Space Power Resources, Manufacturing and Development Volume 9 Number 1 1990
SPACE POWER Published under the auspices of the SUNS AT Energy Council EDITOR Andrew Hall Cutler, Space Studies Institute, and the Space Engineering Research Center, University of Arizona ASSOCIATE EDITORS R. A. Binot, European Space Technology Centre Eleanor A. Blakely, Lawrence Berkeley Laboratory Richard Boudreault, Oerlikon Aerospace Lars Broman, Solar Energy Research Centre, Borlange, Sweden William C. Brown, Consultant Gay Canough, Extraterrestrial Materials, Inc. Lucien Deschamps, Electricite de France Ben Finney, University of Hawaii Josef Gitelson, BIOS Project, Krasnoyarsk, USSR Peter E. Glaser, Arthur D. Little, Inc Praveen K. Jain, Canadian Astronautics Ltd Mikhail Marov, Soviet Academy of Sciences Gregg Maryniak, Space Studies Institute Michael Mautner, University of Canterbury, New Zealand Rashmi Mayur, Global Futures Network, Bombay Makoto Nagatomo, Institute of Space and Astronautical Science, Tokyo Mark Nelson, Institute of Ecotechnics John R. Page, University of New South Wales Geoffrey Pardoe, Brunel Science Park Tanya Sienko, Association for Research on Japanese Space Development Gerry Webb, Commercial Space Technologies, UK Ray A. Williamson, Office of Technology Assessment/US Congress Andrew R. Wolff, Master Builders, Cleveland, OH Space Power is an international journal for the presentation, discussion and analysis of advanced concepts, initial treatments and ground-breaking basic research on the technical, economic and societal aspects of large-scale, space-based solar power, space resource utilization, space manufacturing, and other areas related to the development and use of space for the long-term benefit of humanity. Papers should be of general and lasting interest and should be written so as to make them accessible to technically educated professionals who may not have worked in the specific area discussed in the paper. Editorial and opinion pieces of approximately one journal page in length will occasionally be considered if they are well argued and pertinent to the content of the journal. Submissions should represent the original work of the authors and should not have appeared elsewhere in substantially the same form. Proposals for review papers are encouraged and will be considered by the Editor on an individual basis. Editorial Correspondence: Dr Andrew Hall Cutler can be reached by telephone at (602) 322-2997, by Facsimile at (602) 326-0938 and by mail at 4717 East Fort Lowell, Tucson, AZ 85712, USA. Dr Cutler should be consulted to discuss the appropriateness of a given paper or topic for publication in the journal, or to submit papers to it. Questions and suggestions about editorial policy, scope and criteria should initially be directed to him, although they may be passed on to an Associate Editor. Details concerning the preparation and submission of manuscripts can be found on the inside back cover of each issue. Business correspondence, including orders and remittances to subscriptions, advertisements, back numbers and offprints, should be addressed to the publisher: Carfax Publishing Company, P.O. Box 25, Abingdon, Oxfordshire 0X14 3UE, United Kingdom. The journal is published in four issues which constitute one volume. An annual index and titlepage is bound in the December issue. ISSN 0883-6272 © 1990, SUNS AT Energy Council
Volume 9 Number 1 1990 Editorial 2 Irving Weinberg & David J. Brinker. Indium Phosphide Solar Cells—Recent Developments and Estimated Performance in Space 3 G. Y. Eastman, D. M. Ernst, R. M. Shaubach & J. E. Toth. Advanced Heat Pipe Technology for Space Heat Transport and Rejection Technologies 15 Francis Baron, Ralph Philippi & Werner Tillmetz. European Regenerative Fuel Cell Technology for Space Use 27 Gregg E. Maryniak. Nonterrestrial Materials for Space Solar Power Projects 39 James A. Turi & Robert T. Carpenter. Advanced Radioisotope Space Power Systems 49 Barbara I. McKissock & Harvey S. Bloomfield. Space Nuclear Reactor Shields for Manned and Unmanned Applications 57 Jordin Kare. Pulsed Laser Propulsion for Low-cost High-volume Launch to Orbit 67 D. J. McConnell. Safety and Environmental Analyses for Space Nuclear Programs 77 G. B. Sheble’, R. M. Nelms & L. L. Grigsby. Fundamental Concepts for Realtime Spacecraft Power Distribution Control 89
The 1990s hold great potential for progress in space. Space Power will help you keep up with that progress and contribute to it. We certainly hope you will help us out by letting us know what you need from the journal, letting your colleagues know that it is a good place to find thought-provoking articles, and encouraging them to submit papers to it as appropriate—or even submitting the yourself! We have merged with Journal of Lunar Exploration and Development, and Gay Canough, its editor, has joined our editorial board. We have changed our subtitle to reflect this as well as our broadened scope. We are now Space Power—Resources, Manufacturing and Development. The contents of Volume 9 will contain some material that we are still publishing from the Cleveland International Conference on Space Power in June 1989, as well as new submissions. In addition to a wide variety of papers on space power systems and requirements, there will be papers on lunar surface heat rejection, surface navigation on celestial bodies, application of high-temperature superconductors to space power technology, satellite attitude control, satellite interactions with the space plasma environment and papers on other cogent topics. We will be adding emphasis in biospherics and closed environments, and hope to publish some material on radiation biology pertinent to long-term space activities as well. We are working on a special issue in 1992 or 1993 on climate remediation and modification from space (proposed by our new Associate Editor Michael Mautner). While building solar power satellites early enough to avoid intolerable greenhouse warming would be quite sensible, it may not occur. When considered in light of the scale of program necessary to build an SPS, the added effort to block out a little sunlight to compensate for the Greenhouse Effect in manageable—and the added effort to block out ultraviolet of we have not controlled CFC emissions adequately is also within reason. The side effects and implications of this are of great concern and will be explored in future issues. We also expect to publish papers in the areas Space Power traditionally concentrated in during the 1980’s—microwave and other power transmission in space and from space to the ground, solar-power satellite design tradeoffs, the economic, political and social climate for large space projects, the climatic effects of large-scale space activities, biological effects of power beaming, space resource utilization, space commercialization, as well as the societal effects of advanced space activities. We will publish book and article reviews for material we believe it may be worth your while reading, and we will occasionally publish the proceedings of conferences of interest to our readership. We hope that we are running thought-provoking articles. If they do inspire you to do a bit more work on one of the points, or to examine a point the author passed by, please remember that you can submit a comment to the journal. If you have fleshed out an idea but don’t have the time or resources to turn it into a full paper, remember that we also accept notes. Don’t forget to let us know of anyone who needs a bit of encouragement to write and submit that paper we all should read that they haven’t quite gotten around to yet. If you wish to tell us what you like, or what you would like to change about the journal, please feel free to get in touch with Dr Cutler or any of the editorial board. Also feel free to nominate associate editors to Dr Cutler at any time. Editorial
Indium Phosphide Solar Cells—Recent Developments and Estimated Performance in Space IRVING WEINBERG & DAVID J. BRINKER Summary The current status of indium phosphide solar cell research is reviewed. In the US program, mainly under the aegis of the NASA Lewis Research Center, efficiencies of 18.8% were achieved for standard n/p homojunction InP cells while 17% was achieved for ITO/InP cells processed by sputtering n-type indium tin oxide onto p-type indium phosphide. The latter represents a cheaper, simpler processing alternative. Computer modeling calculations indicate that efficiencies of over 21% are feasible. Initial efforts to produce cheaper, lighter weight and stronger cells are focused toward epitaxial deposition of InP on cheaper, more durable, substrates such as Si. InP solar cells on board the LIPS III satellite show no degradation after more than a year in orbit. Calculated array specific powers are used to estimate the relative performance of arrays containing InP, GaAs and Si, in polar and geosynchronous orbits. Relatively large area cells are produced in Japan with a maximum efficiency of 16.6%. Over 1000 of these latter cells have been manufactured to power a small lunar orbiter on board the Japanese MUSES A satellite. Introduction Most of the satellites currently in space receive their electrical power from silicon solar cells. On the other hand, several satellites have been launched using power obtained from gallium arsenide solar cells [1]. Although these latter cells are more expensive and heavier than the silicon type, their demonstrated higher efficiencies and superior radiation resistance makes their use attractive in specific orbits. Most recently, indium phosphide solar cells have emerged as candidates for use in the space radiation environment. This follows from their significantly increased radiation resistance when compared with gallium arsenide and silicon [2, 3]. In addition, InP cells have been observed to anneal at room temperature under dark conditions and under the influence of incident light [3, 4]. Furthermore, computer modeling calculations show that InP solar cells have theoretical efficiencies above that of Si and slightly below that of GaAs [5, 6]. Thus InP solar cells, when fully developed, have a strong potential to outperform the cells currently used in space. For this reason, in 1985, the NASA Lewis Research Center initiated a program aimed at developing high efficiency, radiation resistant indium phosphide solar cells. In the present case, we review recent significant developments in this program including the experience gained with InP Irving Weinberg & David J. Brinker, NASA Lewis Research Center, Cleveland, OH 44135, USA.
cells on board the LIPS III satellite. We also include recent significant developments in other countries, notably in Japan. Background Most of the early research on InP solar cells was concerned with their possible use in the terrestrial environment [7]. Since no radiation damage data emanated from the terrestrial program, there was little or no interest in developing these cells for use in space. However, in 1984, it was demonstrated in Japan that InP solar cells had radiation resistance under 1 MeV electron irradiation which was significantly greater than that exhibited by either Si or GaAs [8, 9]. Subsequent to this, the superior radiation resistance of InP was demonstrated under proton irradiation [10]. In addition, it was oberved that radiation damage in InP could be annealed at low temperature and under the influence of light [3, 8], The major problems encountered at this relatively early stage of development were the low efficiencies obtained and the relatively high cost of the InP wafers. To date, significant progress has been made in achieving relatively high efficiencies and a modest beginning has been made to solve the problem of cost. Cell Research and Development The major emphasis in InP solar cell research since the discovery of its high radiation resistance has been largely concerned with increasing cell efficiency and determining its properties in laboratory radiation environments. Figure 1 shows the progress attained in increasing cell efficiency since 1984. As one can see from the figure, the highest AMO total area efficiency achieved has risen from 13% in 1984 to 18.8% by the end of 1987 [11]. The progress toward higher efficiency has followed from increased experience and refinement of processing techniques. In 1984, the best cells were processed in Japan by closed tube diffusion [12], Over the course of this program, other processes were investigated. These principally included open tube diffusion, liquid phase epitaxy and metal organic chemical vapor deposition (MOCVD) [13, 14, 15]. The geometrical configuration of the best cell, which attained an efficiency of 18.8%, is shown in Fig. 2 [15]. In attaining this efficiency, an addition to the MOCVD process entailed formation of the silicon doped w-region by ion-implantation [15]. Noting that AMO efficiencies of over 21% are predicted by computer modeling calculations [5], the highest efficiency achieved represents an encouraging step towards this goal. A comparison of achieved and predicted efficiencies for InP, GaAs and Si is shown in Fig. 3, where the solid line represents theoretical AMO efficiencies due to an earlier calculation by Loferski [16], From the figure, the predicted maximum efficiencies for GaAs, InP and Si are 23% 22% and 19%, respectively. Both silicon and GaAs efficiencies achieved to date are close to their maxima [17, 18], while for InP there is obviously more room for improvement. It should be noted that there have been approximately 30 years of R&D on silicon solar cells and 19 years on GaAs cells. Since only six years have been expended on InP space solar cell research, it is anticipated that with continued research InP solar cells will achieve efficiencies much closer to the maximum value shown. The two efficiencies shown for silicon require further comment. The highest efficiency shown refers to a cell processed from low resistivity silicon which degrades excessively under irradiation [2]. On the other hand, the lower efficiency shown for silicon refers to efficiencies achievable in production for higher resistivity silicon solar cells. These latter cells exhibit greater inherent radiation
resistance than their lower resistivity-higher efficiency counterpart. The effect of 1 MeV electron irradiation on all three cell types is shown in Fig. 4 where the relatively poor radiation resistance exhibited by low resistivity silicon cells is clearly shown [2]. The figure illustrates the superior radiation resistance of InP solar cells over both GaAs and Si, and, in effect, supplies the motivation for considering InP as a potentially important space solar cell. Another advantage of InP arises from its propensity to anneal under the influence of light [4]. Due to the relatively high cost of InP wafers, all of the cells whose efficiencies are depicted in Fig. 1 have been relatively small, with areas varying from 0.25 to 0.3 cm2.
Obviously, much larger cell areas are needed for inclusion in practical spacecraft arrays. In this regard, the Nippon Mining Corporation in Toda, Japan, has produced InP cells with areas of 2 cm2 and 4 cm2, respectively [19, 20]. These latter cells were processed by closed tube diffusion [12, 19, 20] on a production basis. They are intended to power a small lunar orbiting satellite, carried piggyback on board the Japanese MUSES A satellite [21]. The satellite is scheduled for launch in 1990 [21]. The spacecraft, on which the small InP powered satellite is mounted, will perform periodic lunar swingbys. At the first swingby, the small piggyback lunar orbiter will be injected into a lunar orbit. Power for the lunar orbiter will come from 1000 InP solar cells with areas of 2 cm2. Because the moon has no magnetic field, the InP cells will not be subjected to a severe ambient radiation environment. In fact, radiation from solar flares would appear to be the major predictable cause of cell degradation. Hence, rather than serving as a severe test of InP in a space radiation environment, the mission will serve mainly to space qualify the cells. It should be mentioned that, in a production run of approximately 1300 cells, more than 1000 had efficiencies over 15%, the highest efficiency being 16.6% [20]. However, past experience could lead one to predict that higher production efficiencies would be attained using the MOCVD process [11]. The preceding cells are monolithic n on p homojunctions processed by techniques which require operating temperatures of between 600 and 700°C. On the other hand ITO/InP cells are processed at room temperature [22]. These latter cells consist of a layer of w-type indium tin oxide sputtered onto a /’-type InP substrate. Cells of this type have achieved AMO efficiencies of 17%, with no visible barrier toward achievement of higher efficiencies [23], Since we were unable to find any published data concerning their behavior under irradiation, we have irradiated several ITO/InP cells
produced by SERI, and observed their performance under 10 MeV proton irradiation. The principal results of such irradiations are shown in Fig. 5 [24], The larger area cells used in these irradiations are identical to those used to power the previously described small lunar orbiter [19, 20, 21]. The fact that the ITO/InP cells exhibit radiation resistance comparable to the n/p homojunction cells is encouraging and serves as motivation for additional research efforts. With the exception of the Japanese large area cells, the InP cells described here are produced in small lots in a research environment. When placed in production, cell
efficiencies depend on the production method and the price that the customer is willing to pay for the delivered cells. For one thing, the yield of acceptable cells usually decreases as the efficiencies required by the customer increases. Hence, the efficiencies quoted for production cells are usually less than those achieved in a research environment. Performance in Space 1. The LIPS III Flight Experiment The LIPS III spacecraft was launched, in the spring of 1987, into a nearly circular 60+ degree orbit whose altitude was 1100 km. ‘LIPS’ is an acronym for ‘living plume shield’. The main purpose of the shield lies in protecting the satellite’s primary payload from rocket engine plumes. Since one surface of the plume shield is unaffected by the plumes, a variety of solar cell experiments were mounted on the unexposed surface. Participants include groups from the USA, England, France and West Germany. Further details are contained in publications by the Naval Research Laboratory which is responsible for the satellite [25, 26]. The planned mission lifetime is three years; however, it is hoped that this can be extended to five years. There are two types of InP cells on board the satellite. One experiment, under the aegis of NASA Lewis, contains a four-cell module composed of monolithic n/p InP solar cells. The second experiment, by the Royal Aircraft Establishment, Farnborough, Hants, UK, includes ITO/ InP cells produced by the research group at Newcastle upon Tyne Polytechnic. At present, we have no details from the RAE flight experiment. Hence, we confine ourselves to discussion of the NASA Lewis experiment. The n/p homojunction cells used in our experiment were fabricated at the Renselaer Polytechnic Institute using the open tube diffusion process [13]. Fifteen cells were received, on extremely short notice, with AMO efficiencies ranging from 11.4 to 14.3%. The two highest efficiency cells were retained as standards while two four-cell modules were assembled by Spectrolab [27]. One module was placed on board the satellite, the other being retained as backup. A photograph of the completed four-cell module is shown in Fig. 6. A platinum resistance thermometer was attached to a fifth cell which was used solely for temperature sensing. During more than one year in orbit, module temperatures varied between 1° and 34°C. No 1 MeV damage equivalence data exists for InP. However, a rough estimate concerning the effects of the space radiation environment can be obtained by using 1 MeV damage equivalents for Si [28]. Noting that the present LIPS InP cells are covered with 12 mils of CMX microsheet cover glass, the 1 MeV electron damage equivalent fluence for silicon is 3.5 X 1013/cm2 per year for the LIPS orbit [28], Further details of the Lewis module are contained in Ref. 27. Summaries of flight data obtained after more than one year in orbit are shown in Figs 7 and 8. The plot of short circuit current versus time in orbit (Fig. 7) shows essentially no degradation in this cell parameter after 370 days. The voltage-current data shown in Fig. 8 is more complex. It is noted that cell maximum power on this curve occurs at approximately 0.7 V. Examination of the curves indicates no degradation in Pmm when compared to pre-flight simulator measurements at Lewis. However, the currents at voltages below the maximum power point are uniformly low with respect to the pre-flight simulation. In addition, significant variation in the data occurs near the maximum power point. Further analysis is required to determine if the variation is influenced by the data acquisition system or is inherent in the cells.
2. Estimated Performance in Space A comparison of the estimated specific power of photovoltaic arrays containing InP, GaAs and Si is shown in Figs 9 and 10. The calculations were performed using silicon 1 MeV electron damage equivalent data [28], This is, admittedly, a rough approximation for GaAs and InP. However, it is felt that the use of silicon data tends to overestimate the degradation in the present III-V solar cells. Specific powers were calculated using published data for the JPL/TRW advanced photovoltaic solar array (APSA) [29]. The 5.3 kW wing developed under the APSA program achieved a BOL
specific power, excluding storage, of approximately 132 W/kg using 2.2 mil 13.5% silicon solar cells, a 2 mil cover glass and a 10% weight add-on for contingencies. The APSA wing is an ultra-lightweight flexible foldout blanket with a weight optimized deployment mechanism consisting of a fiberglass trilongeron lattice mast deployed from a cylindrical aluminum cannister and deployment actuator [29]. In the present case, specific powers were computed using the relation where ps is the array-specific power in W/kg, is the cell efficiency, I is AMO solar
intensity in W/m2, Afsa is array-specific mass in kg/m2 and D is a derating factor which accounts for the cell packing factor, space occupied by the wiring harness, diode losses, etc. A value of 0.7 was used for the latter quantity. Thicknesses used in our calculations were 10 mils for the cover glass, 2.2 mils for Si, and 3 mils for both InP and GaAs. For cell efficiencies at 25°C we used 15% for Si, 18% for InP and 19% for GaAs. These values are our best estimates of efficiencies eventually achievable in production for both GaAs and InP. Efficiencies at 60°C were obtained using the temperature coefficients —4.6X 10-2, 3.2 X 10-2 and 6.6 X 10-2%/°C for InP, GaAs and Si, respectively. The effects of storage were included by using a battery-specific energy of 100 Wh/kg and a nominal half-hour eclipse time. Such specific energies are deemed achievable using advanced sodium-sulfur batteries. In making our calculations for a geosynchronous and polar orbit, the 1 MeV performance data for both InP and GaAs were obtained from Fig. 4, while the data for silicon was obtained from reference 28. In each case, it is noted that the BOL specific power of the system containing GaAs is greater than that of InP, but that the specific power of InP soon becomes greater than that of GaAs. In addition, the specific powers of both these latter cells is significantly greater than that of Si over the entire duration of the times shown in the figures. Conclusion The results of the preceding sections illustrates that, with further development, InP solar cells have the potential to outperform both Si and GaAs in specific space radiation environments. To attain this goal, increased efficiencies are needed for InP. In addition, larger area-lower cost cells must be produced in quantity. Considering the data of Figs 1 and 3, the prospects for attaining higher efficiencies appear promising. The major cost in cell processing lies in the InP wafer used to process the cell. For example, 2 in. diameter silicon wafers can be obtained for under US$10. On the other hand, the present cost of InP wafers ranges between US$200 and US$350, depending on customer specifications. These costs should decrease when a considerable number of wafers are required for a single order. This would be the case when the cells move from research into production. Another approach lies in processing cells from extremely thin layers of InP epitaxially grown on silicon wafers. However, lattice and expansion coefficient mismatches between these two semiconductors present severe problems. Despite this, modest efforts are under way, both in the USA and Japan [30, 31], Initial experiments have produced, as expected, low efficiency cells [30, 31]. At present, it appears that transition layers are required between the InP and Si. In addition, specialized annealing techniques are expected to yield greatly increased efficiencies in these heteroepitaxial cells. At present, however, it is premature to judge the prospects for success in fabricating reasonably efficient cells from InP epitaxially deposited on silicon. However, since the long term result would be a reasonably high- efficiency, cheaper and mechanically sturdier cell, research along these lines appears to be well worth the effort. The Japanese MUSES A satellite was successfully launched in late January 1990. With reference to Fig. 3; a silicon solar cell with AMO efficiency of 20.2% (measured at NASA Lewis) was recently produced at the University of New South Wales, Kensington, Australia [32].
REFERENCES [1] Yoshida, M. (1988) Mitsubishi Electric Corp., Itama City, Japan, private communication. [2] Weinberg, I., Swartz, C.K., Hart, R.E. Jr & Staebler, R.L. (1987) Proceedings of the 19th IEEE Photovoltaic Specialists Conference, p. 548 (New York, IEEE). [3] Yamaguchi, M., Itoh, Y. & Ando, K. (1984) Applied Physics Letters, 45, p. 206. [4] Ando, K. & Yamaguchi, M. (1985) Applied Physics Letters, 47, p. 846. [5] Goradia, C., Geier, J.V. & Weinberg, I. (1987a) Solar Cells, 25, p. 235. [6] Goradia, C., Geier, J.V. & Weinberg, I. (1987b) Proceedings of the 3rd Photovoltaic Science and Engineering Conference, p. 207 (Tokyo, Society of Applied Physics). [7] Weinberg, I. & Brinker, D.J. (1980) Proceedings of the 21st Intersociety Energy Conversion and Engineering Conference, p. 1431. [8] Yamaguchi, M., Uemura, C. & Yamamoto, A. (1984) Journal of Applied Physics, 55, p. 1429. [9] Yamaguchi, M., Uemura, C., Yamamoto, A. & Shibukawa, A. (1984) Japanese Journal of Applied Physics, 23, p. 302. [10] Weinberg, I., Swartz, C.K. & Hart, R.E., Jr (1985) Proceedings of the 18th IEEE Photovoltaic Specialists Conference, p. 1722. [11] Weinberg, I. & Brinker, D.J. (1988) Proceedings of the 23rd IECEC, Vol. II, p. 121 (New York, NY, American Institute of Aeronautics and Astronautics). [12] Weinberg, I., Swartz, C.K., Hart, R.E. Jr., Ghandhi, S.K., Borrego, J.M., Parat, K.K. & Yamaguchi, M. (1987) Solar Cells, 22, p. 213. [13] Parat, K.K., Bothra, S., Borrego, J.M. & Ghandhi, S.K. (1987) Solid State Electronics, 30, p. 283. [14] Choi, K.Y. & Shen, C.C. (1988) Journal of Applied Physics, 63, p. 1198. [15] Keavney, C.J. & Spitzer, M.B. (1988) Applied Physics Letters, 52, p. 1439. [16] Loferski, J. (1972) An introduction to the physics of solar cells, in: Solar Cells, Outlook for Improved Efficiency, p. 25 (Washington, DC, National Academy of Science). [17] Ogasawara, N., Ochi, S., Hayafuji, N., Kato, M., Mitsui, K., Yamanaka, K. & Murotani, T. (1987) Proceedings of the 3rd International Photovoltaic Science and Engineering Conference, Tokyo, p. 477. [18] Green, M., Blakers, A.W., Wenham, S.R., Narayan, S., Willison, M.R., Taouk, M. & Spitalek, T. (1987) Proceedings of the 18th IEEE Photovoltaic Specialists Conference, p. 39. [19] Okazaki, H., Takamoto, T., Takamura, H., Kamei, T., Ura, M., Yamamoto, A. & Yamaguchi, M. (1987) Proceedings of the 3rd Photovoltaic Science and Engineering Conference, Tokyo, p. 791. [20] Okazaki, H., Takamoto, T., Takamura, H., Kamei, T., Ura, M., Yamamoto, A. & Yamaguchi, M. (1988) Proceedings of the 20th IEEE Photovoltaic Specialists Conference, p. 886. [21] Nagatomo, M. (1988) Institute of Space and Astronautical Science, Tokyo, private communication. [22] Coutts, J. & Naseem, S. (1985). Applied Physics Letters, 46, p. 164. [23] Coutts, T.J., Wanlass, M.W., Li, X., Gessert, T.A. & Weinberg, I. (1989) Proceedings of the 4th International Photovoltaic Science and Engineering Conference, Sydney, Australia, March, Vol. 1, p. 377 (Edgecliff, NSW, Australia, Institution of Radio and Electronics Engineers).
[24] Weinberg, L, Swartz, C.K., Hart, R.E. Jr & Coutts, T.J. (1988) Proceedings of the 20th IEEE Photovoltaic Specialists Conference, p. 893. [25] Severns, J., Hobbs, R.M., Elliot, N.P., Towsley, R.H. & Virshup, G.F. (1988) Proceedings of the 9th Space Research and Technology Conference, NASA Lewis Research Center, p. 331. [26] Severns, J.G., Hobbs, R.M., Elliot, N.P., Towsley, R.H., Conway, R.W. & Virshup, G.F. (1988) Proceedings of the 20th Photovoltaic Specialists Conference, p. 801. [27] Brinker, D.J., Hart, R.E. Jr, Weinberg, I. & Smith, B.J. (1988) Proceedings of the 20th IEEE Photovoltaic Specialists Conference, p. 819. [28] Tada, H.Y., Carter, J.R. Jr, Anspaugh, B.E. & Downing, R.G. (1982) Solar Cell Radiation Handbook, 3rd edn (JPL Publication 82-69). [29] Kurland, R.M. & Stella, P. (1988) Proceedings of the 9th Space Photovoltaic Research and Technology Conference, NASA Lewis Research Center, p. 128. [30] Yamaguchi, M., Yamamoto, A., Uchida, N. & Uemura, C. (1986) Solar Cells, 19, p. 85. [31] Keavney, C.J., Vernon, S.M., Haven, V.E. & Al-Jassim, M.M. (1989) Proceedings of the 1st International Conference on InP and Related Materials for Electronic and Optical Devices, p. 434. (Bellingham, WA, International Society for Optical Engineering). [32] Blakers, A.W., Wang, A., Milne, A.M., Zhao J. & Green, M.A. (1989) Applied Physics Letters, 55, p. 1363.
Advanced Heat Pipe Technology for Space Heat Transport and Rejection Technologies G. Y. EASTMAN, D. M. ERNST, R. M. SHAUBACH & J. E. TOTH Heat transfer considerations have a major impact on the viability and practicality of spacecraft and future space missions. Heat transport subsystems will carry primary heat from a heat source (nuclear or solar) to an electrical energy converter, unconverted heat to a radiator, and accomplish the radiant loss of this heat to space. Similarly, as the electrical power from the converter is changed back to heat in the power consuming systems, this low temperature heat must be transported and radiated. In systems envisioned to date, the mass of the heat transport subsystem has ranged from 25% to more than 50% of the power generation system mass. For example, in SP-100 thermoelectric power system currently in ground demonstration, approximately 1350 kg of the total system mass of 4580 kg is attributed to the heat transport and rejection subsystem. A failure of the heat transport subsystem either constitutes, or results in, a complete spacecraft powerplant failure. In short, the heat transport subsystem is one of several critical components that will determine the efficiency, life, reliability and success of a space power system. We believe that heat pipes will withstand their attack by the naysayers, promoting liquid droplet radiators, rotating belt radiators, balloons and other heat transport and radiator concepts and will increasingly be employed in future spacecraft to accomplish heat transport and heat rejection with low mass and low delta-T (high effectiveness). Furthermore, the state-of-the-art of heat pipes is in a period of rapid change, so that it is important to impart knowledge to a broad audience beyond the narrow community of heat pipe developers. The purpose of this paper is to present some of these recent developments and comment on their potential for future applications. High Evaporator Power Density An improved understanding of liquid and vapor flow within heat pipe wick structures, especially nonhomogeneous ones, has led to the demonstration of evaporative power density capability of tens of kilowatts per square centimeter with liquid metals as working fluids, and tens to hundreds of watts per square centimeter with less potent, lower temperature fluids. In general, the new trends of performance are from two to a hundred times those previously demonstrated and accepted. For example, the tungG. Y. Eastman, D. M. Ernst, R. M. Shaubach & J. E. Toth, Thermacore, Inc., 780 Eden Road, Lancaster, PA 17601, USA.
sten-lithium heat pipe seen in Fig. 1. is shown operating at a working fluid temperature of 1200°C under a heat load simulating that which would be encountered by the National Aerospace Plane (NASP) engine cowl leading edge at Mach 20 [1]. The heat fluxes achieved during this and preliminary tests were about 100 times that previously attained with heat pipe technology [2].
Having said, and shown pictorially, that there is a way to accommodate two-phase flow in wick structures, it is necessary to explain the new understanding. As we all know, radial heat flow into the heat pipe evaporator can be limiting. Every vapor heat transfer device has a maximum input power density beyond which vapor generation rates become dominant and seriously reduce liquid flow into the input area. It was once thought that the onset of nucleate boiling in a wick was the point at which vaporblocking of the liquid flow passage became limiting. This is true in most conventional wick structures, especially the homogeneous ones. However, it is not the case with nonhomogeneous wicks such as sintered powdered metals and metal felts. In these structures, there is a range of pore sizes. At some heat flux, vapor generation commences within the wick. The vapor replaces liquid within the wick, starting with the largest pores. Because the capillary pressure is greatest in the smallest pores, these give up their liquid last. Therefore, in a sintered powdered metal wick with a range of particle (and pore) sizes, it is possible to accommodate simultaneous liquid and vapor flow. Vapor generation within the wick is no longer as limiting as it once was. Accordingly, it now lies in the hands of the theorists to include this wick vapor pressure drop term in the treatment of the pressure balance within the heat pipe. A model of this wick vapor pressure drop term has been generated and is being correlated with the experimental data. These new advances will have two principal effects on system designers. First, the emphasis in thermal system design will tend to be away from the accommodation of heat pipe power-input limits and toward geometries, materials and interfaces that can handle the new power densities without excessive temperature losses and stresses. Second, for heat pipe radiators fed by pumped loops, the loop-to-heat pipe heat exchanger introduces significant mass due to the large heat transfer area dictated by heat pipe power-input limits. By using the new capability to increase the power density in these exchangers, important size and mass reductions may now be possible. This potential mass reduction can be seen by looking at the SP-100 and other advanced radiator systems. Heat to be radiated must be conducted across the contact interface between the heat exchanger loop and the heat pipe evaporators. The current SP-100 radiator heat pipes experience boiling limitations when this interface heat flux exceeds 30 W/cm2. As a result, the heat exchange loop ducts must be sized to accommodate this heat pipe performance limit. In the SP-100 design, over 7 m2 of heat exchanger surface area, along with the associated armor, account for approximately 30% of the total radiator subsystem mass. By increasing the heat flux capabilities of the radiator heat pipes, the required size and mass of these loops can be reduced. This trend is illustrated graphically in Fig. 2(a). Any increase in contact interface heat flux results in a greater temperature drop from the heat exchanger loop to the radiator heat pipes. For example, the SP-100 heat exchange loop operates at 820 K. At a heat flux of 30 W/cm2, the temperature drop is 20 K. This results in the radiator heat pipes operating at 800 K. If the heat flux was doubled to 60 W/cm2, the temperature drop would double to 40 K. With the heat exchange loop operating at 820 K, the radiator heat pipes would then operate at 780 K. The result is a decrease in radiator surface operating temperature. As the radiator surface temperature decreases, a larger surface area (leading to greater mass) is required to radiate equivalent power. The effect of the increased interface heat flux is increased radiator panel mass. This trend is illustrated graphically in Fig. 2(b). The optimum interface heat flux is found by superposing these two effects. Figure 2(c) shows graphically this optimum heat flux at 60 W/cm2. By removing the heat flux
limitation design constraints, system designers can optimize for lower heat rejection radiator subsystem mass. Another example for the use of the high heat flux capability of the heat pipe is in electronics cooling. High heat flux evaporators of the new breed of heat pipes are capable of handling the power densities occurring on semiconductor chips. This, coupled with the ability to fabricate heat pipe envelope and wick structures from silicon and silicon powder, respectively, presents the possibility of direct low mass, low temperature-loss cooling of power electronics with a potentially important gain in reliability as well. Silicon heat pipes have been tested to evaporator heat fluxes of in the tens of watts per square centimeter range using methanol and ammonia as working fluids at 20°C. As a general conclusion, recent advances in heat pipes are giving spacecraft designers a much broader spectrum of options. High Axial Heat Transport Rates One immediate application of the new-found increased evaporator heat flux capability can be realized within the heat pipe itself to make improvements in the axial heat transport capability. The new understanding and acceptance of stable two-phase flow within wick structures leads one to want to use and take advantage of this knowledge. The higher local power densities now achievable also imply a requirement for higher liquid flow-rates to feed these thirsty evaporators. Beginning with open grooves, various low-drag liquid condensate return passages have been devised. When covered to prevent adverse vapor shear effects and increase performance, they are usually referred to as arteries or tunnels. When arteries are allowed to penetrate into the evaporator of a heat pipe and vapor generation begins within the wick, the arteries are prone to vapor penetration and become vapor locked, effectively shutting down heat pipe operation. Many practitioners have abandoned arteries for this reason. However, recent work has shown that there exist several techniques for minimizing vapor effects in arteries. First, the arteries are located out of the immediate heat flow zone. Second, the artery wall is formed from material with smaller pores than the balance of the wick [3]. Thus, vapor tends to flow around the artery in the larger pores, rather than penetrate it. Third, the artery can be lined with a smooth surface with small nucleation sites requiring a large superheat to initiate boiling [4]. Fourth, boiling in the artery can be suppressed by subcooling of the returning condensate. Subcooling will also condense vapor which may enter the artery anywhere along its length when an isolated ‘pore’ is overstressed and breached. This subcooling can be achieved passively and simply by designing the heat pipe with the artery located in an area where it can reject heat by radiation, or, in higher performance applications, the artery may be coupled to a small radiator assembly. A small amount of subcooling (by a degree or two) can be sufficient to double power throughput in some cases, depending on the range of pore sizes within the wick structure. Using a combination of these techniques, large increases in power throughput have been achieved. As an example, the data seen in Fig. 3 is from tests performed on a sintered aluminum powder metal wick heat pipe of dimensions 48 ft long by 1.25 in. diameter, using ammonia as the working fluid at 300 K [5]. The lower performance curve was generated without artery subcooling, the higher performance curve was with artery subcooling. The subcooling in this case was done on an ‘artery extension’, which was a 15 cm long fill-tube which extended from the evaporator artery. Accordingly, as
vapor penetrated the artery through a ‘breached’ overstressed large pore, the vapor was swept to the evaporator end and condensed in the subcooled region before large vapor blocking pockets could coalesce. The ultimate limit was reached when a large number of smaller pores became overstressed and breached, and the ensuing vapor coalesced before it could be condensed. The measured non-subcooled horizontal performance was 1800 W and the measured subcooled performance at an adverse tilt of 10 cm was 3200 W. These translate into a power-length product of 1.04 million Win and 1.8 million Win, respectively, both believed to be records for a pipe of these dimensions. A second, and more subtle, improvement in axial heat transport is being realized because nonhomogeneous wicks, with their boiling tolerant capability and small effective capillary radius, can provide the capillary forces necessary to operate heat pipes at their viscous limit over a wide range of temperatures and, therefore, pressures. The viscous limit of a heat pipe is that non-isothermal and normally not desirable condition when the far end of the condenser is ‘cold’. Accordingly, the pressure drop in the vapor is the difference between the saturated vapor pressures of the working fluid at the evaporator and condenser temperatures. For example, with sodium at an evaporator temperature of 700°C and condenser temperature of 400°C this difference is 0.15 bar. Accordingly, the heat pipe wick structure must be capable of supporting this vapor pressure difference plus all of the liquid pressure drops in the wick for a sodium heat pipe to operate with a 700-400°C temperature profile. Conditions such as these are not normally expected, and one tries to design around them. However, recently in making heat pipes for NASA LeRC to provide heat to a RE-1000 Stirling engine, such a temperature profile was encountered in testing the heat pipes in a water cooled gas gap calorimeter which was used to simulate the oscillating flow of helium gas in the Stirling engine heater head. The heat pipes are 16 in. long and have a nominal inside diameter of 0.75 in.. Normal operation is with the evaporator above the condenser.
These pipes were required to operate at two steady state conditions of 2 kW with evaporator temperatures of 600°C and 700°C. Because of size restrictions, the 600°C condition was on the sonic curve and the 700°C condition was pushed into the viscous condition. Accordingly, performance was obtainable only because of the high performance capability of the nonhomogeneous sintered stainless wick structure. To date, these heat pipes are performing as designed in the LeRC test facility [6]. Pumped Two-phase Heat Exchanger Heat pipes are passive devices that can provide for the exchange of heat at relatively high heat flux rates, low delta-T over areas in the tens to hundreds of square centimeters. Pumped convection loops are active systems that can provide for the exchange of heat over larger surface areas at equivalent heat fluxes and delta-Ts but at the expense of electrical pump power, system vibration, higher mass and lower reliability. Properly designed pumped two-phase capillary heat exchangers can offer the advantages of both. Capillary forces are small. If higher heat transfer rates are desired than can be accommodated by the flow rates sustainable with capillary pumps, a capillary evaporator can be aided by external pumping. Such a hybrid system can have appreciable advantages to trade off against the undesirable addition of the pump. The pump returns slightly subcooled condensate from the condenser and feeds a capillary evaporator which assures even distribution of liquid over the heat input surface. By relieving the capillary structure of the need to offset the pressure drop in the liquid return section, more pressure is available to produce local flow for liquid distribution purposes (high power density) or to resist vapor penetration into the wick. Very attractive performance levels have been demonstrated. Because the pump is feeding an evaporative heat transfer system which takes advantage of the latent heat of vaporization, the mass flow is low and, consequently, the pumping power and system vibration are low. Accordingly, the parasitic drain of the pump is low. A general statement on the improvement achievable with pumped capillary evaporators is at best argumentative since, to be correct, one needs to present all of the necessary qualifiers and design parameters along with the data. However, a discussion of what makes for a high Q/A, low delta-T evaporator can be generalized. Capillary evaporators, whether pump augmented or not, will have the highest Q/A at lowest delta-T if the liquid film thickness is held to a minimum. This is well known in heat pipes in the fact that vee grooves generally produce the lowest delta-T at a given Q/A. However it is also known that supplying the grooves with enough liquid to sustain operation is difficult, i.e. bulk liquid distribution. Once the film thickness and Q/A reach the point where boiling, rather than evaporation, is taking place, it has been found that enhanced performance can be obtained and modeled by treating the capillary wick as an extended fin surface [7]. Nucleate boiling takes place on the surface of the wick particles but is entirely driven by capillary forces. Rosenfeld [8] has extended his spreading model [7] to show that heat transfer in wicks in self-pumped and pump-augmented wick structures can behave like extended fin heat transfer surfaces and can be used in series, i.e. a fin on a fin on a fin. Accordingly, high Q/A low delta-T evaporators will be obtained if one can supply the working fluid in multiple closely spaced points to a capillary structure with large surface area that has short heat flow lengths. Tests in the 50-350 W/cm2 range are
currently underway in both capillary and pump augmented evaporators using water, ammonia and methylamine in the 0-100°C temperature range. Start-Up of Liquid Metal Heat Pipes From time to time, undue concern is expressed by those not in daily contact with heat pipes about the start-up ability of long heat pipes with a frozen liquid metal working fluid. It is, indeed, possible to design a test in which such a device cannot start properly. The important point, however, is that it is almost always possible to design a heat pipe which will start and operate properly in a given application. Two conditions usually must be met: the heat pipe must have sufficient fluid inventory to sustain both the start-up and operating conditions, and the capillary pumping capability of the wick must have sufficient pressure head to sustain liquid flow under a full set of start-up conditions. The first condition implies a modest excess of liquid during normal operation. This is because the volumetric thermal expansion of liquid metals is greater than that of likely heat pipe envelope materials. Thus, the fluid volume to saturate the wick during the cold condition of start-up results in an excess at the operating temperature. If the vapor velocity is moderate during normal operation, the excess working fluid will distribute itself uniformly along the heat pipe, and its presence is generally not observable. If the vapor velocity is high, the excess liquid will be swept to the end of the condenser, filling the vapor space. Accordingly, extra space should be allowed beyond the intended condensing area to accommodate the excess working fluid. The need for the second start-up requirement is less obvious. When heat is added to a frozen heat pipe, the working fluid first melts and later begins to vaporize. Since the vapor pressure of the liquid metals (K, Na and Li) is essentially zero at their melting points, their exists a temperature conduction front extending from the evaporator which melts the frozen working fluid at a rate which exceeds the progression of the vapor viscous front down the length of the heat pipe. Obviously, if the heating rate is greater than the time constant for the conduction front progression, liquid return will be diminished and the evaporator can be emptied of liquid working fluid. In addition to the conduction front effect, the heat of vaporization is much larger than the heat of fusion, so that a given mass flow of vapor will thaw a larger mass of solid material, thus assuring an increasing reservoir of liquid. Accordingly, start-up will progress provided the thermal mass of the system and the radiant losses are not an effectively infinite heat sink—a condition which will always prevent any heat pipe from starting up. Note that this is an unlikely situation in a space power system which is designed for low mass. As discussed earlier, the large pressure gradients seen in a heat pipe on start-up and as it goes through the sonic and viscous conditions also must be considered in designing the heat pipe’s wick structure. Whether a given heat pipe will or will not start-up is a function of the size of the pipe, its wick structure and the operating conditions. In general, these factors must be taken into consideration when designing the pipe in the first place. Figure 4 is a graph of data taken on a 5.5 m long X 5 cm diameter semicircular cross-section titanium/potassium radiator heat pipe, developed jointly by Thermacore and Los Alamos National Laboratory for the precursor to the SP-100. These start-up curves were taken over the course of several days, allowing thermal equilibrium to be achieved and stable operation maintained for at least 2 h at each power level. From these data, one can see that, at the lowest listed power data, more than one-half of the
heat pipe was below the melting temperature of potassium (85°C). However, start-up was rational and well-behaved, and was accomplished without any special adjustments. A similar start-up test at full power of 1900 W was performed with the pipe initially at room temperature (frozen K) with no evidence of heat pipe malfunction or evaporator burnout. We have taken a molybdenum-lithium heat pipe from room temperature (frozen lithium) condition to 1900°C in 90 s. The conduction and sonic fronts visibly propagate down the heat pipe until they reach the end of the condenser, where upon the heat pipe temperature rises isothermally to the desired operating condition. New Materials In an effort to reduce the mass of the heat rejection radiator subsystem, alternate materials of construction are constantly being evaluated. Cursory analyses [8, 9] have shown that by simply integrating the new composite materials which are available into existing radiator designs, mass reductions of over 50% are achievable. The unique properties of these new materials also make revolutionary design configurations feasible, in turn bringing order of magnitude mass reductions within reach. Current radiator systems exhibit mass densities on the order of 0.5-1 g/cm2 of radiator area. If the radiator is hardened against potential military threats, or even severe space debris threats, the mass value quickly approaches 2-3 g/cm2. Objectives are to reduce this number to 0.05-0.10 g/cm2 (hardened or unhardened). To meet these demanding objectives, new materials are being investigated. The leading advanced material contender is carbon-carbon. Carbon-carbon offers the potential of (i) exhibiting intrinsic hardness against space debris threats as well as hostile military threats (ii) having very low mass and (iii) being optimized, both thermally and mechanically, through specific orientation of its fibers. In order to enlist this material for use as a heat pipe envelope, a method must be identified to form a
RkJQdWJsaXNoZXIy MTU5NjU0Mg==