Preface With several Space-Shuttle experiments that have been conducted during the past two decades, significant new insights have been gained into various physical phe- nomena under microgravity conditions. There is at present a need to provide direc- tion for microgravity research in which fluid flow and other transport processes form a common basis for many investigations. For example, in biomedical engineering and materials processing, fluid, thermal, and mass-transport aspects have gained pri- macy among various researchers, and the sharing and complementing of expertise have become a necessity for technical progress. In addition, with long-term manned space missions in the near future, technical problems encompassing several of these disciplines have been envisioned. The Banff conference provided a forum for the free flow of ideas under this common theme of transport phenomena and focused on interdisciplinary aspects. One of the main objectives of the conference was to promote the exchange of technical information and ideas among the various scientists and engineers working in microgravity fluid, thermal, biological, and materials sciences. These areas have become of importance to various disciplines within the broader realm of micrograv- ity research. The conference, therefore, addressed the cross-cutting aspects of micro- gravity science and technology and provided a forum for the synthesis of knowledge that has been acquired during the past several years. It was intended to effectively cover the growing interdisciplinary aspects of microgravity science and technology. As a result, many of the participants were able to gain considerable knowledge and insight into new areas, and it is expected that such sharing of expertise will create new frontiers in the emerging science. We were indeed fortunate to have scientific diversity among the participants, and the high level of interdisciplinary activity that emanated from interaction among the participants. By crossing traditional boundaries of scientific expertise, it is our hope that new grounds for collaborative research will continue to emerge, and healthy research programs will develop. The scientific community has indeed recognized that transport phenomena as a science has reached a point whereupon it has a very significant role in many different areas. The conference organizing committee expresses its sincere gratitude to Professor Frank Schmidt who, as the United Engineering Foundation’s technical liaison, gave enormous support. In addition, we appreciate the valuable input from various mem- bers of the scientific committee in putting the conference together, and we are grate- ful to all the participants for sharing their technical expertise. Thanks are also due to NASA and the U.S. National Science Foundation for financial support. On behalf of all the participants, we thank Barbara Hickernell and the rest of the Foundation’s staff in New York who, in spite of the catastrophic tragedy of September 11, 2001, remained resilient in carrying out the final planning for the conference. Finally, we thank the editorial staff of the New York Academy of Sciences for their role in pub- lishing these keynote articles and scientific papers in the Annals of the New York Academy of Sciences in archival form. SATWINDAR SINGH SADHAL (USA), Chair & Scientific Secretary VIJAY K. DHIR (USA), Co-Chair HARUHIKO OHTA (Japan), Co-Chair REGINALD W. SMITH (Canada), Co-Chair JOHANNES STRAUB (Germany), Co-Chair Ann. N.Y. Acad. Sci. 974: xi–xii (2002). ©2002 New York Academy of Sciences. Use of an Electric Field in an Electrostatic Liquid Film Radiator S.G. BANKOFF,a E.M. GRIFFING,a AND R.A. SCHLUTERb aChemical Engineering Department, Northwestern University, Evanston Illinois, USA bPhysics and Astronomy Department, Northwestern University, Evanston Illinois, USA ABSTRACT: Experimental and numerical work was performed to further the understanding of an electrostatic liquid film radiator (ELFR) that was origi- nally proposed by Kim et al.1 The ELFR design utilizes an electric field that exerts a normal force on the interface of a flowing film. The field lowers the pressure under the film in a space radiator and, thereby, prevents leakage through a puncture in the radiator wall. The flowing film is subject to the Tay- lor cone instability, whereby a cone of fluid forms underneath an electrode and sharpens until a jet of fluid is pulled toward the electrode and disintegrates into droplets. The critical potential for the instability is shown to be as much as an order of magnitude higher than that used in previous designs.2 Further- more, leak stoppage experiments indicate that the critical field is adequate to stop leaks in a working radiator. KEYWORDS: electric field; electrostatic liquid film radiator INTRODUCTION Various possible applications have spurred a general interest in electrohydrody- namics. Examples include separation enhancement, electrostatic spraying to pro- duce fine droplets or powders, ink-jet printers, spray painting, electrohydrodynamic pumps, and a variety of MEMS devices. Electrostatic fields produce tensile stresses at fluid interfaces. Typically, these forces are smaller than gravitational forces on Earth. However, in a microgravity environment, electrostatic fields can be quite important. The principal subject of this review is the application of strong electro- static fields to flowing thin liquid films (usually liquid metal), including the stability of the free surface. This is in connection with a very lightweight enclosed radiator concept by the author, termed the electrostatic liquid film radiator (ELFR). An exten- sive review of free-surface, thin-liquid-film dynamics and long-wave instability has been given by Oron, Davis, and Bankoff.3 Comprehensive treatments of electrohy- drodynamics were provided by Melcher4 and by Woodson and Melcher.5 A more recent review of the literature, including details of the leaky dielectric model, is giv- en by Saville.6 Lightweight radiator concepts include droplet,7–10 controlled free-stream,11,12 and falling film.13–19 Generally, however, losses over time, in an unenclosed radiator of hot coolant, by evaporation into free space and/or droplet collision have not been Address for correspondence: S.G. Bankoff, Chemical Engineering Department, Northwest- ern University, Evanston, IL 60208, USA. [email protected] Ann. N.Y. Acad. Sci. 974: 1–9 (2002). ©2002 New York Academy of Sciences. 2 ANNALS NEW YORK ACADEMY OF SCIENCES quantified. Two experimentally observed modes of droplet breakup are (1) the for- mation of conical points (Taylor cones) that eject a fine spray of liquid droplets along field lines, and (2) the separation of one drop into two drops that are temporarily con- nected by a thin thread of fluid. Miksis,20 Sherwood,21 and others calculate a critical value of the dielectric constant that separates the two modes of breakup. When the dielectric constant is greater than the critical value, the charge accumulates on the fluid, and tangential stresses induce circulatory flow. Both authors predict the forma- tion of conical interfaces, but the methods break down due to a singularity when the cone tip is too sharp. More recently several authors have used finite element methods to more accurately calculate droplet break-up and Taylor cone dynamics. Notz and Basaran,22 measured droplet formation from a capillary tube in the presence of an electric field and used a finite element method to calculate droplet breakup. Far less work has been done involving the deformation of a planar fluid interface. An important difference between these problems is the disparity in length scale that is often encountered in planar interface problems. For instance, typical length scales are 15cm, whereas the Taylor cones often form over less than 1cm. This presents an added difficulty in obtaining numerical solutions. Taylor and McEwan23 observed Taylor cones on flat non-moving fluids. They showed that linear stability theory could accurately determine the threshold potential for cones. Mayberry and Donohoe24 measured the wavelength of the instability of liquid gallium, and found it to be consistent with linear stability predictions. Melcher and Smith25 investigated the effect of surface charge on the stability of non-conducting fluids by turning on a dc field, both slowly and quickly. The dc fields had similar effects regardless of how quickly they were administered. ELECTROSTATIC LIQUID FILM RADIATOR DESIGN The EFLR design is a pumped-loop, double-membrane radiator in which a thin coolant film runs along the inside of an outer membrane that radiates heat into space. The inner membrane, which is very thin, has an array of metallized rectangular elec- trodes, any one of which can be charged to a high voltage if a leak is detected by means of a conductive layer on the exterior surface. The electric field pulls the liquid surface away from the puncture and can overcome the combined vapor pressure of the coolant and the hydrostatic head of the thin coolant layer. The voltage is supplied by a small van de Graaf generator in the interior of each module. Because of thin walls (about 100 µm) and a large view factor (0.85–0.95), the ELFR is extremely light in weight. According to preliminary calculations, the spe- cific weight for a lithium-cooled radiator, in kg/kW thermal power, for a rotating disk radiator with inflow temperature of 800K, outflow temperature of 725K, and inlet Reynolds number of 300, is estimated to be 0.094kg/kW without the circulating coolant, and 0.141kg/kW including the coolant. Note that all radiator types require a circulating coolant to remove waste heat from the power cycle. This implies that a 1-MW thermal radiator would weigh only about 150kg, in part because of its excel- lent view factor. Even this number can be substantially reduced if a thinner niobium lining to protect against corrosion by the lithium can be used (possibly 25, rather BANKOFF et al.: ELECTROSTATIC LIQUID FILM RADIATOR 3 than 50 microns). If proven practicable, the ELFR could provide dramatic cost sav- ings for a variety of missions. Several ultralight systems, including droplet, conveyor belt, and thin film radia- tors have been proposed and studied in recent years to accomplish this goal. Tagliafi- co and Fossa suggest liquid sheet or droplet radiators.18,19 They report design weights as low as 1kg/kW at 300–400K. The international space station generates about 75kW of power. More ambitious designs might require one megawatt of power or more, necessitating the rejection of two or more megawatts of waste heat. Practical radiators for this amount of radiation might use temperatures of 700 to 1,000K. We have made two designs to satisfy both power generation and housekeeping operating conditions. A high temperature radi- ator would use a liquid metal such as lithium and operate at about 700K, and a low temperature radiator utilizing either oil, NaK, or gallium would operate at about 300K. Lithium and NaK have both been used in space designs to date. They are very light (Li, about 0.45g/cm3 and NaK, about 0.85g/cm3) and have good heat transfer properties. A carbon coating on the exterior of the radiator would be used to decrease the wettability (and hence the tendency to leak), and at the same time increase the emissivity of the surface. With oil, a Teflon coating could be used to obtain non-wet- ting, but in general a liquid metal coolant is preferred because its high surface ten- sion and conductivity. However, the long-term chemical reactions, and hence wettability, of the exterior surface of the radiator after leakage is arrested from a par- ticular puncture are still unknown. Note that with a liquid metal all punctures are ini- tially non-wetting, and probably remain so, because with a continuous meniscus, the required driving pressures are many orders of magnitude greater than that available. Previous work by Kim, Bankoff, and Miksis13–15 showed fluid film profiles for flow under finite electrodes (see FIGURE 1). They derived lubrication and Kármán- Pohlhausen models to calculate the height profiles and compared the results to a numerical solution of the Navier-Stokes equations as calculated with the SOLA code.26 They solved for the height profile and pressure under the film for conical and cylindrical radiator geometries. Designs were then optimized based on minimum specific weight with constraints on fluid depth and maintaining the ability to gener- ate a negative pressure on the wall at the leak location with an electric field. In the current work, we consider an electrohydrodynamic instability similar to Taylor cones, leak dynamics within the hole, and electrical breakdown. Critical fields for the Taylor cone type instability are obtained experimentally for mercury and oil films. These fields are representative for the liquid metal and hydrocarbon type radiators suggested. A large margin is observed for the liquid metal, which is greatly stabilized by surface tension. However, the critical field for Taylor cones on oil films is comparable to that used in previous designs. Our leak stoppage experiments were very encouraging. For a non-wetting sub- strate, the positive pressure required to start or stop a leak was simply the capillary pressure, 2 σ/r, for small holes. For holes that were significantly larger than the fluid depth, an even greater pressure was required to start a leak. The capillary pressure was very large compared to the hydrostatic head and internal pressure that force leaks. Therefore, choosing a non-wetting substrate would be a great advantage in an ELFR. For a wetting substrate, the pressure required underneath the film was nearly 4 ANNALS NEW YORK ACADEMY OF SCIENCES FIGURE 1. Fluid height profiles for liquid lithium (700K) flowing in the presence of a fixed normal electric field. The field is taken to be Gaussian centered at 0, and the height is calculated using the Reynolds lubrication model. Taken from Kim et al.,13 who provide further details. zero, implying that leaks could be stopped in an organic, as well as metal, based radi- ator with wetting problems. Gas phase breakdown can be avoided by going to lower pressures. On the low pressure (P < 10−1mTorr) side of the Paschen curve, vacuum breakdown takes over as the pressure is reduced. Fields on the order of (200kV/cm) can be exerted in this region, making the Taylor cone threshold (8 to 70kV/cm) the limiting factor on the electric field. From these experiments, we were able to impose additional constraints on the previous design work, specify new design parameters, and calculate new esti- mates of weight/power. PREVIOUS WORK Kim et al.13 numerically investigated the effect of a fixed electric field or pressure profile on a fluid flowing down an inclined plane. Using lubrication and Kármán- Pohlhausen models, as well as a Navier-Stokes solver, they showed steady-states with a depression in the fluid interface underneath the leading edge of the electrode and a peak in the interface underneath the trailing edge. The unsteady solutions exhibited the formation of a traveling wave that washed away downstream. This work neglected surface tension, as well as the coupling of the electric field with the fluid interface, thereby precluding the formation of short wave instabilities. BANKOFF et al.: ELECTROSTATIC LIQUID FILM RADIATOR 5 When fully developed, these are similar to the well-known Taylor cones that have been observed on a horizontal plate and droplets. Infinite height values were observed under the trailing edge of a finite electrode as the field was increased. How- ever, the fields required for this behavior were significantly greater than that measured experimentally. Kim et al.14,15 and Bankoff et al.2 also investigated other geometries that are more applicable to an ELFR. A spinning conical design, in which the coolant flowed from the apex of a cone to the wide end, had a fluid height solution that typically thinned as the fluid moved along the radiator wall. The magnitude of the depth change depended on the cone angle and the rotation rate. A stationary cylindrical design, in which the fluid thickened as it moved tangentially along the radiator wall, was also studied. In both cases, the effect of the field on the interface was qualita- tively the same as that for flow down an inclined plane. CURRENT WORK In the current work, we compared the height and pressure calculations of Kim et al.13 to coupled solutions of the fluid and electric field equations and experimentally demonstrated the feasibility of the ELFR by using the electric field to stop a leak. We used diffusion pump oil to simulate the low temperature design and mercury to sim- ulate the high temperature design. The oil flowed down an inclined plate, passing in front of a glass electrode, which allowed the quantified fluorescence imaging method27 to be used for local film thickness measurements. Pressure and stability measurements were made with ethanol and diffusion pump oil, and stability mea- surements were made with flowing mercury films. The low temperature design could be used for electronics and habitat cooling. The high temperature design for power generation, which incorporates lithium, appears to be quite robust but may present handling problems. Taylor cones pose a potential problem in a radiator. Taylor and McEwan23 showed that the stability of a horizontal layer of fluid under an electrode could be calculated with linear stability theory. When the potential is increased beyond a crit- ical value, the fluid forms waves that develop into a cone and emit a jet of fluid. Each cone shorts the electrode requiring it to be recharged. Fluid would possibly leak dur- ing the recharge time, which is determined by the power (or weight) of the recharg- ing device. The energy produced during the short could cause irreversible damage to the electrode as material evaporates. This was observed in a tin-oxide coated glass electrode in our laboratory. The liquid film could also ionize, enhancing corrosion chemistry or carbonization. Taylor cones were investigated on a horizontal non-moving layer and on an inclined flowing film (see FIGURES 2 and 3). In both cases, we used a finite electrode to determine edge effects. In the horizontal case and small electrode heights, the cones preferentially formed under the corner of the electrode. This edge effect was previously noted by Taylor and McEwan.23 We found that the critical potential for cone formation was not significantly altered and that the stability results can be pre- dicted well using the linear analysis. Non-linear stability checks with the lubrication theory also proved to be effective in predicting cone formation. 6 ANNALS NEW YORK ACADEMY OF SCIENCES FIGURE 2. Experimental values of the critical field are compared to linear stability predictions for a finite rectangular electrode held above a horizontal mercury film. H′ is the height of the electrode above the static film height under the electrode corner. FIGURE 3. Experimental values of the critical field for a flowing mercury film pass- ing in front of a charged electrode. The data series are for various electrode lengths in the direction of flow. The critical field is not strongly dependent on the length above 3cm. The decrease in critical field with increasing Reynolds number (Re) is thought to be due to sur- face waves. BANKOFF et al.: ELECTROSTATIC LIQUID FILM RADIATOR 7 Taylor cones were observed on a vertical oil film when the electric field exceeded a critical value, E ≈ 13kV/cm. Below this value, the height profile showed less than 1% variation along the direction of fluid flow, which was consistent with lubrication theory. Liquid metals have much higher surface tension than oils. Lithium, which was considered in Bankoff et al.,2 has a surface tension of 395dynes/cm at 700K compared to about 30dynes/cm for oil. This allows for a much higher critical poten- tial and a more robust design. A dimensional balance of the surface tension pressure with the electric field pressure shows σ ≈ εε E2/2κ, where σ is the surface tension, 0 E is the electric field, ε is the dielectric constant, ε is the permittivity of free space, 0 and κ is the curvature of the conical interface. Neither theoretical work nor experi- mental data giving Taylor cone thresholds on flowing films exist in the literature. Therefore, we measured Taylor cone thresholds for flowing mercury. Stable films were observed with fields of 55 to 90kV/cm, a factor of two higher than the square root dependence predicted by the dimensional analysis. This is not surprising con- sidering that the Reynolds number (Re) changed by a factor of 103, the Taylor cones observed on the oil film were three-dimensional, and the wave number is likely to vary. The relative stability of the liquid metal film to Taylor cones allows for a large margin of error. The following results support design feasibility. • Fluid film profiles for flow under an electrode were calculated by Kim et al. for a variety of design types and parameters. • Film stability has been demonstrated at sufficient electric fields to stop leaks. In our experiments, vertical flowing mercury films were stable to electric fields of E > 50kV/cm, compared to 8kV/cm, used in previous design work. (Values of the field listed in these works are higher due to an error in units.) • Leaks have been started and stopped, even with the complication of Earth gravity, for wetting and non-wetting substrates. • Electrical breakdown in the vapor phase has been considered and can be avoided without large increases in weight. Before such a radiator is considered safe, further work should be done concerning the nature of the leak holes, long-term chemistry of realistic materials, the effect of reduced gravity, and electrical breakdown in a realistic design. However, it is thought that these issues will affect design details rather than present significant problems. Since liquid lithium is toxic, explosive in contact with water and flammable in contact with air, the recommended experiments should be done in a properly equipped facility with experienced operators. ACKNOWLEDGMENTS This work was supported by the Microgravity Division at NASA with Allen Wilkinson as the program manager. Helpful comments were made by Professor M.J. Miksis. 8 ANNALS NEW YORK ACADEMY OF SCIENCES REFERENCES 1. KIM, H., S.G. BANKOFF & M.J. MIKSIS. 1991. Lightweight space radiator with leakage control by internal electrostatic fields. Proc. Eighth Symp. On Space Nuclear Power Systems, University of New Mexico, Albuquerque, NM. 1280–1285. 2. BANKOFF, S.G., M.J. MIKSIS, H. KIM & R. GWINNER. 1994. Design considerations for the rotating electrostatic liquid-film radiator. Nuc. Eng. Des. 149(1–3): 441–447. 3. ORON, A., S.H. DAVIS & S.G. BANKOFF. 1997. Long-scale evolution of thin liquid films. Rev. Mod. Phys. 69(3): 931–980. 4. MELCHER, J.R. 1981. Continuum Electrohydromechanics. MIT Press. 5. WOODSON, H.H. & J.R. MELCHER. 1968. Electromechanical Dynamics. Wiley, New York. 6. SAVILLE, D.J. 1997. Electrohydrodynamics: the Taylor-Melcher leaky dielectric model. Annu. Rev. Fluid Mech. 29: 27–64. 7. MATTICK, A.T. & A. HERTZBERG. 1982. The liquid droplet radiator—an ultra-light- weight heat rejection system for efficient energy-conversion in space. Acta. Astro- naut. 9(3): 165–172. 8. MATTICK, A.T. & A. HERTZBERG. 1985. Liquid droplet radiator performance studies. Acta. Astronaut. 12(7–8): 591–598. 9. ORME, M. & E.P. MUNTZ. 1987. New technique for producing highly uniform droplet streams over an extended range of disturbance wave-numbers. Rev. Sci. Instrum. 58(2): 279–284. 10. PERSSON, J. 1990. The liquid-droplet radiator—an advanced future heat-rejection sys- tem. ESA J.–EUR Space Agency 14(3): 271–288. 11. MUNTZ, E.P. & M. DIXON. 1986. Applications to space operations of free-flying con- trolled streams of liquids. J. Spacecraft Rockets 23(4): 411–419. 12. MUNTZ, E.P. & M. ORME. 1987. Characteristics, control, and uses of liquid streams in space. AIAA J. 25(5): 746–756. 13. KIM, H., S.G. BANKOFF & M.J. MIKSIS. 1991. The effect of an electrostatic field on film flow down an inclined plane. Phys. Fluids 4(10): 2117–2130. 14. KIM, H., S.G. BANKOFF & M.J. MIKSIS. 1993. Interaction of an electrostatic field with a thin film flow within a rotating conical radiator. J. Propul. Power 9(2): 245–254. 15. KIM, H., S.G. BANKOFF & M.J. MIKSIS. 1994. Cylindrical electrostatic liquid film radi- ator for heat rejection in space. J. Heat Trans. 116: 441–447. 16. GONZALEZ, A. & A. CASTELLANOS. 1996. Nonlinear electrohydrodynamic waves on films falling down an inclined plane. Phys. Rev. E 53(4): 3573–3578. 17. GONZALEZ, A. & A. CASTELLANOS. 1998. Nonlinear electrohydrodynamics of free sur- faces. IEEE T. Dielect. El. In. 5(3): 334–343. 18. TAGLIAFICO, L.A. & M. FOSSA. 1997. Lightweight radiator optimization for heat rejec- tion in space. Heat Mass Transfer 32(4): 239–244. 19. TAGLIAFICO, L.A. & M. FOSSA. 1999. Liquid sheet radiators for space power systems. P.I. Mech. Eng. G. 213(G6): 399–406. 20. MIKSIS, M.J. 1981. Shape of a drop in an electric field. Phys. Fluids 24(11): 1967– 1972. 21. SHERWOOD, J.D. 1988. Breakup of fluid droplets in electric and magnetic fields. J. Fluid Mech. 188: 133–146. 22. NOTZ, P.K. & O.A. BASARAN. 1999. Dynamics of drop formation in an electric field. J. Col. Int. Sci. 313(1): 218–237. 23. TAYLOR, G.I. & A.D. MCEWAN. 1965. The stability of a horizontal fluid interface in a vertical electric field. J. Fluid Mech. 22(1): 1–15. 24. MAYBERRY, C.S. & G.W. DONOHOE. 1995. Measurements of the electrohydrodynamic instability in planar geometry using gallium. J. Appl. Phys. 78(9): 5270–5276. 25. MELCHER, J.R. & C.V. SMITH. 1969. Electrohydrodynamic charge relaxation and inter- facial perpendicular-field instability. Phys. Fluids 12(4): 778–786. 26. HIRT, C.W., B.D. NICHOLS & N.C. ROMERO. 1975. SOLA—a numerical solution algo- rithm for transient fluid flows. Technical Report. Los Alamos Scientific Laboratory of the University of California. Los Alamos.