Dr. Tamara Isaacs-Smith in fabricating the test sections. Thanks are extended to Mr. Travis Wheeler for his precious assistance during microgravity testing. Much gratitude is extended to my colleagues at the Heat Transfer Research laboratory for their help, support, and company. Above all, the author would like to thank his parents and sister for their endless love, support, encouragement and sacri ces. v Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xx 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Thermally actuated pumping systems: Literature review . . . . . . . . . . . 7 1.1.1 Self-propulsion in single phase liquids . . . . . . . . . . . . . . . . . . 7 1.1.2 Self-propelled ow in phase change systems . . . . . . . . . . . . . . . 9 1.1.3 Fabrication of re-entrant cavities and sloped surfaces: Background . . 14 1.1.4 Bubble dynamics in pool boiling on plain and enhanced surfaces . . . 19 1.1.5 Bubble dynamics under reduced gravity . . . . . . . . . . . . . . . . 23 2 Fabrication of Silicon Test Devices . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.1 Fabrication of silicon ratchet array . . . . . . . . . . . . . . . . . . . . . . . 35 2.2 Fabrication of re-entrant cavities . . . . . . . . . . . . . . . . . . . . . . . . . 39 2.3 Fabrication of heater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 2.4 Bonding of saw-tooth ratchet wafer to the heater wafer . . . . . . . . . . . . 46 3 Terrestrial Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 3.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 3.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 3.2.1 Bubble Dynamics using High Speed Photography . . . . . . . . . . . 55 3.2.2 Bubble growth - Experiments . . . . . . . . . . . . . . . . . . . . . . 56 3.2.3 Bubble departure frequency and diameter . . . . . . . . . . . . . . . 67 vi 3.2.4 Bubble growth - Comparison of experimental data with models . . . 70 3.2.5 Heat Dissipation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 3.2.6 Lateral Liquid Velocity Measurements Using Hot Wire Anemometer: Parametric E ects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 3.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 4 Microgravity Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 4.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 4.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 4.2.1 Bubble Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 4.2.2 Bubble Transit Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . 96 4.2.3 Heat Dissipation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 5 Conclusions and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . 106 5.1 Bubble dynamics: Terrestrial experiments . . . . . . . . . . . . . . . . . . . 106 5.2 Net lateral liquid velocity under terrestrial gravity . . . . . . . . . . . . . . . 108 5.3 Bubble dynamics: Microgravity . . . . . . . . . . . . . . . . . . . . . . . . . 108 5.4 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 A Thermophysical Properties of FC-72 and Water . . . . . . . . . . . . . . . . . . 119 B Preliminary Closed Loop Experiments Under Terrestrial Gravity Using FC-72 . 120 C Calibration of Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 C.1 Calibration of thermistor and thermocouple . . . . . . . . . . . . . . . . . . 124 C.2 Calibration of test section heater . . . . . . . . . . . . . . . . . . . . . . . . 125 C.3 Calibration of hot wire anemometer . . . . . . . . . . . . . . . . . . . . . . . 126 D Emperical Model for Lateral Liquid Velocity in Nucleate Boiling . . . . . . . . . 128 E Particle Image Velocimetry (PIV) . . . . . . . . . . . . . . . . . . . . . . . . . . 131 F Microgravity Experiments: Test Equipment Data Package (TEDP) . . . . . . . 137 G Microgravity Experiments: Sample Calculations for the Bubble Velocity Model . 186 vii H Design Draft Files for Boiling Chamber Parts and Other Experimental Fixtures 188 viii List of Figures 1.1 Evolution of supercomputers since its introduction illustrating the increase in number of Floating point operations per second (Flops) and the reduction in cost per operation. The corresponding evolution of cooling systems used in these supercomputers from a simple liquid cooling system such as in car radiators to advanced phase change cooling systems is illustrated. A popular current laptop is used for comparison which is much faster than the supercomputers in 1985. . 2 1.2 Chart illustrates the adaptation of advanced cooling techniques for high heat density applications compared to the simple, lower heat transfer air cooling tech- niques for low power, low heat density applications . . . . . . . . . . . . . . . . 3 1.3 Schematic diagram of the concept under 1g showing the cross-section of the saw- toothed structure. Bubbles grow and depart from the re-entrant cavities at an angle that is normal to the surface. This is hypothesized to result in a net lateral velocity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.4 Schematic diagram of the concept under 0g showing the larger size of the bubble due to lack of buoyancy. Bubble slides from left to right due to pressure di erence in the liquid lm underneath, between the crest and trough, that arises due to di erences in radii of curvature (r) . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.5 (a) Image illustrates the local convective ow at the liquid surface and the result- ing convective cells (b) The surface tension gradients due to alternating di erence in temperatures of warmer and cold liquids cause de ection of the free surface. Image was acquired from Carey [6]. . . . . . . . . . . . . . . . . . . . . . . . . . 8 ix 1.6 (a) Oval ow loop with saw-toothed heated ratchet sections used by Jo [1] (b) Asymmetric temperature gradients induced by the shape of the ratchet pro le, which leads to surface tension gradients leading to liquid movement. The direc- tion of liquid movement is marked in (a) . . . . . . . . . . . . . . . . . . . . . . 10 1.7 Flow pattern in oil observed by Stroock et al. [3] over asymmetric mini-ratchets at velcities up to 2mm/min. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.8 Surface tension induced ow towards the trough of a symmetrical triangular groove which is marked by the movement of glass particles. Alexeev et al. [2] reported that this results in liquid ow towards the wall at the trough and away from the wall at the crest, thus setting up a vortex ow. . . . . . . . . . . . . . 10 1.9 (a) Leidenfrost droplet motion over a saw-toothed surface demonstrated by Linke et al. [8] (b) Droplet propelled by pressure gradients in the thin liquid layer induced by the di erences in radii of curvature of liquid-vapor interface (c) The direction of liquid movement due to the pressure gradient is shown. R-134a droplet velocities of up to 5 cm/s were reported . . . . . . . . . . . . . . . . . . 13 1.10 Bubble based micropumps that rely on asymmetry in surface structure, surface wetting characteristics and heater locations. (i) Meng and Kim [10] demonstrated preferential gas bubble motion due to surface asymmetry in the form of a neck in the microchannel structure as seen in (a). The gas bubble translation was promoted due to a hydrophilic-hydrophobic junction causing bubble translation to the right as seen in (b). Liquid ow rates of up to 65 nL/s were reported. (ii) Jun and Kim [11] demonstrated bubble motion due to surface tension gradients induced by sequential powering of the heaters. Liquid (isopropanol) velocities of up to 160 m/s were observed. . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 x 1.11 (a) Pyramidal re-entrant cavities used in immersion cooled surfaces studied by Nimkar et al. [14]. Such re-entrant cavities are formed by anisotropic potassium hydroxide etch (b) Bulb shaped re-entrant cavity used in immersion cooled heat sinks by Baldwin et al. [13] and Bhavnani et al. [17] . . . . . . . . . . . . . . . 16 1.12 (a) Arti cial cavities for nucleation formed by punch marks that were used in microchannel heat sinks by Kandlikar et al. [15] (b) Inter-connected re-entrant cavities to form slots that were used in silicon microchannel heat sinks by Ko sar et al. [16] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 1.13 Steps involved in deep reactive ion etching (DRIE) (a) Formation of a photomask using photolithography (b) Etching of silicon using a plasma containing SF6 (c) Passivation cycle involving the deposition of a polymer material which protects the side walls from being etched in the subsequent etching cycles (d) Etching cycle is repeated. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.14 Steps involved in gray-scale lithography as illustrated in Waits et al. [23] (a) Design of gray-scale mask (b) Exposure and development of photoresist to obtain a gradient height in photoresist (c) DRIE to obtain a gradient height (slope) in silicon. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.15 Types of bubble growth models (a) Uniform superheat theory, involving spherical vapor growth in an uniformly superheated liquid (b) Bubble growth in non- uniform temperature eld; temperature highest at the wall[25]. . . . . . . . . . 21 1.16 Regimes of bubble growth illustrated as shown in Carey [6] (a) Ambient cooler liquid is brought in to contact with the heater surface (b) Growth of boundary layer (c) Bubble nucleation and inertia controlled regime (d) Relaxation micro- layer thickness decreases (d) Bubble departure. . . . . . . . . . . . . . . . . . . 22 xi 1.17 Bubble growth process from a cylindrical cavity on a silicon surface using FC-72 - Hutter et al. [34] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 1.18 Bubble growth rate from a cylindrical cavity on a silicon surface using FC-72 - Hutter et al. [34]. The asymptotic growth relationship is marked as the equation of t in the gure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 1.19 (a) Parabolic trajectory of a C-9B ight and the achieved reduced gravity periods (b) G-jitters in a parabolic ight due to disturbances such as air turbulence - Straub [37]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 1.20 Large bubble of diameter 2.5 mm in pool boiling of FC-72 under reduced gravity at a wall superheat of 30 C - Kim and Benton[38]. . . . . . . . . . . . . 28 1.21 (a) Lower heat transfer observed under microgravity in pool boiling of FC-72 at higher wall superheat. Critical heat ux (CHF) also reduced compared to earth gravity - Kim and Benton[38] (b) Heat transfer increases with increasing heater sizes and liquid subcooling - Henry and Kim [41] . . . . . . . . . . . . . . . . . 29 1.22 Large hovering bubble in R-113; smaller adjacent bubbles near the larger bubble were noticed to migrate towards the larger bubble at high subcooling - Lee and Merte Jr. [43]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.1 Illustration of steps involved in fabrication of the silicon test devices to be used in the large and small array experiments . . . . . . . . . . . . . . . . . . . . . . 34 2.2 Steps involved in the fabrication of silicon saw-tooth ratchets . . . . . . . . . . 36 2.3 Illustration of DRIE parameters discussed in Table 2.1 . . . . . . . . . . . . . . 37 2.4 SEM image of the ratchets with an angle 19 . The inset shows an SEM image of silicon nano-structures, known as \silicon grass", formed during the DRIE process with Recipe A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 xii 2.5 Cross-section of a ratchet array with a nominal saw-tooth angle of 24 . . . . . 38 2.6 Illustration of fabrication steps involved in etching re-entrant cavities from the back side of saw-teeth to form trapezoidal cavity mouths only on the shallow slope 41 2.7 Isometric and back side view illustrations of small and large array cavities on the shallow slope of saw-toothed surface . . . . . . . . . . . . . . . . . . . . . . . . 42 2.8 SEM image showing the top view of saw teeth and cavity mouth measurements in the test device with a saw-tooth angle of 24 . . . . . . . . . . . . . . . . . . 42 2.9 Sketch of aluminum serpentine heaters used in large and small array experiments that were fabricated on silicon. The width and thickness of heater traces are 500 m and 2 m respectively. The electrical contact pad con guration is labeled in the gure. The arrangement of electrical contact pads for the small array heater is based on the arrangement of electrical contacts in the wafer xture. . . . . . 44 2.10 Fabrication of aluminum serpentine heaters used in large and small array test devices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 2.11 Fabrication of isolation trenches on the back side of heater footprint and redution of wafer diameter to 3 inches. Silicon nitride was coated to eliminate re ection of radiation during infrared measurements made during the small array experiments at OSU. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2.12 Images of fabricated aluminum serpentine heaters used in large and small array experiments. The diameter of heater wafer for the large and small array test experiments are 4 inches and 3 inches respectively. . . . . . . . . . . . . . . . . 47 2.13 Illustration of fusion bonding process used to bond the large array of silicon saw-teeth to the heater wafer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 xiii 2.14 Illustration of gold eutectic bonding process used for bonding the small array saw-tooth ratchets to the 3 inch heater wafer . . . . . . . . . . . . . . . . . . . 50 2.15 X-ray image showing the bond quality of gold eutectic bonding process . . . . . 50 3.1 Silicon test device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 3.2 Boiling chamber assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 3.3 Image showing bubble departure from re-entrant cavities at an angle normal to the sloped surface of saw-teeth during saturated pool boiling at a heat ux of q" = 2.20 W=cm2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 3.4 Bubble growth and departure from re-entrant cavities at q" = 2:0W=cm2 and saturated pool conditions. Bubbles nucleating from neighbouring cavities are marked with a di erent color. Relative time stamps are marked at the bottom right of each image. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 3.5 Ebullition cycle of a bubble at q" = 1.6 W=cm2 and 21 C subcooling . . . . . . 59 3.6 Image processing steps of a single bubble frame at 178.44 ms in Fig. 3.5 . . . . 60 3.7 Estimated bubble growth for images shown in Fig. 3.5 at q"=1.6 W=cm2 and 21 C subcooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 3.8 E ect of heat ux on bubble growth at a liquid subcooling of 21 C . . . . . . . 64 3.9 E ect of subcooling on bubble growth at a nominal heat ux of 1.0 W=cm2 . . . 64 3.10 Growth constant from experiments at a liquid subcooling of 0-20 C. The gure shows a linear curve t used to approximately express the growth constant. The equations 3.3 and 3.5 used for comparison from the studies Fritz and Ende [52] and Cole and Shulman [29] respectively, were calculated for saturated pool conditions. 66 xiv 3.11 E ect of heat ux on bubble growth at a liquid subcooling of 20 C . . . . . . . 68 3.12 E ect of subcooling on bubble growth at a nominal heat ux of 1.0 W=cm2 . . 68 3.13 E ect of subcooling and heat ux on bubble departure frequency . . . . . . . . 69 3.14 E ect of subcooling and heat ux on bubble departure diameter . . . . . . . . . 69 3.15 Comparison with existing models for bubble growth . . . . . . . . . . . . . . . . 73 3.16 E ect of subcooling on pool boiling curve for the test device with a saw-tooth angle of 24 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 3.17 (a) and (b) TSI constant temperature anemometer and power supply used for liquid velocity measurements. Electrical circuit and connections are shown in (c) and (d) [55] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 3.18 Illustration of the probe set up over the test device with a saw-teeth angle of 24 79 3.19 Raw and temperature corrected voltage measurements using hot wire anemome- ter over the test device with a saw-tooth angle of 24 . . . . . . . . . . . . . . . 81 3.20 Temperature corrected liquid velocity measurements using hot wire anemometer over the test device with a saw-tooth angle of 24 . . . . . . . . . . . . . . . . 82 4.1 Zero gravity ight used for parabolic maneuvers - (a) A NASA Zer0-g ight on it upward ascent. Boeing 727 was used for the ight experiments (b)-(c) Gravity pro le and recorded accelerometer readings of the achieved parabolic maneuvers 85 xv 4.2 Experimental setup for reduced gravity experiments - (a) 31 test device with two re-entrant cavities per saw-tooth (b) Assembled test device with the polycar- bonate ow channel. Co-ordinate system shows the orientation of the test device with respect to the aircraft. (c) Boiling chamber used for the reduced gravity experiment. Bellows assembly with double containment is used to maintain a constant chamber pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 4.3 Assembled view of the experimental structure used for microgravity experiments. Details of the individual components in the structure and the related structural analysis are provided in Appendix F. . . . . . . . . . . . . . . . . . . . . . . . . 89 4.4 Bubble dynamics in water and FC-72 under g (a) Vapor bubble 10 times larger than at 1g was observed at microgravity. Image inset shows vapor nucleation from re-entrant cavities, under the foot print of the larger bubble, illustrating the presence of a liquid lm. (b) Sliding motion of vapor bubbles at a velocity of 10 mm/s, across the saw-teeth (left to right in images) at q00 = 0.5 W/cm2. Sliding motion was observed at all tested conditions. (c) Sliding motion of vapor bubbles (left to right in the images) at q00 = 1.4 W/cm2. . . . . . . . . . . . . . 91 4.5 Bubble dynamics in FC-72 under 1:8g. Bubble departure diameters are very small compared to those at 1g ( 0.25 D1g. Bubbles were observed to grow and depart at an angle normal to the shallow slope of the saw-teeth, a phenomenon that was previously demonstrated under 1g. . . . . . . . . . . . . . . . . . . . 92 4.6 Comparison of experimental data with departure diameter estimated using Equa- tion 4.1 for water and FC-72 under g . . . . . . . . . . . . . . . . . . . . . . . 93 4.7 Sliding velocity of bubbles in FC-72 under g. Uncertainty in bubble velocity is 0:5%. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 xvi 4.8 Schematic diagram of bubble motion over a saw-tooth in FC-72 under g (not to scale). The arrows marked between the saw tooth and the vapor bubble indicate the direction of forces acting on the bubble due to pressure di erences. . . . . . 97 4.9 Liquid-vapor interface of a sliding vapor bubble in FC-72 under g . . . . . . . 98 4.10 Estimated bubble sliding velocities for liquid lm thickness varying from 0-25 m. The experimental bubble velocity of 27.4 mm/s is predicted closely for a H value of 17 mm. From bubble images, a value of H. . . . . . . . . . . . . . . . . 104 4.11 Boiling curves for large array experiments with FC-72 under g and 1g. Tsub = Tsat Tpool. Both experiments were performed with a highly subcooled pool and pool temperature was not controlled. Uncertainties in q00 and (Tw-Tsat) are 1% and 0:3% respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 B.1 Bottom plate used in the closed loop experiments with FC-72. The \race-track" is the ow channel in which ow is intended to occur in the clockwise direction. 120 B.2 Assembly of the closed loop experimental set up showing the polycarbonate ow loop and the assembly of the test device. . . . . . . . . . . . . . . . . . . . . . . 121 B.3 Schematic representation of the closed ow loop used in the study. . . . . . . . 122 B.4 Schematic representation of the net ow e ected mainly by the condenser location.123 C.1 Calibration of Omega ON44007 thermistor and a K-type thermocouple using a NIST certi ed thermistor as a standard . . . . . . . . . . . . . . . . . . . . . . 124 C.2 Calibration of test section heater using a NIST certi ed thermistor as a standard 125 C.3 Flow loop for the calibration of hot wire probe . . . . . . . . . . . . . . . . . . . 126 C.4 Flow loop for the calibration of hot wire probe . . . . . . . . . . . . . . . . . . . 127 xvii D.1 Illustration of the angular momentum imparted by a growing bubble normal to the shallow slope, leading to a net lateral liquid velocity . . . . . . . . . . . . . 129 D.2 Control volume de ned for estimation of resulting net lateral liquid velocity due to bubble growth at an angle normal to the shallow slope of the ratchet . . . . . 129 E.1 Illustration of a PIV system showing the laser sheet that is used to illuminate the particles. The image of the particles is recorded at two time instances using a camera to measure velocity [59] . . . . . . . . . . . . . . . . . . . . . . . . . . 131 E.2 Assembled view of the PIV chamber - basic parts include borosilicate glass cham- ber, and a top polycarbonate lid with ports for sensors . . . . . . . . . . . . . . 133 E.3 Exploded view of the PIV chamber showing the assembly of the copper heating block, borosilicate glass chamber, and a top polycarbonate lid with ports for sensors133 E.4 Velocity vectors obtained from a PIV experiment that are overlaid on top of the captured image. In this image, the shallow long slopes of the saw teeth face towards the left side of the image, and any expected ow due to asymmetric bubble growth will be from right to left in the image. Vectors colored in green and yellow represent measured and interpolated velocities, respectively. . . . . . 134 E.5 Magnitude of velocity vectors obtained from the PIV experiment. Color chart indicates liquid velocities of up to 17 mm/s, although scattered in di erent di- rections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 xviii List of Tables 2.1 DRIE etch parameters and resulting saw-tooth angle . . . . . . . . . . . . . . . 37 2.2 Characteristics of di erent types of test sections fabricated . . . . . . . . . . . . 51 3.1 Summary of curve tting parameters from Fig. 3.8 and Fig. 3.9 . . . . . . . . . 65 3.2 Summary of available models and experimental data for bubble growth during inertia controlled (IC) and heat transfer controlled (HTC) regimes. . . . . . . . 70 A.1 Thermophysical properties of FC-72 and saturated water at 1 atm . . . . . . . . 119 xix List of Abbreviations Acronyms CVD Chemical Vapor Deposition CHF Critical Heat Flux DAQ Data Acquisition System DRIE Deep Reactive Ion Etching HWA Hot Wire Anemometry LUT Look Up Table MABE Microheater Array Boiling Experiment PECVD Plasma Enhanced Chemical Vapor Deposition PID Proportional Integral Di erential Controller PAE Phosphoric Acid and Acetic Acid Etchant PIV Particle Image Velocimetry TEDP Test Equipment Data Package Symbols A constant in Eq. 3.7, mm=s A area, m2 Cp speci c heat, J=kgK xx CD drag co-e cient D diameter, m F Force, N g gravity, m=s2 H height, m hlv latent heat of vaporization, J=kg I supply current, A Ja Jacob number, m k thermal conductivity, W=mK L length, m m mass, kg m exponent in power law curve t (Table 3.1) P pressure, Pa P momentum, kg:m=s Pr Prandtl number q power supplied to the heater, W q00 heat ux, W=cm2 qloss heat lost to the ambient, W R radius, m r radius of curvature, m xxi Re Reynolds number, m Rh heater resistance, t time, s T temperature, C v velocity, m=s V voltage, V Greek symbols thermal di usivity, m2=s growth constant 4P pressure drop, kPa 4Tsub inlet subcooling, C dynamic viscosity, Nsm 2 kinematic viscosity, m2=s density, kg=m3 surface tension, N=m saw tooth angle, degree shear stress, N=m2 Subscripts 1 ambient c cavity xxii h heater l liquid le equilibrium liquid superheat p particle sat saturation w saw tooth surface v vapor xxiii Chapter 1 Introduction Advances in electronics processing and fabrication techniques have led to signi cant miniaturization, allowing manufacturers to pack a billion transistors, carrying out a billion calculations into an area smaller than a thumbnail. Consequently, current supercomputers, which are indispensable for storm tracking, DNA mapping, social networking, and numerous other scienti c research applications, are more than 20 billion times faster than the rst ones built in 1960s. During the same period, the cost per million operations in a supercomputer has reduced by more than a billion fold. Central to these developments has been the evolu- tion in cooling technologies for the dissipation of resultant high heat densities. Liquid-vapor phase change (boiling) of dielectric uids has been at the forefront of cooling techniques starting with the immersion cooling technique employed in Cray 2 supercomputers to the ow boiling technique used in the current TITAN supercomputers. These advances in com- puting technologies aided largely by miniaturization are not limited to supercomputers, but have translated to laptops and hand-held devices. The microprocessors used in the latest computers are much faster than the supercomputers in 1960s and often uses a combination of air cooling and liquid cooling techniques such as forced convection heat sinks and heat pipes, respectively, to dissipate the resulting high heat density. A summary of this evolution in advances in computing power and cooling techniques is shown in Fig. 1.1 and 1.2. To further reduce the footprint of the billions of transistors, signi cant advances are required in the eld of electronics cooling to push the limits of super-computing and bring even more powerful computing to desks and palms. Such advances in electronics cooling, especially liquid cooling, will also be essential and critical to modern space electronics, like Martian rovers, and critical electronics including life support systems in space. 1 Figure 1.1: Evolution of supercomputers since its introduction illustrating the increase in number of Floating point operations per second (Flops) and the reduction in cost per oper- ation. The corresponding evolution of cooling systems used in these supercomputers from a simple liquid cooling system such as in car radiators to advanced phase change cooling systems is illustrated. A popular current laptop is used for comparison which is much faster than the supercomputers in 1985. 2 Figure 1.2: Chart illustrates the adaptation of advanced cooling techniques for high heat density applications compared to the simple, lower heat transfer air cooling techniques for low power, low heat density applications 3 Boiling or liquid-vapor phase change, as a mode of high heat ux dissipation technique can be realized in two di erent ways, namely, pool boiling or ow boiling. Pool boiling involves immersing the electronics in a dielectric uid with a low boiling point, like in the early supercomputers such as Cray 2. The heat from the microprocessor converts the liquid to vapor providing very high heat removal rates. These rates are increased even further if the liquid is owing such as in ow boiling. Flow boiling involves heat rejection to a liquid owing in a closed loop causing phase change to vapor as shown in Fig. 1.2. Such a cooling technique o ers a number of advantages over single phase ow of liquid such as higher heat transfer, lower ow rates required for the same heat transferred, and chip temperature uniformity. The heat transfer rate involved in phase change liquid cooling is also dependent on the regime of boiling. The highest heat transfer rates in boiling are associated with the lower surface temperature nucleate boiling regime which involves the cyclic nucleation, growth and departure of individual vapor bubbles. In contrast, a lm boiling regime, which involves the formation of a thin vapor layer between the surface and liquid, results in lower heat transfer compared to nucleate boiling but higher heat transfer than natural convection, at temperatures high enough to cause failure of any electronic components. Such advanced cooling techniques, however, consume power to facilitate pumping of uid through narrow passages. Reduction of required power to cool an electronic component is a critical factor to reduce the operational cost of the system. Further, it is of paramount importance in space systems, as power is a precious commodity. To further illustrate this, the current martian rover, Curiosity, uses a \pumped uid loop" to e ciently cool the on- board electronic components. If the pumping of uid in a an electronics cooling system can be achieved by devising a technique to achieve self-propelled ow of liquid, the operational cost of the cooling system could potentially be reduced signi cantly. Such a self-propelled liquid cooling system also o ers other advantages such as reduction in system weight by the removal of prime movers such as pumps. 4 The study reported in this dissertation, conducted in collaboration with Oregon State University, describes an innovative concept wherein a heat sink with microscopic asymmetric surface features in silicon is developed to create a self-propelled ow of a dielectric liquid, dissipating heat e ectively while operating in the high heat transfer nucleate boiling regime. Nucleate boiling is sustained by the presence of re-entrant cavities, which are very e ective vapor trapping sites. Under terrestrial conditions (1g), in such a heat sink, heat from the microprocessor (heater) causes nucleation of vapor bubbles (boiling) from the fabricated microscopic cavities located only on the shallow slopes of the silicon saw-tooth structures as shown in Fig. 1.3. The bubbles grow and depart from the sloped asymmetric surface of the ratchets at an angle normal to the shallow slope of the ratchets with signi cantly high velocities. This asymmetric bubble growth imparts an angular momentum to the uid thus e ecting liquid motion in the adjacent liquid. Under microgravity, due to lack of buoyancy, bubbles generated tend to reside on the surface and grow to several times larger than under terrestrial conditions. In a plain at surface, this would cause temperature to rise under the bubble leading to the failure of electronic device being cooled. However, using the micro-structured surface, the residing bubble which conforms to the shape of the structured surface is propelled due to pressure di erences in the trapped thin liquid lm underneath. This prevents any temperature rise and also leads to a self-propelled ow. This phenomenon is illustrated in Fig. 1.4 Primary objectives of the current study include: MEMS fabrication of the novel heat sink with an asymmetric saw-toothed surface with nucleation sites (re-entrant cavities) on only the long slope of the ratchets analysis of bubble growth and departure from such asymmetric silicon structures using high speed imaging environments analysis of e ects of asymmetric bubble growth on net lateral liquid movement using experimental techniques such as hot wire anemometry and particle image velocimetry 5 Net Liquid F low Vapor bubble Liquid Re-entrant Cavity Saw-t ooth Ratchet Heater Figure 1.3: Schematic diagram of the concept under 1g showing the cross-section of the saw-toothed structure. Bubbles grow and depart from the re-entrant cavities at an angle that is normal to the surface. This is hypothesized to result in a net lateral velocity. Figure 1.4: Schematic diagram of the concept under 0g showing the larger size of the bubble due to lack of buoyancy. Bubble slides from left to right due to pressure di erence in the liquid lm underneath, between the crest and trough, that arises due to di erences in radii of curvature (r) 6 experimentation aboard reduced gravity ight to analyze e ect of microgravity on bubble dynamics and thereby its e ects on net liquid movement 1.1 Thermally actuated pumping systems: Literature review Thermally actuated pumping of a working uid has been demonstrated in the past, in both single phase liquid and liquid-to-vapor phase change systems aided by di erent convective mechanisms. The current section reviews the applicability of both these types of systems for cooling of electronics. 1.1.1 Self-propulsion in single phase liquids In the case of heating a liquid, a convective ow can be obtained by Marangoni con- vection and Rayleigh-Benard convection. Marangoni convection [1{5] is a surface tension driven ow while Rayleigh-Benard convection is driven by buoyancy. As the liquid is heated, warmer liquid rises to the top and can move either to the left or right along the surface of the liquid, as illustrated in Fig. 1.5a. In this process, the liquid loses heat to the colder gas. This di erence in temperature along the surface of the liquid causes a gradient in surface tension, since surface tension is inversely proportional to temperature in pure liquids. The colder liquid moves down due to di erence in density. This causes the colder descending liquid to be surrounded by warmer liquid. The alternating pattern between colder liquid with high surface tension and warmer liquid with low surface tension, causes the de ection of liquid surface as shown in the Fig. 1.5b. This local convective ow due to surface tension gradients is known as the Marangoni e ect [6]. A large number of studies have been conducted to analyze the aforementioned concept of Marangoni convection and to understand the e ects of surface geometry [1{4], and the thickness of liquid layer [1, 3, 4] on the obtained convective transport. The surface geometry, especially, plays a very signi cant role since having a asymmetric heated surface geometry aids in setting up temperature gradients in the uid which in turn aids Marangoni convection. 7 Figure 1.5: (a) Image illustrates the local convective ow at the liquid surface and the resulting convective cells (b) The surface tension gradients due to alternating di erence in temperatures of warmer and cold liquids cause de ection of the free surface. Image was acquired from Carey [6]. This was demonstrated by Jo [1] where an asymmetric periodic saw tooth pattern was used for the heated surface and it was observed that the ratchets created asymmetric surface tension forces which induced a pumping e ect in a thin layer of silicone oil over the ratchets. The e ect of ratchet surface on the temperature gradients as observed by Jo [1] is shown in Fig. 1.6. Velocity measurements in oil using die-tracking demonstrated a maximum velocity of 0.86 mm/min at an optimum lm thickness of 1 mm. Stroock et al. [3] analyzed the e ect of liquid thickness and temperature gradient across the liquid on the Marangoni-Benard convective ow over similar saw-tooth pro led ratchets. It was reported that the direction of ow was dependent on the lm thickness and the temperature gradient. The maximum ow rate of oil observed was 2 mm/min at a temperature gradient of 130 C across the oil layer. The observed ow pattern over the asymmetric structures is shown in Fig. 1.7. In a thin lm of oil over symmetrical triangular grooves, it was observed by Alexeev et al. [2] that surface tension gradients induce a ow 8 at the free surface of the lm towards the trough of the groove and the ow is directed downwards at the trough as seen in Fig. 1.8, thus setting up symmetrical vortices over the period of a groove. In addition to Marangoni convection, surface geometry also a ects the heat transfer from the surface. In experiments conducted with 0.5mm of silicone oil lm heated on parallel symmetrical grooves, it was observed that the heat transfer rate was 30% more in the case of sinusoidal surface shape compared to a trapezoidal surface. From the aforementioned studies, it can be inferred that Marangoni convection leads to a very low ow rate and operates in the natural convection regime which may not be applicable for high heat ux applications. The next section will continue the review of phase-change propulsion systems. 1.1.2 Self-propelled ow in phase change systems A number of studies have been conducted in the recent past to develop self-propelled ow systems utilizing liquid-vapor phase change and asymmetry in heated surface structure or surface wetting characteristics. A self-propelled ow obtained during the nucleate boiling regime holds signi cant potential to be used as a high heat transfer, pump-less, phase change cooling system which is one of the objectives of the current study. In phase change systems, bubble dynamics play a signi cant role in the promotion of the pumping e ect. During phase change of liquid on a plain surface, vapor bubble ebullition cycle consisting of nucleation, growth and departure of the vapor bubble has been shown to induce a pumping e ect by driving the superheated liquid away from the surface and bringing colder liquid to the surface [6]. The kinetic energy of the liquid due to spherical bubble growth on a plane heated wall has been studied in the past by Witze et al. [7], and 9 Figure 1.6: (a) Oval ow loop with saw-toothed heated ratchet sections used by Jo [1] (b) Asymmetric temperature gradients induced by the shape of the ratchet pro le, which leads to surface tension gradients leading to liquid movement. The direction of liquid movement is marked in (a) Figure 1.7: Flow pattern in oil observed by Stroock et al. [3] over asymmetric mini-ratchets at velcities up to 2mm/min. Figure 1.8: Surface tension induced ow towards the trough of a symmetrical tri- angular groove which is marked by the movement of glass particles. Alexeev et al. [2] reported that this results in liq- uid ow towards the wall at the trough and away from the wall at the crest, thus setting up a vortex ow. 10 the liquid kinetic energy was expressed as, K:E: = 9:35 l dR dt 2 R3 where, K:E: = kinetic energy of liquid dR dt = bubble growth rate (1.1) From this equation it is also evident that if the bubble dynamics could be altered to cause bubble growth at an angle inclined to the vertical, then the velocity of the resulting liquid motion has a horizontal component which contributes towards the net lateral horizontal velocity. Linke et al. [8] demonstrated that liquid R-134a droplets over a surface with periodic saw-tooth structures, similar to the structure used in the current study, were propelled at velocities as high as 5 cm/s in the lm boiling regime at a signi cantly high temperature. The propulsion of the droplet was attributed to pressure di erences under the droplet in the thin layer of vapor as observed in Fig. 1.9. Pressure di erences arise due to di erence in radii of curvature of the liquid-vapor interface at the trough and crest of the saw-teeth. This is also known as the Leidenfrost phenomenon and the droplets are called Leidenfrost droplets. Theoretical models developed by Yuan and Prosperetti [9] have shown that phase change in uids could be utilized for producing a pumping e ect. It was shown that repetitive growth and collapse of a vapor bubble nucleating at heaters located in a small liquid- lled tube, that connects two reservoirs, can lead to a net lateral liquid ow as long as the bubble is not generated from the mid-point of the tube. The net ow was attributed to unequal liquid columns on both sides of the bubble. Net pumping velocities of 500 cm/s were reported when the bubble is at the ends of the tubes and the velocity approached zero as the bubble move towards the center of the tube. Such systems in the past have led to the development of 11 bubble based micropumps based on pumping due to surface asymmetry which are attractive for micro uidic biological applications. Bubble-based micropumps have also been developed by using asymmetry in structure and surface wetting characteristics. In experiments conducted by Meng and Kim [10], gas bubbles were generated in an asymmetric microchannel structure as shown in Fig. 1.10(i)a. The directional growth of gas bubbles due to asymmetry and bubble displacement using a hydrophobic-hydrophilic junction resulted in a net liquid ow. Liquid ow rates of up to 65 nL/s was reported. A similar bubble based micro-pump involving phase change was demonstrated by Jun and Kim [11]. A net pressure di erential was obtained as a result of surface tension imbalance arising from asymmetric heating in a microchannel. A bubble produced by a heater as shown in Fig. 1.10(ii)a can be translated to the right by switching on the heater to its right, while switching the previous (upstream) heater o . This sequential powering up of the heaters, causes a pressure gradient due to gradients in vapor pressure of the bubble and surface tension. This causes the bubble occupying the entire cross-section to move towards the location of the heated area, thereby sweeping the liquid along. Liquid velocities of up to 160 m/s were reported using isopropanol as the uid in microchannels of hydraulic diameter 3.4 m and a length 726 m. All the aforementioned bubble based micropumps involving single phase ow of liquid and multi-phase ow of gas-liquid can nd applications in electronics cooling applications as a way to pump uid over the hot surface. As described above, activation of liquid motion at the micropump may require an additional mechanism. Also, the discussed mechanisms were not designed for e ective heat transfer from the hot surface. The current study aims at developing a self-propelled ow mechanism, in which the heat transfer from a hot surface, such as a microprocessor, when immersed in a dielectric liquid leads to nucleate boiling which in turn propels the liquid. The current study aims at constructing a heat sink, as illustrated in Fig. 1.3, with periodic, asymmetric, saw-tooth structures on a silicon surface with a pro le that is similar to the surface used by Linke et al. [8]. Nucleate boiling is promoted 12 Figure 1.9: (a) Leidenfrost droplet motion over a saw-toothed surface demonstrated by Linke et al. [8] (b) Droplet propelled by pressure gradients in the thin liquid layer induced by the di erences in radii of curvature of liquid-vapor interface (c) The direction of liquid movement due to the pressure gradient is shown. R-134a droplet velocities of up to 5 cm/s were reported Figure 1.10: Bubble based micropumps that rely on asymmetry in surface structure, surface wetting characteristics and heater locations. (i) Meng and Kim [10] demonstrated preferen- tial gas bubble motion due to surface asymmetry in the form of a neck in the microchannel structure as seen in (a). The gas bubble translation was promoted due to a hydrophilic- hydrophobic junction causing bubble translation to the right as seen in (b). Liquid ow rates of up to 65 nL/s were reported. (ii) Jun and Kim [11] demonstrated bubble mo- tion due to surface tension gradients induced by sequential powering of the heaters. Liquid (isopropanol) velocities of up to 160 m/s were observed. 13 by fabricating re-entrant cavities only on the shallow slope of the saw-tooth structure to promote asymmetric bubble growth which leads to a net lateral ow of liquid. 1.1.3 Fabrication of re-entrant cavities and sloped surfaces: Background The surface structure of the designed silicon heat sink to be used in the current study is unique and hence literature corresponding to its fabrication methods are not very widely available. Hence the fabrication methods that were devised for obtaining the sloped surfaces and re-entrant cavities that open up on the sloped surfaces, required extensive calibration. In the past, fabrication of re-entrant cavities in silicon that are very similar to the pro les developed in the current study, and other bulb shaped re-entrant cavities, have been demonstrated in studies conducted by Goyal et al. [12] Baldwin et al. [13] and Nimkar et al. [14]. The aforementioned studies investigated bubble dynamics involved in nucleation from re-entrant cavities on plain surfaces, associated heat transfer enhancement and optimum spacing between the cavities for phase-change immersion cooling of electronics. An image of the re-entrant cavity used in [14] and a bulb shaped cavity used in [13] are shown in Fig. 1.11. The pyramidal re-entrant cavities used in [12, 14] were fabricated by transferring the patterns of the cavity base that are printed on a glass mask to a silicon wafer coated with photoresist using standard photolithographic procedure. The developed wafer is then dipped in a solution of potassium hydroxide that etches the exposed silicon along the grain angle of silicon, which is 54.7 , to result in a pyramidal structure that opens a mouth on the other side of the silicon wafer. In immersion cooling applications with dielectric liquids such as FC72, due to the low contact angle and highly wetting nature of the uid, the naturally existing cavities or im- perfections on a plain surface are ooded and hence vapor nucleation is delayed, thereby causing signi cant overshoot in wall temperature. One of the ways to mitigate overshoot as shown by Nimkar et al. [14] is to use re-entrant cavities of the shape shown in Fig. 1.11a, 14 that trap vapor more e ciently due to the presence of four corners. Re-entrant cavities also lead to signi cant reduction in onset of nucleate boiling as demonstrated by Baldwin et al. [13]. The mouth diameter of the cavities also signi cantly a ect the bubble dynamics such as bubble growth rate, bubble departure diameter and departure frequency. Numerous other shapes and forms of cavities have been investigated by researchers in an e ort to enhance boiling characteristics and thereby the performance of the phase-change cooled heat sink. Kandlikar et al. [15] studied the e ect of cavities, that are 100 m punch marks as shown in Fig. 1.12a on ow instabilities during ow boiling in microchannels. Ko sar et al. [16] investigated the performance of circular inter-connected re-entrant cavities as shown in Fig. 1.12. It was concluded that such re-entrant cavities decrease the heat ux required to initiate boiling and mitigate ow instabilities in microchannels. Fabrication of sloped surfaces in silicon, similar to the saw-tooth structures in the current study, has been performed in the past primarily for di ractive optical elements such as deep phase Fresnel lens [18]. Such sloped surfaces in silicon are developed by gray-scale lithography and deep reactive ion etching (DRIE) which was rst demonstrated by Whitley et al. [19]. Gray scale lithography is mainly used to develop structures in silicon that have a gradient in height. To achieve such structures in silicon, gray scale lithography is followed by DRIE which is a very common process to etch deep anisotropic structures such as channels or trenches in silicon. To better understand gray-scale lithography, the DRIE process and the factors a ecting the etching process have to be understood. The DRIE process has four main steps as illustrated in Fig. 1.13[20, 21]. The pattern to be etched is transferred on to a photoresist layer on silicon. The photoresist is then developed to form a photomask and expose the silicon layer to be etched Fig. 1.13a. The DRIE process in carried out in an STS ICP etcher, where silicon is etched in two steps - etching and passivation. In the etching step, silicon is etched using SF6 gas, as shown in Fig. 1.13b, and in the passivation step, a Te on like lm is deposited along the side walls and oor of the channel (Fig. 1.13c). In the subsequent etching cycle, the polymer lm only on the oor of the channel is removed by 15 Figure 1.11: (a) Pyramidal re-entrant cavities used in immersion cooled surfaces studied by Nimkar et al. [14]. Such re-entrant cavities are formed by anisotropic potassium hydroxide etch (b) Bulb shaped re-entrant cavity used in immersion cooled heat sinks by Baldwin et al. [13] and Bhavnani et al. [17] Figure 1.12: (a) Arti cial cavities for nucleation formed by punch marks that were used in microchannel heat sinks by Kandlikar et al. [15] (b) Inter-connected re-entrant cavities to form slots that were used in silicon microchannel heat sinks by Ko sar et al. [16] 16 SF6 ion impingement, thereby protecting the walls of the channel from being etched in the subsequent cycles. This process is repeated until the required depth is achieved. It has to be noted that in the etching cycle, the plasma inside the etching chamber consists of a mixture of gases, usually SF6 and O2. The increased concentration of SF6 in the plasma makes the etch more isotropic, thereby lending a slight concave shape to the surface. Increased concentration of O2 in the plasma leads to an increase in anisotropy but also increases the rate at which the photoresist is etched [21]. The ratio of silicon to photoresist rate is called etch selectivity. A typical photolithography process involves transferring a pattern in a mask to the photoresist deposited on a silicon layer, as discussed above. The patterns form either the transparent or the opaque region of the mask. Ultra-violet light passes through the transpar- ent regions of the mask to expose the photoresist layer on the mask, which is later developed to remove the exposed regions of the photoresist, to expose the silicon. In comparison, a gray-scale mask consists of a pattern that ranges from being opaque to transparent as shown in Fig. 1.14a. When ultra-violet light is passed through such a mask, the photoresist de- posited on silicon is exposed to varying depths (Fig. 1.14b). This leads to a gradient in height in photoresist. The next step is DRIE (Fig. 1.14c) which transfers the slope in pho- toresist to a slope in silicon, although the angle of slope achieved in silicon is a function of etch selectivity. Hence, the DRIE process parameters such as the ow rate of gases, and du- ration of etching and passivation cycles have to modi ed signi cantly to achieve the required angle. This step requires extensive calibration, as it involves optimization of a number of parameters. This step of tailoring the etch selectivity by manipulating the DRIE parameters has been investigated by Waits et al. [22, 23]. Although the steps of gray-scale lithography for optical elements, DRIE for forming channels, and wet etching for the formation of re-entrant cavities have been performed in the past separately for individual applications, the combination of three to form a large array of periodic delicate saw-tooth pattern to form a heat sink has never been attempted. 17 Figure 1.13: Steps involved in deep reactive ion etching (DRIE) (a) Formation of a photomask using photolithography (b) Etching of silicon using a plasma containing SF6 (c) Passivation cycle involving the deposition of a polymer material which protects the side walls from being etched in the subsequent etching cycles (d) Etching cycle is repeated. Figure 1.14: Steps involved in gray-scale lithography as illustrated in Waits et al. [23] (a) Design of gray-scale mask (b) Exposure and development of photoresist to obtain a gradient height in photoresist (c) DRIE to obtain a gradient height (slope) in silicon. 18 1.1.4 Bubble dynamics in pool boiling on plain and enhanced surfaces Studies of bubble dynamics involved in liquid-vapor phase change and their interactions with the heated surface and the liquid is critical for understanding the underlying mech- anism responsible for the enhanced heat transfer associated with boiling or phase change. Knowledge of bubble dynamics provides signi cant advantages in the design of phase-change heat exchangers and engineered surfaces. Bubble dynamics, in general, over the past decades were investigated for common uids such as water and/or refrigerants such as R-113, R-134a and surfaces made of common metals. However, with the rising demands of high heat dissi- pation in modern high power electronics, phase-change cooling is well placed as an attractive cooling technique involving a range of surfaces such as silicon with micro-scale features and carbon nanotubes, and dielectric uids such as FC-72 which promote direct contact cooling at low surface temperatures. These new combinations of surface structures, surface scales, and liquids alter the bubble dynamics signi cantly necessitating further study. Investigations on this front have led to signi cant developments in phase change heat transfer for microelectronics such as ow boiling in microchannels, direct contact liquid immersion cooling and enhancements to engineered micro-surfaces. In the recent past, bubble dynamics have also been manipulated for the development of bubble based micro-pumps which are attractive for applications in electronics cooling since the latent heat required for the bubble production is acquired from the microelectronics leading to pump-less, power-free pumping with high heat dissipation. It has been discussed earlier that the structure of heat sink used in the current study alters the bubble dynamics signi cantly to aid in asymmetric bubble growth and departure. This was also illustrated in the schematic diagram shown in Fig. 1.3. To analyze the e ects of structural asymmetry and asymmetry in bubble growth on the net lateral liquid motion, it is important to characterize the bubble dynamics for the surface considered in this study and the experimental conditions tested. Analysis of bubble dynamics, which is one of the primary objectives of the study, includes the study of bubble growth rate, bubble departure 19 frequency and bubble departure diameter as a function of heat ux, subcooling, and type of uid. Such information would also aid in the development of a model for the liquid velocity due to asymmetric bubble growth. Apart from contributing to the model developed, the analysis contributes to the scarce data available for bubble growth from re-entrant cavities and highly wetting uids such as FC-72. Analytical bubble growth models developed in the past can broadly be classi ed into two types - bubble growth in uniform and non-uniform temperature elds. In the former, a vapor bubble grows in a uniformly superheated pool of liquid where the bubble growth is symmetrical, as illustrated in g. 1.15. In models using the uniform superheat theory, bubble growth is mainly attributed to the pressure di erence between vapor inside the bubble (Pv) and the surrounding liquid (P1), which in-turn is related to Tv T1 using the Clausius Clapeyron equation. On the other hand, a non-uniform temperature eld results from the presence of a heated surface at a temperature Tw which is greater than Tsat, leading to bubble nucleation and growth as the bubble stays attached to the surface. Compared to the uniform superheat theory, bubble growth on a heated surface is often not spherical and growth is not symmetrical. Bubble growth from a naturally occurring or a structured cavity on a heated surface, as illustrated in Fig. 1.16, can be classi ed into two types - inertia controlled and heat transfer controlled bubble growth [6]. After the departure of a bubble, colder ambient liquid is drawn on to the surface, during when heat transfer from the surface to the liquid is by transient conduction and this period is known as waiting period which lasts until the boundary layer thickness grows enough to sustain nucleation. During the initial stages of bubble growth, the growth rate is quite high as the radius of curvature increases suddenly, due to the presence of a highly superheated liquid layer adjacent to the vapor interface. This rapid bubble growth is only limited by the inertia of the liquid and is hence known as inertia controlled growth. As the bubble grows radially, a very small thin lm of liquid underneath the bubble, known as the evaporation microlayer exists which evaporates at the vapor interface. With further growth, the superheated liquid lm surrounding the bubble, 20 Figure 1.15: Types of bubble growth models (a) Uniform superheat theory, involving spher- ical vapor growth in an uniformly superheated liquid (b) Bubble growth in non-uniform temperature eld; temperature highest at the wall[25]. known as the relaxation microlayer, depletes during which the bubble growth rate reduces. This regime of bubble growth is known as heat transfer controlled bubble growth. This process is illustrated in Fig. 1.16. Models developed for the case of uniform temperature eld can be divided into inertia controlled growth (Rayleigh [24]) and heat transfer controlled growth (Mikic et al. [25], Plesset and Zwick [26], Scriven [27]). These models were later extended for the case of bubble growth in a non-uniform temperature eld like those of Zuber [28], Cole and Shulman [29], Van Stralen [30] and Mikic and Rohsenow [31]. However, the results in this study will show the inadequacy of these models to predict bubble growth for structured surfaces and highly wetting uids such as FC-72. Models for experimental conditions that are similar to those in the current study are mostly empirical in nature and often present asymptotic growth relationships as a function of time (Lee at al. [32], and Ramaswamy et al. [33]). Such models are, however, simplistic in nature and do not adequately explain the e ects of liquid, and surface such as roughness or re-entrant cavities. The study presented here is an e ort to contribute to the fundamental understanding of the bubble dynamics such as growth rate, departure frequency and diameter experimentally using high speed imaging for the unique surface structure with re-entrant 21 Figure 1.16: Regimes of bubble growth illustrated as shown in Carey [6] (a) Ambient cooler liquid is brought in to contact with the heater surface (b) Growth of boundary layer (c) Bubble nucleation and inertia controlled regime (d) Relaxation microlayer thickness decreases (d) Bubble departure. 22 cavities that was tested with FC-72. Some of the studies that are similar to the conditions of the current study, in the aspect of pool boiling on silicon surfaces using FC-72 include (Hutter et al. [34], Nimkar et al. [14], Ramaswamy et al [33], Moghaddam and Kiger [35], and Demiray and Kim [36]). Of the aforementioned studies, [14, 17, 33, 34] conducted pool boiling studies on silicon surfaces with structured cavities. No studies have been conducted for the nucleation of FC-72 on sloped surfaces with re-entrant cavities. Growth rate of FC-72 was analyzed by Hutter et al. [34] on silicon surfaces with cylin- drical cavities. The observed bubble growth and bubble shape is shown in Fig. 1.17. It was observed that the growth rates vary with time on the order of t 1=2 as noticed in Fig. 1.18, which is similar to the trends reported by Ramaswamy et al. [33] who conducted pool boiling experiments using FC-72 on structured surfaces with square cavities. Nimkar et al. [14] and Bhavnani et al. [17] used pyramidal re-entrant cavities and bulb shaped re-entrant cavities, respectively, during pool boiling with FC-72. E ects of heat ux, cavity spacing, bubble departure frequency were studied. It was observed that the pyramidal re-entrant cavities completely avoided nucleation hysteresis compared to the bulb-shaped cavities. Also, bubble departure diameter and frequency were found to increase with wall superheat. The data thus collected on bubble dynamics at 1g will be corroborated with data from microgravity experiments conducted using the same set up. 1.1.5 Bubble dynamics under reduced gravity It was earlier stated that one of the objectives of the study would be to develop and test a pump less ow loop utilizing surface asymmetry and re-entrant cavities for space electronics applications. This requires a better understanding of vapor bubble dynamics under reduced gravity and hence its e ect on pumping. Due to the complexity of such experiments, literature on micro gravity experiments and the related bubble dynamics is very limited. 23 Figure 1.17: Bubble growth process from a cylindrical cavity on a silicon surface using FC-72 - Hutter et al. [34] Figure 1.18: Bubble growth rate from a cylindrical cavity on a silicon surface using FC-72 - Hutter et al. [34]. The asymptotic growth relationship is marked as the equation of t in the gure. 24 Microgravity testing methods Microgravity, or reduced gravity, conditions can be produced in a number of ways by compensating for the earth?s gravity. Some of the common methods to produce \weightless- ness" or microgravity include a free fall in di erent trajectories, a freely drifting spacecraft or space station, and ballistic rockets. A free fall could be conducted in drop towers and drop shafts where the free falling object on guide rails experiences microgravity due to the absence of forces that oppose gravity [37]. The relationship between free fall height and time can be expressed as, t = s 2H g (1.2) where, t = time of free fall H = free fall height Key drop tower and drop shaft facilities in United States include NASA Glenn Research Center in Cleveland, Ohio, which has a drop tower of 24 m and a draft shaft of 132 m providing a microgravity duration of 2.2 and 5.2 seconds respectively. Marshall Space Flight Center in Huntsville, AL, can produce 4.6 s of microgravity with a drop tower of 106 m. The drawback of testing in a drop tower or drop shaft is the low duration of microgravity and the quality. The low duration signi cantly a ects analysis of processes with longer time scales such as bubble growth in microgravity, which is considerably slower compared to that under earth gravity. Also, in a drop tower or a drop shaft the highest quality of reduced gravity is achieved during the initial period of fall, extending less than 2 seconds, after which the free falling object experiences a residual gravity due to the drag forces of surrounding gas. However, residual gravity can be reduced using evacuated conditions. The short duration of microgravity is also a drawback since any natural convection that started before the drop will not be eliminated completely within the short period of microgravity. 25 Figure 1.19: (a) Parabolic trajectory of a C-9B ight and the achieved reduced gravity periods (b) G-jitters in a parabolic ight due to disturbances such as air turbulence - Straub [37]. A similar method to produce microgravity involves free fall along parabolic trajectories that are own by special aircraft such as KC-135, DC-9, C-9B, and recently Boeing 727- 200F, by NASA from the Johnson Space Center in Houston (Fig. 1.19a). Advantages of such a method include longer durations of microgravity, up to 20 seconds, and the volume of experiments that can be tested. Also, di erent gravity levels can be achieved by altering the trajectory of the ight to simulate lunar and martian gravities. However, the quality of microgravity could be a ected by air turbulence and ight maneuvers, producing gravity uctuations on the order of 10 2g as shown in Fig. 1.19b, known as g-jitters. However, such g-jitter could be dampened by free oating the experiment inside the ight. Parabolic trajectories to produce reduced gravity are also achieved in ballistic missiles known as \sounding rockets" which are categorized as sub-orbital ights. The sounding rockets are powered by a solid propellant rocket motor to achieve the trajectory and the rockets reach altitudes of up to 165 km where the air density is very low, thereby reducing the air drag to produce a high quality ( 10 6g), long duration ( 200 seconds) microgravity regime [38]. The payloads are retrieved using a parachute for re-use and data retrieval. Some 26 of the commonly used facilities include the sounding rockets under the German TEXUS program and NASA?s Terrier-Orion sounding rockets. The best method for microgravity testing is aboard the International Space Station (ISS) which provides unlimited duration of high quality microgravity. The ISS facility used to conduct boiling experiments is known as Microheater Array Boiling Experiment (MABE) located in a module known as the Microgravity Science Glovebox (MSG) [39]. Bubble dynamics and heat transfer under microgravity In boiling at terrestrial gravity, bubble departure from the heated surface is primarily a balance between surface tension which causes the bubbles to remain attached to the surface, and, drag and buoyancy forces which cause the bubbles to detach from the surface. But under reduced gravity, the forces due to buoyancy are negligible and hence the bubble tends to remain at the surface for longer periods resulting in larger bubble departure diameters as has been widely reported in previous studies [40{43]. An image of the large bubble diameters observed by during pool boiling of FC-72 under microgravity is shown in Fig. 1.20. The dependence of departure diameter on gravity can be expressed as, Dd; g Dd;1g = a g 1/2 (1.3) where, Dd = bubble departure diameter g = earth gravity a = tested gravity Because of the bubble size being as large as the heater as seen in Fig. 1.20, the heat transfer is signi cantly reduced and many studies have reported reduction in heat transfer compared to terrestrial gravity using a variety of uids such as FC-72 [38, 41] and R-113 [40]. Heat transfer from a 2.7 mm x 2.7 mm array of heaters to FC-72, studied in [38] shows 27 Figure 1.20: Large bubble of diameter 2.5 mm in pool boiling of FC-72 under reduced gravity at a wall superheat of 30 C - Kim and Benton[38]. that at high wall temperatures heat transfer at microgravity is signi cantly a ected (Fig. 1.21a). It was also demonstrated by Henry and Kim [41] that heat transfer in microgravity is a function of heater size and liquid subcooling. In a large heater, the bubbles are distributed and the presence of a large number of smaller satellite bubbles lead to a higher heat transfer compared to that with smaller heater sizes (Fig. 1.21b). Also, high liquid subcooling resulted in a higher heat transfer compared to lower subcooling. Bubbles of larger sizes, observed under microgravity, also cause dry-out of the surface under the bubble which leads to critical heat ux (CHF). CHF is the maximum heat ux condition beyond which heat transfer decreases and wall temperature increases signi cantly (Fig. 1.21a). Pool boiling experiments on at heater surface using R-113 was performed in space shuttle ights by Lee and Merte Jr. [43]. E ects of heat ux and liquid subcooling were studied. While observations of larger bubble diameters similar to a number of other studies were reported at all tested conditions, an interesting observation of bubble migration was reported. It was observed that a number of small bubbles coalesced to form a very large bubble which departed from the surface and hovered close to the surface. Bubble departure was mainly attributed to the momentum and coalescence of smaller bubbles. Also, the migration of smaller bubbles towards the larger bubble at velocities of 2.5 cm/s noticed at 28 Figure 1.21: (a) Lower heat transfer observed under microgravity in pool boiling of FC-72 at higher wall superheat. Critical heat ux (CHF) also reduced compared to earth gravity - Kim and Benton[38] (b) Heat transfer increases with increasing heater sizes and liquid subcooling - Henry and Kim [41] 29 high subcooling led to a 40% increase in heat transfer compared to similar conditions at earth gravity. An image of the large hovering bubble and adjacent smaller bubbles are shown in Fig. 1.22 Figure 1.22: Large hovering bubble in R-113; smaller adjacent bubbles near the larger bubble were noticed to migrate towards the larger bubble at high subcooling - Lee and Merte Jr. [43]. Similar migration of vapor bubbles on plain heated surfaces has also been reported previously under reduced gravity attributed to g-jitters [42]. Sliding velocities of up to 2.25 cm/s were reported in pool boiling of distilled water. Marangoni and thermocapillary convection [38, 44{46] have also been shown to cause bubble migration due to boiling in a subcooled pool of liquid. From the aforementioned studies it could be observed that the bubble dynamics in pool boiling are considerably altered under microgravity. Also, literature for such microgravity studies is extremely limited owing to the experimental constraints involved in a microgravity experiment, such as duration of microgravity, number of experiments, and the complicated experimental setup required for safety. Data is often limited to pool boiling on at sur- faces and wires in uids such as water, R-113 and, FC-72. However, no studies have been conducted for surface pro les like the one used in current study, and hence no information is available on the e ects of surface asymmetry and re-entrant cavities on bubble dynam- ics. Hence, the current study to analyze the e ect of surface asymmetry at reduced gravity 30 on bubble dynamics and liquid pumping will be a novel attempt and will provide valuable information about bubble dynamics. 31 Chapter 2 Fabrication of Silicon Test Devices The silicon heat sinks used in this study consist of saw-tooth cross-sectioned, asymmetric surfaces and re-entrant cavities that are located only along the long slope of the surface. The heat sinks fabricated in this study are of two types and were designed to be be used in two di erent sets of experiments. One set of experiments, termed as large array experiments, were designed for global measurements of liquid velocity, bubble dynamics and heat transfer. The test devices used in large array experiments consisted of 80 ratchets in silicon. The other set of experiments were designed to study local bubble dynamics and resulting e ects on the adjacent liquid using a small array of silicon ratchets, termed as small array experiments. The small array test device consisted of only eight ratchets of smaller footprint compared to the large array experiments. The small array test devices were fabricated for use at Oregon State University, a partner institution on this research project. While the overall objective of the study, which is to obtain net lateral liquid ow using structural asymmetry, is the same in both sets of experiments, there are di erences in the scale of the experimental setup, methods of experimentation, and techniques used. Because of these di erences, the test devices used were designed and fabricated to cater to speci c requirements of the techniques used and experiments conducted. Some of the important characteristics of the test device required for the di erent experiments are outlined below: Fabricate saw-tooth cross-sectioned ratchets on silicon to obtain a nominal pro le of 30 - 60 - 90 . The surface of the resulting ratchets required a smooth nish as the presence of any surface imperfections will act as nucleation sites. 32 Etch re-entrant cavities from the back side of the ratchets to open up trapezoidal mouth on the sloped face of the saw-teeth. The presence of these features only on the shallow slope is to cause an asymmetry in vapor bubble nucleation thereby leading to an asymmetry in momentum imparted to the liquid. Design and fabricate aluminum serpentine heaters on separate silicon wafers. The traces were designed to prevent electromigration while providing adequate power to induce vapor nucleation. Electromigration is the severance of current carrying metal paths due to excessive current density in a metal of given cross-section [47, 48]. The heaters for the small array experiments were also designed for infrared imaging of cavities from the back side of test devices, requiring optimum spacing between the aluminum heater traces and anti-re ective coating on the heater to prevent re ection from the aluminum traces. Dice the silicon wafer consisting of saw-tooth ratchets into large and small array of ratchets for the two di erent sets of experiments. Bond the silicon wafer consisting of ratchet array to the heater wafer with minimal thermal interface resistance. Due to the di erence in the footprint area of the large and small array test devices di erent bonding schemes were used. Because of the unique cross-section of the test structures utilized, literature correspond- ing to the fabrication processes was limited and fabrication steps involved elaborate calibra- tion and trials to precisely obtain the required structure. Broadly, the fabrication of the heat sink consists of four steps. Gray-scale photolithog- raphy to obtain the saw-tooth ratchets in silicon, etching re-entrant cavities on the shallow slope of the saw-tooth ratchets, fabrication of serpentine aluminum heater on silicon, and nally dicing and bonding the two silicon layers using techniques such as fusion and gold eutectic bonding. These basic fabrication steps are represented in Fig. 2.1. 33 Figure 2.1: Illustration of steps involved in fabrication of the silicon test devices to be used in the large and small array experiments 34 2.1 Fabrication of silicon ratchet array The saw-tooth structures were fabricated on a 100 mm diameter and 0.5 mm thick double side polished <100>silicon wafer. Gray-scale photolithography process involves deposition of photoresist with a gradient in height. The height gradient in photoresist is obtained by allowing UV light to pass through a gray-scale optical mask consisting of patterns ranging from being completely transparent to opaque. This causes light penetrating to increasing depths in photoresist, which when developed results in a gradient in height [18, 22, 23]. The photoresist used in the current study is SPR 220-7. This asymmetric photoresist structure is transformed into a silicon structure of required dimensions using DRIE, as illustrated in Fig. 2.2. The angle of surface etched in silicon is dependent upon the etch selectivity (ratio of silicon to photoresist etch rate) of the photoresist used, the duration of the etching and passivation cycles, and the ratio of gas ow rates in the plasma during DRIE. Extensive calibration was conducted to analyze the e ects of ow rate of the constituent gases in the plasma during the DRIE etching cycle and cycle duration, on the etched angle of the silicon structure and the surface quality. Prior to DRIE, the wafer with the photoresist structure was subjected to an oxygen plasma, known as the plasma de-scum process, to remove a thin layer of photoresist so as to expose a few micrometers of silicon at the trough. Without this step, etching would begin unevenly across the wafer leading to a poor saw-tooth pro le. All the DRIE processes were performed using an STS ICP etcher. As discussed earlier, the DRIE process consists of an etching cycle, during which silicon is etched using SF6 ow, and a passivation cycle which involves deposition of a Te on layer (C4F8) along the side walls. The rst wafer was etched using a standard Bosch process with the ow rate of SF6 set at 130 SCCM and O2 at 13 SCCM in the etching cycle, and 85 SCCM of C4F8 in the passivation cycle. The time duration used for etching and passivation cycles were 13s and 7s respectively. This resulted in a very high etch selectivity of 100 which in turn resulted in a very high angle of the saw-tooth and a rough surface. Hence to reduce the selectivity i.e., increase the photoresist 35 Silicon Photoresist Silicon Silicon 1. Double side polished 100mm silicon wafer 2. Gray scale deposition of photoresist at a desired angle 3. Deep Reactive Ion Etching (DRIE) to obtain the saw tooth angle of 30 deg. 30 Figure 2.2: Steps involved in the fabrication of silicon saw-tooth ratchets removal rate, the oxygen ow rate was increased sequentially in the subsequent cycles as oxygen has been shown to etch the photoresist faster [20, 23]. Reduction of the duration of passivation cycles also has the e ect of increasing the photoresist etch rate and hence in all subsequent recipes the duration of passivation cycle was reduced to 3s. The recipes followed for DRIE in the subsequent trials have been tabulated in Table 2.1. In recipe A, the ow rate of O2 was increased drastically which decreased the selectivity resulting in an angle of 19 . However, short passivation cycles resulted in poor surface quality due to the formation of black oxide or silicon grass as shown in the SEM image in Fig. 2.4. These nanostructures in silicon or grass are generally formed when the polymer material is not completely etched due to short etching cycles or low concentration of etching species (F radicals) in the plasma [20]. This scenario of low concentration of etching species arises when O2 ow rate is increased in the plasma. It has also been reported that high concentration of oxygen forms micromasks which results in grass like structures due to the directional etching [21]. Hence, to prevent the silicon grass formation, the concentration of 36 the etching species can be increased by increasing SF6 ow rate. But increasing SF6 also leads to deterioration of anisotropy [21]. Hence, in all the subsequent recipes B, C and D short periodic etching lasting not more than 10 seconds using SF6 in the absence of O2 and C4F8 was introduced in between sets of etching/passivation cycles. Figure 2.3: Illustration of DRIE parameters discussed in Table 2.1 Table 2.1: DRIE etch parameters and resulting saw-tooth angle Recipe Flow Rate in the Etching Cycle O2=SF6 (SCCM) Duration of Etching/ Passivation Cycles (s) Angle ( ) Depth d ( m) A 80/100 13/3 19 235 B 65/100 13/3 24 305 C 50/100 13/3 32.15 440 D 53/100 13/3 30.5 415 Since the objective is to achieve an angle of 30 for the saw-tooth, it was required to slow down the photoresist etching rate and hence in recipe B, O2 ow rate was decreased to 65 SCCM keeping the other parameters constant along with periodic 10 second SF6 etching. This procedure resulted in a saw tooth angle of 24 without the silicon grass. In recipe C, by further decreasing the O2 ow rate to 50 SCCM an angle of 32.5 was achieved and nally with 53 SCCM of O2, noted as recipe D, an angle of 30.5 was achieved without any grass formation. An image of a cross-section of the ratchet array with a nominal saw-tooth angle of 24 and without any silicon grass is shown in Fig. 2.5. 37 Figure 2.4: SEM image of the ratchets with an angle 19 . The inset shows an SEM image of silicon nano-structures, known as \silicon grass", formed during the DRIE process with Recipe A. Figure 2.5: Cross-section of a ratchet array with a nominal saw-tooth angle of 24 38 2.2 Fabrication of re-entrant cavities The next step in the process was to etch the re-entrant cavities from the back side of the saw-tooth structure using potassium hydroxide (KOH), known as anisotropic wet etching process. The fabrication steps involved in etching the re-entrant cavities are illustrated in Fig. 2.6. Before the etching process, a 2000 A silicon nitride layer was deposited using low pressure chemical vapor deposition (LPCVD), which acts as a barrier layer preventing the KOH from etching silicon. To prevent any damage to the delicate ratchets during the process of etching cavities and subsequent handling of the wafer, a 500 m thick, 100 mm diameter Pyrex wafer was stuck on top of the ratchets. The Pyrex wafer was coated with photoresist AZ 5214 after treatment with hexamethyldisilazane (HMDS) vapor for 5 minutes which improves adhesion of photoresist to the surface. The Pyrex wafer with the photoresist is stuck to the ratchet side of silicon wafer and baked on a hot plate at 100 C. The back side of the saw teeth was then coated with photoresist AZ 5214 after treatment with HMDS. This is followed by patterning of the photoresist using an optical mask to de ne an array of squares of side 0.655 mm and 0.246 mm for each pair of cavities - large and small respectively as shown in Fig. 2.7. Each saw tooth had eight such pairs spaced equidistantly along the transverse length of the saw tooth. The pattern of cavities on the optical mask held on the back side of the saw teeth was aligned with the saw teeth on the other side. This ensured that the cavities were etched only on the shallow slope. The alignment process was performed using a mask aligner consisting of two microscopes - one on the side of ratchets and other on the side of cavities, and the position of mask was adjusted until the cavity pattern was aligned to the ratchets. The photoresist was then developed to expose the nitride layer as shown in Fig. 2.6(iii). This exposed silicon nitride is etched to transfer the pattern to silicon. The wafers were then dipped into a solution consisting of 30% KOH in DI water with 20% isopropyl alcohol. The temperature of the solution was maintained at 80 C during the etching process. The beaker containing the mixture is covered by a re ux condenser that was mounted on top to prevent evaporation of the solution thereby maintaining the 39 integrity of mixture. The etching process occurs along the grain angle of silicon, which is 54.6 . The anisotropic etching process was periodically monitored by measuring the depth of etch under the microscope until the cavities breached the other side of the wafer on the long slope of the saw teeth to form a trapezoidal mouth Fig. 2.6(vi). Once the cavities were formed, the photoresist was stripped by cleaning with acetone, isopropyl alcohol and DI water. The remaining silicon nitride was removed using a phosphoric acid etch at 154 C. After the completion of all the steps, the Pyrex wafer was removed by dipping the wafer in acetone until the Pyrex wafer parted from the silicon wafer. It has to be noted that test devices with a saw-tooth angle of 24 has only one cavity along the shallow slope of the ratchet compared to the two cavities along the long slope in other test devices. An SEM image with cavity mouth measurements for a 24 test device is shown in Fig. 2.8. This can be attributed to misalignment of the mask used to pattern the cavities on the back side of saw teeth. During DRIE, due to non-uniformity in etching, the saw tooth angle across the wafer varies between the outer periphery and the center of the wafer. This also leads to a variation in cavity mouth sizes. So even a slight misalignment could cause huge shifts in the cavity mouth diameter. 2.3 Fabrication of heater An aluminum serpentine heater was used in the study to supply heat for the pool boiling experiments. Two di erent heater designs were used for the study - one for the large array test device and another for the small array test device. The two di erent heater designs are shown in Fig. 2.9. The steps involved in the fabrication of heater are shown in Fig. 2.10. All the heaters were fabricated on a 100 mm diameter and 0.5 mm thick double side polished < 100 > silicon wafer. The wafer was oxidized in a furnace to deposit 5000 A of silicon oxide. This is followed by evaporating a 2 m thick layer of aluminum on to the SiO2 surface using e-beam evaporation process. Photoresist AZ 5214 was deposited on the aluminum layer for photolithography. The heater design was then transferred to the photoresist from an optical 40 Silicon Silicon Nitride Silicon Silicon Nitride Silicon (ii) Chemical vapor deposition of nitride Silicon Nitride Silicon Nitride Silicon Photoresist Photoresist Silicon Nitride Silicon Nitride Silicon Silicon (iv) Nitride etch to expose silicon (vii) Nitride etch using phosphoric acid Photoresist 56 56 Silicon Nitride Silicon Nitride Silicon (vi) Strip photoresist using acetone (i) Saw-tooth ratchet after DRIE (iii) Patterning of cavities using photolithography 56 56 Silicon Nitride Silicon Nitride Silicon (v) Anisotropic wet etching of silicon using KOH Figure 2.6: Illustration of fabrication steps involved in etching re-entrant cavities from the back side of saw-teeth to form trapezoidal cavity mouths only on the shallow slope 41 Figure 2.7: Isometric and back side view illustrations of small and large array cavities on the shallow slope of saw-toothed surface Figure 2.8: SEM image showing the top view of saw teeth and cavity mouth measurements in the test device with a saw-tooth angle of 24 42 mask using a mask aligner. This is followed by the development of photoresist to expose the aluminum layer as per the heater design. The heater was formed by etching the exposed aluminum in a solution of phosphoric acid and acetic acid etchant (PAE) for 60 minutes. The heater was then diced to the size of the footprint of large array test section. The fabrication of heater for small array test section involved additional steps. A thermal isolation trench was etched using DRIE along the perimeter of the heater to promote one dimensional heat transfer across the heater wafer to the silicon ratchets. This process is conducted along with the process to reduce the diameter of the heater wafer to three inches to t in the test setup for the small array experiments. The wafer diameter was reduced by etching a ring of diameter equal to three inches through the silicon using DRIE. Prior to the etching of isolation trenches and the three inch ring, photoresist was coated on the side of heater and patterned using one mask for both the aforementioned steps. Since the trench was only designed to be 350 m deep, after etching to this depth the trench was covered with photoresist to prevent any further etching. However, etching along the three inch ring continued until the silicon wafer was completely etched through to reduce the wafer to a diameter of three inches. The photoresist was stripped from the wafer using acetone. Further, to facilitate IR temperature measurements the entire surface of the heater was coated with a 1.1 m thick layer of silicon nitride using plasma enhanced chemical vapor deposition (PECVD). This layer of silicon nitride serves to prevent the re ection of infrared radiation. Photoresist was spun on top of the silicon nitride layer and patterned using a mask in a mask aligner to expose the nitride layer over the electrical contacts. This was followed by etching the nitride layer from the electrical contact pads of the heater to enable power supply as silicon nitride acts as an insulator. The additional steps required in the heater for small array experiments are illustrated in Fig. 2.11. Images of fabricated aluminum serpentine heaters on silicon for the large and small array test devices are shown in Fig. 2.12. 43 A D CB B A D C Figure 2.9: Sketch of aluminum serpentine heaters used in large and small array experiments that were fabricated on silicon. The width and thickness of heater traces are 500 m and 2 m respectively. The electrical contact pad con guration is labeled in the gure. The arrangement of electrical contact pads for the small array heater is based on the arrangement of electrical contacts in the wafer xture. 44 SiO2 Silicon SiO2 (iii) E-beam evaporation of aluminum of thickness 2 m Aluminum SiO2 Silicon SiO2 (iv) Deposition of photoresist AZ 5214 Aluminum Photoresist AZ 5214 SiO2 Silicon SiO2 (v) Exposure of development of photoresist using the heater mask Aluminum SiO2 Silicon SiO2 (vi) Aluminum etch by immersing in PAE for 60 min SiO2 Silicon SiO2 (vii) Photoresist stripped by immersing in acetone Aluminum heater traces Silicon (i) Double side polished 100mm diameter, 500 m thick silicon wafer SiO2 Silicon SiO2 (ii) Wet oxidation of wafer at 1000 C to obtain 5000A of oxide Figure 2.10: Fabrication of aluminum serpentine heaters used in large and small array test devices. 45 Silicon SiliconNitride SiliconSilicon Silicon SiO2Al heater Silicon SiliconNitride SiliconSilicon Silicon Al heater Figure 2.11: Fabrication of isolation trenches on the back side of heater footprint and redution of wafer diameter to 3 inches. Silicon nitride was coated to eliminate re ection of radiation during infrared measurements made during the small array experiments at OSU. 2.4 Bonding of saw-tooth ratchet wafer to the heater wafer Bonding of the silicon wafer containing the saw-tooth ratchet array to the heater wafer was performed di erently for the large array and small array test structures. Large array test structures were bonded to the heater wafer using fusion bonding where the bonding between two ultra- at extremely clean surfaces is due to bonding bridge re- placements and bonds between OH groups [49]. Surfaces to be fusion bonded are required to be extremely clean, particle free, at, smooth and hydrophilic. The surfaces to be bonded were rendered hydrophilic using an RCA-1 clean which is used to remove organic residues from the silicon wafer. The recipe for RCA-1 clean is 5 parts of DI water, 1 part hydrogen peroxide (H2O2), and 1 part ammonium hydroxide (NH4OH). The solution was heated up to 70 C when it starts boiling. The surfaces were immersed in to the solution for not more than 5 minutes. This was followed by rinsing the surfaces with DI water and thoroughly drying it. The surfaces to be bonded were immediately brought into contact with each other 46 (a) Fabricated aluminum serpentine heater on a silicon wafer for large array test devices (b) Diced aluminum serpentine heater (c) Fabricated aluminum serpentine heater on a 3" silicon wafer for small array test devices (d) Back side of the small array heater showing the isolation trenches (350 m deep) along the perimeter of the heater footprint Isolation trenches Figure 2.12: Images of fabricated aluminum serpentine heaters used in large and small array experiments. The diameter of heater wafer for the large and small array test experiments are 4 inches and 3 inches respectively. 47 mechanically at room temperature. Under the right conditions, the two surfaces would bond together instantly, although weakly. This was followed by placing the weakly bonded surfaces in an oxidization furnace under pressure at 1100 C for 2 hours. The resulting bonds have a very high fracture strength of up to 20 MPa [49]. Because of the high temperature, any aluminum on the surface would soften or melt away. Hence, for the large array test devices, the aforementioned heater fabrication process was carried out after the bonding process. The fusion bonding process required multiple trials at di erent oven temperatures and durations before achieving one successfully bonded test device. By conducting initial calibration tests for fusion bonding at two di erent temperatures of 450 C and 1100 C, it was learned that the voids were signi cantly reduced at higher temperature. The fusion bonding process of large array silicon saw-tooth ratchet array with the heater is schematically represented in Fig. 2.13 Silicon ratchet Silicon heater wafer Flat, ultra clean, hydrophilic surfaces mechanically brought into contact at room temperature. Bond strength increased by heating at 1100 C in a furnace for 2 hours Fusion bonding Figure 2.13: Illustration of fusion bonding process used to bond the large array of silicon saw-teeth to the heater wafer. The bonding of the small array test devices by fusion bonding process could not be carried out due to the very small area available for bonding. For the small array test devices, bonding of the saw-teeth to the heater was achieved using gold eutectic bonding. Gold eutectic bonding is a low temperature process which involves the use of a solder made of gold-tin alloy which when held between the surfaces to be bonded and heated to the eutectic temperature of the solder, the materials fuse to provide a high strength bond. In this process, both the bonding surfaces were prepared by evaporating 1000 A titanium followed by 2700 A 48 of nickel and nally 1500 A of gold. Titanium improves the adhesion of gold to silicon [49] and nickel serves as a di usion barrier between gold and silicon [50]. The deposition of layers of Ti-Ni-Au on the back of heater wafer was performed after coating photoresist and patterning the back side to only expose the silicon spanning the footprint of the heater. After the deposition of Ti-Ni-Au a lift-o process is performed by immersing the heater wafer in acetone. This causes the thin layer of photoresist under the layers of Ti-Ni-Au to dissolve, thereby causing the lift-o of Ti-Ni-Au from the surface except for the area where the layers are directly in contact with silicon, thus leaving layers of Ti-Ni-Au to span only the footprint of the heater. The bonding process was carried out using an FC150 automated die bonding machine where an upper arm holds the die and the substrate (heater) rests on a stage. A 1 mil thick alloy of 80% Au 20% Sn was used as a solder in the bonding process. The solder with a eutectic temperature of 280 C was advantageous since this temperature is well below the melting point of aluminum used in the heater. The bonding process was carried out at a temperature of 300 C by applying a weight of 200g for a period of 100 s. This process is illustrated in Fig. 2.14. The bonded test structures were inspected for voids using x-ray imaging. Presence of voids at the interface would appear dark in an x-ray image. From the x-ray images shown in Fig. 2.15 it can be noticed that the bonding was void free. The dark square regions in the x-ray image are the re-entrant cavities. It can also be noticed that some of the cavities appear lighter as the solder melted into the cavities. In summary, two types of large array test sections were fabricated - one with a saw tooth angle of 24 and the other with an angle of 31 . The 24 test section has one cavity along the long slope between the crest and trough of each saw tooth, and 31 test section has two cavities. The small array test sections that were fabricated, similar to the large array test section, di ered in saw tooth angle and number of cavities on the long slope of the saw tooth. Additionally, small array test sections also di ered in the number of rows of cavities along the transverse length of saw-tooth. In only one type of small array test sections, the 49 Figure 2.14: Illustration of gold eutectic bonding process used for bonding the small array saw-tooth ratchets to the 3 inch heater wafer (a) X-ray image of bonded 24 small array silicon saw-tooth test section with 3" heater wafer. Voids appear dark in the image. The dark square spots in the image are the re-entrant cavities. No voids in the bond were noticed. Some cavities appear lighter since the solder melts and flows into the cavities partly. (b) X-ray image of bonded 31.5 small array silicon saw-tooth test section with 3" heater wafer. No voids in the bond were noticed. A few smaller cavities appear to be filled partly by the solder Figure 2.15: X-ray image showing the bond quality of gold eutectic bonding process 50 number of rows of cavities were reduced from four to one to reduce the nucleation activity. Characteristics of all the test devices built are summarized in Table 2.2 Table 2.2: Characteristics of di erent types of test sections fabricated Test section parameters Large array test section Small array test sectionA B A B C Number of saw teeth 80 80 7 7 7 Saw tooth angle 24 31 24 31 31 Number of cavities along the long slope of the saw tooth 1 2 1 2 2 Number of rows of cavities along the transverse length of saw tooth 8 8 4 4 1 Foot-print area (mm x mm) 80 x 20.32 80 x 20.32 8 x 11.75 8 x 11.75 8 x 11.75 51 Chapter 3 Terrestrial Experiments 3.1 Experimental Setup The silicon test device used for the large array, terrestrial, pool boiling experiments shown in Fig. 3.1a, is made of two layers - the asymmetric saw-toothed heat sink, and a serpentine aluminum heater. The heat sink has an asymmetric saw tooth cross-section with a 24 90 66 pro le and 1 mm pitch. The heat sink constitutes 80 such saw-teeth spanning a foot print of 80 mm x 20.5 mm. The asymmetric cross-section of the saw teeth was obtained using a combination of novel gray-scale photolithography and Deep Reactive Ion Etching (DRIE). The long, shallow slope of each saw tooth is structured with re-entrant cavities, or vapor trapping sites, that aid in triggering bubble nucleation. The pyramidal re-entrant cavities, fabricated using a wet etching procedure, have a trapezoidal mouth of size ranging between 200 250 m. An SEM image of the cavities is shown in Fig. 3.1b. Each saw tooth has 8 cavities spaced equally along the ratchet on the long slope and in total the heat sink consists of 640 cavities acting as nucleation sites. Heat was provided using an aluminum serpentine heater shown in Fig. 3.1c. The voltage leads on the heater also aid in surface temperature measurements. The fabrication details of the asymmetric structure with re-entrant cavities and heater were discussed in the previous chapter. The bonded test section is soldered on to a printed circuit board for electrical connections and under- lled for structural integrity. The board also serves as the base of a transparent polycarbonate open channel which is 130.8 mm long, 25.4 mm wide and 7 mm deep, and mounted right on top of the test device as illustrated in Fig. 3.1d. The channel serves to con ne the volume of uid for realization of net liquid ow. The test board was suspended from the lid into a pool of FC-72 (C6F14), a highly dielectric highly wetting uid, contained in 52 Figure 3.1: Silicon test device Figure 3.2: Boiling chamber assembly 53 an aluminum boiling chamber (Fig. 3.2) of dimensions 33 cm x 23 cm x 22 cm (L x W x H). Properties of FC-72 are summarized in Appendix A. A thermistor and a pressure transducer were used to measure liquid temperature and absolute pressure respectively. A re ux coil condenser open to the atmosphere helped maintain atmospheric pressure inside the boiling chamber and condense the vapor. Prior to experiments, FC-72 was thoroughly degassed by boiling using the submerged cartridge heaters in the boiling chamber. The cartridge heaters were controlled using a PID controller to obtain the required pool temperature along with chilled water circulation through copper cooling coils immersed in the pool. Note that an earlier embodiment of the uid chamber, in the form of a closed loop was built, but had to be replaced in light of issues associated with vapor management in the small volume. Details of that ow loop are in Appendix B. The boiling chamber was also equipped with quartz windows for high speed imaging. A Phantom V310 high speed camera was tted with K2SC long working distance microscope lens for capturing bubble images. Also, to obtain di erent levels of magni cation close up objectives CF-1, CF-2 and CF-3 were tted to the lens. Images were captures at 3200 fps with a resolution of 1280 x 800. A 250 W halogen lamp was used as a light source. The test section heater was powered by an AMREL 220V programmable DC power supply. Data including the supplied current, voltage measurements from the test section heater, thermistor and pressure transducer was recorded using NI data acquisition system and LabView. The four slot NI data acquisition system chassis consisted of three data ac- quisition modules - NI 9201 for voltage measurements from the pressure transducer, NI 9219 (universal input module) for anemometry, heater voltage, and thermocouple measurements, and NI 9227 for measurements of current supplied to the test section heater by the DC power supply. 54 3.2 Results and Discussion Pool boiling experiments were conducted using FC-72 at atmospheric pressure. Exper- imental parameters tested include a heat ux range of 0 - 4.5 W=cm2 and pool subcooling ranging from 0 C to 20 C. After degassing the uid, the pool temperature was increased us- ing the cartridge heaters until the required subcooling was achieved. The applied heat ux was increased and once steady state was achieved data were acquired and high speed images of bubbles were recorded. This procedure was repeated for a cycle comprising increasing heat ux followed by decreasing heat ux. It is to be noted that the low heat ux range is mainly due to constraints in heater design and not limited by critical heat ux. The results from the large array experiments discussed in this section include bubble dynamics such as bubble growth, bubble departure frequency and diameter, heat dissipation characteristics of the surface, and liquid velocity measurements. 3.2.1 Bubble Dynamics using High Speed Photography Bubble dynamics were studied using the high speed images of bubble growth recorded during the experiments. The captured high speed images were processed and analyzed using NI Vision Assistant software to estimate bubble characteristics. Fig. 3.3 shows bubble departure from the re-entrant cavities on the shallow slope of the saw-teeth at a heat ux of 2.2 W=cm2. It was observed that the bubbles grew and departed at an angle normal to the slope of the ratchets. The angle of departure could be noted by observing the angled line of departed bubbles in the saturated pool at the left end of saw-tooth array. To further illustrate the bubble dynamics at the saw-toothed surface, Fig. 3.4 shows a close-up image of bubble growth and departure from re-entrant cavities on the shallow slope of the saw-teeth. As discussed earlier, it can be observed that the bubble growth is normal to the slope of the saw-teeth and the bubbles depart in a direction normal to the sloped surface at high velocities. This asymmetric growth with respect to the vertical axis, was consistent across the entire surface and at all tested conditions. The growth of 55 Angle of bubble departure Saw-toothed surface Figure 3.3: Image showing bubble departure from re-entrant cavities at an angle normal to the sloped surface of saw-teeth during saturated pool boiling at a heat ux of q" = 2.20 W=cm2. the bubble continues until the buoyancy force overcomes the surface tension force to cause bubble departure. The departed bubble, as noticed in Fig. 3.4 transits in the pool at an angle that is approximately normal to the sloped surface. In the sections to follow, it will be demonstrated that such asymmetric growth of bubbles has the potential to impart a net lateral motion of liquid in the immediate vicinity using a semi-empirical momentum transfer model. 3.2.2 Bubble growth - Experiments Fig. 3.5 shows one complete cycle of bubble ebullition cycle, consisting of bubble nu- cleation, growth and departure from a re-entrant cavity at 2.0 W=cm2 and saturated pool conditions. The vapor bubble nucleating from the cavity is initially hemispherical in shape. The hemispherical bubble grows rapidly, entirely in a direction normal to the shallow slope of the saw-toothed surface and not laterally. In a naturally occurring nucleation site on a planar surface, bubble growth would be concentrated in both lateral and vertical directions. This 56 1 mm 22.188 ms 17.188 ms 10.625 ms 4.062 ms 0 ms 1 mm 1 mm 1 mm 1 mm Figure 3.4: Bubble growth and departure from re-entrant cavities at q" = 2:0W=cm2 and saturated pool conditions. Bubbles nucleating from neighbouring cavities are marked with a di erent color. Relative time stamps are marked at the bottom right of each image. 57 initial rapid bubble growth is inertia controlled where the driving potential for bubble growth is the temperature di erence between the thin superheated liquid layer (relaxation micro- layer) surrounding the bubble, and the vapor. During this initial rapid growth of the bubble normally into the pool, the bubble remains pinned to the cavity as the bubble foot-print remains constant while the dome becomes more spherical. With increasing time, the dome grows larger in size, although at a considerably slower rate as the microlayer depletes. The growth continues until the buoyancy force is large enough to overcome the surface tension force that holds the bubble. As the bubble begins to depart, the neck of the bubble narrows as it breaks and leads to bubble departure normal to the surface. It was also observed that the waiting time, the period between bubble departure and nucleation of the next vapor bubble, is less than 312.5 s which is the time interval between two successive image frames. An interesting observation from these bubble images was the shape of the bubble as it is attached to the surface. Vapor bubbles in highly wetting uids such as FC-72 are generally observed to be more spherical and possess a very low contact angle due to the low surface tension of the liquid. However, in all the bubble images captured in this study the bubbles are consistently \light-bulb" shaped and the contact angles appear to be very high between 60 90 . Similar bubble shapes (Fig. 1.17) were reported by Hutter et al. [34] during pool boiling of FC-72 on a plain silicon surface with cylindrical cavities. This high contact angle also suggests that the evaporation microlayer, which often exists as a thin liquid layer underneath a bubble, as shown in Fig. 1.16, is less signi cant or nearly absent . It is the evaporation of this microlayer which has been shown in the past to contribute signi cantly to the rapid bubble growth in the inertia controlled regime and high heat transfer rates involved in nucleate boiling [30, 51]. However, Demiray and Kim [36] studied the nucleate boiling heat transfer under a single FC-72 bubble and observed that the microlayer contribution is not signi cant compared to transient conduction and microconvection contributions, which also supports the earlier observation about the evaporation microlayer in this study. Similar conclusions were also drawn in [35] in pool boiling of FC-72. 58 Figure 3.5: Ebullition cycle of a bubble at q" = 1.6 W=cm2 and 21 C subcooling In the current study, bubble diameter was estimated from the captured images utilizing standard image processing techniques using NI Vision Assistant software. The main steps involved in processing a single captured frame is shown in Fig. 3.6. The raw image is rst enhanced using a Look Up Table (LUT) with a power (1/x) function to increase the contrast. This is followed by a ltering technique using a convolution theorem to highlight the details such as bubble edges. Next, image thresholding is performed to binarize the image using a threshold value of image intensity. In this process, the pixels with an intensity above the threshold are identi ed (these appear white in the image). This aids in eliminating the noise emanating from light re ection such as at the center of the bubble. An image mask is applied to remove the saw-tooth from the images. A clear image of the saw-tooth without nucleation is selected to be used as a mask. This step is followed by other ltering techniques such as removing smaller particles and regions near the border. The hole in the bubble due to image thresholding is lled and the nal image in the process is used to calculate projected bubble area and perimeter among other parameters. Using the projected area of the bubble from the nal step of image processing shown in the gure, bubble diameter of the non-spherical bubble is calculated as the equivalent diameter of a projected circular disk with the same area as that of the bubble. This estimate of diameter for non-spherical entities is also known as the Waddel disk diameter which can be expressed as D = 2 r A (3.1) 59 This process is repeated for hundreds of frames involved in a single bubble growth cycle and multiple bubbles at di erent experimental conditions. Figure 3.6: Image processing steps of a single bubble frame at 178.44 ms in Fig. 3.5 The bubble diameter estimated using this process during a single growth cycle is shown in Fig. 3.7 as a function of time. As described earlier, it could be noted that the growth of bubble involves a rapid initial inertia controlled growth phase right after nucleation followed by a slower heat transfer controlled growth phase leading to departure. This observation agrees with the results reported in studies by Ramaswamy et al. [33] and Hutter et al. [34] for FC-72, and Lee et al. [32] for R-11 and R-113. Using a linear curve t for the inertia controlled growth and a power law t for heat transfer controlled growth for the single bubble shown in Figures 3.5, the asymptotic bubble growth rate relationship between bubble diameter and time can be obtained. For the inertia controlled growth, the bubble diameter was observed to scale linearly with time expressed as [24, 25], D = At (3.2) where A = 2 7 hlv v T lTsat 1=2 This linear relationship is similar to the experimental growth rate reported in a number of other studies in the literature for the inertia controlled regime . In the heat transfer controlled regime, the asymptotic relationship is of the form, D tm where, and m 60 D 1.43t 0.24D 150.8t Figure 3.7: Estimated bubble growth for images shown in Fig. 3.5 at q"=1.6 W=cm2 and 21 C subcooling are curve tting parameters. The parameter is also termed as a growth constant [27, 29] for the heat transfer controlled regime which is given as a function of wall superheat and uid properties. The bubble growth expression and growth constant as reported in Cole and Shulman [29] is, D = 4 p t (3.3) where, = cp Ja c = experimentally determined constant Ja = Jacob number = (Tw Tsat)Cp;l l vh lv In the above equation, the value of c was experimentally determined to be 1, =2, p3 by Fritz and Ende [52], Forster and Zuber [53], and Plesset and Zwick [26] for bubble growth in a uniformly superheated liquid. 61 The value of m for the experimental conditions shown in Fig. 3.7 is 0.24 which is much lower than the t1=2 relationship reported by a large number of studies in the literature for water [25, 29, 54] and for FC-72 [33]. In a recent study conducted with FC-72 and structured cavities in silicon, Hutter et al. [34] reported values of 0.37 and 0.53 for m, and in studies with saturated R-11 and R-113 Lee et al. [32] observed m to be between 1/5 and 1/3. The lower value of m reported in the current study could be due to the lower liquid equilibrium superheat, 0.1 K for the conditions shown in Fig. 3.7 , experienced due to larger cavities. In comparison, an equilibrium superheat of 0.92 K was reported by Hutter et al. [34]. The equilibrium liquid superheat is the temperature rise of the liquid above the saturation temperature (Tle Tsat) which is a requirement for vapor nucleation to be sustained. It is de ned as, Tle Tsat = 2 Tsat vhlvrc (3.4) where, rc =cavity mouth radius Fig. 3.8 shows the e ect of heat ux at a liquid subcooling of 21 C. As the heat ux increases, the wall superheat increases which vaporizes the liquid near the bubble interface at a faster rate and hence the bubble growth rate increases with heat ux. Also as the wall superheat increases, surface tension of pure uids which holds the bubble on the heated surface decreases. Additionally, since the bubble grows at a faster rate, at a given time the bubble size is larger at a higher heat ux thereby leading to increased buoyancy force which is responsible for bubble detachment. This increasing magnitude of buoyancy forces and the decreasing magnitude of surface tension forces, lead to faster bubble detachment, or smaller overall growth periods as heat ux increases which is evident in Fig. 3.8. These observations corresponding to the in uence of heat ux, and hence wall temperature, on growth rate accord well with studies conducted especially with FC-72 such as Hutter et al. 62 [34] and Ramaswamy et al. [33]. The increase in rate of bubble growth with increasing heat ux was also noticed in the inertia controlled regime which is supported by the increasing value of A in the linear relationship, D = At. The experimental values of A for di erent heat ux values are summarized in Table 3.1. From Fig. 3.8, it could also be noticed that, at a constant subcooling of 21 C, the transition from inertia to heat controlled regime occurs within a narrow range of bubble diameter, 0.45 to 0.55, and observed to be una ected by heat ux. Similarly, when the liquid subcooling is decreased (increasing pool temperature), the wall superheat increases leading to faster bubble growth rates as seen in Fig.3.9 for a nominal heat ux of 1.0 W=cm2. But unlike Fig. 3.8, with decreasing subcooling the inertia controlled regime was sustained for higher bubble diameters which could be attributed to the increasing thermal boundary layer thickness with decreasing subcooling. The e ect of independent parameters - heat ux and subcooling, on the curve tting parameters and m in the asymptotic relationship for heat transfer controlled regime can be studied from Figures 3.8 and 3.9, which are summarized in Table 3.1. In Table 3.1, the units for ?A? and ?B? are mm=s and mm=sm, if the diameter is expressed in mm and time in s. The value of m ranges between 0.13 and 0.27 without being a ected much by heat ux or subcooling, and it is also similar to the range reported by Lee et al. [32]. The value of was found to increase with increasing heat ux and decreasing subcooling and hence , increases with increasing wall superheat which agrees well with the observation by Scriven [27] for bubble growth in a uniformly superheated liquid (Equation 3.3). Similarly, in the experiments conducted by Cole and Shulman [29] with R-113, it was observed that the growth constant increased with increasing wall superheat. The values of c used in Equation 3.3 using the uniform superheat theory signi cantly over estimated the growth constant. Cole and Shulman [29] reported better predictions of growth constant by using a modi ed Jacob number with the wall superheat approximated as 0:5(Tw Tsat), which is valid since 63 D a1 1.72t 0.24 D a11.43t 0.24 D a11.09t 0.19 D a1 1.11t 0.27 D=At A 2.19 W/cm2 = 175.36 A 1.60 W/cm2 = 150.79 A 1.33 W/cm2 = 94.46 A 1.07 W/cm2 = 48.03 Figure 3.8: E ect of heat ux on bubble growth at a liquid subcooling of 21 C D a1 1.11t 0.27 D a11.22t 0.14 D a11.95t 0.21 D=At A 0 oC = 181.32 A 10.92 oC = 165.96 A 21.04 oC = 48.03 Figure 3.9: E ect of subcooling on bubble growth at a nominal heat ux of 1.0 W=cm2 64 the liquid superheat away from the heated wall approaches zero. Cole and Shulman [29] reported a modi ed equation for growth constant ( ) as, = 54Ja34 (3.5) In comparison, the experimental values of growth constant from the current study are plotted in Fig. 3.10 as a function of wall superheat. It can be observed that the modi ed equation (Eq. 3.5) for bubble growth near heated surfaces reported by Cole and Shulman [29] still over-estimates the growth constant. The equation for growth constant obtained using curve tting the experimental data from the current study can be expressed as, = 0:034Ja (3.6) Table 3.1: Summary of curve tting parameters from Fig. 3.8 and Fig. 3.9 Experimental conditions Bubble growth regime Fig. #Inertia controlled Heat transfer controlledD = At (mm) D = tm (mm) Tsub q00 A m ( C) (W=cm2) (mm=s) (mm=sm) 21.04 2.19 175.36 1.72 0.24 Fig. 3.8 21.04 1.6 150.79 1.43 0.24 Fig. 3.8 21.04 1.33 94.46 1.09 0.19 Fig. 3.8 21.04 1.06 48.03 1.11 0.27 Fig. 3.8, 3.9 10.92 0.94 165.96 1.22 0.14 Fig. 3.9 0 1 181.32 1.95 0.21 Fig. 3.9 Figures 3.11 and 3.12 also represent the e ect of heat ux and subcooling on growth rate but with overall asymptotic relationships for growth rate obtained by a power law curve t for both the inertia and heat transfer controlled regimes combined. The growth relationships are only slightly di erent from Figures 3.8 and 3.9 in the heat transfer controlled regime. In comparison with Hutter et al. [34], who conducted a similar analysis for bubble growth rate 65 30 40 50 60 70 80 900 5 10 15 20 25 30 35 40 45 50 Wall Superheat, Tw ? Tsat (oC) Growth Constant, ? Experimental ? Subcooling 0?20oC ?=0.034*Ja Eq. 3.3 with ?c=1 Eq. 3.5 Figure 3.10: Growth constant from experiments at a liquid subcooling of 0-20 C. The gure shows a linear curve t used to approximately express the growth constant. The equations 3.3 and 3.5 used for comparison from the studies Fritz and Ende [52] and Cole and Shulman [29] respectively, were calculated for saturated pool conditions. 66 in FC-72, observed the values of m to be in the range of 0.37 - 0.53 which is slightly higher than those reported in this study. 3.2.3 Bubble departure frequency and diameter As discussed earlier, bubble departure from a nucleation site is a balance between the surface tension and buoyancy forces, in the absence of forces due to liquid ow. As surface temperature increases, surface tension decreases and hence buoyancy forces tend to cause the bubbles to depart from the nucleation site earlier than at lower surface temperatures or wall heat ux, thereby leading to increased bubble departure frequencies or truncated bub- ble growth period. The average bubble departure frequency was estimated by calculating the average time period between a bubble?s nucleation and departure. As shown in Fig. 3.13, bubble departure frequency is a function of both subcooling and heat ux. As heat ux increases the bubble departure frequency also increases, although with a few anoma- lies. Similarly, the bubble departure frequency increases with decreasing subcooling and the highest frequency of 60 Hz was observed under saturated pool conditions and high heat ux. The frequencies observed under saturated conditions agree well with other pool boiling experiments conducted with FC-72 [14, 34]. However, in those studies, the bubble departure diameter was not observed to be a ected by subcooling or heat ux as seen in Figures 3.11, 3.12 which was unexpected. In other pool boiling studies conducted with FC-72 [14, 17, 33, 34], the bubble departure diameter was observed to increase with heat ux, or wall superheat. In [17], the bubble departure diameter increased between 0.26 mm to 0.38 mm over a heat ux range of 0 - 40 W/cm2. However, in the current study the range of heat ux tested was very low, and this could be a reason for the absence of any trend in the variation of bubble departure diameter. 67 D a11.07t 0.24 D a11.39t 0.22 D a11.63t 0.23 D a11.08t 0.18 Figure 3.11: E ect of heat ux on bubble growth at a liquid subcooling of 20 C D a11.23t 0.14 D a11.07t 0.24 D a12.03t 0.22 Figure 3.12: E ect of subcooling on bubble growth at a nominal heat ux of 1.0 W=cm2 68 0 0.5 1 1.5 2 2.5 30 10 20 30 40 50 60 70 80 Wall Heat Flux, q" (W/cm2) Bubble Departure Frequency, f (1/s) Saw teeth angle = 24oCavity mouth radius = 96.7 ?m?Tsub=21.04oC ?Tsub=16.04oC ?Tsub=10.92oC ?Tsub=5.31oC ?Tsub=0oC Figure 3.13: E ect of subcooling and heat ux on bubble departure frequency 0 0.5 1 1.5 2 2.50.75 0.8 0.85 0.9 0.95 1 Wall Heat Flux, q" (W/cm2) Bubble Departure Diameter, D d (mm) Saw teeth angle = 24oCavity mouth radius = 96.7 ?m?Tsub=21.04oC ?Tsub=16.04oC ?Tsub=10.92oC ?Tsub=5.31oC ?Tsub=0oC Figure 3.14: E ect of subcooling and heat ux on bubble departure diameter 69 3.2.4 Bubble growth - Comparison of experimental data with models The experimental results presented in this study show that the bubble growth rates are dependent on the regime of growth. Models developed in the past to predict bubble growth rates, in general, are also regime speci c and universal analytical solutions are only available for very speci c experimental conditions. As discussed earlier models for bubble growth can be classi ed into two types - one developed for the growth of a spherical symmetric bubble in a uniformly superheated pool and the other for bubble growth near heated walls where the temperature eld is non-uniform. Models for inertia controlled regime are very limited [24, 30] and most of the models developed apply mainly for the heat transfer controlled bubble growth. Models developed for the heat transfer control regime are often empirical in nature [27{29] and depend on the power law relationship of bubble growth with time (D t1=2). However, no analytical models are available for experimental conditions similar to those in the current study such as bubble growth in highly wetting uids on surfaces with structured cavities. Bubble growth for such conditions is often expressed as an asymptotic relationship of bubble diameter as a function of time. Analytical models and empirical relationships discussed in the literature that are used for comparison with the current study are summarized in Table 3.2 Table 3.2: Summary of available models and experimen- tal data for bubble growth during inertia controlled (IC) and heat transfer controlled (HTC) regimes. Model Conditions Bubble growth correlation Rayleigh [24] Uniform super- heat theory D = At where A = 2 7 hlv v T lTsat 1=2 (3.7) 70 Model Conditions Bubble growth correlation Mikic and Rohsenow [31] Non-uniform temperature eld D = 4Ja p3 ltp ( 1 Tw T1T w Tsat " 1 + twt 1=2 t w t 1=2#) where tw = 14 l 8 < : rc erfc 1 h Tsat T1 Tw T1 + 2 Tsat(vv vl) (Tw T1)hlvrc i 9 = ; 2 rc = cavity mouth radius (3.8) Van Stralen [30] Bubble growth due to evap- oration and relaxation mi- crolayer near heated surfaces. Applicable for both IC and HTC. D = D1 (t)D2 (t)D 1 (t) +D2 (t) D1 (t) = 1:633 vu ut vhlv (Tw Tsat) exph (t=td)1=2i lTsat t D2(t) = 3:9088 ( b exp " t td 1=2# + T1 TsatT w Tsat ) Jap lt+ 0:746 1=6 Pr l( exp " t td 1=2# + T1 TsatT w Tsat ) Jap lt b = 0:6964 D2 (t)Jap lt 0:1908 1=6 Pr l (3.9) Cole and Shulman [29] Non-uniform temperature eld D = 5Ja3=4p t (3.10) 71 Model Conditions Bubble growth correlation Zuber [28] Non-uniform temperature eld D = c 4 Jap t 1 qw p t 2k(Tw Tsat) (3.11) Lee et al. [32] Non-uniform temperature eld - R11 and R113 R+(t+) = 2 t+1=5 tanh(t+1=5) +R+0 where R+ = R=Rc, t+ = t=tc Rc = p27 2 Ja r lRd tc = 94Ja lRd , = tting parameters R +0 = dimesionless critical radius (3.12) Experiments with R11 and R113 D/t (IC) D/t1=2 (HTC) Ramaswamy et al. [33] Experiments with FC-72 on structured surfaces D/t (IC) D/t1=2 (HTC) Hutter et al. [34] Experiments with FC-72 on silicon with cylindrical cavities D/t1=2 72 0 0.005 0.01 0.015 0.02 0.025 0.03 0.0350 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Time, t (s) Bubble Diameter, D (mm) Experimental ?Tsub=0oC; q = 1.9 W/cm2 Mikic and Roshenow ? Eq. 3.8Van Stralen ? Eq. 3.9 Lee et al. ? Eq. 3.12Cole and Shulman ? Eq. 3.10 Zuber ? Eq. 3.11Rayleigh ? Eq. 3.7 Figure 3.15: Comparison with existing models for bubble growth From Fig. 3.15, it can be observed that all the tested models deviate signi cantly from the experimental results shown for saturated conditions. Rayleigh?s equation (Equation 3.7) captures the high growth rate involved in the inertia controlled growth regime adequately. All the other equations for the heat transfer controlled growth, although they represent the shape of the bubble growth rate, deviate signi cantly from the experimental data. The model developed for refrigerant R-113 by Lee et al. [32] showed signi cant deviations although the asymptotic growth relationships represented in that study were similar to those reported in the current study. These results suggest that all the available models perform poorly for the experimental conditions tested in the current study. This re-emphasizes the need for further studies and models to analytically explain the e ects of structured surfaces and low surface tension uids. 73 3.2.5 Heat Dissipation Heat transfer characteristics of the surface can be analyzed by plotting wall superheat (Tw Tsat) against wall heat ux (qw), which is known as the boiling curve. Wall temper- ature was calculated from the test section heater resistance measurements using a linear relationship that was previously obtained by calibration. Calibration of heater resistance was performed in a convection oven using an NIST calibrated thermistor as a standard, by varying the oven temperature from 25 C to 85 C. Heater calibration data are presented in Appendix C. Saturation temperature was calculated at the measured pressure inside the tank using a pressure transducer. Heat ux was calculated as, q" = q qlossA st (W=cm2) (3.13) where, q = V I (W) Ast = Surface area of saw-toothed surface, cm2 V = Applied voltage, V I = Applied current, A Heat loss (qloss) was quanti ed by attaching a thermocouple to the back of the test device. Heat loss, which is heat transfer from the back side of the test device, was estimated by assuming 1D steady conduction across the device. It has to be noted that such an estimate of heat loss is only a conservative measure as it does not account for any spreading associated within the PCB. Fig. 3.16 shows the boiling curve at di erent liquid subcooling values. At a constant subcooling, as heat ux increases, heat transfer from the surface is by natural convection de ned by Newton?s law of cooling. As wall superheat increases high enough to trigger vapor nucleation from the cavities, the heat transfer increases sharply, marked by the change in 74 slope of the curve, without causing a signi cant increase in wall temperature. This can be attributed to the e ects of microconvection due to mixing induced by bubble growth and departure, transient conduction due to the constant disruption of the boundary layer and also due to the microlayer evaporation which is less signi cant in the current study owing mainly due to the shape of the bubble [6, 36]. As heat ux increases further, more nucleation sites become active increasing the heat transfer further. It has to be noted that the highest heat ux applied to the device was only 4.5 W=cm2, which was not limited by critical heat ux but only due to limitations of the designed heater, as discussed earlier. Also, at the highest tested heat ux the wall temperature was 92 C and observations of bubble dynamics showed that the boiling regime was still not fully developed suggesting that the CHF was much higher than the highest heat ux tested. This could also be attributed to the re- entrant cavities which prevents bubble coalescence by controlling the location of nucleation sites. Another important characteristic of boiling curve for a highly wetting uid such as FC-72 is the nucleation overshoot which is due to the ooding of cavities caused by the low contact angle of the the liquid. However, in the current study nucleation overshoot was not observed mainly because of the wide variation in cavity mouth radius. According to Hsu?s analysis [6], cavities with larger mouth radius are activated rst and as wall temperature increases cavities with smaller radius nucleate. This causes nucleation to happen in stages with increasing wall superheat and hence overshoot was not observed. As subcooling increases, pool temperature decreases causing increase in heat transfer in the natural convection regime. Also, in the partially developed boiling regime heat transfer is higher as subcooling increases. These observations accord well with the results of number of other studies in the literature. As heat ux increases, it could also be observed that the boiling curves at di erent subcooling tend towards merging with each other, as fully developed nucleate boiling is independent of both liquid subcooling and mass ux and mainly controlled by bubble ebullition cycle. 75 Increasing Subcooling Figure 3.16: E ect of subcooling on pool boiling curve for the test device with a saw-tooth angle of 24 3.2.6 Lateral Liquid Velocity Measurements Using Hot Wire Anemometer: Parametric E ects In the previous sections it has been demonstrated that asymmetry in the location of re-entrant cavities and asymmetry in surface structure can be used to cause bubble growth and departure at an angle normal to the slope of the surface. Using a semi-empirical model it was shown that such asymmetric bubble growth has the potential to impart an angular momentum to the surrounding liquid, thereby causing a net lateral ow. It was shown that in water, the asymmetric growth of bubbles can lead to liquid velocities up to 20 mm/s and in FC-72 the velocities are much lower at 1-2 mm/s due to higher liquid inertia associated with FC-72. In this section, experimental results of liquid velocity measurements over the saw- toothed surface using hot wire anemometry are reported. Hot wire anemometry involves a micro-wire or lm made of platinum or tungsten that is electrically heated to a constant 76 temperature. The heat convected from the probe due to liquid ow causes a change in cur- rent which in turn causes a change in voltage at the anemometer output. The measured voltage output is directly proportional to the liquid velocity (V). In the current study, a TSI Model 1750 constant temperature anemometer was used for the liquid velocity measurements. The anemometer setup includes a power supply, a probe, probe support and an angle adapter. An image of the anemometer along with the power supply is shown in Fig. 3.17a and b. Current supply to the probe and voltage measurements are made using a Wheatstone bridge circuit shown in Fig. 3.17c which is built in to the anemometer. The probe used in the experiments is a TSI Model 1210-20 straight probe with a platinum lm. The probe was held inside the polycarbonate ow channel as shown in Fig. 3.18 using a Model 1152 angle adapter which aids in positioning the probe inside the ow channel horizontally. The angle adapter is connected to a probe support which is connected to the anemometer power supply connection pins. One of the pins in the anemometer is connected to a control resistor which is used to set the operating temperature of the probe. The resistance of the control resistor is calculated based on the prescribed operating resistance for the probe and the cable resistance. The operating temperature of the probe 1210-20 is 67 C and the operating resistance is 48 Ohms. Based on these speci cations the control resistor used has a resistance of 34 Ohms. The selection of control resistor is explained in the manual [55]. The bridge output from the anemometer ranging between 0-5V is connected to the data acquisition system. The assembly of the probe setup is shown in Fig. 3.18. The probe is positioned 4mm over the saw-toothed surface and midway along the length of the test device facing the direction of liquid ow. The hot lm of the probe is enclosed in an aluminum sleeve with a slit at the top. The aluminum sleeve serves to prevent the contact of the vertical convection currents from the surface with the probe, so that the probe measures only the horizontal component of velocity. The heat transfer from the probe (hot wire/ lm) maintained at a constant tempera- ture, apart from liquid velocity, is also strongly a function of liquid temperature and liquid 77 (a) TSI Constant Temperature Anemometer (Model 1750) (a) Power supply used with the anemometer (Model 1751) (c) Wheatstone bridge circuit used for voltage measurement (d) Block diagram of the anemometer showing the pin connections Voltage Output Figure 3.17: (a) and (b) TSI constant temperature anemometer and power supply used for liquid velocity measurements. Electrical circuit and connections are shown in (c) and (d) [55] 78 (b) Exploded view of the probe assembly. Probe assembly includes the Model 1210-20 hot wire probe, Model 1152 angle adapter, and a sleeve 4 mm (c) An image showing the position of the probe over the surface in the experimental setup (a) Illustration of the anemometry setup inside the flow channel. Image inset shows a close-up illustration of the hot wire probe enclosed in an aluminum sleeve Aluminum sleeve slitted at the top Probe support with electrical connector Saw-teeth surface Hot wire probe Aluminum sleeve Hot wire Figure 3.18: Illustration of the probe set up over the test device with a saw-teeth angle of 24 79 properties. Hence for a constant liquid and operating temperature of the probe, the voltage output of the anemometer is primarily a function of liquid velocity and liquid temperature. To estimate the velocity of the liquid from the bridge output of the anemometer, the probe was calibrated in FC-72. A metered magnetic drive gear pump was used to pump the liquid upwards through a vertical tube in which the probe was placed. The calibrated velocity ranges from 1 - 50 mm/s at room temperature. By plotting output voltage against velocity, a fth order polynomial expressing velocity as a function of output voltage was obtained by curve tting. Calibration data for the hot wire probe are presented in Appendix C. The data from the anemometer was also corrected for actual temperature of liquid in the experiment by using the expression, E2cor = E2meas CF CF = Ts Te;calT s Te (3.14) where, Ecor = corrected output of the anemometer, V Ecor =measured output of the anemometer, V CF =correction factor Ts = lm operating temperature, C Te =liquid temperature during the experiment, C Te;cal =liquid temperature during calibration, C Experiments were conducted at a pool subcooling range of 0-20 C and a heat ux range of 0 - 4.5 W=cm2. For a constant liquid subcooling, power to the test device was increased periodically and the bridge output from the anemometer was measured after steady state conditions were achieved. This step was repeated while decreasing the applied heat ux and for experiments at other liquid subcooling values. Figure 3.19 shows the raw and temperature corrected bridge output of the anemometer. The output of the anemometer showed a lot of uctuations due to two reasons. One, was 80 due to operation at the lower limit of hot wire anemometer?s output. The lower limit of velocity measurement using a hot wire anemometer is 10 mm/s for liquid ows. Secondly, the uctuations are also caused by contact between departing vapor bubbles and the probe. The voltage measurements shown in Fig. 3.19 represent the peak amplitude of the recorded data.Using the temperature corrected voltage, liquid velocity can be measured using the calibration equation which is represented in Fig. 3.20. It can be noticed that at high subcooling, the measured velocity increases with heat ux thus indicating the pumping potential although at low velocities. However, at low subcooling the velocity was observed to be independent of heat ux and the measured velocities were very low. This net horizontal velocity component increased slightly at higher subcooling but increased consistently with heat ux. 0 0.5 1 1.5 2 2.51 1.1 1.2 1.3 1.4 1.5 1.6 Wall heat flux q??, W/cm2 Anemometer Voltage, V Saw teeth angle = 24oSubcooling = 21.04oC Raw HWA OutputTemperature Corrected HWA Output Figure 3.19: Raw and temperature corrected voltage measurements using hot wire anemome- ter over the test device with a saw-tooth angle of 24 81 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 50 1 2 3 4 5 6 7 8 9 Heat Flux, W/cm2 Velocity, mm/s ?T sub=0oC?T sub=5.31oC?T sub=10.92oC?T sub=16.04oC?T sub=21.04oC Figure 3.20: Temperature corrected liquid velocity measurements using hot wire anemometer over the test device with a saw-tooth angle of 24 3.3 Summary Pool boiling experiments were conducted on asymmetric silicon heat sinks using FC- 72. The independent parameters controlled in the experiment include heat ux and liquid subcooling. In the experiments that were discussed in this section, the parameters measured include bubble diameter, bubble departure frequency, wall surface temperature and liquid velocity. High speed imaging was used to capture bubble images and image processing techniques were used to estimate the characteristics of vapor bubble such as shape, diameter, and departure frequency as a function of time, heat ux and inlet subcooling. Based on the observed bubble characteristics, bubble growth was characterized as inertia and heat transfer controlled and asymptotic relationships for bubble growth were presented as a function of time, for both inertia and heat transfer controlled regimes. Based on the wall temperature measurements and heat ux, pool boiling curves were presented which illustrated the heat dissipation characteristics of the surface. Net lateral liquid velocity due to asymmetric bubble 82 growth, measured using hot wire anemometry, showed potential pumping velocities of up to 6 mm/s. To further corroborate the results from hot wire anemometry, PIV experiments were conducted with FC-72 which are described in Appendix E. Due to light re ection from bubbles and non-availability of uorescent particles with densities similar to that of FC-72, results from PIV experiments were inconclusive about lateral velocity of liquid. 83 Chapter 4 Microgravity Experiments In pool boiling experiments conducted at 1g, it was observed that bubble growth and departure were normal to the sloped surface of the saw teeth. The buoyancy force acting on the asymmetrically growing and departing bubble, thereby, has a horizontal and a vertical component. The vertical component of buoyancy at 1g acting on a an asymmetrically growing bubble is less than the buoyancy force acting on a vertically growing bubble on a plain surface. In the previous sections, it was observed that this led to signi cant change in bubble dynamics. To further understand the role of buoyancy, it is important to study the bubble dynamics and the resulting lateral motion in the absence of buoyancy which is only possible under microgravity. Hence, microgravity experiments were conducted to study bubble dynamics during phase change on an asymmetric surface in the absence of buoyancy and its e ect on lateral velocity of liquid. The experiments were conducted aboard a Boeing 727 aircraft (Zero-g Inc.) (Fig. 4.1) carrying out parabolic maneuvers to achieve reduced gravity. The maneuvers were carried out on two separate days. Each day consisted of 40 consecutive parabolas with each parabola consisting of a 1.8g pull up, 0g and a 1.8g pull out with periods of 60 seconds for hyper-gravity ( 1:8g) and up to 17 seconds for microgravity ( g) as shown in Fig. 4.1b. Fig. 4.1c shows the recorded accelerometer readings during a single parabolic maneuver consisting of 1.8g, g, and 1.8g. The ight operations initiated and terminated at Ellington eld, Johnson Space Center in Houston, Texas. 4.1 Experimental Setup Fig. 4.2 shows the experimental setup used in the study. The silicon test device used for the pool boiling experiments, shown in Fig. 4.2a, is made of two layers - the asymmetric 84 (b) Parabolic maneuver of the aircraft and period of gravity regimes (c) Accelerometer data of the gravity profile (a) A NASA Zero-g flight on a 1.8g pull up maneuver 1.8g Figure 4.1: Zero gravity ight used for parabolic maneuvers - (a) A NASA Zer0-g ight on it upward ascent. Boeing 727 was used for the ight experiments (b)-(c) Gravity pro le and recorded accelerometer readings of the achieved parabolic maneuvers 85 saw-toothed heat sink, and a serpentine heater layer. The heat sink has an asymmetric saw tooth cross-section with a 31 90 59 pro le and 1 mm pitch (hereafter referred to as the 31 test section). The heat sink used for large array experiments with FC-72, shown in Fig.4.2a consists of 80 such saw-teeth spanning a foot print of 80 mm x 20.3 mm. Each saw tooth has a long slope that is structured with re-entrant cavities. The cavities have a trapezoidal mouth of size ranging between 50 100 m. Each saw tooth has 8 pairs of cavities spaced equally along the transverse length of ratchet on the long slope, with each pair consisting of a large and small cavity (marked as (a) and (b) in Fig.4.2a) spaced between the crest and the trough. In total, the heat sink consists of 1280 cavities acting as nucleation sites. Heat was provided using an aluminum serpentine heater fabricated on silicon. The voltage leads on the heater also aid in surface temperature measurements. The test device is mounted on a printed circuit board for electrical connections. The board also serves as the base of a transparent polycarbonate open channel which is 130.8 mm long, 25.4 mm wide and 7 mm deep, and mounted right on top of the test device as shown in Fig. 4.2b. The channel serves to con ne the volume of uid for realization of net liquid ow. The test board was suspended from the lid into a pool of FC-72 contained in an aluminum boiling chamber (Fig. 4.2c) of dimensions 33 cm x 23 cm x 22 cm (L x W x H). A thermistor and a pressure transducer were used to measure liquid temperature and pressure respectively. Prior to ight experiments, the uid was charged into the system and degassed thoroughly at the ground station. Once the uid was charged, the bellows on top of the boiling chamber were compressed and the valve on top of it was closed with pressure inside the chamber remaining at atmospheric conditions. During the ight experiments, the boiling chamber was completely sealed and any increase in pressure due to the boiling process was compensated by the expansion of rubber bellows, thereby maintaining ground atmospheric pressure within the boiling chamber. A similar experimental apparatus was used by Oregon State University research group for small array experiments with deionized water as working 86 uid. The test section used for small array experiments with water is similar in cross-section consisting of only 8 saw-teeth ratchets with a footprint of 8 mm x 11.3 mm. During the experiment, data from sensors were recorded using NI Compact DAQ, and a Phantom high speed camera tted with an In nity K2SC microscope lens and a CF2 close-up objective was used to record bubble images. The electrical equipment in the setup were grouped in to 110 V and 220 V equipment and power was drawn from separate sources on the ight. Emergency cut-o was setup for disconnecting both the 110 V and 220 V equipments from the power source. The other signi cant di erence between terrestrial and microgravity gravity experimental setup was that the entire boiling chamber compartment was covered by polycarbonate sheets to form a double containment to prevent the liquid from entering the aircraft cabin in the event of a leak. The assembled view of the experimental setup is shown in Fig. 4.3. Experimentation aboard the ight also required adherence to strict mechanical design guidelines which required a factor of safety of 4 while withstanding forces up to 9g at every joint and all the load bearing members of the experimental setup. Details of the experimental design and structural analysis were furnished in a report, Test Equipment Data Package (TEDP), that was submitted to NASA. The submitted version of TEDP is presented in Appendix F. 4.2 Results and Discussion The data reported in this section were collected from experiments on large and small arrays of saw-teeth ratchets with highly subcooled FC-72 and 10 C subcooled deionized water respectively (subcooling, Tsub, is the di erence between saturation and pool temperature, Tsat Tpool) at g ( 10 2 g). The following sections carry discussion of bubble dynamics such as bubble diameter, bubble sliding motion, heat dissipation and a model for the estimation of bubble sliding velocity. 87 Figure 4.2: Experimental setup for reduced gravity experiments - (a) 31 test device with two re-entrant cavities per saw-tooth (b) Assembled test device with the polycarbonate ow channel. Co-ordinate system shows the orientation of the test device with respect to the aircraft. (c) Boiling chamber used for the reduced gravity experiment. Bellows assembly with double containment is used to maintain a constant chamber pressure. 4.2.1 Bubble Dynamics At reduced gravity, due to lack of buoyancy it was observed that the bubbles reside on the heater surface and grow to several diameters larger than those in 1g, as previously reported by a number of studies discussed in the literature. In experiments with water, because of the small size of the heat sink the entire surface of the test device was covered by a single large vapor bubble of diameter as large as 8.5 mm. It was observed that the single large bubble was pinned to the edges of the test section and the bubble interface appeared undulated at all tested heat ux values. Smaller vapor bubbles nucleating from the cavities under the footprint of the large bubble, at an angle normal to the shallow slope of the surface, were observed as shown in the image inset shown in Fig. 4.4a. This phenomenon is only 88 Figure 4.3: Assembled view of the experimental structure used for microgravity experiments. Details of the individual components in the structure and the related structural analysis are provided in Appendix F. 89 possible by the presence of a thin liquid lm between the larger bubble and the surface; the presence of this thin liquid lm is supported by the low measured surface temperature of 113 C, which is well below the Leidenfrost temperature. Straub [37] discusses a similar phenomenon with circulating vapor inside a bubble which was attributed to the shear forces induced by thermocapillary convection at the interface. In the case of large array experiments with FC-72, the bubbles were = 6 times larger compared to 1g. Interestingly, in the large array experiments, the bubbles departed the nucleation site laterally and continued to slide at high velocities across the saw teeth and along the length of test device as shown in Fig. 4.4b and 4.4c. At a low heat ux of 0.5 W/cm2, the interface of sliding bubbles appeared smooth and hemispherical as observed in the time sequence of images shown in Fig. 4.4b, and no departure from the surface was observed. As the heat ux was increased the bubble interface exhibited an unstable behavior as the surface appeared wrinkled, corrugated and more spherical. The corrugated sliding bubbles lifted-o and hovered over the surface which could be attributed to the momentum associated with bubble coalescence. This was also reported in studies with R-113 by Lee and Merte Jr. [43]. Fig. 4.4c at time 325 ms shows a corrugated bubble hovering over the surface at a heat ux of 1.4 W/cm2. Lee and Merte Jr.[43] observed similar corrugated rough bubble surface which was to Rayleigh-Taylor instabilities. The smaller bubbles that appear in Fig. 4.4c resulted from bubble departure during the preceding 1.8g regime that remain suspended in the pool during the next microgravity regime. At 1.8g (Fig. 4.5) with FC-72, due to increased buoyancy forces, the bubble departure diameters (Dd;1:8g) were very small compared to microgravity or 1g, Dd;1:8g 0 and P2;v >P2;l. The pressure inside the vapor bubble is assumed to be same throughout (P1;v = P2;v) Equations 4.2 and 4.3 can be combined to obtain an expression for pressure drop along the saw-tooth (Equation 4.4) P1;l P1;v = 2 r 1 (4.2) P2;v P2;l = 2 r 2 (4.3) P1;l P2;l = 2 1 r1 + 1 r2 (4.4) The shear force acting on the bubble due to the pressure gradient in the liquid lm can be calculated by approximating the ow to be a fully developed plane Couette-Poiseuille ow 97 r1 r2 vapor (v) liquid (l)H x y (a) Liquid-vapor interface of a sliding bubble at 0.45 W/cm2 (b) Schematic diagram of the liquid-vapor interface Figure 4.9: Liquid-vapor interface of a sliding vapor bubble in FC-72 under g with no slip. Assuming the bubble to be initially moving at a constant velocity V, for a ow of liquid between between parallel plates (bubble and surface of saw tooth) at zero gravity under a constant pressure gradient (@P@x), the momentum equation can be expressed as u(x) = 12 @P@xy2 +c1y +c2 (4.5) Boundary conditions for the ow represented in Fig. 4.9 are: ux = 0 at y = 0 (no slip) ux = V at y = H (no slip) where H is the thickness of the liquid lm between the vapor bubble and surface of saw tooth, as shown in Fig. 4.9. By using the boundary conditions in Equation 4.5 c2 = 0 c1 = VH 1 @P@xH2 98 Using the constants c1 and c2, the equation for velocity pro le is u(x) = VHy 12 @P@x Hy y2 (4.6) Shear stress exerted by the liquid lm on the bubble wall at y = H is given by w = yxjy=H = dudy y=H = H2 @P@x - VH (4.7) The total shear force acting on the bubble is then F W = LZ 0 wdx = V H + H 2 @p @x L (force per unit width) F P = V H + H 2 @p @x LW (4.8) where L is the length of the liquid lm that exists between the surface of saw tooth and vapor bubble. The horizontal component of this force acting on the liquid along the shallow slope of the saw tooth is, F P = V H + H 2 @P @x LW cos (4.9) where, = angle of saw tooth = 31 Equation 4.8 represents the force due to pressure drop over the long slope of a single saw tooth. The same form of the equation also applies for the short steep slope of the saw tooth, which is a force opposing the sliding motion of bubble in the direction marked. For a bubble moving at a velocity V with a foot print Ab spanning over m number of long slopes and n number of short slopes of the saw-teeth, the total force due to pressure di erence is 99 expressed by F P;tot = ( m V H + H 2 @P @x LW cos long slope n V H + H 2 @P @x LW sin short slope ) (4.10) Force due to Marangoni convection Marangoni forces arise due to presence of a surface tension gradient caused by a gradient in temperature. In the case of a heated wall in a pool of liquid, if the wall temperature Tw is larger than the pool temperature, a temperature distribution exists in the liquid close to the wall with temperatures ranging between Tw close to the wall and Tpool far away from the surface. A pure liquid, unlike a binary mixture, exhibits a negative surface tension gradient with respect to temperature, meaning, the surface tension decreases with increasing temperature. This surface tension gradient gives rise to a ow from high surface tension (cold liquid) to low surface tension (warmer liquid). When the temperature gradients in the liquid exist vertically and if the bubble is attached to a sloped wall, Marangoni convection will lead to a net force acting on the bubble which could result in its displacement. The force due to Marangoni convection, FM, for an hemispherical bubble is, FM = 2 R2 d dT dT dy (4.11) where d dT = surface tension gradient between top and base of the bubble, N=mK dT dy =temperature gradient between top and base of the bubble, K=m R =radius of bubble, m For a sliding bubble shown in Fig. 4.9a, Marangoni force acts vertically downwards and does not lead to a net lateral force. However, during the initial stages of bubble growth the Marangoni force may play a vital role in displacing the bubble from the nucleation site 100 laterally. Since, the model is developed for a sliding bubble the Marangoni force will be neglected. Force due to vapor nucleation from cavities In the case of a large vapor bubble hovering over a bubble growing over a cavity, the growth and departure of the bubble from the nucleation site exerts a force on the bubble hovering close to the surface. This force calculated by using momentum balance was given by Lee and Merte[43] Fv = ntX i=1 fi i v;i 3 1/2 D 5/2 i (4.12) where fi = mean bubble departure frequency, s 1 Di =bubble departure diameter, m i =surface tension, N=m Due to lack of information about bubble growth and departure from the nucleation site, the force due to nucleation from cavities using Equation 4.12 cannot be determined. Also, since the bubble growth rates are signi cantly lower in microgravity compared to earth gravity, the momentum transferred to the surrounding liquid is not expected to be a signi cant factor. Viscous drag on the vapor bubble The drag force experienced by the bubble as it moves against the inertia of liquid is given by the Stokes drag [56], Fd = 4 RV (4.13) where R is the radius of bubble. The overall force balance using Newton?s second law is F P;tot Fd = 0 (4.14) 101 By substituting Equations 4.10 and 4.13 in Equation 4.14, an expression for velocity of bubble is given by V = ( m V H + H 2 @P @x LW cos long slope n V H + H 2 @P @x LW sin short slope ) 1 4 R (4.15) where m =number of long shallow slopes under the bubble n =number of short steep slopes under the bubble W Diameter of bubble, m H =thickness of liquid lm under the bubble, m L =length of liquid lm under the bubble, m =dynamic viscosity of liquid, Nsm 2 @P @x =Pressure drop per unit length across the liquid lm, N=m 3 R = radius of bubble, m In the derivation of Equation 4.15, the thickness of liquid lm under the bubble is assumed to be a constant and to be same over both the short and long slope of the saw teeth. For simplicity and to check the adequacy of the model, it could also be assumed that the parameters of liquid lm and the radius of curvatures of bubble (r1;r2;r3) repeat across every saw tooth over which the bubble interface resides. As an example, for the sliding bubble shown in Fig. 4.9a, which was experimentally observed to be moving at 27 mm/s, the parameters such as the radii of curvature and thickness of lm could be estimated by image processing. The vapor bubble properties estimated are: r1 = 0:26 mm; r2 = 0:35 mm; D = 5:3 mm 102 It is assumed that the radii of curvature of the liquid-vapor interface over every saw tooth is repeated. This also leads to the conclusion that r2 = r3. The estimated values for the liquid lm parameters are shown below: Llongslope = 0:85 mm; Lshortslope = 0:53 mm The thickness of the lm is assumed to be a constant over both the short and long slope of the saw tooth. The parameter L is the length of the lm that is trapped between the vapor bubble and the the saw tooth, and it is assumed that the vapor bubble occupies the entire slopes of the saw tooth. The uid properties are estimated at the saturation temperature of the uid. By using all the aforementioned parameters in Equation 4.15, and varying the thickness of the liquid lm between the vapor bubble and saw tooth, H, from 1-25 m, the velocity of the bubble can be calculated as a function of H. The e ect of H on the velocity of bubble is illustrated in Fig. 4.10. The other liquid and bubble parameters which were held constant are represented in the gure. It can be observed that the experimentally observed velocity of 27.4 mm can be obtained from the model for H 17 mm. By observing the bubble images, it can be inferred that the values of H plotted in the Fig. 4.10 are plausible. This agreement between the model and the experiments proves that pressure di erences in the liquid lm is a potential driving force for the sliding motion of bubbles under microgravity. In a similar study conducted by Linke et al. [8], for R-134a droplet velocities of 5 cm/s, the value of vapor lm thickness was 10 mm. In the current analysis, the values of r1, r2, and r3 were estimated from images. However, those values were varied to analyze their e ect on bubble velocity and it was observed that small changes in the radii of curvature did not a ect the velocity much. Similarly, the values for the length of liquid lm, L, were also varied and the bubble velocities were not a ected signi cantly. Sample calculations are provided for bubble velocity are provided in Appendix G. 103 Figure 4.10: Estimated bubble sliding velocities for liquid lm thickness varying from 0-25 m. The experimental bubble velocity of 27.4 mm/s is predicted closely for a H value of 17 mm. From bubble images, a value of H. 4.2.3 Heat Dissipation The boiling curve shown in Fig.4.11 represents the heat transfer from the large array experiments with FC-72 at reduced gravity. Wall superheat de ned as (Tw-Tsat), represents the di erence between the saw teeth surface temperature and liquid saturation temperature. It was observed that heat dissipation rates over 2 W/cm2 were achieved at a wall superheat as low as 0 C without reaching critical heat ux (CHF) or dry-out under the bubble. Henry and Kim [41] who conducted similar pool boiling experiments with FC-72 under reduced gravity conditions with a silicon heater of size 2.7 mm x 2.7 mm reported similar heat dissipation rates although at much higher wall temperatures of 15 C. By comparison with the current study, the higher heat transfer at lower heater temperatures could be partly attributed to sliding bubbles leading to net lateral liquid velocity in the channel which also prevents CHF due to dry-out under a stagnant vapor bubble. In comparison with experiments under 1g performed with the 24 test device, it was observed that heat dissipation was higherunder g in the nucleate boiling regime. This 104 ?40 ?30 ?20 ?10 0 10 20 30 40?0.5 0 0.5 1 1.5 2 2.5 Tw ? Tsat ,oC Heat flux, W/cm 2 ?Tsub,avg=21.5oC | 0g |Saw teeth angle = 31o ?Tsub,avg=27.7oC | 1g |Saw teeth angle = 24o Figure 4.11: Boiling curves for large array experiments with FC-72 under g and 1g. Tsub = Tsat Tpool. Both experiments were performed with a highly subcooled pool and pool temperature was not controlled. Uncertainties in q00 and (Tw-Tsat) are 1% and 0:3% respectively. enhancement in the nucleate boiling regime is partly due to the drop in chamber pressure during the ight experiment which led to a drop in saturation temperature to 50.5 C. This drop in pressure caused early nucleation which could be observed as a shift in slope in Fig. 4.11. Also, the 31 test device, which was used in the microgravity experiments has 2 cavities per saw-tooth whereas the 24 test device has only one per saw-tooth. However, the enhancement may also be attributed to the the sweeping motion of sliding bubbles which can enhanced heat transfer. In pool boiling experiments conducted with R-113, Lee and Merte Jr. [43] reported heat transfer enhancement of up to 40% under microgravity compared to 1g, and the enhancement was attributed to vapor bubble migration along the surface. 105 Chapter 5 Conclusions and Recommendations A novel concept of using surface asymmetry to develop self-propelled ow of liquid has been studied for electronics thermal management applications. The driving potential for the ow is the heat dissipated from the hot surface by nucleate boiling. Surface asymmetry arises from two sources - one, a saw tooth pro le, and two, re-entrant cavities that are located only on the shallow slope of the saw tooth. Such a heat sink was designed to achieve bubble nucleation only from the shallow face of the saw teeth. The asymmetric surface with re- entrant cavities was fabricated using gray-scale lithography, DRIE, anisotropic etching and other photolithographic processes. The control of etch selectivity and the surface quality was achieved by varying the gas ow rates in DRIE. Pool boiling experiments with FC-72 were conducted on a heat sink with asymmetric silicon ratchets and pyramidal re-entrant cavities. Integrated aluminum serpentine heaters deposited in silicon, were used to supply heat to the heat sink. Experiments were conducted over a subcooling range of 0 C - 20 C and heat ux ranging from 0 - 4 W=cm2 under both terrestrial gravity (1g) and reduced gravity ( g). The following conclusions can be drawn from the experiments conducted: 5.1 Bubble dynamics: Terrestrial experiments In FC-72, under terrestrial gravity, the designed surface pro le and asymmetry in location of re-entrant cavities caused vapor bubbles to nucleate, grow and depart at an angle normal to the shallow slope of the saw teeth, at all tested conditions. Using image processing techniques, the high speed images were processed to obtain bubble growth data which is unique to the surface being tested. The bubble shape, 106 interestingly, resembled a \light-bulb" and was not circular as would be expected for a low surface tension uid. This also suggests that the evaporation microlayer may play a less signi cant role than it would in water, as reported by Demiray and Kim [36]. Bubble growth rate was observed to be both inertia and heat transfer controlled. The inertia controlled regime witnessed a linear rapid growth rate of the form D t. In the heat transfer controlled regime, growth was observed to increase at the rate of t1=4 which is less than the t1=2 relationship reported in a number of studies. Bubble growth was studied as a function of heat ux and subcooling and it was observed that the growth rate increased with increasing heat ux and reducing liquid subcooling. For the inertia controlled regime, in the growth rate relationship expressed as D = At, A was noticed to increase with increasing heat ux between 48 and 175. Similarly, A increased between 48 - 181 with reducing subcooling. In the heat transfer controlled regime where the growth rate is expressed as D = tm, while m was 0.20-0.25, (growth constant) was observed to increase with increasing heat ux and reducing subcooling. This suggests that the growth constant increases linearly with increasing wall superheat which was also observed by Cole and Shulman [29]. Bubble departure frequency increased with increasing heat ux (or wall temperature) and reducing subcooling as reported by number of other studies. However, subcool- ing or heat ux did not signi cantly a ect the bubble departure diameter which was attributed to the low range of tested heat ux. Available analytical models for bubble growth performed poorly for the experimental conditions of the current study emphasizing the need for further studies. Heat dissipation of up to 2.5 W/cm2 was obtained from the surface while keeping the surface temperatures less than 90 C. Heat transfer increased with increasing subcool- ing. 107 5.2 Net lateral liquid velocity under terrestrial gravity Using the bubble growth model developed earlier by Kapsenberg et al. [57] in a com- plementary study at Oregon State University, it was demonstrated that asymmetric bubble growth can lead to a net lateral liquid velocity of up to 5 mm/s in FC-72 and 18 mm/s in water. Hot wire anemometry was used to measure liquid velocity over the surface. While no velocity was detected at saturated conditions, liquid velocity of up to 5 mm/s was noticed at subcooled conditions. However, the output from the hot wore probe uctuated a lot owing to contact with the vapor bubbles and operation at the lower limit of the anemometer. 5.3 Bubble dynamics: Microgravity Pool boiling experiments with FC-72 on silicon saw teeth were conducted aboard NASA?s reduced gravity ight to investigate the e ects of surface asymmetry, re-entrant cavities on bubble dynamics and lateral liquid velocity under microgravity. The following conclusions can be drawn from the reduced gravity experiments aboard the parabolic ights: In FC-72, bubbles were six times larger under g than under 1g. Under 1.8g bubbles were very small ( 1/4 D1g). Under g, bubbles were observed to slide across the saw teeth at velocities as high as 27 mm/s. Bubbles were driven across the saw teeth by forces such as pressure drop in the thin liquid under the bubble, similar to the sliding motion of liquid droplet over a saw-toothed surface observed by Linke et al. [8]. Pressure di erences arise due to di erences in the radii of curvature of liquid-vapor interface at the crest and trough of the saw teeth. 108 In the absence of asymmetry (i.e. a plain heater surface), the vapor bubble would be stationary as observed in a number of studies discussed in the literature, which would lead to heat transfer deterioration owing to a premature dry-out under the bubble. The ability to move the bubble passively, as demonstrated, leads to a self-propelled ow resulting in heat transfer enhancement rather than deterioration, which underscores the potential of the study to cool high heat ux microelectronics under reduced gravity conditions. A model was proposed to study the e ect of pressure di erences in the liquid lm on vapor bubble propulsion. It was inferred that pressure di erence in the liquid lm is a potential force to cause bubble propulsion. Other forces include Marangoni convection which could not be tested due to the very limited number of microgravity parabolas available. 5.4 Recommendations Based on the observations from the experiments, asymmetric bubble growth from sloped surfaces has the potential to e ect lateral motion of liquid in the bubble?s im- mediate vicinity. Also, microgravity testing proved the importance of inducing a thin liquid lm between the saw-toothed surface and the vapor bubble. Both of these could be achieved by constructing a closed loop microchannel test device with saw-toothed heated side walls, which while keeping the liquid volume low, will also promote the existence of a thin liquid lm along the walls. This would also result in heat trans- fer enhancement due the annular ow regime that would prevail in such a scenario. This con guration, however, will be very sensitive to gravity and hence may be a very e ective thermal management solution for space electronics. To further study the role of forces due to Marangoni convection in bubble propulsion, it may be very useful to conduct experiments with saturated pool conditions which would 109 eliminate temperature gradients in the pool, and thereby surface tension gradients. Also, high speed imaging to further study the liquid-vapor interface may help improve the model. It also has to be made sure that the exposure time for the high speed images is high enough to capture both the bubble nucleation and its transit across the surface. The current study was mainly focused on heat dissipation from the electronic com- ponent to the liquid using an asymmetric heat sink, that was fabricated to achieve self-propelled ow of liquid. While heat transfer from the surface to the liquid is im- portant, it is also necessary for the heat to be rejected to another sink, so that the liquid could be continuously used in a closed loop. A complimentary process to the one discussed in the current study is condensation heat transfer, which is often used in waste heat recovery systems. A surface similar to the one constructed in the current study, coupled with asymmetry in surface tension induced by hydrophilic-hydrophobic junctions, could e ectively be used in self-propelled transport of the condensate. 110 Bibliography [1] M. C. Jo, Thermally actuated pumping of a single-phase uid using surface asymmetry, Master?s thesis, Oregon State University, OR, 2008. [2] A. Alexeev, T. Gambaryan-Roisman, P. Stephan, Marangoni convection and heat trans- fer in thin liquid lms on heated walls with topography: experiments and numerical study, Physics of Fluids 17 (2005) 062106. [3] A. D. Stroock, R. F. Ismagilov, H. A. Stone, G. M. Whitesides, Fluidic ratchet based on marangoni-b enard convection, Langmuir 19 (2003) 4358{4362. [4] R. F. Ismagilov, D. Rosmarin, D. H. Gracias, A. D. Stroock, G. M. Whitesides, Competi- tion of intrinsic and topographically imposed patterns in b enard{marangoni convection, Applied Physics Letters 79 (2001) 439{441. [5] S. J. VanHook, M. F. Schatz, W. D. McCormick, J. Swift, H. L. Swinney, Long- wavelength instability in surface-tension-driven b enard convection, Physical review let- ters 75 (1995) 4397. [6] V. Carey, Liquid-vapor phase-change phenomena: an introduction to the thermophysics of vaporization and condensation processes in heat transfer equipment, Taylor & Francis London, 1992. [7] C. Witze, V. Schrock, P. Chambre, Flow about a growing sphere in contact with a plane surface, International Journal of Heat and Mass Transfer 11 (1968) 1637{1652. [8] H. Linke, B. Alem an, L. Melling, M. Taormina, M. Francis, C. Dow-Hygelund, V. Narayanan, R. Taylor, A. Stout, Self-propelled leidenfrost droplets, Physical re- view letters 96 (2006) 154502. 111 [9] H. Yuan, A. Prosperetti, The pumping e ect of growing and collapsing bubbles in a tube, Journal of Micromechanics and Microengineering 9 (1999) 402. [10] D. Meng, C.-J. Kim, Micropumping of liquid by directional growth and selective venting of gas bubbles, Lab Chip 8 (2008) 958{968. [11] T. K. Jun, C.-J. Kim, Valveless pumping using traversing vapor bubbles in microchan- nels, Journal of Applied Physics 83 (1998) 5658{5664. [12] A. Goyal, R. C. Jaeger, S. H. Bhavnani, C. D. Ellis, N. K. Phadke, M. Azimi-Rashti, J. S. Goodling, Formation of silicon reentrant cavity heat sinks using anisotropic etching and direct wafer bonding, Electron Device Letters, IEEE 14 (1993) 29{32. [13] C. Baldwin, S. Bhavnani, R. Jaeger, Toward optimizing enhanced surfaces for passive immersion cooled heat sinks, Components and Packaging Technologies, IEEE Transac- tions on 23 (2000) 70{79. [14] N. Nimkar, S. Bhavnani, R. Jaeger, Benchmark heat transfer data for microstructured surfaces for immersion-cooled microelectronics, Components and Packaging Technolo- gies, IEEE Transactions on 29 (2006) 89{97. [15] S. G. Kandlikar, D. A. Willistein, J. Borrelli, Experimental evaluation of pressure drop elements and fabricated nucleation sites for stabilizing ow boiling in minichannels and microchannels, Sat 8 (2005) 1{1. [16] A. Ko sar, C.-J. Kuo, Y. Peles, Boiling heat transfer in rectangular microchannels with reentrant cavities, International Journal of Heat and Mass Transfer 48 (2005) 4867{ 4886. [17] S. H. Bhavnani, G. Fournelle, R. C. Jaeger, Immersion-cooled heat sinks for electronics: insight from high-speed photography, Components and Packaging Technologies, IEEE Transactions on 24 (2001) 166{176. 112 [18] B. Morgan, C. M. Waits, J. Krizmanic, R. Ghodssi, Development of a deep silicon phase fresnel lens using gray-scale lithography and deep reactive ion etching, Microelectrome- chanical Systems, Journal of 13 (2004) 113{120. [19] M. Whitley, R. Clark, J. Shaw, D. Brown, P. Erbach, G. Dorek, Deep grayscale etching of silicon, 2002. [20] K.-S. Chen, A. A. Ay on, X. Zhang, S. M. Spearing, E ect of process parameters on the surface morphology and mechanical performance of silicon structures after deep reactive ion etching (drie), Microelectromechanical Systems, Journal of 11 (2002) 264{275. [21] H. Jansen, M. de Boer, R. Legtenberg, M. Elwenspoek, The black silicon method: a universal method for determining the parameter setting of a uorine-based reactive ion etcher in deep silicon trench etching with pro le control, Journal of Micromechanics and Microengineering 5 (1995) 115. [22] C. M. Waits, A. Modafe, R. Ghodssi, Investigation of gray-scale technology for large area 3d silicon mems structures, Journal of Micromechanics and Microengineering 13 (2003) 170. [23] C. Waits, B. Morgan, M. Kastantin, R. Ghodssi, Microfabrication of 3d silicon mems structures using gray-scale lithography and deep reactive ion etching, Sensors and Actuators A: Physical 119 (2005) 245{253. [24] L. Rayleigh, On the pressure developed in a liquid during the collapse of a spherical cavity, The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science 34 (1917) 94{98. [25] B. Mikic, W. Rohsenow, P. Gri th, On bubble growth rates, International Journal of Heat and Mass Transfer 13 (1970) 657{666. 113 [26] M. Plesset, S. A. Zwick, The growth of vapor bubbles in superheated liquids, Journal of Applied Physics 25 (1954) 493{500. [27] L. Scriven, On the dynamics of phase growth, Chemical Engineering Science 10 (1959) 1{13. [28] N. Zuber, The dynamics of vapor bubbles in nonuniform temperature elds, Interna- tional Journal of Heat and Mass Transfer 2 (1961) 83{98. [29] R. Cole, H. L. Shulman, Bubble growth rates at high jakob numbers, International Journal of Heat and Mass Transfer 9 (1966) 1377{1390. [30] S. Van Stralen, M. Sohal, R. Cole, W. Sluyter, Bubble growth rates in pure and binary systems: combined e ect of relaxation and evaporation microlayers, International Journal of Heat and Mass Transfer 18 (1975) 453{467. [31] B. Mikic, W. Rohsenow, Bubble growth rates in non-uniform temperature eld, Progress in Heat and Mass Transfer 2 (1969) 283{293. [32] H. C. Lee, B. D. Oh, S. W. Bae, M. H. Kim, Single bubble growth in saturated pool boiling on a constant wall temperature surface, International journal of multiphase ow 29 (2003) 1857{1874. [33] C. Ramaswamy, Y. Joshi, W. Nakayama, W. Johnson, High-speed visualization of boiling from an enhanced structure, International journal of heat and mass transfer 45 (2002) 4761{4771. [34] C. Hutter, D. Kenning, K. Se ane, T. Karayiannis, H. Lin, G. Cummins, A. Walton, Experimental pool boiling investigations of fc-72 on silicon with arti cial cavities and in- tegrated temperature microsensors, Experimental Thermal and Fluid Science 34 (2010) 422{433. 114 [35] S. Moghaddam, K. Kiger, Physical mechanisms of heat transfer during single bub- ble nucleate boiling of fc-72 under saturation conditions-i. experimental investigation, International Journal of Heat and Mass Transfer 52 (2009) 1284{1294. [36] F. Demiray, J. Kim, Microscale heat transfer measurements during pool boiling of fc-72: e ect of subcooling, International journal of heat and mass transfer 47 (2004) 3257{3268. [37] J. Straub, Boiling heat transfer and bubble dynamics in microgravity, Advances in heat transfer 35 (2001) 57{172. [38] J. Kim, J. Benton, Highly subcooled pool boiling heat transfer at various gravity levels, International journal of heat and uid ow 23 (2002) 497{508. [39] R. Raj, J. Kim, J. McQuillen, Pool boiling heat transfer on the international space station: Experimental results and model veri cation, Journal of Heat Transfer 134 (2012) 101504. [40] M. Zell, J. Straub, A. Weinzierl, Nucleate pool boiling in subcooled liquid under mi- crogravity results of texus experimental investigations, in: 5th European Symp. on Material Science under Microgravity, 1984, pp. 327{333. [41] C. Henry, J. Kim, A study of the e ects of heater size, subcooling, and gravity level on pool boiling heat transfer, International journal of heat and uid ow 25 (2004) 262{273. [42] D. Qiu, V. Dhir, D. Chao, M. Hasan, E. Neumann, G. Yee, A. Birchenough, Single- bubble dynamics during pool boiling under low gravity conditions, Journal of thermo- physics and heat transfer 16 (2002) 336{345. [43] H. Lee, H. Merte Jr, Pool boiling mechanisms in microgravity, Ann Arbor 1001 (1999) 48109{2125. 115 [44] R. Thompson, K. DeWitt, T. Labus, Marangoni bubble motion phenomenon in zero gravity, Chemical Engineering Communications 5 (1980) 299{314. [45] P. Hadland, R. Balasubramaniam, G. Wozniak, R. Subramanian, Thermocapillary migration of bubbles and drops at moderate to large marangoni number and moderate reynolds number in reduced gravity, Experiments in uids 26 (1999) 240{248. [46] C. Henry, J. Kim, J. McQuillen, Dissolved gas e ects on thermocapillary convection during boiling in reduced gravity environments, Heat and mass transfer 42 (2006) 919{928. [47] J. R. Black, Electromigration failure modes in aluminum metallization for semiconduc- tor devices, Proceedings of the IEEE 57 (1969) 1587{1594. [48] J. R. Black, Electromigrationa brief survey and some recent results, Electron Devices, IEEE Transactions on 16 (1969) 338{347. [49] R. Wol enbuttel, K. Wise, Low-temperature silicon wafer-to-wafer bonding using gold at eutectic temperature, Sensors and Actuators A: Physical 43 (1994) 223{229. [50] S. Lani, A. Bosseboeuf, B. Belier, C. Clerc, C. Gousset, J. Aubert, Gold metallizations for eutectic bonding of silicon wafers, Microsystem technologies 12 (2006) 1021{1025. [51] M. Cooper, A. Lloyd, The microlayer in nucleate pool boiling, International Journal of Heat and Mass Transfer 12 (1969) 895{913. [52] W. Fritz, W. Ende, Uber den verdampfungsvorgang nach kinematographischen aufnah- men an dampfblasen, Phys. Z 37 (1936) 391{401. [53] H. K. Forster, N. Zuber, Growth of a vapor bubble in a superheated liquid, Journal of Applied Physics 25 (1954) 474{478. [54] H. S. Lee, H. Merte Jr, Hemispherical vapor bubble growth in microgravity: experiments and model, International journal of heat and mass transfer 39 (1996) 2449{2461. 116 [55] Model 1750 Constant Temperature Anemometer - User Manual, revision n ed., TSI, 2008. [56] T. Papanastasiou, G. Georgiou, A. N. Alexandrou, Viscous uid ow, CRC Press, 2010. [57] F. Kapsenberg, N. Thiagarajan, V. Narayanan, S. Bhavnani, Lateral motion of bub- bles from surfaces with mini ratchet topography modi cations during pool boiling- experiments and preliminary model, in: Thermal and Thermomechanical Phenomena in Electronic Systems (ITherm), 2012 13th IEEE Intersociety Conference on, IEEE, 2012, pp. 165{175. [58] T. G. Beckwith, R. D. Marangoni, J. H. Lienhard, Mechanical measurements, Pearson Prentice Hall, 2007. [59] J. Westerweel, Fundamentals of digital particle image velocimetry, Measurement Science and Technology 8 (1997) 1379. 117 Appendices 118 Appendix A Thermophysical Properties of FC-72 and Water Thermophysical properties of the working uid used in this study, FC-72, are listed in the table below, along with properties of water for comparison. Table A.1: Thermophysical properties of FC-72 and saturated water at 1 atm Properties FC-72 Water Saturation temperature, Tsat ( C) 56.6 100 Speci c heat, Cp (J/kg-K) 1097.41 4190 Latent heat of vaporization, hfg (J/kg) 84510.9 2257030 Liquid density, l (kg/m3) 1619.73 957.85 Vapor density, v (kg/m3) 13.396 0.598 Thermal conductivity, k (W/m-K) 0.05216 0.68 Dynamic viscosity, (Ns/m2) 0.000454 0.000279 Prandtl number, Pr 9.555 1.76 Surface tension, (N/m) 0.008273 0.0589 From Table A.1, it can be noticed that FC-72 boils at 56.6 C, which is one of the signi cant advantages compared to water as this property allows the temperatures of surface being cooled by phase change to be maintained at a level considerably lower than that with water. Also, FC-72 is heavier and more viscous than water. Especially, for the application of liquid pumping due to nucleate boiling, the velocities obtained in FC-72 will be lower owing to the higher liquid inertia. Lower surface tension of FC-72, compared to water, leads to very low contact angles and smaller vapor bubbles. The drawbacks of FC-72 include poor heat transfer properties such as lower thermal conductivity and latent heat of vaporization. 119 Appendix B Preliminary Closed Loop Experiments Under Terrestrial Gravity Using FC-72 Prior to conducting experiments in a large chamber using an open loop channel, ex- periments were conducted to test the concept in a closed loop set up. The closed loop experimental set up has a \race-track" shaped ow channel as shown in Fig. B.1. The two part experimental setup was made of polycarbonate. The top plate served as a lid with ports for with ports for condenser and an accumulator. The bottom plate comprises the ow channel with ports for temperature and pressure measurements at both ends of the test device. The slot on the surface of the ow channel serves to hold the three layered test section. On the inlet side of the test section, a ceramic heater of capacity 1000W is tted on to the surface of a copper block to preheat the liquid upstream of the test section, hence providing the required inlet subcooling. The exit of the test section is connected to an accumulator which serves to control the pressure at the exit. The degassing operation was accomplished by heating the uid in the ow channel using both the pre-heater and the test section heater. The dissolved gases will be vented out through the Graham condenser tted to the top plate. Figure B.1: Bottom plate used in the closed loop experiments with FC-72. The \race-track" is the ow channel in which ow is intended to occur in the clockwise direction. The silicon test device is the same as the one use for open channel experiments. However, additional layers of a glass plate and a Te on block of the same footprint as that of the test 120 Figure B.2: Assembly of the closed loop experimental set up showing the polycarbonate ow loop and the assembly of the test device. device were attached. The Te on layer was added to increase the structural rigidity of the entire test device, and the glass layer acts as an insulator. The three layered test device was mounted on pogo pins that served as electrical contacts. The electrical contacts, in turn were connected to the power supply and NI DAQ. An assembled view of the experimental set up is shown in Fig. B.2. The ow loop is schematically represented in Fig. B.3 In the experiments conducted with the closed ow loop held at close to 0, an interesting phenomena was noticed which actually caused a net ow although in the direction opposite to the intended direction. As the power supplied to the test section was increased, the vapor cloud over the test device increased as illustrated in Fig. B.4. As the growth increased, the liquid vapor interface on the left reached the condenser and stopped growing. But the interface on the right continued its growth to the right of the test device and reached the subcooled pool of liquid 121 Figure B.3: Schematic representation of the closed ow loop used in the study. 122 Figure B.4: Schematic representation of the net ow e ected mainly by the condenser loca- tion. on the other side of the ow track. This resulted in the condensation of vapor on the other side of race-track, where the liquid-vapor vapor blanket breaks in to smaller bubbles that circulate in an anti-clockwise direction. As the supplied power increased this mechanism increased in strength and frequency and resulted in a net ow as the vapor pockets reached the condenser by owing in a counter clockwise direction. However even as this mechanism existed, a portion of vapor over the test device also moved to the left into the condenser, but there is nothing to suggest that this was being caused by the bubbles departing at an angle. The ow was mainly dictated by the position of the condenser. By shifting the condenser to the other side of the condenser, ow in the opposite direction, clockwise, was achieved. However, as seen in all the previous experiments conducted in a chamber in an open channel, the bubbles departed at an angle normal to the saw-teeth. 123 Appendix C Calibration of Sensors This section discusses the calibration of sensors used in the experiments and the asso- ciated uncertainty in the measurements. C.1 Calibration of thermistor and thermocouple An Omega thermistor ON44007 was used in the experiments for pool temperature mea- surements and a K-type thermocouple was used for the measurement of heat loss from the back of the test section. Both sensors were calibrated using an NIST certi ed thermis- tor as a standard in an oven, where the temperature was periodically increased from room temperature to 80 C. The calibration plot from the experiment is shown in Fig. C.1. Figure C.1: Calibration of Omega ON44007 thermistor and a K-type thermocouple using a NIST certi ed thermistor as a standard From Fig. C.1, the calibration equations for thermistor and K-type thermocouple are, Tstd;thermistor = 1:0127Tthermistor 0:2579 (C.1) Tstd;thermistor = 0:9389Tthermocouple + 0:5714 (C.2) 124 Based on the calibration equations, the experimental uncertainty [58] can be expressed as, Ua = t =2; Sy=xS xx (C.3) where, t =2; = student T-distribution Sy=x = data?s vertical standard deviation about the tted line Sxx = mean horizontal variation of data The uncertainty values were estimated to be 2.12% and 1.81% in the thermistor and thermocouple, respectively. C.2 Calibration of test section heater As discussed in chapters 2 and 3, the aluminum serpentine heater also serves as a temperature sensor by using the resistance measurement from the heater. This is made possible since the resistance of the current carrying material is a function of temperature. Resistance can be estimated by measuring the voltage from the heater and the current supplied (50 mA) using an ammeter. The calibration was performed using an NIST certi ed thermistor as a standard, in an oven where the temperature was periodically increased from room temperature to 85 C in 5 C increments. The calibration plot from the experiment is shown in Fig. C.2. Figure C.2: Calibration of test section heater using a NIST certi ed thermistor as a standard From Fig. C.2, the calibration equation for the heater is, T = 13:404Rheater 280:62 (C.4) where, 125 T = Temperature, C Rheater = Resistance of heater, Based on the calibration equations, and the procedure stated previously in the calibra- tion of temperature sensors, the uncertainty in heater temperature measurement is estimated to be 6.4% of measured temperature. C.3 Calibration of hot wire anemometer Hot wire anemometer was used in the experiments to estimate the net lateral liquid velocity of the surface. Hot wire anemometer involves the use of a probe which is maintained at a constant temperature. When the probe is immersed in a liquid, the heat transfer from the probe causes a change in the bridge output of the anemometer which is a measure of liquid velocity. While it is a relatively simple process to calibrate the probe for use in air, the calibration for use in a uid such as FC-72 was extensive and required building a separate ow loop as shown in Fig. C.3. Figure C.3: Flow loop for the calibration of hot wire probe The probe used in the experiment is TSI 1210-20 and the pump?s meter is used as the known velocity of ow. Velocity was increased periodically from 1 mm/s to 130 mm/s and the output from hot wire anemometer was sent to the data acquisition system. From Fig. C.4, the calibration equation for the hot wire anemometer probe is, v = 0:1521V 4HWA 1:0704V 3HWA + 2:8182V 2HWA 3:2705V + 1:4083 (C.5) where, v = velocity, m=s VHWA = voltage output of anemometer, V Based on the calibration equation, the estimated uncertainty is 0.2% of the measured velocity. 126 1.4 1.5 1.6 1.7 1.8 1.9 2 2.10 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02 Bridge Output (V) Velocity, V (m/s) Upward Flow Tube diamater = 1 in. Fluid temperature = 22.12oC Experiment ? Probe 1210?20?7122011 ? w/o L?adapter y=0.15213x4?1.0704x3 + 2.8182x2?3.2705x +1.4083 Figure C.4: Flow loop for the calibration of hot wire probe 127 Appendix D Emperical Model for Lateral Liquid Velocity in Nucleate Boiling From chapter 3, it is evident that the asymmetry in surface cross-section and location of re-entrant cavities results in bubble growth and departure to take place at an angle, approximately, normal to the shallow slope of the ratchet. It was earlier hypothesized, that such angled growth and departure leads to an angular momentum imparted to the adjacent liquid, which in turn leads to a net lateral liquid velocity. Based on the available data for bubble dynamics such as growth rate and bubble departure frequency, an expression for the net lateral liquid velocity was derived [57] by estimating the momentum imparted by the growth of the vapor bubble while being attached to the re-entrant cavity on the sloped surface as illustrated in Fig. 3.15a. The drag force due to bubble growth, Fg, can be estimated as, !Fg = d !P l dt = 1 8 lCDD 2 dD dt 2 (D.1) where, !F g = force due to bubble growth ! Pl = momentum imparted to the liquid dD dt = bubble growth rateC D = co-e cient of drag = 24Re For the model, the entire surface of the the heat sink is divided into equal number of regions, with each region representing an area around the mouth of a single re-entrant cavity as shown in Fig. 3.15b. The volume projected by this area will be treated as the control volume for the analysis. By re-arranging Equation D.1, the momentum imparted by the bubble is estimated for a bubble growth cycle starting with nucleation at time, t = 0, to the time frame penultimate to bubble departure, t = td, as shown in Equation D.2 The bubble diameter, D, in Equation D.2 is the average diameter of bubble in successive time frames, and dDdt is the experimental bubble growth rate. !Pl = 18 l i=td 1X i=t0 CD D i +Di+1 2 2 D i Di+1 dt 2 dt (D.2) where, t0 = time at bubble inception td = time at bubble departure D = average bubble diameter = Di+Di+12 dD dt = bubble growth rate = Di Di+1 dt 128 Figure D.1: Illustration of the angular momentum imparted by a growing bubble normal to the shallow slope, leading to a net lateral liquid velocity Figure D.2: Control volume de ned for estimation of resulting net lateral liquid velocity due to bubble growth at an angle normal to the shallow slope of the ratchet 129 Momentum imparted to the liquid for the control volume shown in Fig. 3.15 can be expressed as, !P l = ml !vl = lAtsN c cos !vlf 1d !vl (D.3) where, Ats = area of footprint of the test section Nc = number of cavities = angle of saw - tooth fd = bubble departure frequency ! vl = net liquid velocity Using Equations D.2 and D.3, the equation D.4 for net liquid velocity is expressed as, !vl = vu ut Nc fd 8Ats cos i=td 1X i=t0 CD D i +Di+1 2 2 D i Di+1 dt 2 dt (D.4) From the above equation, the net lateral or horizontal component of liquid velocity is vl;H = !vl cos(90 ) (D.5) Pool boiling experiments using water on a small array test device was conducted by the research group at Oregon State University [57]. Using the above equation, net horizontal lateral liquid velocities of 7 mm/s and 18 mm/s were estimated for low and high subcooling cases, respectively. These velocities were compared with experimental observation of bubbles during nucleate boiling from the asymmetric structure. It was observed that the bubble growth and departure from the re-entrant cavities were asymmetric in nature. As the bubbles departed in to the subcooled pool, the volume of bubbles reduced signi cantly. These small bubbles in the far- eld, known as plume bubbles, were observed to be moving horizontally. It was hypothesized that the lateral motion of the far- eld bubbles were e ected by the asymmetric growth of bubbles on the surface. The experimentally observed velocities of the plume bubbles were between 30 and 70 mm/s. Since, the velocities predicted by the model were close to the experimentally observed plume bubble velocity, it was inferred that, asymmetric bubble growth is a primary force in e ecting lateral motion of liquid in its vicinity. 130 Appendix E Particle Image Velocimetry (PIV) Particle image velocimetry (PIV) is an optical technique to measure uid ow. In PIV, uid is seeded with particles which are imaged at two instances of time. By estimating the displacement of the particles, the instantaneous velocity of the ow can be measured, pro- vided the selected tracer particles track the ow adequately [59]. A schematic arrangement of a PIV system is shown in Fig. E.1, which illustrates the sheet of light formed by a laser used for illuminating the seeding particles. The image of the light re ected from the particles is recorded using a digital camera which is synchronized with the laser. Figure E.1: Illustration of a PIV system showing the laser sheet that is used to illuminate the particles. The image of the particles is recorded at two time instances using a camera to measure velocity [59] Since the measured uid velocity using this technique is that of the particle velocity, it has to be determined that the particles follow the ow. Particles in a ow can be in uenced by gravitational forces if the density of particles and density of the uid are di erent. Di erence in densities will cause the particles to settle or rise in a uid. The settling velocity can be expressed as, Vsettling = 2R 2 ( p f)g 9 (E.1) 131 where, Vsettling = settling velocity of tracer particles, m=s R = mean radius of tracer particles, m p = density of tracer particles, kg=m3 f = density of uid, kg=m3 g = acceleration due to gravity, m=s2 = viscosity of uid, Ns=m2 In chapter 3, liquid velocity measurements using hot wire anemometry showed that asymmetric bubble growth and departure can, potentially, induce net lateral velocities of liquid of up to 6 mm/s. In the current study, PIV is used to measure the ow eld of liquid in the immediate vicinity of the bubbles growing and departing at an angle normal to the surface. Measured velocities will be used to corroborate the results from hot wire anemometry. The PIV system used in the current system consists of a dual ND:YAG 532 nm, 200 mJ, pulsed laser that is used to produce a laser sheet to illuminate the particles. The imaging system consists of a PowerView Plus (Model no. 63002) 11 megapixel camera with a pixel resolution of 4008 x 2672, pixel size of 7.4 m x 7.4 m, and a 12-bit output. Silver coated hollow glass particles with a mean diameter of 14 m, and a speci c gravity of 2.2, were used as tracer particles. The settling velocity of these particles in FC-72 calculated using Eq. E.1 is 5.4 m/s. The boiling chamber used is made up of 2 inch x 2 inch borosilicate glass as shown in Figures E.2 and E.3. The borosilicate glass tube containing FC-72 was covered by a polycarbonate lid at the top with ports for various sensors used. A thermistor was used for pool temperature measurements, and a cartridge heater was used for heating the uid. A central port was used to connect a Graham condenser for condensing the FC-72 vapor. One of the ports was also used for introducing the tracer particles and another port was used for thermocouples that were used for heater temperature measurements. The heater used is cartridge type, 50 W, 3.17 mm diameter, and 31.7 mm in length. The heater was enclosed in a copper block of 7.9 mm square cross-section, and power to the heater was controlled using an AC variac. The copper block was equipped with three 1 mm holes for thermocouples that are spaced equidistantly, and the temperature measurements were used to calculate the applied heat ux. The copper heating block was covered by a 19.0 mm diameter Te on block that serves in promoting one-dimensional heat transfer through the copper block, by insulating the sides. A small array silicon test device with a saw-tooth angle of 24 was xed to the surface of the copper block using a colloidal silver paste. Before the PIV experiment, the pre-processor, processor, and post processor settings were adjusted in the software, Insight 4G. In the pre-processor settings, image calibration was performed to convert pixel values to mm using the distance between the peaks of the saw teeth. Also, an image mask was applied to set the processing area to cover the region right over the test section. In the processor settings, the size of the interrogation window was set to be 128 pixels x 128 pixels. Post-processor settings were adjusted to display the velocity vector statistics. During the experiment, the laser was triggered from the software in a continuous mode to align the laser sheet with the test section. During the alignment process, the aperture of the camera was adjusted to remove any light saturation in the captured images. Once the camera was aligned, the laser was triggered in a single mode in 132 Figure E.2: Assembled view of the PIV chamber - basic parts include borosilicate glass chamber, and a top polycarbonate lid with ports for sensors Figure E.3: Exploded view of the PIV chamber showing the assembly of the copper heating block, borosilicate glass chamber, and a top polycarbonate lid with ports for sensors 133 Figure E.4: Velocity vectors obtained from a PIV experiment that are overlaid on top of the captured image. In this image, the shallow long slopes of the saw teeth face towards the left side of the image, and any expected ow due to asymmetric bubble growth will be from right to left in the image. Vectors colored in green and yellow represent measured and interpolated velocities, respectively. which the laser is pulsed twice between a set time interval (in this case 10,000 s) to obtain a set of two images. The captured images were processed to obtain the velocity vectors in the uid. Fig. E.4 shows the obtained velocity vectors overlaid on top of the captured image, and Fig. E.5 shows the magnitude of velocity. The color chart indicates the velocity in m/s. In the images, saw teeth were oriented with the shallow long slope facing the left side of the image and the expected liquid ow direction is from right to left in the image. From the gures, while liquid velocities of up to 17 mm/s were observed, the velocity vectors were scattered in di erent directions. Hence, no inference could be made about the e ect of asymmetric bubble growth on the liquid ow around the bubble. The scatter in vectors can be caused due to multiple factors: Particles used in the study were re ective in nature. In two-phase processes, bubbles re ect the light signi cantly, and hence distinction between the light from the particles and bubbles is very di cult. To avoid this problem in PIV for two-phase ows, uo- rescent particles with a 532 nm lens lter for the camera is often used. While such a system was tested in the current study, nding uorescent particles of a speci c grav- ity value that matches the speci c gravity of FC-72 was di cult. Hence, PIV using uorescent particles of speci c gravity 1.2 did not yield good results. Although e orts were taken to eliminate bubbles from the edge of the test section, there were a few bubbles nucleating from locations other than re-entrant cavities on the saw- toothed surface which could alter the ow around the bubbles nucleating from cavities. To get an accurate measure of the ow eld around the bubble, it is important to have 134 Figure E.5: Magnitude of velocity vectors obtained from the PIV experiment. Color chart indicates liquid velocities of up to 17 mm/s, although scattered in di erent directions 135 just one single nucleating bubble, which is a very di cult objective to accomplish during phase-change. 136 Appendix F Microgravity Experiments: Test Equipment Data Package (TEDP) Design of pool boiling experiments that were conducted aboard Zero-g ight required strict adherence to mechanical, electrical and uid containment guidelines. Structurally, every joint and load bearing member of the structure required a factor of safety of at least 4 for load conditions of up to 9g. This required a thorough structural analysis of the structure using ANSYS and the factor of safety was calculated based on the stresses calculated at every joint and structural member. Some of the electrical design guidelines include the design of an emergency stop for every current drawing device on the set up and adherence to proper wiring and grounding procedures. Since the experiments involved the use of uids, the experimental setup had to be double contained to prevent any uid leak in to the aircraft cabin. Also, any drop in cabin pressure could cause the welded joints of aluminum boiling chamber to leak due to the increase in pressure di erential. Hence, pressure testing of the boiling chamber was performed to withstand a pressure di erential of up to 12 psi between the inside and outside of the boiling chamber. Prior to experimentation aboard the aircraft at Johnson Space Center, a document consisting of all the design data, structural analysis, electrical design and experimental pro- cedures, known as the Test Equipment Data Package (TEDP) was submitted for approval, which was tested against the actual set up for conformance to design speci cations. This section includes the submitted version of TEDP to the Reduced Gravity Program at NASA Johnson Space Center. 137 ii DEPARTMENT OF MECHANICAL ENGINEERING AUBURN UNIVERSITY Sushil H. Bhavnani Principal Investigator Professor, Mechanical Eng., Auburn University, AL ? 36849. bhavnsh@auburn.edu Naveenan Thiagarajan Author Doctoral Candidate Mechanical Eng., Auburn University, AL ? 36849 thiagna@auburn.edu Fluid Lateral Motion using Surface Microstructures-Channel Flow from a Large Array Test Equipment Data Package 4/27/2012 138 iii CHANGE PAGE Revision Description of changes Date TEDP-1.0 Initial release 02/27/12 TEDP-2.0 Updated release after review of initial release 03/13/12 TEDP-3.0 Updated release incorporating joint structural analysis, grounding method, and other minor changes 04/06/12 TEDP-4.0 Updated structural analysis for the LCD monitor 04/12/12 TEDP-5.0 Added structural analysis for DAQ, power strip and keyboard docking tray and updated overall weight. Flight manifest has been updated. Camera assembly to be stowed during take-off and landing. Weight of the equipment frame has been updated and structure analysis on the floor attachment has been modified to reflect this. 04/27/12 139 iv QUICK REFERENCE DATA SHEET Principal Investigator: Prof. Sushil H. Bhavnani Contact Information: bhavnsh@auburn.edu, (334) 844-3303 Experiment Title: Boiling Heat Transfer Mechanism During Thermally Actuated Pumping by Asymmetric Surface Structures Work Breakdown Structure (WBS): Flight Date(s): April 30th ? May 4th, 2012 Overall Assembly Weight: 205.64 [lbs.] (187.64 lbs during take-off and landing) Assembly Dimensions (L, W, H): 26.0, 40, 48 [in] Equipment Orientation Requests: None Proposed Floor Mounting Strategy (Bolts/Studs or Straps): Bolts (AN-6) Gas Cylinder Requests (Type & Quantity): No Overboard Vent Request (Yes or No): No Power Requirement (Voltage and current required): 220VAC and 10A; 115 VAC and 20 A, see the electrical analysis section for power breakdown. Free Float Experiment (Yes or No): No Flyer Names for Each Proposed Flight Day: Prof. Sushil Bhavnani, Naveenan Thiagarajan, and Travis Wheeler on all flight days Camera Pole and/or Video Support: No 140 v TABLE OF CONTENTS 1. FLIGHT MANIFEST???????????????????????????????????....1 2. EXPERIMENT BACKGROUND?????????????????????...................................1 3. EXPERIMENT DESCRIPTION??????????????????????????????...3 4. EQUIPMENT DESCRIPTION??????????????????????...................................4 5. STRUCTURAL VERIFICATION??????????????.??????....????????.14 6. ELECTRICAL ANALYSIS????????????????????????????????...32 7. PRESSURE VESSEL/SYSTEM??????????????????????????????35 8. LASER CERTIFICATION????????????????????????????????...36 9. PARABOLA DETAILS & CREW ASSISTANCE REQUIRED??????????????.....36 10. INSTITUTIONAL REVIEW BOARD???????????????????????????36 11. HAZARD ANALYSIS???????????????????????????????????36 12. TOOL REQUIREMENTS?????????????????????????????????.37 13. PHOTO REQUIREMENTS????????????????????????????????.37 14. AIRCRAFT LOADING??????????????????.............................................................37 15. GROUND SUPPORT REQUIREMENTS???????????.....................................................37 16. HAZARDOUS MATERIALS???????????????????????????????.37 17. MATERIAL SAFETY DATA SHEETS??????????????????................................38 18. EQUIPMENT PROCEDURE DOCUMENTATION????????????????????.43 19. BIBLIOGRAPHY????????????????????????????????????....44 141 1 1. FLIGHT MANIFEST Flight One (Day 3): Sushil H. Bhavnani, Naveenan Thiagarajan, and Travis Wheeler Flight Two (Day 4): Sushil H. Bhavnani, Naveenan Thiagarajan, and Travis Wheeler 2. EXPERIMENT BACKGROUND Liquid-vapor phase change systems are now being considered as an attractive solution for the heat dissipation problems in modern computers and high power electronics. Several investigations in the last couple of decades have been aimed at providing a solution for the problem of rising demands of heat flux dissipation. Boiling as a mode of high heat flux dissipation can be realized in two different ways, namely, pool boiling which involves heat rejection to a stagnant pool of boiling liquid as in the case of liquid immersion cooling techniques, and flow boiling which involves heat rejection to a liquid flowing in a closed loop such as in a microchannel which has a very high heat dissipation capabilities. However, microchannels are associated with high pumping requirements owing to the high pressure drops associated with flow in narrow passages. This challenge only grows in cases of high power electronics and even more in thermal management of electronics in space since power availability is very limited. So the challenge remains to be the development of a heat dissipation module utilizing the high heat transfer rates involved in boiling without the pumping requirements involved in a flow boiling system. Such a challenge can be confronted only by the design of a self propelled fluid flow system i.e. a pump-less flow loop, where liquid flow will be thermally induced by vapor bubbles departing from an asymmetric surface. Studies in the past have shown that liquid flow can be thermally induced by the repeated growth and collapse of a bubble asymmetrically in a tube [1]. Bubble micro-pumps have also been demonstrated where the pumping effect was due to asymmetry in surface tension forces created by the shape of the flow channel [2]. Experiments conducted by Linke et al. [3] demonstrated that in a two-phase film boiling regime, droplet motion as shown in Figure 1 was observed due to the asymmetry of the surface like the one used proposed in this study. However, the droplet motion occurred at Liedenfrost temperatures and the net motion was attributed to the pressure differences in the vapor layer between the droplet and the surface. Figure 1: Liedenfrost droplet movement due to surface asymmetry demonstrated by Linke et al. [1] 142 2 Hence, from the aforementioned studies it can be understood that surface asymmetry along with controlled ebullition cycle can be effectively used to alter the bubble dynamics to obtain a net pumping effect, so as to develop a pump less flow boiling thermal management device for electronics. The current study derives its inspiration from the work of Linke et al. [3] and pool boiling enhancement using artificial re-entrant cavities [4]. It is hypothesized that surface asymmetry along with controlled nucleation can be effectively used to alter the bubble dynamics to obtain a net pumping effect, so as to develop a pump less flow boiling thermal management device for electronics. Figure 2 shows a schematic of the proposed thermally-actuated, phase- change pumping concept. In the proposed study, surface asymmetry is obtained by use of an array of miniature saw-tooth ratchets and the bubble ebullition cycle is controlled by the use of re-entrant cavities fabricated on one of the sloped walls of the ratchets. As a first step in the development of such a technique, pool boiling experiments were conducted at 1g with de-ionized water using the test device shown in Figure 2. The surface asymmetry resulted in net force imbalance acting on the vapor bubbles departing from the re-entrant cavities, thereby resulting in bubbles departing at an angle to the vertical with instantaneous velocities in excess of 600 mm/s near the surface [5]. This movement of bubbles demonstrates the significant potential of such a heat sink to be used for thermally actuated pumping of liquids contained in a narrow channel as illustrated in Figure 3 (a) and (b). Figure 3: Schematic of the induced flow in (a-b) terrestrial gravity and (c- d) microgravity Figure 2: (a) Heat sink structure (b) Schematic representation of the concept 143 3 One of the primary objectives of the microgravity experiments is to develop and test the effects of surface asymmetry and re-entrant cavities on boiling mechanisms under reduced gravity and analyze the potential of the developed micro phase change heat sinks for space electronics applications. This however requires a better understanding of vapor bubble dynamics under reduced gravity, but due to the complexity of such experiments, literature corresponding to micro gravity experiments and the related bubble dynamics is very limited. In flow boiling at terrestrial gravity, bubble departure diameter is primarily a function of surface tension (causes the bubbles to remain attached to the surface), drag and buoyancy forces (causes the bubbles to detach from the surface). But at reduced gravity, the forces due to buoyancy are negated and hence the bubble tends to remain at the surface for longer periods and also results in larger bubble departure diameters. This phenomenon was observed by Zhang et al. [6] and Celata [7] in their microgravity experiments performed on FC72. Under such conditions, the vapor slug does not stratify and hence conforms to the shape of the ratchet as illustrated in figure# c and d. The pressure at location 1 is greater than the pressure at 2 or 3, thus resulting in a favorable pressure gradient. Due to asymmetry, and due to the inertial force of the liquid when the cavities are nucleating, the liquid below the slug will cause the vapor slug to move from location 1 to 2. This is similar to the flow of the Leidenfrost droplet observed by the aforementioned study conducted by Linke et al. [3] 3. EXPERIMENT DESCRIPTION Objective: The primary objective of the experiment is to study the effects of microgravity on thermally actuated pumping effects in pool boiling on micro-asymmetric structures. High speed imaging will be used to study the bubble dynamics and the recorded temperature and heat flux data will be used to quantify the heat transfer characteristics. Description: The asymmetric silicon surface used for boiling will be suspended into a rectangular aluminum sealed boiling chamber containing 13.31 lb of FC72 (boiling point - 56.6?C). The required heat input for boiling the fluid will be supplied using a 220VDC power supply with a maximum required current of 10A. All the boiling experiments will be conducted at 1 atm which will be maintained in-flight using bellows. Thermistors and pressure transducers measure the temperature and pressure inside the chamber. The measured data from the sensors will be recorded using NI Compact DAQ. A Phantom high speed camera will be used to record images of departing bubbles under microgravity. All acquired data will be logged into a computer and the data will be monitored on the mounted LCD display. All the components will be mounted onto a frame built using Aluminum t-slot extrusions. 144 4 4. EQUIPMENT DESCRIPTION Table 1 shows the physical dimensions and weight of all the components and assemblies used in the experiment. Table 1: Physical properties of experimental equipment Equipment Physical Dimensions (H, W, D) [in] Weight [lbs.] Equipment Frame 48.0 x 26.0 x 38.625 44.53 Power Supply 3.94 x 17 x 16.94 29.37 Data Acquisition System (DAQ) 2.3 x 6.28 x 3.5 2.26 Boiling Chamber (incl. weight of fluid, and other accessories) 8.75x 13 x 9 34.50 Bellows Assembly 6 Diameter x 18 Height 6 Outer Containment 33 x 24 x 9 18.50 High Speed Camera 5.5 x 12.25 x 5 12 Microscope lens 9.436 3.5 Lights 1.32 Emergency kill switch 4.75 x 11.5 x 11.5 9.92 CPU 13.25 x 4 x 14.94 7.32 Monitor 12.7 x 20.5 x 1.8 7.1 Power strip 12.5 x 2.5 x 1.5 1.45 Keyboard tray 2 x 13.3 x 21.8 2.76 Total 205.64 Total (without camera assembly)* 187.64 *Represents the actual weight during take-off and landing as the camera assembly will be stowed Equipment Frame The support structure for the experiment was assembled using Aluminum t-slot extrusions. All the electrical and electronic equipments required for the experiment will be mounted onto this support structure shown in Figure . The vectors marked in the figure represent the forces acting at the center of mass of the frame. The physical dimensions of the structure are 26 x 38.625 x 48 in (L x W x H). The space required around the structure for operations during the experiment has been marked in Figure 5. The structure will be mounted on to the aircraft floor using six AN-6 bolts using the floor attachment hardware as shown in Figure 6. The floor attachment hardware will be attached to the structure at six points configured in a 20 x 20 in. fashion. The structure is also equipped with steel handles along the length on both the sides which will be used to load and unload the structure. The various components attached to the equipment frame include the boiling chamber with the 145 5 outer fluid containment box, a power supply, data acquisition system, camera, computer and a monitor for data visualization. Figure 4: Equipment frame and the forces acting. 9g represents the forward force in the direction of the aircraft. 9-g 6-g 2-g 2-g 3-g 2-g 3 ft 2 ft 2 ft Figure 5: Image showing top view of the experimental structure with minimum space required around the setup 146 6 Figure 6: Isometric view of the equipment frame. Inset shows the floor mounting attachement for the AN-6 bolt. 147 7 Figure 7: Front view of the equipment frame Boiling Chamber The boiling chamber shown in Figure 8 consists of a rectangular aluminum tank of dimensions 13in x 9in x 8.75in (L x W x H). The side walls of the tank, each 0.5 in. thick, and the bottom plate are welded together. The top lid is mounted on the tank using socket screws and sealed with an EPDM rubber gasket in between. The silicon test section consists of two layers of silicon ? the top layer constituting the asymmetric ratchet structure and the bottom layer consists of a serpentine aluminum heater. An AMREL SPS series power supply is used to power up the test section heater. The silicon test section mounted on a PCB is suspended from the lid using threaded rods into a pool of FC 72. The liquid is heated to the required pool temperature by two cartridge heaters (Omega model# EMH-060- 120V) located on opposite ends of the tank to avoid temperature gradients. Temperature of the pool is measured using two pipe plug thermistors (Omega model# ON-970-44006) probes that are suspended in the liquid. The cartridge heaters with ? NPT thread and the thermistors with 1/8 NPT thread are mounted on to the lid. To facilitate high speed imaging of the boiling process, the front wall of the tank is fitted with a 6? diameter quartz window to facilitate high speed imaging. Lights for imaging will be through two 4? diameter quartz windows located on the other two sides of the tank. All quartz windows are 0.25? thick and will be mounted on the shoulders on the side walls using polyurethane and also Bellow Assembly Boiling Chamber Computer LCD Monitor High Speed Camera Emergency Stop Switch High Speed Camera Polycarbonate outer containment 148 8 held against the walls by a 0.25? polycarbonate ring that is fastened on the wall using eight ?-20 socket screws as shown in Figure 9. Figure 8: Boiling chamber shown without the lid Figure 9: Exploded view of the quartz window assembly 0.25?deep shoulder for placement of quartz window 0.25?thick 4? dia quartz window EPDM gasket 0.25? thick lexan support Side wall of the tank ?-20 socket screws All dimensions in inches 149 9 Figure 10: Boiling chamber assembly During the experiment, as the fluid is heated, increase in fluid volume will cause the bellows to expand thereby maintaining a constant pressure. The outer acrylic vessel is held on to the top lid of the boiling chamber using an aluminum flange. An assembly of the boiling chamber is shown in Figure 10. The whole assembly will be enclosed by 1/8? thick polycarbonate sheets sealed on to the frame using an EPDM rubber gasket around the boiling chamber as shown in Figure 9. To further prevent leaks in the outer containment, the corners of all the polycarbonate sheets are sealed with a polyurethane adhesive. Pressure Transducers The experiment uses a miniature absolute pressure sensors with an amplified output in the range of 0 ? 4V corresponding to a pressure range of 0.3 psi to 100 psi. The sensor utilizes silicon, micro-machined, stress concentration enhanced structure that provides a linear output. The sensor is temperature compensated and has a calibrated zero and span. The sensors require an excitation voltage of 5VDC for operation. Electrical connections and the physical dimensions of the sensor are shown in Figure 11. Neoprene bellows Outer acrylic jacket Lid 6? quartz window for imaging 4? quartz windows for lighting Mounting holes Cartridge heater port Cartridge heater port 150 10 (c) Power Supply The experiment uses a programmable DC power supply to power the test section heater. The power supply used is an AMREL e-POWER SPS series 300V, 3 KW, single phase unit (Model # SPS300-10-K0E1). During the in-flight experiment, the power supply which is facilitated by RS232 interface will be remotely controlled using LABVIEW. The physical dimensions of the power supply are show in Figure 12. (a) (b) (c) Figure 11: (a) Absolute pressure sensor (b) Pin-out diagram (c) Physical dimensions (c) 151 11 Figure 12: Physical dimensions of the power supply POWER SUPPLY SPECIFICATIONS: Maximum Voltage (VMAX): 300 V Maximum Current (IMAX): 10 A Programming Accuracy: Voltage: 0.05%*VMAX + 0.1% of FS Current: 0.01%*IMAX + 0.1% of FS Over-voltage: 0.2% of Vout + 0.3% of FS Measurement Accuracy: Voltage: 0.1% of RDG + 0.1% of FS Current: 0.1% of RDG + 0.2% of FS 152 12 Load Regulation: Voltage: 0.01%*VMAX + 2 mV Current: 0.01%*IMAX + 2 mA Line Regulation: Voltage: 0.001%*VMAX + 2 mV Current: 0.001%*IMAX + 2 mA Transient Response Time: 3ms Programming/Measurement Resolution: 14-bit OVP Programmable Range: 5% - 110% of VMAX Drift7 (8 Hours): CV Mode: 0.5%*VMAX CC Mode: 0.5%*IMAX Temp. Cofficient: CV Mode: 0.02%*VMAX / ?C CC Mode: 0.03%*IMAX / ?C ac Input: 1? 240Vac 50/60Hz dc Output Isolation: ?400Vdc: ?600Vdc ?600Vdc: ?1000Vdc ?800Vdc: ?1500Vdc Data acquisition system (DAQ) The data acquisition chassis as shown in Figure 13 consists of embedded modules ? NI 9219 and NI 9202 that acquire signals from the various sensors in the experiment. The data acquired include voltage from heater, temperature, pressure and accelerometer signals which are monitored and recorded using a desktop computer. Computer A HP Compaq CPU with a quad core processor and 16 GB RAM will be used for monitoring, controlling and recording data from the experiment. The dimension and weight of the mini- tower have been tabulated in Table 1. All data logging and experimental controls will be handled using a wireless keyboard. Figure 13: NI Compact DAQ chassis 153 13 High Speed Camera A Phantom V310 high speed camera will be used to capture high speed videos of the boiling process. Images will be recorded at a frame rate of 3250 fps with a resolution of 1280 x 800. A K2SC Infinity microscope lens with a CF3 objective will be used along with the camera. Both the camera and the lens are supported on a sliding plate attached to the frame on two linear motion bearings which can be locked using brakes as illustrated in Figure 15. The camera and the lens will be stowed during take-off and landing. Figure 14: Phantom V310 high speed camera Figure 15: Camera plate mounted on two linear motion bearings Sliding plate Linear motion bearing K2/SC Infinity Microscope lens 154 14 5. STRUCTURAL VERIFICATION The equipment frame has been designed to withstand the g-load specifications during take- off and landing. The various loads acting on the assembly during takeoff/ landing has been represented in Figure . Stress analysis was performed on every component and assembly placed on the t-slot extrusions and the factor of safety was determined. Factor of safety (FS) can be expressed as, where the maximum allowable stress is the material?s yield strength. The T-slot extrusions used in the frame is made up of Aluminum alloy 6105-T5 and its properties are tabulated in Table 2. Table 2: Properties of T-slot extrusion material Material Aluminum alloy 6105-T5 Modulus of Elasticity, E 10,200,000 psi Moment of Inertia, I 0.0442 in4 Yield Strength (Max. Allowable Stress) 35000 psi Using the properties, the allowable stress was calculated for the extrusions on which the components were placed. Analysis of loads on components: Analysis of stress due to the components placed on the equipment frame at 6g and 9g forces was simulated using Ansys. For the ease of analysis, the cross section of the aluminum extrusion used in the equipment frame was simplified by removing edges of least importance on the face of the extrusion. For the 6g downward acting force, loads were modeled as pressure distributed on the contact area with the extrusion. For the 9g forward acting force, the forces due to the individual components were located at the center of mass of the individual components placed on the frame. The results of the analysis shown in Figures 16 ? 20 illustrate the deformation in the members and the Von Mises equivalent stress. To analyze the structural integrity of the brackets at the joints, the forces acting at the joints were manually compared with the maximum load specifications provided by the manufacturer to obtain a measure of factor of safety for the joints. 155 15 Figure 16: Image showing the location of the components for comparison with results of stress analysis 9g Forward 6g Downward 156 16 Figure 17: Total deformation due a 6g force acting downwards 157 17 Figure 18: Von-Mises equivalent stress due to 6g downward force 158 18 Figure 19: Total deformation due a 9g force acting forward 159 19 Figure 20: Equivalent (Von-Mises) stress due to forward acting 9g force 160 20 Analysis of Joints: The frame made up of several aluminum extrusions are joined together using different types of joints. The various joints used in the experiment and its maximum allowable loads are tabulated in Table 3. The factor of safety for each joint is evaluated as shown below: Joint factor of safety, FS = The maximum allowable load for each of the joints under a direct force (A), as shown in the Figure 21a, is provided by the manufacturer (8020). For other loading conditions such as a normal force acting on the face of the bracket (shown in 21 b), it was assumed that the maximum allowable load is comparable to the direct force scenario. At each of the joints marked in Figure 22, the forces acting in three directions were compared manually to the allowable loads of the bracket. Table 3: Types and specifications of brackets used at the joints Joint Type A B C D E Max. allowable load under direct force (A) (lbs) 175 175 325 325 325 Figure 21: Illustration of forces acting on joint brackets A B (b) (a) 161 21 Figure 22: Illustration of joints whose factors of safety were analyzed Table 4: Factor of safety for the joints marked in Figure # Joint No. Joint Type 6g 9g FSx FSy FSz FSx FSy FSz 1 A & D 20.29 54.56 12.92 9.42 7.12 7.90 2 D & D 294.45 64999.00 8.63 4.29 18.66 3.46 3 A & E 65.76 11.59 9.60 4.69 1427.57 5.05 4 B 42.97 32.46 10.64 88.29 1.59 11.92 5 B 42.97 32.46 11.57 3.43 1.59 11.69 6 B 10.41 65.54 4.36 9.57 43.19 10.89 7 B 10.41 65.54 46.68 6.09 43.19 11.09 8 B 55.27 19.02 10.02 22.71 1.39 13.07 9 B 55.27 19.02 12.39 2.62 1.39 12.80 10 A & E 90.91 13.37 9.78 4.85 33.67 4.26 11 A & D 7.00 184.19 7.74 3.16 8.31 27.84 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 19 21 22 20 18 162 22 12 D & D 51.97 105.21 10.80 3.78 14.01 4.01 13 B 21874 109.06 5.70 3.56 9.34 20.58 14 B 17.96 64.79 11.49 5.24 80.78 13.60 15 B 17.98 64.79 3.78 4.69 80.78 13.60 16 B 349.00 471.97 3.38 3.14 8.19 8.68 17 B & C 891.86 67.40 4.74 12.33 32.99 4.36 18 B & C 441.48 44.58 5.11 15.40 74.41 3.73 19 B & C 260.78 128.53 1.54 7.69 13.21 2.85 20 B & C 134.50 205.61 7.86 6.68 14.79 16.22 21 B & C 358.71 53.29 9.90 11.68 67.78 1281.05 22 B & C 461.96 80.17 5.84 13.33 90.74 8.32 From the above analysis, for the joints with a factor of safety less than 4.0 additional brackets of type C will be added. Analysis of components: In this section stress analysis on the frame members and the fasteners holding the components placed on the frame was performed at various load conditions. The factor of safety results are summarized in Table 5. Table 5: Summary of results from stress analysis Component/Assembly G-Load Load effect Weight (lbf) Factor of Safety (FS) Boiling Chamber 9g forward Shearing of screws 408.87 322.57 6g downward Bending of extrusion 272.58 15.75 Camera 9g forward Shearing of screws 153 96.06 6g downward Bending 102 7.00 Power supply 9g forward Shearing of screws 265.5 41.96 6g downward Bending of extrusion 176.22 13.73 163 23 2g lateral Tension in screws 58.74 257.84 Kill switch assembly 9g forward Shearing of screws 89.28 664.32 6g downward Bending 59.52 28.48 CPU 6g downward Bending of extrusion 43.92 10.78 9g forward Bending of extrusion 65.88 249.48 2g lateral Tension in screws 14.64 6035.21 Monitor 9g forward Shearing of screws 63.9 82.85 Boiling Chamber: The aluminum tank used for the boiling experiments is supported on the frame using the bottom polycarbonate sheet used for the secondary containment (26in x 9in). The tank is also fastened on to five t-slot extrusions using twenty ?-20 socket screws through the polycarbonate sheet as shown in the Figure 23. The load acting on the extrusions includes the weight of the tank, fluid contained in the tank, the sensors and the bellow assembly. Figure 23: Location of boiling chamber. Red shaded area shows contact area with the extrusion. The stress and deformation due to 6g forces due to the assembly has been illustrated in Figures 17 and 18. The 6g forced causes a deformation of 0.03 inches in the extrusion 164 24 member leading to a bending stress of 2089.1 psi in the extrusion. Safety factor during bending due to 6g downward force can be evaluated as follows: Factor of safety, FS = Shear force due to 9g forward force (significantly larger than shear due to 2g lateral force): Weight of boiling chamber, W =45.43 lbs (including fluid, polycarbonate plate and bellows) Force due to 9g force on each bolt, F = (45.43 x 9)/20 = 20.44 lbs Bolt diameter = 0.25 in; Bolt cross-sectional area = 0.049 in2 Bolt ultimate shear strength = 135,000 psi Shear stress on each bolt, ? = Force on each bolt/ Bolt area = 417.21 psi Factor of safety, FS = Power Supply: The power supply is supported on horizontal t-slot extrusion members and fastened to two vertical extrusions using four 8-32 screws as shown in Figure 24. The 6g downward force causes bending stress in the extrusion while the 9g forward force acts to shear the four 8- 32 screws supporting the power supply. Forces acting in the lateral direction at 2g will cause tensile forces on the fasteners. For the downward force of 6g acting on the extrusions, The bending stress due to the load, (from ANSYS analysis) The maximum yield strength of the extrusion is 35000 psi. Factor of safety, FS = (35000/ 2376.2) - 1 = 13.73 Shear stress due to 9g forward force: Force due to 9g force on each bolt, F = (29.37 x 9)/4 = 66.08 lbs Bolt diameter = 0.164 in Shear stress on each bolt, ? = Force on each bolt/ Bolt area = 3146.67 psi Figure 24: Red shaded area shows the location of power supply and its contact area with the extrusion Location of screws 165 25 Factor of safety, FS = Axial tensile stress due to 2g lateral force: The power supply is held on to the vertical frame using four 8-32 alloy steel fasteners on the front face of the power supply. The lateral axial force of 2g induces a tensile stress in the fasteners. Weight of the power supply, W = 29.37 lbs Force due to 2g force on each bolt, F = (29.37 x 2)/4 = 14.69 lbs Bolt diameter = 0.164 in Bolt ultimate tensile strength = 180,000 psi Axial stress on each bolt, ? = Force on each bolt/ Bolt area = 695.42 psi Factor of safety, FS = CPU assembly: The CPU is supported on three t-slot extrusion members as shown in Figure 25. Downward force of 6g induces bending stress in the extrusion members. The CPU is restrained against the 2g lateral force by two L-joining plates with five ?-20 bolts each. In the forward direction, the 9g forces induce a bending stress in the two t-slot extrusions restraining the movement of CPU in that direction. Figure 25: Red shaded area shows the location of the CPU and its contact area with the extrusion 166 26 For the download force of 6g: Bending Stress, ? = 2970.4 psi Factor of safety, FS = (35000/ 2970.4)-1 = 10.78 For the lateral force of 2g (tensile stress): Weight of CPU, W = 7.32 x 2 = 14.64 lbs Tensile force acting on each ?-20 fastener, F = W /10 = 1.464 lbs Tensile stress, ? = Force on each bolt/ Bolt area = 1.464/0.049 = 29.82 psi Tensile strength of bolt = 180000 psi Factor of safety, FS = For the forward force of 9g (Bending stress in the extrusion): Weight of CPU, W = 7.32 x 9 = 65.88 lbs Force acting on one extrusion, P = 65.88/2 = 32.94 lbf Length of extrusion, L = 4.5 in Uniformly distributed load, w = 7.32 lb/in The bending stress due to uniformly dist. load can be expressed as, Where, = = 0.0884 in3 Bending Stress, ? = 139.73 psi Factor of safety, FS = (35000/ 139.73)-1 = 249.48 LCD Monitor: The monitor is supported on a 100 x 100mm VESA mount adapter which is fastened to a vertical t-slot extrusion using three ?-20 socket screws as shown in Figure 26. The forward force of 9g induces a shear stress in the screws (larger than stress due to 6g). Weight of the monitor at 9g, W = 63.9 lbs Shear force acting on each screw, Fs = 63.9/3 = 21.3 lbs 167 27 Figure 27: Schematic top view of the monitor assembly Figure 26: Backside of the LCD with the attached VESA adapter and mounting screws Shear stress, ? = Force on each bolt/ Bolt area = 21.3/0.049 = 434.69 psi Tensile force on each bolt, = = 97.55 lbs Tensile stress on each bolt, ? = Ft/A = 1990.0/0.049 = 1990.9 psi Factor of safety for the combined loading of shear and tensile stress FS = ?-20 screws VESA Adapter 168 28 Factor of safety for the bolt pull out The factor of safety for the case of pull out of the bolt (1/4 -20) from the nut can be evaluated as, FS = In the case of t-slot extrusions, a load is fastened on to the t-slot in the frame due to the spring-locking action of the nut against the aluminum slot as the bolt is tightened. In this scenario, since the aluminum has lower yield strength than the steel fasteners, the aluminum slot will fail before the bolt pulls out of the nut. Hence, the bolt pull out strength is assumed to be yield strength of aluminum, 35000 psi. FS = = 16.58 High Speed Camera: The camera and the lens are supported on two t-slot extrusion members. The t-slot extrusions are fixed at both the ends by a T-joining plate using five ?-20 bolts. For the analysis only the load on one of the extrusions will be considered. For the download force of 6g acting on the extrusion: Bending Stress, ? = 4370.0 psi Factor of safety, FS = (35000/ 4370.0)-1 = 7.00 Shear stress due to 9g forward force: The camera is held on to the camera plate using one 3/8-24 alloy steel fastener from the bottom of the camera plate. The maximum stress acting on the fasteners is due to the forward 9g force which induces a shear in the fastener. Hence the analysis is only performed for the worst case scenario of 9g force. Weight of the camera assembly, W = 17 lbs Force on the bolt due to 9g force, F = (17 x 9) = 153 lbs Bolt diameter = 0.375 in Bolt max shear strength = 135,000 psi Shear stress, ? = Force on each bolt/ Bolt area = 153/0.11 = 1390.91 psi Factor of safety, FS = 169 29 Kill switch unit: The kill switch box is supported on t-slot extrusions using four ?-20 socket screws. The forward force of 9g causes a shear force in the screws (significantly greater than 2g lateral force). Weight of kill switch unit, W = 9.92 lbs Shear force acting on each screw, Fs = (9.92 x 9)/4 = 22.32 lbs Shear stress, ? = Force on each bolt/ Bolt area = 22.32/0.11 = 202.91 psi Factor of safety, FS = For the download force of 6g acting on the extrusion: Bending Stress, ? = 3496.2 psi Factor of safety, FS = (35000/ 3496.2)-1 = 9.01 Floor attachment: The frame is attached to the cabin floor using six floor attachment brackets. Since the structure is firmly mounted on the floor, only the forward acting 9g force is considered for the structural analysis which acts at the center of mass of the frame. Total weight of the structure (without camera assembly) at 9g, W = 187.61 x 9 = 1688.49 lb Assuming the load acts equally on all the six brackets, Load on each bracket = 281.42 lb Bolt 3 ? Shear stress Shear stress acting on bolt 3 (3/8?) = 281.42/.11 = 2558.32 psi Factor of safety = (135,000/2558.32) ? 1 = 51.77 Bolt 1, 2 ? Tensile stress Tensile force acting on a selected bolt, n, can be evaluated as, Fnt = (FvRv + FhRh). Vn/(V12+ V22 +...+ Vx2) For bolt 1, F1t = (281.42 x 18.55) x 4.25/(2.752 + 4.252) = 865.81 lbf Tensile stress on bolt 1 = 865.81/ 0.049 = 17669.65 psi 170 30 Tensile factor of safety = (180,000/17669.65) ? 1 = 9.19 Figure 28: Schematic of the floor mounting bracket For bolt 2, F2t = (281.42 x 18.55) x 2.75/(2.752 + 4.252) = 560.23 lbf Tensile stress on bolt 1 = 560.23/ 0.049 = 11433.3 psi Tensile factor of safety = (180,000/11433.3) ? 1 = 14.74 Data Acquisition (DAQ) Module: The National Instruments DAQ module is attached to the extrusion using four ?-20 screws. The forward force of 9g induces a shear stress in the screws (larger than stress due to 6g). Weight of the module at 9g, W = 2.26 x 9 = 20.34 lbs Shear force acting on each bolt, Fs = 63.9/4 = 5.09 lbs Shear stress, ? = Force on each bolt/ Bolt area = 5.09/0.049 = 103.78 psi Factor of safety Bolt 1 ? ?- 20 screw Bolt 2 ? ?- 20 screw Bolt 3 ? AN6 Fh = 230.15 lb V1 = 4.25 in V2 = 2.75 in Pivot point Rh = 18.55 in 171 31 FS = Power Strip: The 9 outlet power strip equipped with circuit breaker is fastened on to the extrusion using two ?-20 screws as shown in Figure #. The forward force of 9g induces a shear stress in the screws (larger than stress due to 6g). Weight of power strip at 9g, W = 1.45 x 9 = 13.05 lbs Shear force acting on each bolt, Fs = 13.05/2 = 6.53 lbs Shear stress, ? = Force on each bolt/ Bolt area = 6.53/0.049 = 133.16 psi Factor of safety FS = Keyboard tray: The keyboard docking tray used to mount the keyboard during the flight is attached to the extrusion using a sliding bracket which is fastened using four ?-20 screws. The forward force of 9g induces a shear stress in the screws (larger than stress due to 6g). Weight of the tray including the bracket at 9g, W = 2.76 x 9 = 24.84 lbs Shear force acting on each bolt, Fs = 24.84/4 = 6.21 lbs Shear stress, ? = Force on each bolt/ Bolt area = 6.21/0.049 = 126.73 psi Factor of safety FS = 172 32 6. ELECTRICAL ANALYSIS/VERIFICATION Figure 29: Electrical circuit diagram for components connected to 115 VAC Figure 30: Electrical circuit diagram for components connected to 230 VAC The experimental frame will be grounded to the aircraft frame using a 10 gage braided grounding strap. All the exposed metal surfaces will be connected to this strap. 173 33 (i) LOAD TABLES Table 6: 115VAC Load Table Power Source Details Electronic Equipment Current Drawn (A) Name : Power Cord A Data Acquisition 1.25 Voltage : 115 VAC, 60 Hz Desktop Computer 1.3 Monitor 0.24 Wire Gauge : 12 High Speed Camera 2.0 Max Outlet Current : 20 A Light 2.1 Transformer ? kill switch 1.29 Total current draw (A) 8.18 Table 7: 220VAC Load Table Power Source Details Electronic Equipment Current Drawn (A) Name : Power Cord B Power Supply 5 Voltage : 230 VAC, 60 Hz Wire Gauge : 12 Max Outlet Current : 30 A Total current draw (A) 5 (ii) STORED ENERGY No stored electrical energy. (iii) ELECTRICAL KILL SWITCH There is an electrical kill switch that will cut the power to all the electrical and electronic equipments. The electrical circuit for the kill switch circuit is shown in Figure 31. The kill switch, when pressed, de-energizes the contactor coil thereby losing contact with the components connected to the contactor. The contactor is energized using a 24 V ac transformer. Two contactors are used ? one rated for 20A for components connected to 110 VAC and the other rated for 30A for the power supply connected to 230VAC. The kill 174 34 switch assembly also has a NEMA 5-20 R socket connecting a power strip, and a NEMA L14 20R socket for the power supply. Two wires from the kill switch terminate in plugs NEMA 5-20P and NEMA L14-20P to connect to the receptacles from the aircraft. Figure 31: Electrical circuit for the emergency kill switch (iv) ELECTRICAL SHUTDOWN PROCEDURE ? Verify that the test sequence has been completed. ? Turn off the camera and the lights. ? Set the voltage and current to 0Volts and 0 Amps respectively. Turn off the power supply. ? Monitor the temperature and pressure in the boiling chamber until they return to ambient conditions ? Turn off the data acquisition system ? Turn off the computer (v) EMERGENCY SHUTDOWN PROCEDURE Push the electrical kill switch button to kill the power supply to all the equipments. To return to normal operating conditions twist and release the emergency stop button and make sure the power supply returns to ?zero? state. (vi) LOSS OF ELECTRICAL POWER In the event of electrical power loss, the experiment will fall to a safe position with the electrical power to the heaters set to zero. Upon the return of electrical power, the experiment will be in an idle state with the power to the heaters off. 175 35 7. PRESSURE VESSEL/SYSTEM The boiling chamber used in the current experimental set up, as mentioned earlier, will be maintained at atmospheric pressure during the experiment. However, loss of flight cabin pressure might lead to a pressure differential of about 12 psia between the pressures inside and outside the tank. To test the structural integrity of the tank and the weld joints to withstand the pressure differential, a pressure test was conducted as shown in Figure 32. Figure 32: Pressure testing of boiling chamber The tank was completely filled with water and air was supplied to the tank from an overhead supply station. A pressure gauge was used to monitor the pressure in the tank. The air supply valve was opened to increase the pressure up to 30 psia and no leaks were detected in the tank. Boiling chamber Air supply station Pressure gauge 176 36 8. LASER CERTIFICATION This experiment does not use a laser system 9. PARABOLA DETAILS AND GROUND CREW ASSITANCE Ground Crew Assistance: None Required In flight crew assistance: None Required Free floating requirements: None 10. INSTITUTIONAL REVIEW BOARD This experiment does not use human, animal or any biological test specimen. 11. HAZARD ANALYSIS Table 8: Hazard analysis prepared in accordance with JPR-1700.1 HAZARD CAUSE EFFECT CONTROLS VERIFICATION Electrical Shock Short circuit in test equipment Burn, shock and/or death All electrical circuits grounded Review of Electrical drawing. Hot surface burns Cartridge heaters operate at more than 60?C Burn Isolated from operators Review of Test Equipment. Weight to be lifted (>40 lbs) Test structure weighs more than 100 lbs Physical injury Handles and casters installed More than 2 people to lift Toxic gas Decomposition of FC72 at temperatures greater than 200?C Blindness. Tissue damage Test section heater not designed to reach such high temperatures Heater surface temperature will be monitored 177 37 HAZARD CAUSE EFFECT CONTROLS VERIFICATION Pressure differential Cabin pressure loss Liquid leak Secondary containment. Chamber has been pressure tested. Monitor chamber pressure 12. TOOL REQUIREMENTS The following tools will be required for ground operations. The tools will be brought by the investigators and no other special tools are required. ? Allen keys ? Phillips and slotted screw driver ? Adjustable crescent wrench ? Needle nose pliers 13. PHOTO REQUIREMENTS No special photo or video supports equipments are required for this experiment. 14. AIRCRAFT LOADING The aluminum structure is equipped with handles and wheels. The wheels will be disassembled before the flight and the assembly will be carried into the aircraft manually. 15. GROUND SUPPORT REQUIREMENTS 220 V and 115 V AC power outlets will be required on site in order to test the experimental setup and degas the fluid prior to the flight experiment. 16. HAZARDOUS MATERIALS The working fluid used in this experiment, FC 72, decomposes into Hydrogen Fluoride and Perfluoroisobutylene (PFIB) at elevated temperatures - greater than 200 ?C. However, in the current experimental setup the heater is not designed to reach such elevated temperatures. The supply current required to reach such high temperatures would cause electromigration in the heater leading to loss of electrical power supply. Also, the heater 178 38 temperature is continuously monitored in the experiment and in the event of any undesired temperature rise the power supply will be turned off manually. 17. MATERIAL SAFETEY DATA SHEET 3M MATERIAL SAFETY DATA SHEET FC-72 FLUORINERT Brand Electronic Liquid 04/26/2005 SECTION 1: PRODUCT AND COMPANY IDENTIFICATION PRODUCT NAME: FC-72 FLUORINERT Brand Electronic Liquid MANUFACTURER: 3M DIVISION: Electronics Markets Materials Division ADDRESS: 3M Center, St. Paul, MN 55144-1000 EMERGENCY PHONE: 1-800-364-3577 or (651) 737-6501 (24 hours) Issue Date: 04/26/2005 Supercedes Date: 03/09/2005 Document Group: 10-3789-4 Product Use: Intended Use: For industrial use only. Not intended for use as a medical device or drug. Specific Use: Testing Fluid or Heat Transfer Fluid for Electronics. SECTION 2: INGREDIENTS Ingredient C.A.S. No. % by Wt PERFLUORO COMPOUNDS, (PRIMARILY COMPOUNDS WITH 6 CARBONS) 86508-42-1 100 SECTION 3: HAZARDS IDENTIFICATION 3.1 EMERGENCY OVERVIEW Specific Physical Form: Liquid Odor, Color, Grade: Colorless, odorless liquid. General Physical Form: Liquid Immediate health, physical, and environmental hazards: None known. 3.2 POTENTIAL HEALTH EFFECTS Eye Contact: Contact with the eyes during product use is not expected to result in significant irritation. Skin Contact: Contact with the skin during product use is not expected to result in significant irritation. Inhalation: If thermal decomposition occurs: May be harmful if inhaled. Ingestion: No health effects are expected. 3.3 POTENTIAL ENVIRONMENTAL EFFECTS This compound is completely fluorinated (perfluorinated), or it contains perfluorinated portions. Perfluoroalkyl groups resist degradation in most natural environments. This low-solubility substance has insignificant toxicity to aquatic organisms (Lowest LL50 or EL50 is >1000 mg/L). LL50 (Lethal Level) and EL50 are similar to LC50 and EC50, but tests the water phase from incompletelymiscible mixtures. Take precautions to prevent direct release of this substance to the environment. ATMOSPHERIC FATE: Perfluoro compounds (PFCs) are photochemically stable and expected to persist in the atmosphere for more than 1000 years. PFCs have high global warming potentials (GWP), exceeding 5000 (100-yr-ITH). The Ozone Depletion Potential (ODP) is Zero. SECTION 4: FIRST AID MEASURES 4.1 FIRST AID PROCEDURES 179 39 The following first aid recommendations are based on an assumption that appropriate personal and industrial hygiene practices are followed. Eye Contact: Flush eyes with large amounts of water. If signs/symptoms persist, get medical attention. Skin Contact: Wash affected area with soap and water. If signs/symptoms develop, get medical attention. Inhalation: If signs/symptoms develop, remove person to fresh air. If signs/symptoms persist, get medical attention. If Swallowed: No need for first aid is anticipated. SECTION 5: FIRE FIGHTING MEASURES 5.1 FLAMMABLE PROPERTIES Autoignition temperature Not Applicable Flash Point Not Applicable Flammable Limits - LEL [Details: Nonflammable] Flammable Limits - UEL [Details: Nonflammable] 5.2 EXTINGUISHING MEDIA Material will not burn. 3M MATERIAL SAFETY DATA SHEET FC-72 FLUORINERT Brand Electronic Liquid 04/26/2005 5.3 PROTECTION OF FIRE FIGHTERS Special Fire Fighting Procedures: Wear full protective equipment (Bunker Gear) and a self-contained breathing apparatus (SCBA). Water may be used to blanket the fire. Exposure to extreme heat can give rise to thermal decomposition. Unusual Fire and Explosion Hazards: No unusual fire or explosion hazards are anticipated. No unusual effects are anticipated during fire extinguishing operations. Avoid breathing the products and substances that may result from the thermal decomposition of the product or the other substances in the fire zone. Keep containers cool with water spray when exposed to fire to avoid rupture. Note: See STABILITY AND REACTIVITY (SECTION 10) for hazardous combustion and thermal decomposition information. SECTION 6: ACCIDENTAL RELEASE MEASURES Accidental Release Measures: Observe precautions from other sections. Call 3M- HELPS line (1-800-364- 3577) for more information on handling and managing the spill. Evacuate unprotected and untrained personnel from hazard area. The spill should be cleaned up by qualified personnel. Ventilate the area with fresh air. Contain spill. Working from around the edges of the spill inward, cover with bentonite, vermiculite, or commercially available inorganic absorbent material. Mix in sufficient absorbent until it appears dry. Collect as much of the spilled material as possible. Clean up residue with an appropriate organic solvent. Read and follow safety precautions on the solvent label and MSDS. Place in a metal container approved for transportation by appropriate authorities. Seal the container. Dispose of collected material as soon as possible. In the event of a release of this material, the user should determine if the release qualifies as reportable according to local, state, and federal regulations. SECTION 7: HANDLING AND STORAGE 7.1 HANDLING Avoid skin contact with hot material. For industrial or professional use only. No smoking: Smoking while using this product can result in contamination of the tobacco and/or smoke and lead to the formation of the hazardous decomposition products mentioned in the Reactivity Data section of this MSDS. Store work clothes separately from other clothing, food and tobacco products. Use general dilution ventilation and/or local exhaust ventilation to control airborne exposures to below Occupational Exposure Limits. If ventilation is not adequate, use respiratory protection equipment. 7.2 STORAGE 180 40 Store away from heat. Keep container tightly closed. Keep container in well-ventilated area. SECTION 8: EXPOSURE CONTROLS/PERSONAL PROTECTION 8.1 ENGINEERING CONTROLS Provide appropriate local exhaust when product is heated. Provide appropriate local exhaust ventilation on open containers. For those situations where the fluid might be exposed to extreme overheating due to misuse or equipment failure, use with appropriate local exhaust ventilation sufficient to maintain levels of thermal decomposition products below their exposure guidelines. 8.2 PERSONAL PROTECTIVE EQUIPMENT (PPE) 8.2.1 Eye/Face Protection Avoid eye contact. The following eye protection(s) are recommended: Safety Glasses with side shields. 3M MATERIAL SAFETY DATA SHEET FC-72 FLUORINERT Brand Electronic Liquid 04/26/2005 8.2.2 Skin Protection Avoid skin contact with hot material. Wear appropriate gloves, such as Nomex, when handling this material to prevent thermal burns. Avoid skin contact. Select and use gloves and/or protective clothing to prevent skin contact based on the results of an exposure assessment. Consult with your glove and/or protective clothing manufacturer for selection of appropriate compatible materials. Gloves made from the following material(s) are recommended: Nitrile Rubber. 8.2.3 Respiratory Protection Under normal use conditions, airborne exposures are not expected to be significant enough to require respiratory protection. Avoid breathing of vapors, mists or spray. Select one of the following NIOSH approved respirators based on airborne concentration of contaminants and in accordance with OSHA regulations: Half facepiece or fullface air-purifying respirator with organic vapor cartridges. Consult the current 3M Respiratory Selection Guide for additional information or call 1-800-243-4630 for 3M technical assistance. If thermal degradation products are expected, use fullface supplied air respirator. 8.2.4 Prevention of Swallowing Do not eat, drink or smoke when using this product. Wash exposed areas thoroughly with soap and water. 8.3 EXPOSURE GUIDELINES None Established SECTION 9: PHYSICAL AND CHEMICAL PROPERTIES Specific Physical Form: Liquid Odor, Color, Grade: Colorless, odorless liquid. General Physical Form: Liquid Autoignition temperature Not Applicable Flash Point Not Applicable Flammable Limits - LEL [Details: Nonflammable] Flammable Limits - UEL [Details: Nonflammable] Boiling point 50 - 60 ?C Density 1.7 g/ml Vapor Density Approximately 11.7 [@ 20 ?C] [Ref Std: AIR=1] Vapor Pressure Approximately 232 mmHg [@ 20 ?C] Specific Gravity Approximately 1.7 [Ref Std: WATER=1] pH Not Applicable Melting point Not Applicable Solubility in Water Nil Evaporation rate > 1 [Ref Std: BUOAC=1] Volatile Organic Compounds [Details: Exempt] 181 41 Percent volatile Approximately 100 % VOC Less H2O & Exempt Solvents [Details: Exempt] Viscosity Approximately 0.42 centistoke [@ 20 ?C] SECTION 10: STABILITY AND REACTIVITY Stability: Stable. Materials and Conditions to Avoid: Finely divided active metals; Alkali and alkaline earth metals; Heat(greater than 200 ?C) Hazardous Polymerization: Hazardous polymerization will not occur. Hazardous Decomposition or By-Products Substance Condition Hydrogen Fluoride At Elevated Temperatures - greater than 200 ?C Perfluoroisobutylene (PFIB) At Elevated Temperatures - greater than 200 ?C Hazardous Decomposition: Hydrogen fluoride has an ACGIH Threshold Limit Value of 3 parts per million (as fluoride) as a Ceiling Limit and an OSHA PEL of 3 ppm of fluoride as an eight hour Time-Weighted Average and 6 ppm of fluoride as a Short Term Exposure Limit. The odor threshold for HF is 0.04 ppm, providing good warning properties for exposure. SECTION 11: TOXICOLOGICAL INFORMATION Product-Based Toxicology Information: A Material Toxicity Summary Sheet (MTSS) has been developed for this product. Please contact the address listed on the first page of this MSDS to obtain a copy of the MTSS for this product. Please contact the address listed on the first page of the MSDS for Toxicological Information on this material and/or its components. SECTION 12: ECOLOGICAL INFORMATION ECOTOXICOLOGICAL INFORMATION Test Organism Test Type Result Fathead Minnow, Pimephales promelas 96 hours Lethal Concentration 50% >1000 mg/l Water flea, Daphnia magna 48 hours Effect Concentration 50% >1500 mg/l CHEMICAL FATE INFORMATION Test Type Result Protocol 20 days Biological Oxygen Demand Nil Chemical Oxygen Demand Nil SECTION 13: DISPOSAL CONSIDERATIONS Waste Disposal Method: Reclaim if feasible. As a disposal alternative, incinerate in an industrial or commercial facility in the presence of a combustible material. Combustion products will include HF. Facility must be capable of handling halogenated materials. To reclaim or return, check product label for contact. EPA Hazardous Waste Number (RCRA): Not regulated Since regulations vary, consult applicable regulations or authorities before disposal. SECTION 14: TRANSPORT INFORMATION ID Number(s): 98-0211-0216-9, 98-0211-0217-7, 98-0211-0267-2, 98-0211-1795-1, 98-0211-4860-0, 98-0211-5506-8, 98- 0211-5533-2, 98-0211-7346-7, 98-0211-7992-8, ZF-0002-0305-7, ZF-0002-0321-4, ZF-0002-0354-5, ZF- 0002-0802-3, ZF-0002-1162-1 Please contact the emergency numbers listed on the first page of the MSDS for Transportation Information for this material. SECTION 15: REGULATORY INFORMATION US FEDERAL REGULATIONS Contact 3M for more information. 182 42 311/312 Hazard Categories: Fire Hazard - No Pressure Hazard - No Reactivity Hazard - No Immediate Hazard - Yes Delayed Hazard - No STATE REGULATIONS Contact 3M for more information. CHEMICAL INVENTORIES The components of this product are in compliance with the chemical notification requirements of TSCA. All applicable chemical ingredients in this material are listed on the European Inventory of Existing Chemical Substances (EINECS), or are exempt polymers whose monomers are listed on EINECS. The components of this product are listed on the Canadian Domestic Substances List. The components of this product are listed on the Australian Inventory of Chemical Substances. The components of this product are listed on Japan's Chemical Substance Control Law List (also known as the Existing and New Chemical Substances List.) Contact 3M for more information. INTERNATIONAL REGULATIONS Contact 3M for more information. This MSDS has been prepared to meet the U.S. OSHA Hazard Communication Standard, 29 CFR 1910.1200. SECTION 16: OTHER INFORMATION NFPA Hazard Classification Health: 3 Flammability: 0 Reactivity: 0 Special Hazards: None National Fire Protection Association (NFPA) hazard ratings are designed for use by emergency response personnel to address the hazards that are presented by short-term, acute exposure to a material under conditions of fire, spill, or similar emergencies. Hazard ratings are primarily based on the inherent physical and toxic properties of the material but also include the toxic properties of combustion or decomposition products that are known to be generated in significant quantities. HMIS Hazard Classification Health: 0 Flammability: 0 Reactivity: 0 Protection: X - See PPE section. Hazardous Material Identification System (HMIS(r)) hazard ratings are designed to inform employees of chemical hazards in the workplace. These ratings are based on the inherent properties of the material under expected conditions of normal use and are not intended for use in emergency situations. HMIS(r) ratings are to be used with a fully implemented HMIS(r) program. HMIS(r) is a registered mark of the National Paint and Coatings Association (NPCA). Revision Changes: Section 16: HMIS hazard classification for health was modified. Section 3: Potential effects from eye contact was modified. Section 3: Potential effects from inhalation information was modified. Section 4: First aid for inhalation - medical assistance - was modified. Section 10: Hazardous decomposition or by-products phrase was added. Section 10: Hazardous decomposition or by-products comment was deleted. DISCLAIMER: The information in this Material Safety Data Sheet (MSDS) is believed to be correct as of the date issued. 3M MAKES NO WARRANTIES, EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE OR COURSE OF PERFORMANCE OR USAGE OF TRADE. User is responsible for determining whether the 3M product is fit for a particular purpose and suitable for user's method of use or application. Given the variety of factors that can affect the use and application of a 3M product, some of which are uniquely within the user's knowledge and control, it is essential that the user evaluate the 3M product to determine whether it is fit for a particular purpose and suitable for user's method of use or application. 3M provides information in electronic form as a service to its customers. Due to the remote possibility that electronic transfer may have resulted in errors, omissions or alterations in this information, 3M makes no representations as to its completeness or accuracy. In addition, information obtained from a database may not be as current as the information in the MSDS available directly from 3M. 183 43 18. EQUIPMENT PROCEDURE DOCUMENTATION (i) EQUIPMENT SHIPMENT The experimental equipments will be transported to the test site by the investigators. (ii) GROUND OPERATIONS A dry run of the experiment will be conducted on ground prior to the flight experiments and the fluid may also have to be degassed. These operations require both 115VAC and 220V AC power outlets. (iii) LOADING/STOWING The experiment support structure is equipped with handles and hence will be used to manually carry the structure into the aircraft. (iv) PREFLIGHT ? Preflight operations include degassing the fluid to remove the dissolved gas content. 115VAC power outlets will be required for this operation. ? Heat the liquid to required temperature ? Charge the tank with fluid; compress the bellows and close the valve atop to maintain 1 atm pressure ? Set up camera and adjust focus and lighting requirements ? Turn on the computer to monitor data (v) IN-FLIGHT CHECKLIST Pre-parabola: ? Set required heat flux ? Monitor fluid temperature ? Check camera focus Parabola: ? Record data at required g-force level ? Trigger high speed camera Post-parabola: ? Change heat flux to test device ? Monitor fluid temperature 184 44 (vi) POST FLIGHT CHECKLIST ? Copy data from the computer hard drive to laptop for further analysis ? Change test section if required for the next experiment ? Charge the tank with more fluid if required (vii) OFFLOADING OF EXPERIMENT ? Handles will be used to manually offload the experimental structure\ (viii) EMERGENCY PROCEDURES ? Power down the power supply unit in case of problems with boiling experiment ? Press the emergency stop button to kill the process and power down all electronic equipments 19. BIBLIOGRAPHY [1] Yuan, H., and Prosperetti, A., 1999, ?The Pumping Effect of growing and Collapsing Bubbles in a Tube?, J.Micromech. Microeng. 9, pp. 402-413. [2] T. Jun and C.J. Kim, ?Micropumping of liquid by directional growth and selective venting of gas bubbles?, Proc. IEEE Solid-State Sensor Actuator Workshop, pp. 144?147, 1996. [3] Linke, H., Aleman, B., Melling, L., Taornima, M., Francis, M., Dow-Hygeland, C.,Narayanan, V., Taylor, R. P., and Stout, A., 2006, ?Self-propelled Liedenfrost Droplets?, Physical Review Letters, Vol. 96, 154502. [4] Nimkar, N.D., Bhavnani, S.H., Jaeger, R.C., 2006, ?Effect of Nucleation Site Spacing on the Pool Boiling Characteristics of a Structured Surface?, International Journal of Heat and Mass Transfer, Vol. 49, pp. 2829-2839. [5] Thiagarajan, P., Kapsenberg, F., Narayanan, V., Bhavnani, S.H., Ellis, C.D., 2011, ?Effect of Nucleation Site Spacing on the Pool Boiling Characteristics of a Structured Surface?, Interpack 2011, Portland, Oregon, 2011. [6] Zhang, H., Mudawar, I., Hasan, M.M., 2005, ?Flow Boiling CHF in Microgravity?, International Journal of Heat and Mass Transfer 48, pp. 3107 ? 3118. [7] Celata, G.P., 2007, ?Flow Boiling Heat Transfer in Microgravity: Recent Results?, Microgravity Sci. Tech. ? 3/4. 185 Appendix G Microgravity Experiments: Sample Calculations for the Bubble Velocity Model This section provides the set of sample calculations to estimate bubble velocity for the 5.3 mm bubble shown in Fig. 4.9, which was experimentally estimated to be sliding at 27.4 mm/s. Fluid properties are estimated at Tsat: = 4:5 10 4Nsm 2; = 0:8273 10 2Nm 1 Properties of liquid lm between the saw tooth and the vapor: Thickness of lm (H) will be assumed to be constant and same over both the long slope and short slope of the saw-teeth. Thickness (H) and length of the lm (L) over the saw teeth were estimated from the bubble image. Llongslope = 0:85mm;Lshortslope = 0:53mm Fig. 4.10 shows bubble velocity values estimated for H values ranging from 1-25 m. Sample calculations provided below are for a H value of 17 m. Vapor bubble properties: r1 = 0:26mm;r2 = 0:35mm;D = 5:3mm Pressure di erence in the liquid lm for the long slope of a single saw tooth is estimated as, P1;l P2;l = 2 1 r1 + 1 r2 = 110:91 Nm2 dP dx longslope = P 1;l P2;l Llongslope = 130486:0 Nm3 The same procedure is repeated for the short slope of the saw-tooth to calculate the pressure di erential. dP dx shortslope = P 1;l P2;l Lshortslope = 209269:0 Nm3 The total force due to pressure di erences in the liquid lm in both short and long slopes of the saw teeth can be calculated using the equation below. For the bubble of diameter, (D=5.3mm), that is under analysis it is approximated to reside over 5 saw-teeth (5 long 186 slopes and 5 short slopes), F P;tot = ( m V H + H 2 @P @x LW cos long slope n V H + H 2 @P @x LW sin short slope ) In the above equation, = 31 ; m= number of long slopes; n = number of short slopes; W Diameter of bubble. The Stokes drag due to inertia on the vapor bubble moving through the liquid is, Fd = 4 RV = 1:5 10 5V The overall force balance can be expressed by, m V H + H 2 @P @x LW cos long slope n V H + H 2 @P @x LW sin short slope 4 RV = 0 Re-arranging the equation, m HLDcos 31 V +mH2 @P@xLDcos 31 longslope n HLDsin 31 V +nH2 @P@xLDsin 31 shortslope 1:5 10 5V = 0 Substituting the properties of uids, liquid lm thickness and length, and the vapor bubble parameters, the above equation can be simpli ed as, 5:11 10 4 1:91 10 4 1:5 10 5 V = 1:3 10 5 + 2:1 10 5 V = 28:06 mm=s 187 Appendix H Design Draft Files for Boiling Chamber Parts and Other Experimental Fixtures 188 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSION S ARE IN INCHES ANGLES ?X. X? 2 PL ?X.X X 3 PL ?X. XXX NAME Nono DATE 04/20/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAM E: Front.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPT ION DATE APPROVE D 11.000 8.000 .500.500 1.500 2.000 2.000 2.000 2.000 1.500 .281 O6.062 O5.5001/4-20 UN C ` .300 x 8 #10-24 UNC ` .500 x 5.250O6.500 45? 45? 4.000 5.500 Boiling C hamber - Front Wa ll 189 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSION S ARE IN INCHES ANGLES ?X. X? 2 PL ?X.X X 3 PL ?X. XXX NAME Nono DATE 04/20/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAM E: Back.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPT ION DATE APPROVE D 11.000 8.000 .500 .500 Boiling C hamber - Back Wal l 7 1.500 2.000 2.000 2.000 2.000 1.500 .250 #10-24 UN C ` .500 x 5 190 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSION S ARE IN INCHES ANGLES ?X. X? 2 PL ?X.X X 3 PL ?X. XXX NAME Nono DATE 04/20/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAM E: Sides.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPT ION DATE APPROVE D 6.000 8.000 .500 .500O4.063 O3. 500 2.000 2.000 1.000 1.000 .250 4.000 3.000 #10-24 UN C ` .500 x 3 1/4-20 UN C ` .300 x 8 45? 45? O4.500 Boiling C hamber - Side Wall s 191 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSION S ARE IN INCHES ANGLES ?X. X? 2 PL ?X.X X 3 PL ?X. XXX NAME Nono DATE 04/20/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAM E: Bottom .dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPT ION DATE APPROVE D 13.000 9.000 .500.500 .500 2. 000 2.000 Boiling C hamber - Base 192 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSION S ARE IN INCHES ANGLES ?X. X? 2 PL ?X.X X 3 PL ?X. XXX NAME Nono DATE 04/20/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAM E: Lid.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPT ION DATE APPROVE D 11.000 7.000 .375 2.000 2.0002.600 2.000 2.000 1.500 2.000 2.000 1.500 7.000 5.000 5.500 3.500 1.500 2.200 1.500 1.500 2.000 1.000 1.5001.000 .196 O.316O.196 1/4-18 NP T ` .375 x 21/8-27 NP T ` .273 x 2 1/2-1 4 NPT ` .375 x 35/16-18 UNC x 4 Boiling Chambe r - Lid #10-32 UN F ` .273 x 1 193 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSION S ARE IN INCHES ANGLES ?X. X? 2 PL ?X.X X 3 PL ?X. XXX NAME Nono DATE 03/06/12 Solid Edge TITLE SIZE DDWG NO REV FILE NAM E: Lid2.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPT ION DATE APPROVE D O.703#4-40 UNC ` .300 x 4 .812 1.265 1.265 .812 .859 .859 1/4-20 UN C ` .300 x 6 O5.500 45? .375 7.000 11.000 Lid - Bellows flange ho les & electrica l connecto r holes 194 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSION S ARE IN INCHES ANGLES ?X. X? 2 PL ?X.X X 3 PL ?X. XXX NAME Nono DATE 04/20/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAM E: Quartz_suppor t_6in.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPT ION DATE APPROVE D O7.000 O5.500 O.261 .250 O6.500 Quartz W indow Retainer RIng 6"- Front Wa ll 195 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSION S ARE IN INCHES ANGLES ?X. X? 2 PL ?X.X X 3 PL ?X. XXX NAME Naveenan DATE 08/05/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAM E: Quartz_suppor t_4in.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPT ION DATE APPROVE D O3.500 O5.000 O.26 1 .250 Quartz W Indow Re tainer Ring 4" - Side Walls O4.500 196 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSION S ARE IN INCHES ANGLES ?X. X? 2 PL ?X.X X 3 PL ?X. XXX NAME Nono DATE 03/06/12 Solid Edge TITLE SIZE DDWG NO REV FILE NAM E: Bellows _flange.d ft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPT ION DATE APPROVE D 1.000 O6.000 O5.000O5.500 R.375 .750.750 45? O.261 ` through x 6 Bellows - Flange 197 Squ are Co pper He ati ng Bl oc k Pol ycar bon ate base Pol ycar bon ate flange Pol ycar bon ate lid Teflon Ja cket Exoloded view of the PIV boiling chamber 198 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSIONS ARE IN INCHES ANGLES ?X.X? 2 PL ?X.XX 3 PL ?X.XXXNAME NonoDATE 04/08/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAME: copper_block.d ft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPTION DATEAPPRO VED .312 .312 1.250 1.000 O.125 .125 .079 .079.039 .053 .156 Copper he ating block .156 Drill size # 60 (1/32) O.040 199 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSIONS ARE IN INCHES ANGLES ?X.X? 2 PL ?X.XX 3 PL ?X.XXXNAME NonoDATE 04/09/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAME: Teflon jacke t with square hole.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPTION DATEAPPRO VED O.750 .312 .312 1.250 .079 .079.039 .375 Drill size # 60 (1/32)O.040 Teflon Jacke t 200 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSIONS ARE IN INCHES ANGLES ?X.X? 2 PL ?X.XX 3 PL ?X.XXXNAME NonoDATE 04/09/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAME: Polycarbonate sleeve.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPTION DATEAPPRO VED 1.750 .500 .313 .199 O.750 O.250 .750 .125 .199THRU x 5 holes fo r socket sc rew size 10-24 .313 Counterbo re for socke t screw siz e 10-24 Polycarbon ate Flange O5.000 72? 201 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSIONS ARE IN INCHES ANGLES ?X.X? 2 PL ?X.XX 3 PL ?X.XXXNAME NonoDATE 04/09/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAME: Polycarbonate base.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPTION DATEAPPRO VED 4.000 .500 4.000 3.000 3.000 2.0791.76 1 .159 O.375 .500.500 .125 `Through Hole Drill Size 49/64 O.766 #10-24 UN C` .350 x 5 Tapped Holes O2.500 72? Polycarbon ate Base 202 DRAWN CHECKED ENG APPR MGR APPR UNLESS OTHER WISE SPECIFIED DIMENSIONS ARE IN INCHES ANGLES ?X.X? 2 PL ?X.XX 3 PL ?X.XXXNAME NonoDATE 04/09/13 Solid Edge TITLE SIZE DDWG NO REV FILE NAME: Polycarbonate Lid.dft SCALE: WEIGHT: SHEET 1 OF 1 REVISION H ISTORY REV DESCRIPTION DATEAPPRO VED .500 2.000 2.000 .375 .375 1.000 1.000 1/8-27 NPT ` .273 x 5 NPT thr eaded throu gh holes fr om the top Polycarbon ate Lid 1.250 203