RASOR ASSOCIATES, INC. S?R lle^^
^
ADVANCED THERMIONIC ENERGY CONVERSION
DOE CONTRACT DE-AC02-76ET11293
JPL CONTRACT 955033
JOINT HIGHLIGHTS AND STATUS REPORT
JULY-SEPTEMBER 1979
-DISCLAIMER .
This book was prepared as an account of work sponsored by an agency of the United States Government Neither the United Stales Government nor any agency thereof nor any of iheir employees makes any warranty express or implied or assumes any legal liability or responsibility for the accuracy rompleleness or usefulness of any information apparatus product or process disclosed or :hat Its use would not infringe privately owned rights Reference herein to any specific product process or service by trade name trademark manufacturer or otherwise does constitute or impiv its endorsement recommendation or favoring by the United States Government or any agency thereof The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or ariy dgency thereof PREPARED BY RASOR ASSOCIATES. INCORPORATED 253 HUMBOLDT COURT SUNNYVALE. CALIFORNIA 9W6
253 HUMBOLDT COURT. SUNNYVALE. CALIFORNIA 94086. (408)734-1622 DISCLAIMER
This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency Thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. DISCLAIMER
Portions of this document may be illegible in electronic image products. Images are produced from the best available original document.
INTRODUCTION AND SUMMARY
The Advanced Thermionic Energy Conversion Program at Rasor Associ ates, Inc. is supported by the U.S. Department of Energy and the National Aeronautics and Space Administration through the Jet Propulsion Laboratory.
The DOE portion of the effort is directed primarily toward terrestrial applications of thermionic energy conversion. It focuses on the develop ment of converters suitable for use with fossil fueled heat sources in power plants. Studies have shown that the successful use of thermionic power modules in coal-fired central station power plants, for example, could improve plant efficiencies from their current average level of 33% to 50% or more. Such an improvement would provide a fuel savings of over 25% while cutting waste heat in half. The objective of the program is to develop such modules in the mid-1980's for commercial applications in the late 1980's and early 1990's.
The NASA program is directed at establishing the technical feasi bility of an advanced, light-weight, long-life thermionic conversion system compatible with a remote nuclear or solar heat source. The principal application foreseen at this time is in nuclear electric propulsion (NEP) missions in the mid-1990's. Both sponsors cooperate in the improvement of converter performance. In addition, the DOE effort includes the evaluations of a preprototype Thermionic Heat Exchanger (THX) designed for central station power plant applications and NASA/JPL is suporting a study of thermionic NEP system performance as a function of converter performance and design.
This report covers progress made during the three month period from July 1, 1979 through September 30, 1979.
Significant accomplishments for the three month period include:
DOE Program: -Successfully operating a thermionic converter using a "cold" in sulator seal (Plexiglass and Viton). -Completed fabrication and testing of SPC-9, a reference planar converter with smooth molybdenum electrodes. -Created a "shooting type analytical ignited mode converter computer model. -Projected the operating conditions needed to achieve advanced con verter performance with a thick cesium oxide collector.
-Invented a cellular ceramic heat exchanger for obtaining high radiant heat flux from a hot gas.
NASA Program: -Achieved over 3100 hours of operation with the cylindrical converter JPL-5 (STR/STR). -Provided guidelines for definition of optimum lead characteristics in the JPL NEP computer program. -Performed a preliminary NEP optimization study which suggests a 400 kWe system with a specific mass of 26 kg/Kwe is possible with present converter performance (Vg =2.0). A collector work func tion reduction to 1.1 eV would yield a 23 kg/kWe specific mass and a further arc drop reduction to 0.1 eV would yield a 16 kg/kWe system. RELEVANT PUBLICATIONS AND PRESENTATIONS IN 1978 AND 1979
"Thermionics and its Application to the SPS" presented at the 3rd Radiation Energy Conversion Conference, 1/78, published in Radiation Energy Conversion in Space, Vol. 61 of Progress in Astronautics and Aeronautics, K. W. Bill man editor
"Thermionic Energy Conversion for Solar Applications" presented at the STTF Users Meeting, 4/78 "Analytical Model for Auxiliary Ion Source Thermionic Converters" presented at the Int. Conference on Plasma Science, 5/78 "Negative Ion Emission from Cesiated Electrodes" presented at the Int. Conference on Plasma Science, 5/78
"Advanced Thermionic Energy Conversion" Biennial Report for Period 9/75 - 9/77, NSR 2-7, COO-2263-7 Sept. 1, 1975 - Sept. 30, 1977 (2 volumes) "A Summary of USSR Thermionic Energy Conversion Activity," N. S. Rasor, presented at the lECEC, 8/78 "Thermionic Power Plant Design Point Selection: The Economic Impact," G. 0. Fitzpatrick and E. J. Britt, presented at the lECEC, 1978 "Analytical Model for the Ignited Mode Thermionic Converter," L. K. Hansen, J. B. McVey, and E. J. Britt, NSR 2-8, COO-2263-8, Aug. 1978
"Negative Ions and the Collector Sheath," L. K. Hansen, NSR 2-9, COO-2263-9, Aug. 1978
"High Current Thermionic Converter Test Facility, and Initial Con verter Test Results," G. 0. Fitzpatrick and L. L. Begg, NSR 2-10, COO-2263-10, Sept. 1978
"Thermionic Converters and Low-Temperature Plasma," edited by L. K. Hansen, DOE-tr-1, Baksht et.al., April 1978 "Analysis of Advanced Thermionic Converters," J. B. McVey, L. K. Hansen, and E. J. Britt, presented at the IEEE Plasma Science Conference, 6/79
"Emission and Space Charge Effects at Cesiated Electrodes," L. K. Hansen and H. Y. Woo, presented at the IEEE Plasma Science Con ference, 6/79 "Low Cost Cylindrical Converters for Measuring Lead Efficiency, G. L. Hatch, L. Nakata, and E. J. Britt, presented at the lECEC, 8/79 "Power Coupling Alternatives for the NEP Thermionic Power System," M. L. Manda, E. J. Britt, G. 0. Fitzpatrick, NSR 7-1, JPL Contract 955121, 12/78
"A Study of Cesium Oxygen Coadsorption on Metallic Substrates," J. L. Desplat, NSR 6-2, NASA CR-152272 5/79. To be published as two papers in Surface Science, Vol. 91, 1980. "Experimental and Analytical Study of an Advanced Mode Thermionic Converter," G. L. Hatch, L. K. Hansen, and E. J. Britt, NSR 5-2, NASA CR-159638, 6/79
"A Clean Source of Metallic Zr for Ultrahigh Vacuum Surface Studies," P. R. Davis and H. R. Poppa, J. of Vac. Sci. and Tech nology, 15, 1771, 1978 "The Adsorption of Zr Onto W(IOO) Surfaces," to be published in Surface Science, Vol. 91, 1980 "A Study of Zirconium-Oxygen Coadsorption on Single Crystal Tungsten Surfaces," Paul R. Davis, Final Report NAS2-9629, 5/78
"ZERO" - The Worlds Largest Thermionic Converter," L. L. Begg and G. 0. Fitzpatrick, presented at the 14th lECEC, 8/79 TABLE OF CONTENTS
Paae
INTRODUCTION AND SUMMARY
RELEVANT PUBLICATIONS AND PRESENTATIONS IN 1978 AND 1979
PART I - DOE PROGRAM TASK 1 - THERMIONIC CONVERTER RESEARCH AND TECHNOLOGY 1 1.1.1 THX Converter Development (L. L. Begg, M. L. Manda, 1 B. Carl smith, M. Smith) 1.1.2 Heat Transfer Technology (L. L. Begg, D. Johnson) 16 1.1.3 System Analysis (G. 0. Fitzpatrick, E. J. Britt) 20 1.2.1 Converter Technology (M. L. Manda) 21 1.3.1 Converter Performance Analysis (E, J. Britt, J. McVey) 37 1.3.2 Plasma Characterization (G. L. Hatch) 49 1.3.3 Electrode Surface Characterization (J.-L. Desplat, 50 L. K. Hansen, H. Woo) 1.3.4 Thermionic Converter Data Review (L. K. Hansen) 70
TASK II - THX POWER MODULE EVALUATION 70 2.1.1 THX Test Module (L. L. Begg, B. Carlsmith) 70 2.1.2 Heat Source (D. C. Johnson, L. L. Begg) 75 2.1.3 Output Power Transfer System (R. S. Dick, E. J. Britt) 79 2.1.4 Heat Transfer System (L. L. Begg) 83 2.2 Conceptual Plant Design (G. 0. Fitzpatrick, 87 E. J. Britt)
PART II - NASA/JPL PROGRAM TASK I - CONVERTER FABRICATION AND TEST (G. L. Hatch, M. L. 96 Manda) TASK 2 - UNINSULATED HEAT PIPE THERMIONIC POWER ANALYSIS 101 (E. J. Britt, M. D. Smith, R. S. Dick)
REFERENCES 110 APPENDIX I: THE IGNITED MODE DIODE ANALYTICAL MODEL "IMD-4" PART I - DOE PROGRAM
TASK 1 - THERMIONIC CONVERTER RESEARCH AND TECHNOLOGY
1.1.1 THX Converter Development (L. L. Begg, M. L. Manda, B. Carlsmith, M. Smith
Introduction: The objective of this task is to develop cost effec tive, reliable, and reproducible converter configurations which can effectively interface with other thermionic power module components. The terrestrial applications for thermionic energy conversion put tight constraints on conversion system costs and operating characteristics, such as part load efficiency, load following capability, start up power requirements, etc. However, constraints which are very important for space power systems, such as system size and weight, are of lesser importance. As a result, new innovations in converter design and opera tion inappropriate for systems designed for use in space--such as the use of high current, all metal seals, distributed leads, and inductive output power coupling—can be used to significantly improve system performance in the new terrestrial applications. The THX power module has been designed to take full advantage of these innovations. Demon stration of their advantages requires detailed analysis and testing of each new design concept to establish both its viability and its impact on the system performance and operation. To accomplish this, a High Current Converter Test Facility has been built to test a series of high current (HC) converters which will sequentially incorporate and demon strate each advance in design. The test last year of HC-1, (ZERO), the first converter in the series, demonstrated the basic feasibility of large converters by stably driving a uniform current exceeding 6500 amperes.^ ' Earlier the feasibility of using inductive power coupling was demonstrated using a "mini-system" which incorporated a small 2 (2) converter (An -10 cm ), operating in air.^ ' The converter provided sufficient power while operating in the flux-reset inductive power coupling mode to drive all the circuitry required for operating in addition to a light bulb load bank. Work during this contract period is concentrating on two attractive approaches for improving THX performance while reducing cost. First, the characteristics of a small-scale, bolt-together "Cold Seal" converter will be determined experimentally. In this concept, a low pressure (~1 Torr) of inert gas is introduced into the interelec- trode space. During operation, the cesium vapor displaces the inert gas to the end of the converter as in a "gas-buffered" heat pipe, establish ing a cesium-free, inert-gas buffer space. Since no cesium is present in this region, it can be below the cesium reservoir temperature, permit ting the use of low-temperature, easily-replaced seals. This, in turn, may permit rapid and inexpensive assembly and disassembly of the conver ter as well as reduced seal costs. Second, the materials and joint designs which will be used in distributed leads will be tested. Based on the result of this work, a second high current converter, HC-2, will be designed, built, and tested to evaluate these concepts at the THX scale. HC-2, will be designed and built to deliver current to an external load, permitting acquisition of current-voltage data and information on dynamic operating characteristics.
Current Effort: Cold Seal Converter (M. L. Manda, M. Smith): The cold seal conver ter task has been divided into three phases. Phase one involves the construction of a converter envelope incorporating a heat pipe section. This will be used to investigate the operational characteristics and stability of cesium-inert gas interfaces. In the second phase, an emitter assembly will be inserted into the heat pipe to demonstrate that polymer 0-rings can be used for power producing thermionic devices. Phase three will further investigate the interactions of the inert gas- cesium interface with the cesium above the inert gas. The design of the cold seal converter is shown in Fig. 1. The fabrication of the conver ter and the gas handling system was completed early in this reporting period. VALVE
THERMOCOUPLE TUBES EMIHER JDQ ca IfTJlfrJIrrJIrnllnollmllcpllcollooJoDll , IITXTII-.Tmondhfrm -xxxs:-: • v' ••'. ••>x: v«;
ALUMINA TUBE — nfrr* i"M'» r^ r\rs riy r ff\ r^ r^f^ i-^r'^v^r^r^'r^'r^'r^ •••I.—. TXT HEATER WIRES COLLECTOR
WICK
PLEXIGLAS SUPPORT WATER LINES PLATE Fig. 1 Cold-Seal Converter Design Fig. 2 shows the critical components. At the bottom is the heat pipe-collector assembly with the Viton O-ring in place. The collector is made of 300 series stainless steel. A smooth molybdenum emitter is used. In the middle is the emitter, welded to a Kovar transition and a stainless steel tube which acts as the emitter lead. The emitter area is 10 cm and the interelectrode space is 20 mils. At the top the coaxial filament used for heating the emitter is shown.
Filament
Mo Emitter Emitter Lead
Plexiglas Emitter plate Collector
O-ring seal
Heat pipe condenser region
Fig. 2 Cold-Seal Converter
The device was first operated without the emitter assembly. A Plexiglas window plate was used in place of the Plexiglas emitter support plate.
-4 Initial operation of the heat pipe was restricted to low tempera tures and low input heat fluxes due to excess cesium. Under certain conditions the cesium flow in the heat pipe produced a visible jet. This occurred under a combination of conditions involving high input heat flux and low argon pressure above the cesium. The jet became visible when the hot cesium flow penetrated the cold argon region, thus exceeding the dew point for the cesium in argon. As the jet was unstable, it usually deposited cesium on the windows and seals. The jet effect could be completely suppressed by increasing the heat rejection in the condenser region of the heat pipe. Approximately 5 gms. of cesium were removed and the device was then operated successfully over a wide range of argon overpressure (-1-100 Torr) and evaporation (collector) wall temperatures (400-1100°K).
After appropriate operating conditions for the heat pipe were determined, a small thermionic diode probe was installed in the heat pipe. This probe can be positioned at any location within the heat pipe to determine the presence or absence of cesium. The diode exhibits vacuum emission characterisitics in the absence of cesium and ignited mode diode characteristics in the presence of cesium. Probe measure ments indicated that the cesium/argon interface was stable and located in the middle of the condenser section of the heat pipe. Following these tests the emitter assembly was installed and conver- 2 ter testing begun. An output of 4 watts (0.4 W/cm ) was achieved with an argon overpressure of 0.4 Torr and an approximate emitter temperature of 1700°K. During operation, however, the emitter was subjected to several sudden temperature excursions causing a leak in the emitter/lead weld joint. The emitter assembly was removed, and the weld repaired. Following reinstallation testing was resumed. The converter was found to respond to overpressure changes as if they were cesium pressure changes, as expected. Testing will continue during the next reporting period.
-5- Distributed Lead (L. L. Begg, B. Carlsmith): During this period a decision was made to emphasize development of the blade-type distributed lead. The typical converter single lead design, the stud-type distributed lead, and the new blade-type lead are shown in Figs. 3 and 4. The primary disadvantage of the single-lead design is the need for a thick and expensive emitter electrode in high-current converters in order to minimize resistive power losses. The distributed lead eliminates this problem by periodically removing current from the emitter and delivering it to a low-cost emitter current bus. The stud-type distributed lead is the most straight-forward way to accomplish this goal, but it is rela tively complex and poorly suited to accommodating large amounts of differential thermal expansion between the emitter and collector. Magnetic field effects near each post can also be troublesome. The blade-type lead reduces the complexity of the design while permitting differential thermal expansion parallel to the blades. In addition it facilitates fabrication of a converter since a collector segment between blades can, in principle, be removed without removing the emitter cur rent leads and emitter bus. Further development of the distributed lead will be reported as part of the following sub-task, "High Current Converters."
High Current Converters: The conceptual design of HC-2 has been completed and is shown in Fig. 5. The design utilizes a solid nickel emitter and segmented nickel collector. Springs press the collectors inward against pads between the emitter and collector. This enables the collector to move with the thermal expansion and creep of the emitter yet maintains a constant interelectrode spacing. A flexible connector conducts current from the collector to a collector bus which in turn carries it to a collector interlink. Blade type distributed leads carry current from the emitter to emitter buses which connect to an emitter interlink. The interlinks are used to deliver current to an external load. Standard, production bellows are used to bridge between the interlinks, sealing the converter. The electrical resistance of the bellows is such that only a small parasitic current loss occurs through the bellows. COLLECTOR EMITTER NETAL INSULATOR SEALS
EMITTER BUSBAR
COLLECTOR BUSBAR SPACER BLOCKS HEAT REJECTION SYSTEM Fig. 3a Conventional Single Emitter Current Lead Design
COLLECTOR EMITTER EMITTER LEADS METAL INSULATOR SEALS
OLLECTOR BUSBAR
EMITTER BUSBAR HEAT REJECTION SPACER BLOCKS SYSTEM Fig. 3b Stud-Type Distributed Emitter Current Lead Design EMITTER BUS
I 00
COLLECTOR EMITTER LEAD EMITTER STEAM PIPES
Fig. 4 New Blade-Type Distributed Lead Design 8"-
EMITTER COLLECTOR INTERLINK
BELLOW: SEAL
EMITTER INTERLINK
HC-2 Fig. 5 Thermionic Converter The design allows the emitter to creep outward under atmospheric pressure to a predetermined point where flexible connectors between the collector and collector bus bottom out. The collector bus is cooled by a water/steam mixture and provides a heat sink, through the flexible connectors, to maintain and regulate collector temperature. A design of the flexible connector between the collector and Collector bus was completed during this period. It consists of sixty pieces of OFHC copper foil, each 0.002" thick, welded to copper blocks. The weld was limited to only that area within the copper blocks leaving the individual sheets unbonded in the distance between blocks. This design provides sufficient cross-sectional area to pass the collector cooling heat flux and yet provide the flexibility needed for outward thermal expansion and creep movement of the collector due to the emitter. The design is shown conceptually in Fig. 6.
It was envisioned that water passing through axial holes in the collector bus cross-section would be used to cool the collector bus and in turn the collector through the flexible connectors. In passing through the collector bus, the water absorbs the heat of vaporization and is converted into steam. The stability of this phenomenon and re sulting collector bus temperature distribution were in question. A basic feasibility test was run using a heated copper block to simulate the collector bus. The copper block was heated to the temperature and heat fluxes to be experienced by the actual collector bus. A cross- section of the block is shown in Fig. 7. Two axial holes were drilled into the copper block. The water would enter through one hole, pass down the length of the block, cross over to the other tube and exit as a water/ steam mixture through the second hole at the top of the block. The test apparatus is pictured in Fig. 8.
-10- EMITTER LEAD
-X>
<1THERKAL EXPANSION & CREEP STAND OFF DISTANCE
Fig. 6 Flexible Converter Concept
•II- TJ/)TtK .^- ^ r-— 7j/>Te»/Sre/>/1 « JVirru/ts Our •
• • J ^ 'Vl'^ If J 4s M •- '-1 i ^ ^--xlS € •• : '" - 5^ •^ A - \ '^ € ., -^ *"\% -/- ^ € " - : ^ ^ r^ V, ; € I) CoiifcroH \'^ ^ *.,"•*'•"•
S ••>'•' SeetnsNT
V:, ; vs "'-v • 3 € ' :«
• >>,: ' \ '^1
' •• P C 1
vv I - 9 (3 " ; ' 5) ® • \ 1 ' 5 ^® . . ^ • 3 HfflTt'i ^/A,. / /^ g ••'' . ' \ 5 e y\, ,^ : , 3 ,1 '*, ''/ K , / '• € ; 1 " / 3 s i ',' ' 3 " •' , e ' \ 1 3
Fig. 7 Collector Cooling Test Apparatus
•12- Fig. 8 Distibuted Lead Test Furnace
Severe axial temperature gradients were initially experienced in the copper test block. It was thought that some water initially boiled downstream of the entrance to the copper block. The resulting pressure rise in the steam would accelerate the steam and entrain the surrounding water. The water would be carried out the exit at such high velocity that its heat of vaporization could not be absorbed from the block. This left the bottom part of the test block running 114°C hotter than the entrance region.
This problem was solved by adding a "bypass" tube within the inlet hole in the copper block. Upon entering the test block, some water would flow inside the bypass tubing while some would flow in the annular space between the tube and block. The water inside the bypass tube was not entrained by the steam forming in the annular space. The bypass
-13- tube would terminate near the bottom thus delivering undisturbed water to this portion of the block. By experimenting with bypass tube lengths, the axial temperature gradient was reduced to 35°C. The test block temperature was controlled by a valve on the water inlet. For the operating temperatures anticipated, the test block temperature could be controlled by increasing or decreasing water flow, as would be expected. If the test block temperature dropped below 250°C, the convective cooling would become unstable and uncontrollably cool the block to ~150°C. Over this range there was no control of temperature or cooling rate. This characteristic is explained by the regions of boiling illustrated in Fig. 9.
• t n — Interloce evoporotion uuuuie* Film- TTT TTT —V-^ vr I a Puft convtclion heot 1 fil m
Uortiftrrci by super- ris e t o ng . n g on d erloc e I / •-/ ticolcd liquid rising .0 O «« •5 S.^ \ . "iB / lolht liquid-»opor f 0 "to _§ / inttrfoce »hfri t»Op- \ 0 ^ 0— / ocrj / '•^ «'' >.« / o'olion lolts ploce. .S.E 0 / — •«« 0 — c 3 / JO — cr S / ** •« ^ z / "5 s. \ c ° 0
'0 03 c5 c3r
€X. / to
^^-"^ 1 1 1 ^T =^ T^air'^sat
Fig. 9 Boiling Water Heat Transfer Characteristics
Region II indicates the beginning of nucleate boiling. In Region III, bubbles form at a faster rate. In Region IV, bubbles are formed so rapidly that they blanket the heating surface and prevent liquid from reaching the surface. This is the initiation of film boiling and is unstable in Region IV. In Region V, the film boiling becomes stable, partially due to the increasing contributions of radiation heat transfer.
-14- Under normal operating temperature, the cooling system is operating in Region V and VI. If the block temperature dropped below ~250°C, it would enter into the Region IV area and become unstable until it reached point "a" at ~150°C. In conclusion, it has been determined that water/steam cooling is a feasible way to cool the collector bus and control its temperature at the temperatures and heat fluxes experienced in operation. The emitter will be machined from 8" Ni 200 pipe. The pipe has been received. Material has been ordered and received for the collector mockup test. This test will use a full scale collector, flexible con nectors and collector bus with steam cooling. The purpose is to develop the necessary fabrication expertise, fine tune the collector cooling system and confirm the operating characteristics of the whole collector sub-assembly.
Tests were performed to determine a suitable method of attaching the 0.010" thick emitter lead to the emitter. Both TIG and plasma welding proved unsuccessful. Studies were made to determine the feasi bility and cost of laser welding and electron beam welding the lead to the emitter. It was discovered that both methods can be used to weld the lead to the emitter but laser welding can be done at one-third of the cost of EB welding.
An emitter mockup test has been designed and the necessary material ordered and received. This test will determine the effects of differ ential thermal expansion on the emitter, the thin emitter lead, the emitter blade and the emitter bus. In addition the test will give information as to the integrity of the designed joints between each of these components.
In order to compensate for the low voltage generated by nickel/nickel electrodes employed in HC-2, a power supply will be used to "drive" the converter. A contract was let to Energy Research Associates to design and build a 10 VDC, 10,000 amp continuous power supply to be used during testing.
-15- The entire ZERO tubulation was removed, cleaned and leak checked. The converter, including the inner and outer bellows, were also leak checked. A significant leak was found across the bellows at the small hot valve in the return line as shown in Fig. 10. Upon disassembly of the still, cesium oxide was found on the low pressure side of the still and the small line joining the high and low pressure sides was completely plugged. The interior of the tubulation on the upstream side of the small hot valve was clean. The downstream side was black on the inside. It is believed that air leaked into the low pressure side of the cesium still through the leak in the bellows of the small hot valve. The resulting cesium oxide plugged up the small interconnecting line. Cesium oxide continued to accumulate in the low pressure side of the still until the reservoir on the high pressure side had run dry.
1.1.2 Heat Transfer Technology (L. L. Begg, D. Johnson)
Introduction: The objective of this task is to provide the heat transfer technology needed to couple the THX power module to the heat source. It is concentrating on development of a high temperature heat pipe for this purpose and methods of protecting it from the combustion environment. Previous development of high temperature heat pipes has established the basic technology, as demonstrated in Mo/Li heat pipe (3) life tests exceeding 15,000 hours at 1700°K.^ ' The more critical task is development of protective hot shells capable of long life times in a combustion environment. Complementary work on the development of pro tective hot shells by Thermo Electron Corporation has concentrated on ceramic (SiC) protective hot shells on refractory metal substrates and free standing superalloy emitters. This task will develop combustion protection techniques based on the use of refractory metals for strength, with protection provided either by superalloy cladding or self-healing coatings which exist in equilibrium with the combustion environment. The latter technique, the maintenance of a protective hot shell in equilibrium with the combustion environment, is one of the most attrac tive approaches to ensuring a reliable hot shell. The isothermal- stationary nature of the THX heat pipe makes it particularly adaptable to this form of protection. -16- f, -a
Z£PO
\j\r\ J
Ssz/o ItJS
V^cuuri Pur?p
V/icu?J.n.
M Significant leak found across valve bellows Upstream Downstream
This line completely plugged PfifSJUfff I P/fi'-SSUfif J
Fig. 10 ZERO Cesium Still with Oxygenation System: Post Test Examination -17- During this contract period a furnace will be built to permit the addition of additives to provide simulated slag, silica, and other desired coatings. The furnace will then be used with suitable test specimens in order to determine the stability of each hot shell over a period of time.
Current Effort: The University of Denver has been selected to perform the explosion bonding of an Inconel 671 protective layer to a molybdenum substrate. Samples of 3"x3"x.060" arc-cast molybdenum and .030" thick Inconel 671 were received and sent to the University for testing.
In order to inhibit the diffusion of nickel into molybdenum at high temperatures, it is necessary to place a chromium diffusion barrier between the two. A literature search was made as to the methodology of chrome plating molybdenum and the plating apparatus was assembled. Attempts were first made to plate V'xl" molybdenum sheets. Uniform plating thicknesses of 1-1/2 - 2 mils were obtained over the entire surface. The 3"x3" molybdenum plates which will eventually be life- tested will be plated next.
Conceptual design of the life-test slag furnace was completed and detailed design was begun. The furnace has two tangentially mounted, gas-fired burners which fire into an annular space. The inside surface of the annular space is made up of low-mass, high surface area porous ceramic. This ceramic radiates into the test samples located in the center of the furnace. The test samples can be mounted entirely within the furnace or, in the case of a heat pipe for example, penetrate through the furnace lid to the outside. The flue gas exits through the bottom thus allowing free access to a heat pipe on the top. A concep tual sketch of the furnace is shown in Fig. 11.
Orders were placed for the furnace burners and related equipment and the furnace refractory material. The steel furnace shell was rolled.
-18- _-*Tr—., .'•.'< J*'I ..•..' ..-^i—^
•'•.'.;V-','.lS".- -:'
BURNER #1
BURNER #2
Fig. 11 Conceptual Design of Slag Furnace
I •19- Work on the facility for the slag furnace was begun, water and low pressure gas lines were run to the approximate location of the furnace. 110 VAC and 480 VAC, 3({) electrical lines, on dedicated circuits were also run.
1.1.3 System Analysis (G. 0. Fitzpatrick. E. J. Britt)
Introduction: There are a variety of applications for which Thermionic Power Modules could provide significant energy and cost savings. These include utility central station power plants, industrial cogeneration systems, systems incorporating advanced combustors (fluidized bed combustors, gasifiers, etc.), and solar thermal power plants. Each application has special requirements for the converters used in the system. For example, central station power applications require low cost converters with high efficiencies. Low operating and maintenance costs are particularly important in industrial applications.
This task reviews system designs for the most attractive of these applications to determine the probable thermionic operating conditions needed to minimize system electricity costs and maximize energy conser vation. The results are then used to guide the converter research and development effort. It has been shown that the use of THX thermionic power modules as a topping system for coal-fired central station power plants could improve plant efficiencies from their present level, near 33% to 50%. Such an improvement for all U.S. coal fired power plants in the year 2000 could save the equivalent of up to 10 million barrels of oil daily. The preliminary studies indicate that such THX-thermionic power plants would have capital costs comparable to conventional steam power plants, thus providing electricity at reduced costs while saving energy (up to 25%) and reducing pollution (25-40%).^ ' ' Repowering or retrofitting existing power plants with thermionic topping systems could substantially increase their capacity (50% or more) with no more require ment for cooling water. Based on these preliminary results a more detailed system study has been initiated. Task 2.2 in this program.
-20- During this contract period a preliminary study of the cogeneration application will be performed as part of this task. The designs used in the DOE/NASA funded "Cogeneration Technology Alternative Study" will be reviewed and the most attractive industrial application will be selected for optimization. Projections of possible energy savings and system costs will be made. Current Effort: Since the "Cogeneration Technology Alternative Study" has not yet been completed no work was accomplished on this task during this period.
1.2.1 Converter Technology (M. L. Manda)
Introduction: The objective of this task is to increase basic con verter performance using improved converter designs, materials, and pro cessing techniques suitable for subsequent incorporation into practical prototype power modules. The effort concentrates both on improved ef ficiency and power density and on reducing the emitter temperature needed for high performance. Reduced emitter temperatures in turn improve compatibility with the heat source and extend operating life.
Basic converter performance can be roughly characterized by a quantity called the "barrier index" Vg. This index is a measure of the voltage loss within a given converter compared to an "perfect" converter operating at the same emitter temperature. The barrier index for such a "perfect" converter is Vn = 0 eV. Useful estimates of both converter efficiency and output power density can be projected once the barrier index of a device is specified. Fig. 12 shows such efficiency projec tions and output power projections are shown in Fig. 13.
The best converter output power performance obtained to date in the U.S. thermionic program is characterized by a barrier index Vg -1.9 eV. This result has been obtained both by LaBc/LaBc electrodes^ ' and (7) •' 6 b W-IIQWOW electrodes.^ ' The workers in the USSR claim barrier indices as low as VR -1.7 eV, but these results have not yet been verified in the U.S.(8>
-21- ' 1 1 ^— '
» ••• •«» «^ «^ ^^^ _ VB =1^0^ ^^<^ 30 ^^ H ^—• >- 15^ ..^<^- o / z 20 UJ X o u. u. ^ UJ 10 -J
1 1 1 1000 1200 1400 1600 1800 2000 EMITTER TEMPERATURE (K) LEGEND: MAXIMUM EFFICIENCY lOAMP/cm^
Fig.12 Projected Converter Efficiency
-22- OUTPUT POWER DENSITY iWcn?)
o o IQ o
CO -o o-i O C_i. ro o o m n- (D Q. c-> I o 4^ no m 00 < o (D -$ o rt- H n> m -s o ^ 0) ^ m g •g 3J O
-$ m Q
to
O While output power densities can be measured in any converter, actual operating efficiencies can best be measured in cylindrical con verters. For comparison Figs. 14 and 15 show the measured electrode efficiency and average operating electrode power densities of the cylindrical converters tested at Gulf General Atomic between 1965 and (9) 1973.^ ' The operating range for the four NASA/JPL cylindrical con verters reported previously^ ' are shown for comparison in both plots. The output power of the planar W/WOy TECO converter with Vg ~1.9 eV is also shown for comparison.^(7) ' As can be seen steady progress is being made on all fronts, reducing Vn, reducing emitter temperatues and improving efficiency and power density. Both analysis and converter tests show that improvements of 0.1 eV to 0.2 eV are possible using structured electrodes, but no attempt has yet been made to use a combination of structure and low collector work function materials. A number of attractive surfaces which in combin ation could reduce the effective collector work function, and barrier index by up to 0.5 eV have been identified. Finally, analysis, sup ported by basic plasma experiments, indicate that barrier index reduc tion of up to 0.4 eV should also be possible by using more efficient auxiliary ion sources to generate the interelectrode plasma. To study and evaluate these promising concepts a series of conver ter tests is being performed using three types of standard test vehicles. Standard variable spacing planar converters (SPC) are used to obtain accurate J-V data, refined cylindrical converters (RCC) are used to obtain energy balance and efficiency data, and an inexpensive standard cylindrical converter (SCC) is used for screening purposes. In addition special converters are designed to meet the specific needs of some advanced concepts.
Current Activity:
Standard Converters (M. Manda): The fabrication of SPC-9 a refer ence converter with a smooth molybdenum emitter and smooth molybdenum collector was completed during this reporting period, and the converter was installed in the test stand. Initial heating of the converter indicated that the emitter thermocouple was operating improperly.
-24- x GA (Ref. 9) Rasor (Ref. 10) + (J G2 SI Si Fig. 14 Cylindrical Converter Efficiency Performance -25- 8 .. 6 .. 4 .. 1965-"* 1968 2 .. 1970-1972 X GA (Ref. 9) o TECC (Ref.7) Rasor (Ref. 10) + (Planar) 1 1 i 1 s s (S ca ca (S ca s S) ca G3 s S2 ta in (O t^ GO O) s »-• 1-1 *-« T-i «-l r-t OJ (M EMITTER TEMPERATURE (°K) Fig. 15 Cylindrical Converter Output Power Performance -26- Although exact temperature measurements during testing are made with an optical pyrometer, the thermocouple signal is utilized for emitter temperature control. The thermocouple was repaired and the system was re-evacuated. Subsequent heating of the converter confirmed correct operation of the emitter thermocouple and controller. The testing of SPC-9 was completed in late September. Figs. 16-20 show plots of the converter output current density, J, versus the elec trode voltage, V. The J-V characteristics for emitter temperatures of 1600, 1650, and 1700°K are optimized with respect to collector tempera- 2 ture, Tp, at 6-10 A/cm . Fig. 21 shows the performance envelopes for this converter at emitter temperatures of 1600, 1650, and 1700°K. Plots of work function c(> versus T/Tp for the emitter and collector are shown in Figs. 22 and 23 respectively. Fig. 22 indicates that the cesiated performance of the emitter was typical of a surface with a bare work function of 4.5 eV, the actual bare work function was not measured. The lector had a cesiated work function of 1.45-1.6 eV as calculated from back emission measurements. The performance of the converter was normal for a device with smooth molybdenum electrodes. Its Barrier Index was typically slightly higher then 2.1 eV. Fabrication of SPC-10 was completed during this reporting period. This converter has a smooth molybdenum emitter (SMo) and a structured (rough) molybdenum collector (RMo) with machined rectangular grooves. The structured collector is shown in cross section in Fig. 24. The area ratio is -3.0. This converter will be installed in the test stand and tested during the next reporting period. Assembly of the next two converters, SPC-ll and SPC-12, was begun. SPC-11 will have a smooth molybdenum emitter (SMo) and a structured CVD molybdenum collector (RMoCVD). SPC-12 will be used to check performance reproducibility and both of its electrodes will be SMo. These conver ters will be completed during the next reporting period. The electrodes for the Zr-0-W/Cs-O-W converter, SPC-13, are being fabricated. The collector has been electropolished and annealed, and will have a W(llO) layer chemically vapor deposited on it. Several -27- T^ = 1500°K Tj, = 900°K* 28 T T^ = VAR d = VAR Em = SMo Coll = SMo *Collector temperature not optimized 0.4 0.6 0.8 ELECTRODE VOLTAGE (VOLTS) Fig. 16 SPC-9 J-V Characteristics -28- T^ = 1600°K r 28 T 1000°K 24 + 20 + 16 + J [k/an) 12 + 0.4 0.6 1.0 ELECTRODE VOLTAGE (VOLTS) I Fig. 17 SPC-9 J-V Characteristics -29- T^ = 1650°K r 28 T J [h/cxn) 0.4 0.6 0.8 ELECTRODE VOLTAGE (VOLTS) Fig. 18 SPC-9 J-V Characteristics \ -30- T^ = 1700°K 28 T 24.. 20-- 16-' J (A/cm'^) 12- 0.4 0.6 0.8 1.0 ELECTRODE VOLTAGE (VOLTS) Fig. 19 SPC-9 J-V Characteristics -31- ^E = 1750 °K 281 = 900° K* \ h = VAR d = VAR Em = SMo Coll = SMo *Collector temperature not optimi zed J (A/cm^) 0.4 0.6 0.8 1.0 ELECTRODE VOLTAGE (VOLTS) Fig. 20 SPC-9 J-V Characteristics -3Z- ^E = 1600^'K , 1650^'K , 1700^' K 28 T ^C = 1000^' K = OPT. \ d = OPT. 24.. Em = SMo Coll = SMo 20-- 16-- J (A/cm^) 12-- 8-- 4.. 0.2 0.4 0.6 0.8 1.0 ELECTRODE VOLTAGE (VOLTS) Fig. 21 SPC-9 J-V Characteristics -33- (}) (eV) 4.4 h^\ Fig. 22 (Juvs 1/1^ for SPC-9 Emitter -34- 2.2 2.0 o 850°K ° 900°K A 950°K o 1000°K 1.8 T^ = 1650°K Coll = SMo 1.6 (eV) o '^''^^ ^ i °-2 s s ° 1.4 1.2 1.0 0.8 _i L 1.0 1.2 1.4 1.6 1.8 2.0 2.2 TC/TR Fig. 23 ^^vs T/Tp for SPC-9 Collector. Measured by back emissi on. •35- .178 mm (.007 in) .078 mm . .103 mm (.0031 in) (.0041 in) AREA RATIO =2.96 Fig. 24 SPC 10 Groove Geometry b -36- fabrication techniques are being investigated for the production of the Zr-O-W emitter. Modifications to the test stand to accommodate this converter are being considered. These may include gas handling and special pumping systems. Work on this converter will continue in the next reporting period. Current Effort: Bloss Tube Experiment (N. Rasor): No work was accomplished on this experiment during this period. 1.3.1 Converter Performance Analysis (E. J. Britt, J. McVey) Introduction: The objective of this task is to develop analytical models of converter characteristics to guide converter development and provide diagnostic tools. Previous efforts in this program have resulted in models which describe the characteristics of diodes operating in the ignited and the unignited modes, and of converters using auxiliary ion sources. These models have been used to identify and better understand the most important physical phenomenon occurring in the converter and their impact on converter performance. The models are continually refined to explore new possibilities for converter improvement. Emphasis during this contract period will be on possibly beneficial two-dimensional effects, such as those found in structured electrode devices or those which may have existed in the very high performance Bloss plasmatron converter. Other possibly beneficial phenomenon, such as thermoelectric effects, will be studied. Current Effort: Additional work on the analytical models for the ignited mode converter have been carried out during the current reporting period. The analytical models, IMD-2 and IMD-3 described in the January- April quarterly report, have been now been replaced with a more general model IMD-4. The IMD-4 analytical model is a complete one dimensional solution of the thermionic converter using the "shooting method." Five simul taneous equations are solved numerically in order to carry out calcula tions with IMD-4. Two transport equations, Eqs. 1 and 2 shown below -37- describe the currents for electrons and ions respectively. The total energy transport Qc is described by Eq. 3. Production of electron-ion pairs is described by Eq. 4. Production of heat within the plasma is described by Eq. 5. Definition of symbols for all of these equations is given in the nomenclature section at the beginning of Appendix A. '^e = -^eKH^'^"''(^''>Hr] ^'^ 'i - -^i [^Ti HJ - ^"E + (1 H- k{)nk ^ - R. j (2) dT, Qp = Jp[(5/2 + kI)kTe'-^'^e +" "-VI] "e dx (3) dJg dJ. 3 dT = dT = ^(^V - «" ) (^) dQ^ dJ^ All terms of the transport equations are retained and included in the solution. This is significant because thermal diffusion effects can be studied without approximation, and the ion drag term is also included. Boundary conditions for these equations have monotonic electron retaining sheaths in the quasi-saturation region of the J-V characteristic and double sheath barrier at the emitter site of the plasma in the obstructed portion of the characteristic. Electrode area ratios and electron reflectivities are included in the boundary conditions also. Electron back emission from the collector is in the collector side boundary conditions, but ion emission from the emitter has been neglected. A flow chart which shows the calculational logic for solution of the IMD-4 computer program is given in Fig. 25. The calculation con tains three nested interation loops. The inner most loop adjusts the value of the electron temperature at the collector edge of the plasma T p by checking the continuity of heat flux. The second loop which adjusts the value of the plasma density at the collector edge ^^ iterates -38- ENTER DATA, COMPUTE FIXED PARAMETERS GUESS Te^. AND n^. SOLVE COLLECTOR BOUNDARY CONDITIONS INTEGRATE EQUATIONS FOR n(x), T^(x) V(x), J (x).\(x) SOLVE EMITTER SOLVE EMITTER BOUNDARY BOUNDARY CONDITIONS CONDITIONS (DOUBLE SHEATH (SCHOTTKY) =u UPDATE T eCr UPDATE n- CALCULATE AND PRINT RESULTS, INCREMENT J Fig. 25 Flow Chart for the Computer Program Logic in IMD-4 Analytical Converter Model -39- 5-15-80-13 until the electron current at the emitter side has the correct value. During each iteration of the plasma density Tir> the inner loop on T p must be reconverged. There is also a small loop on the emitter side boundary condition which is executed during every shooting attempt. This loop tests for the existence of a double sheath and then matches the boundary conditions at the emitter edge. Finally both values of rip and T p achieve sufficiently small errors and the results are calcu lated, plotted and printed for a given J-V point. The next point on the J-V curve is then calculated by incrementing the current density. An improved computer algorithm is used in solving the IMD-4 model to permit increased stability and efficiency. The first feature of the improved algorithm is a "smart guess" which helps choose initial values for T p and r\~ that lead to a rapid solution. This is done by retaining the converged values for the previous two solved points on the J-V curves and making linear extrapolation to determine initial values for the next point to be solved on the J-V characteristic. The second feature is a "smart step." The smart step improves the iterative search on T p by retaining the convergence values of the T p loop as each value of the rip loop is iterated. Linear extrapolation for the next value of T p to begin the T - iteration is made from the previous converged values. If all the iterated variables remain well behaved, the solution is rapidly obtained by a secant method search routine to direct the iter ation efficiently to convergence. However, there is also a sophisti cated default procedure which becomes active if bad values of the iterated variables make it impossible to shoot the Eqs. 1-5 across the gap. Under these conditions T p is set to a value which is known to be too low for convergence. The value of T _ is then incremented until either the solution is crossed or a default occurs. Crossing the solution can be detected because the test function for T p search is backed up one step and the size of the increment is then cut by one- third. .After this the forward stepping procedure for T p begins again looking for either a solution crossing or another default. Proceeding in this way the search is automatically redirected into a region of convergence if one exists. The default procedure remains active until -40- two valid points have been obtained on opposite sides of the solution crossing then the search technique is converted to the secant method which converges more rapidly. The default procedure is seldom used except on first one or two points to begin calculating a J-V curve, or when calculations are performed with an unusual set of conditions for thermionic converter operation. For the very first calculated point, a typical number of 40 to 50 iteration loops is required for solution. When data have been accumulated from previous solutions of two or more points, so that the smart guess and smart step are effective, a solution is typically obtained with less than 10 iterations. Thus the program is able to "learn" how to reduce its iteration time by a factor of 4. Of equal importance to the reduction of iteration time is the increased stability i the iteration search which is available through this type of search logic. The IMD-4 analytical model has been able to obtain solutions of large pd cases which previous shooting models have not been able to do. An example of the results obtained with the IMD-4 model is given in Fig. 26. which shows a computed I-V curve (solid line) compared with an experimental curve (dashed line). The calculated motive diagram at various points along the curves is also shown in the figure as indicated. For points (j) to (4) in the obstructed region of the J-V curve, there is a double sheath barrier at the emitter edge of the plasma. The emitter sheath in this region of the characteristic is approximately constant in size, equal to about 0.81 to 0.83 volts. The collector sheath does not vary much either in this region. It remains at about 0.26 to 0.3 volts. Points (B) to (T) are in the quasi-saturation region of the J-V curve, and monotonic sheaths are present on both sides of the plasma. Schottky effects on the emitter are included in the calculation in this region which, along with the ion currents, give the slope to the characteristic in the quasi-saturation region. In quasi-saturation both the emitter sheath and the collector sheath increase in size as the current is raised and the plasma becomes more dense. The plasma drop is negative in this region of the curve and growing in magnitude, whereas the plasma drop in the obstructed region is very small and actually has a positive value for the very lowest current densities. -41- The slope of the computed curve agrees well with the experimental data in the quasi-saturation region. However the slope of the calculated J-V curve is too small in the obstructed region. The calculated curve approximately has the shape of a Boltzman line in the obstructed part of the characteristic, whereas the experimental curve is steeper. All the one dimensional calculation models which use a double sheath barrier to reduce the current in the obstructed region, that we have investigated, have exhibited a Boltzman line shape in this part of the characteristic and have had too low a slope. Similar results have been obtained by C. C. Wang at Thermal Electron and others. Various effects, which have been neglected in the previous calculation models, have been suggested to account for the lack of agreement between the slope of the experiment and the calculation in this portion of the curve. However the IDM-4 model includes all the terms of the complete transport equations and still has too low a slope. It's proposed that constriction of the dis charge for the experimental data is responsible for the discrepency in the slope. The constriction of the plasma has been observed visually, beginning at the knee, in all the planar converters with windows that have been tested at Rasor Associates, Inc. If the constriction phenomena is responsible for the disagreement in the slope only a multi-dimensional model which has both radial and axial variables can yield complete agreement with the experiment. The plasma density profiles for the various points on the calculated J-V curve in Fig. 26 are shown along with the corresponding electron temperature profiles in Fig. 27. The density profiles for points (s) and (J) are referred to the scale at the right hand side of the graph, the rest of the plasma density profiles are referred to the left hand side. As the current is increased the plasma density also increases monotonically, and the position of the maximum shifts towards the emitter side slightly. The value of the electron temperature is highest overall for the lowest current at (T). As the current is increased the electron temperature at the collector decreases by about 200°K and the slope of electron temperature becomes somewhat larger. For the highest currents the electron temperature at the emitter edge begins to rise as shown for points (T) and \7J. -42- Tp 1700°K Tf 773°K 567°K TR *F 2.654 eV *r 1.560 eV d 10 mils OUTPUT POTENTIAL (VOLTS) Fig. 26 Comparison Between Calculated Results from IMD-4 and Experimental Data from a Planar Converter • -43- 5-15-80-16 to 1 1 1 ^ 3800 . 3600 . 1 y^ >^ 3400 ^r^ / L i>£ ^^^ / rT) f 3\ o c^ y""^ \ iJ UJ 3200 r^-^ T oc •^^ :;^-^..^/ / »— S 3000 . "''^^^^J^^^~^~~~~~-~,.v_^ ,L LaU: UJ ^^--^^^ 1— y^^^^=:^^£^ z: 2800 . g ®< h® ^^^^ 1— U<-J> ^^^^^^^\ UJ 2600 . 2400 . • 2200 . 2000 —1 1 1 1- \n X (mils) Fig. 27 Variation of Plasma Density and Electron Profile Densities as a Function of Position Along the J-V Curve (see the indicated points on Fig. 26) -44- 5-15-80-17 The data in Figs. 26 and 27 correspond to a near optimum value of the pressure spacing product, approximately 20 mil Torr. The IMD-4 model can also be used to solve cases with high pd. An example of this is shown in Fig. 28, which has a plasma density profile plotted for 40 Torr- mil case. The location of the maximum in the plasma density nearest the emitter side of the plasma is clearly obvious in this case. The long LTE region, represented by the nearly straight line portion of the plasma density profile leading toward the collector, is also evident. Despite the existence of this approximately LTE region of the IMD-4 model obtains stable solutions. Ability to obtain solution for a case like this is due mainly to the improved search routine for the iteration logic. The effects of area ratios have also been studied with the IMD-4 model. The results are similar to those obtained with the simplier models IMD-1, IMD-2 and IMD-3. Increasing the collector area ratio has the effect of improving the output voltage by about 0.1 volts at the knee or below, while the current is not changed. This is shown in Fig. 29, which has calculated J-V curves for 3 different area ratios. Calculated J-V curves with the IMD-4 model that have increased emitter area ratios are shown in Fig. 30. Again, these results are similar to previous calculations. Increasing the emitter area ratio raises the saturation current with a small loss of output voltage in the vicinity of the knee. The stability and computing speed of the IMD-4 model have shown it to be a flexible and useful analytical tool. A FORTRAN computer program has been constructed, and is used for most of the calculations. However, a version of the program which runs on our in-house HP 9825A desk top computer has also been developed. The computing time required by the IMD-4 model is too long to be convenient on the HP machine, but it is occasionally used. A J-V curve consisting of about 30 points can be obtained by setting up an overnight calculation on the desk top computer. The desk top computer program has been useful in testing and debugging the calculational logic because the small computer program is extremely flexible and easy to edit. The one to one -45- —1 —1— —1 1 , 1— » 1 1 1700°K 18 . • 773°K 608°K 16 . • 2.457 eV •E 14 . • •c 1.610 eV 10 mils »^^^ d 12. - y ^ ^^X^J = 16 A/cm^ •• 10 . • / N. • • 1 , , 12 A/cm^ ^V 8 . • / •• Mf^ 6. ^^_^_8_A^cm^ ^*'"Ns^ •• 7 / ^K** \ y*"^ 4 . 4 A/cm^ ^''*'^^ / \ v- -•--•••'*' "'"''' ^^**~*''*--t--t.,u,. ^ 2: ^^^ •• -/ C- ^ 0: —1— 1 1 ! 1 1 /^*^^k>. ca (\i en rt in (o i^ 00 O) o X (mils) Fig. 28 Plasma Density Profile for High Pd (40 mil-Torr) -46- 5-15-80-18 r T5 E 1700°K r 773°K R 567°K F 2.654 eV r 1.560 eV d 10 mils 10-- o >- I—I o on a: C_5 = 1 •+• •+• 0.2 0.4 + 0.+6 + 0.8 1.0 1.2 OUTPUT VOLTAGE (VOLTS) Fig. 29 Variation of Calculated Performance with Different Collector Area Ratios -47- 5-15-80-24 f F 1700°K c 773°K R 567°K F 2.654 eV C 1.560 eV d 10 mils CO E o cC >- I— l-H oo LU Q a: +• -h 0.2 0.+4 0.6 0.8 1.0 1.2 OUTPUT VOLTAGE (VOLTS) Fig. 30 Variation of Calculated Perforance with Different Emitter Area Ratios -48- 5-15-80-25 correspondance between these two program has proven very valuable. In fact, it would have been extremely expensive if not intolerably difficult to develop and test the required iteration search logic for IMD-4 with a FORTRAN version of the program only. 1.3.2 Plasma Characterization (G. L. Hatch) Introduction: The objective of this task is to define theoretically and to determine experimentally the nature of the current and voltage losses associated with the interelectrode space and methods of reducing such losses. Earlier results in this program have shown the real advan tages of using inert gases for the interelectrode plasma, as opposed to the cesium plasma used in conventional diodes. Particularly at low current densities (<2 A/cm ) the use of inert gas and auxiliary ion sources can provide substantial reductions in the arc drop. They also permit the use of wider spacings, and lower temperatures for key com ponents of the converter. The current effort addresses the problem of achieving this result at higher, more practical current densities (>5 A/cm ). It also addresses the nature of the plasma-surface inter face, particularly that of structured electrode surfaces. Current Effort: Work during this reporting period concentrated on selecting a spectrometer that would be suitable for measuring plasma 13 14 3 densities of the order of 10 -10 ions/cm and electron temperatures 2000-4000°K in cesium. A test stand to be used in the noble gas aux iliary ion source experiments is being built. The tests will require mixing of low pressure cesium and pure noble gases. Therefore a system with a small cesium containment volume coupled with a gas purifying system is being planned. -49- 1.3.3 Electrode Surface Characterization (J.-L. Desplat, L. K. Hansen, H. Woo) Introduction; The objective of this task is to develop and experi mentally characterize the electrode surface systems required for success ful and improved operation of converters in other tasks. Three approaches which could substantially improve performance are being studied. First, materials which can provide a low collector work function (<1.3 eV) under converter operating conditions are being sought. A variety of materials with such work functions are available, but they do not perform well at collector temperatures or in the presence of cesium. Recently the work by J.-L. Desplat at NASA-Ames Research Center^ ' has shown that a properly processed W-0 surface stable to 1250°K can have a cesiated work function below 1.0 eV. This appears to be a particularly attractive candidate for use in a converter. Second, the development of emitter surface materials capable of operating low cesium pressures are being developed. Operation at low cesium pressure permits wider electrode spacing, permits the use of otherwise unattractive collector surfaces, and permits the use of gases other than cesium for the interelectrode plasma. The recent development of W-Zr-O surfaces stable to 1800°K which have work functions near (12) 2.6 eV has been \/ery encouraging in this area.^ ' Finally, there is a real possibility that the use of very low work function collectors may result in the generation of negative ions at the collector surface. Such ions could eliminate the advantages of such low work function surfaces. Earlier work in this program has shown that both Cs~ and H~ are generated by such surfaces, but probably at a rate too low to offset performance gains. The possibility of cesium molecule complexes or converter contaminants forming harmful negative ions is under continuing study. -50- Two specific subtasks are in progress. The first is to study the characteristics of low work function surfaces in cesium vapor. This effort is being complemented with a joint program with NASA-Ames to study low work function surfaces under nonequilibrium conditions. The latter effort resulted in the discovery of the stable W-0 surface last year. The second subtask is continuing the study of negative ion species using a special quadrupole mass spectrometer-converter test stand developed in FY1978. Surface Study (J.-L. Desplat) Introduction: From the discussions held at the IEEE Conference on Plasma Science (Montreal, June 1979) it appears that one system of special surfaces looks particularly attractive for improving the perfor mances of thermionic converters. This system consists of: for the collector, the surface obtained by Cs adsorption on a mono layer of oxygen adsorbed on W(llO) at 1250°K.^^"^^ The true work function of this Cs-O-W(llO) surface is 1 eV (measured at 300°K by the Kelvin method). The oxygen structure and its coverage are ex pected to be stable up to 1250°K. However, due to the back emission limitation, the maximum allowed collector temperature is 650°K: the Cs pressure required to maintain the right Cs coverage at this particular temperature should be very low (10 Torr). for the emitter, the surface obtained after a sequence of diffusion into the bulk of a W(IOO) crystal, follov/ed by segregation at its surface of a Zr-O complex.^ ' This Zr-O-W(lOO) emitter has a FERP work function around 2.6 eV and the emitter stability is excellent up to 1800°K. Although data for Cs adsorption on this emitter are not yet available, it is likely that the Cs pressure required to obtain 2.3 eV at 1600°K will be low, making these two special sur faces compatible. -51- These special surfaces have been developed on single crystals in UHV chambers in which it is not possible to study their properties under equilibrium conditions (i.e. constant Cs pressure). It is therefore neces sary to get additional informations in environments more representative of a real converter: the characterization of newly developed surfaces will consist in work function measurements under equilibrium adsorbate condi tions in a plasma anode tube and in actual converters. A second topic was extensively discussed in Montreal: the possi bility that a low work function surface developed in UHV conditions cannot act in a converter as a collector with the same low work function. The origin of this phenomenon is not even clear: is it a surface phenomenon or the result of the interaction between the plasma and the surface? In order to examine the first case, the experimental facility at NASA Ames will be used to simultaneously measure emitting, collecting and true work functions of the already mentioned Cs-O-W(llO) surface which has been shown to have a true work function of 1 eV in the submonolayer mode, or even (15) lower in the thick Cs^O mode. ' Interfacial barriers at the junction CSoO-Ag or GaSb were shown to limit the decrease of photoelectric/emitting (16) work function while the Kelvin work function could go as low as 0.6 eV.^ ' If a similar effect is found for the Cs-O-W(llO) in the submonolayer mode, this would bring the conclusion that even a Cs-0 fractional layer can have semiconducting properties. In addition to these experiments, the NASA Ames facility will be used to study the thermal stability of the structure and oxygen coverage of the "two-dimensional oxide" layer on W(llO), which has a cesiated work function minimum of 1 eV. The sensitivity of this 2D oxide layer to con tamination following an exposure to air will also be studied. These experi ments should provide definite answers as to the long term use of the 2D oxide without the need for regeneration, and to the possibility of preparing this layer outside the converter. -52- Later on, other surfaces will be substituted to the W(llO) single crystal, such as CVD W(llO) layers, or Mo(llO). Surface Testing in a Plasma Anode Tube The plasma anode (or Marchuk) tube was assembled, following the design shown in the previous report. The main oven, in vrhich the Marchuk tube is immersed, was assembled and tested: at 200°C, the temperature gradient from the bottom of the tube to its top is around 10°C, The cesium reser voir temperature control allows to cool the reservoir below the average oven temperature and then regulates the cesium temperature by providing extra heating. The temperature stability is better than 1 C. The probe to be immersed in the cesium plasma has the usual looped filament shape, so widely used before. The main advantage of this design is easiness to build and to heat up. There are however serious drawbacks: near impossibility of studying single crystals and of spot welding permanently a thermocouple which would then add to the electronic emission of the sample under study. This last problem could be alleviated by cover ing the thermocouple wires with an insulating layer of zirconia capable of withstanding 2000°C, but this has not been tried yet. The temperature determination is the most crucial parameter, as in all work function measurements based on the thermionic emitted current. There fore a particular effort was made to get a reliable temperature calibration for the first probe which was chosen to be a polycrystalline tungsten 10 mil wire. In this case, the probe resistance was determined versus the tempera ture of the loop according to the following method. The probe v/as mounted inside a vacuum bell jar. A 3 mil W-26% Re wire and 5 mil W extension wires were temporarily spot-welded respectively at the top of the loop and at each end of the probe, and connected to separate feedthroughs at room temperature. Voltage leads were used for the voltage drop across the probe. The temperature of the emitting loop Tr, inferred from the emf of the W/W-26 Re junction, v;as thus calibrated versus the resistance of the probe Rr, averaged over the temperature distribution. The agreement between the "built-in" thermocouple and pyrometer readings was excellent, within 15 C. It is believed that the correlation Rp(Tp) will still be valid when the probe will be immersed in the plasma. -53- Problems were encountered when trying to use a commercial ceramic coating (Aremco zirconia paste able to withstand 2000°C) to insulate the probe holder from the plasma. The curing procedures had to be very care fully followed and curing times extended to avoid bubbling and swelling v/hich would make the coating brittle and porous. Fig. 31 shows the Marchuk tube, with one probe mounted, con nected to the UHV pumping system and to the Cs reservoir. The emission characteristics of the polycrystalline W wire will be measured in October. Then the same wire will be thermally etched in oxygen to reveal {110} facets, to simulate a (110) single crystal on which the 2D oxide struc ture will then be formed, and its emission in Cs measured. Surface Testing in Converters The standard planar converter (SPC) offering more flexibility, it was decided to use its design to test the special surfaces. Then the collector and emitter surface would be processed from CVD W(llO) and W(IOO) layers respectively. A survey of the reports on CVD W layers on Mo substrates, as they appeared in several Thermionic Conversion Specialist Conferences from 1968 to 1971, showed the need for electropolishing the substrate and the deposited layer in order to achieve respectively a better integrity at the interface and a higher bare work function in the case of W(llO) layers. Electropolishing was not part of the standard machining and assembling procedures. A method for electropolishing Mo collectors was successfully tried,^ ' The electrolyte is 12.5% sulfuric acid and the remainder ethanol. The cathode is a nickel foil. The correct voltage, around ^0 V, results in an intense blue layer which is removed by dipping Mo in ammonia. It is better to agitate the Mo anode and for flat surfaces to have them facing up. Electropolishing a CVD W layer was also experimented in a 2% KOH solu tion, with a SST cathode and a 20 V voltage. The W layer was extremely shiny, but big pits subsisted, maybe caused by defects in the layer. -54- r Fig. 31 View of the Marchuk Tube, with One Probe Mounted -55- Examining the assembling procedures of the SPC, it was found that contamination by hydrocarbons or CO can occur. As the effect of C contamination on the 2D-oxide surface has not been yet investigated, it was decided to oxygenate the collector, once the converter is assembled. (The Zr-O-W surface is reportedly insensitive to an air exposure.) This could be done by connecting the converter through a removable 3/4" bakeable UHV val to an oxygen line capable of UHV performances. This design offers the possibility of reactivating the surfaces and of filling the converter with a noble gas if the Cs pressure giving the right work function is too low for creating a plasma. If the experiments to be done at NASA Ames show that the 2D-oxide or a WO2 deposit on W(llO) can be exposed to air, then a much easier way to proceed would be to use a WO^ deposit and anneal it at 1250°K while outgassing the converter. The Zr-O-W(lOO) emitter, however, does not seem as easy to build through Zr evaporation as originally thought and alternate ways such as plasma sputter deposition or decomposition of ZrH might have to be studied. Basic Research at NASA Ames (J.-L. Desplat, C. A. Papageorgopoulos) The equipment is supplied by NASA and has been used previously to study the Cs-O-W(llO) system. Fig. 32 shows the experimental set-up as it will be modified for the present program. The main modification is the addi tion of a FERP gun which will be used to measure collecting work-functions and electron reflectivity. As we also want to determine emitting work func tions by thermionic emission, the Ta mask, which restricts the evaporation of the W0„ source to a small angle, will be used as a collector. The rest ^ (13) of the equipment has been described previously.^ ' Thick Cesium Oxide Collector Converter Concept (L. K. Hansen) Introduction: There are various operational features that would be advantageous for a practical, advanced-performance, thermionic converter. Three features often mentioned for the cesium vapor converter are: -56- AUGER ELECTRON SPECTROMETER MASS SPECTROMETER KELVIN PROBE LEED OPTICS W OXIDE SOURCE FERP GUN OXYGEN INLET CESIUM KNUDSEN SOURCE Fig. 32 NASA-Ames Experimental Set-up used for Basic Surface Physics -57- 1. Wide spacing operation: The converter becomes easier to fabricate and the perfor mance less sensitive to electrode deformation. 2. Minimum barrier index: Minimum voltage drop and minimum collector work function allow maximum power output. 3. Equilibrium, refluxed electrodes: The emitter surface is therefore stable at high tempera ture operation, and the collector surface regenerates to recover from vaporization products from the emitter. One possible way to achieve these features is by using a cesium oxide multilayer surface on the collector, in equilibrium with a cesium oxide vapor. The cesium oxide surface has one of the lowest work functions measured (0.7-1.0 eV). If such a surface is in equilibrium with a vapor it will be refluxed and will be insensitive to emitter vaporization products. Minimum arc drop and wide spacing operation can be achieved also by using a cesium vapor at an appropriate partial pressure. This pressure can be in equilibrium with the cesium oxide vapor by adjusting the temperature and degree of oxidation of the cesium reservoir. For wide spacing operation this pressure will be relatively low (-0.1 Torr). An optimum emitter work function can be achieved with such a low cesium pressure if the cesium oxide vapor delivers oxygen to the emitter at an appropriate arrival rate. In surmary, the cesium oxide converter appears to combine many of the most desired features of a practical, advanced performance converter. An effort was made during this period to scope the design features of such a device and to determine how the concept can be tested experimentally. The Cesium Oxide Multilayer (18 ?4) There is considerable evidence^ ^^ that one of the lowest work functions in the cesium-oxygen coadsorption system occurs for cesium oxide multilayer adsorption. Such a surface has perhaps the lowest work -58- function of any known material. Gregory, et ar^^^ have measured values as low as 0.7 eV. Their results, obtained by photoelectric emission, are shown in Fig. 33, where the work function of a cesium surface is shown as a function of oxygen exposure. Apparently, the lowest work function occurs for a specific stochiometry. I 10' 10» 10* CLEAN OXYGEN EXPOSURE ILs^muiri) Fig. 33 Work Function of Cesium Oxide vs Oxygen Exposure (Ref. 18) .(19) Denisot^^^ and Alleua and Bacar^'',(20^) have also reported very low values for the work function of the cesium-oxygen coadsorption system, measured by thermionic emission. Some of Denisot's results, plotted versus equivalent J/1^, are shown in Fig. 34. Various investigators have suggested that the lowest work functions in this data and the data of Alleau and Bacal are achieved with a cesium oxide surface. Desplat^^^^ has shown conclusively, by measuring the work function as a function of coverage, that a 1.0 eV surface is obtained for a cesium oxide multi layer. These results are shown in Fig. 35. A minimum work function is apparently reached at 3-5 layers of cesium-oxide. The reason Denisot's minimum occurs near 1/1^ = 1 is probably that only there can a multi layer begin to form. -59- TCs = 3tB'K ntissiDii ruaniii d OXYCEIIE Clorrl 01 3.5 . 10-" 02 1.5 . 10" 03 3.9 . 10"* 01 1.7 . 10* 05 2.7 . 10"* 08 5.1 . 10"* 07 6.1 . 10'* 1 08 l.t . 10"^ 09 3 . 10'^ 10 5.5 .10"^ 11 7.5 . 10"^ 12 H . 10* •Results interpreted 3,3 . 10"* as occurring for a " cesium oxide multilayer. T/Tcs i_J I I I L_L Fig. 34 Work Funct ion of Tungsten (100) Surface for c Tp = 348°K and p(0o) = 3.5 x lO-ll - 3.3 x 10"° Torr (Ref.19) -60- CESIUM FLUX ~ 1.8 x lO" cm-^ s""" CODEPOSITION OXYGEN PRESSURE: (1) 5x10"^ torr (2) 1.5x10-8 4 6 8 10 12 APPROXIMATE THICKNESS (LAYERS) Fig. 35 Work Function of Tungsten (110) Versus Thickness of Cesium Oxide Multilayer (Scale of Abscissa Inferred from Experiment) (Ref. 24) -61- Previous Work with Oxide Converters In the past, there have been studies of the effects of oxygen on thermionic converter performance. In the United States, for example, the effects of oxygen dispensed from a collector layer of tungsten , v. I • >.v. «..* v>i \Jt\j a«-" V -62- 5.5 TEMB90*K M &i z o 2050'K 2 •'•'• 3 u- DC o i 5.1 2I75*K cc UJ 55.0 4.9 4.8 10" 10-• 10"' 10-* OXYGEN PRESSURE, Torr Fig. 36 Effect of Oxygen Pressure on Work Function of Tungsten for Several Temperatures (Ref. 25) In none of the above tests was there an attempt, explicitly stated (or that can be inferred from the experimental details) to lay down a cesium oxide multilayer on the collector and to obtain simultaneously: 1) the low collector work function associated with such a multilayer, 2) the optimum oxygen arrival rate at the emitter, and 3) the optimum cesium partial pressure for the discharge. We now discuss some of the principles for obtaining these conditions in a converter. Obtaining a Cesium Oxide Multilayer on the Collector In all of the previous reported studies of oxygen in thermionic converters the cesium reservoir was by far the coldest spot in the converter. Thus, in these cases almost all of the cesium oxide in the converter would have been in the reservoir. There would have been no -63- possibility for more than a sub-monolayer of cesium oxide on the col lector surface, not enough to develop the low work function character istic of the cesium oxide multilayer. A cesium oxide multilayer can be achieved on a surface by adjusting the surface temperature close to the reservoir temperature. Therefore, it is proposed that a thermionic converter be designed to allow the col lector temperature to be carefully adjusted about the reservor tempera ture without having any other converter internal part at a temperature below that of the collector and the reservoir. Such a converter is shown schematically in Fig. 37. Note that in this case the collector structure does not penetrate the vacuum wall. In this way, an attempt to adjust the temperature of the collector surface near to the reservoir temperature does not produce regions in the converter (about the col lector) that are cooler than the collector surface and perhaps cooler than the reservoir. Cs AND EMITTER CS2O VAPOR TEMPERATURE CONTROL AUXILIARY RESERVOIR COLLECTOR Fig. 37 Schematic of a Converter for Obtaining a Multilayer on the Collector -64- The actual temperature differences between the collector and reservoir to obtain the optimum collector coverage, 3-5 molecular layers, is difficult to calculate. This is because the theory of multilayer adsorption is not highly developed and because the adsorption energies of cesium oxide are not well known. However, some estimates can be made using available data and theory. One of the best analyses of multilayer coverages (for e>2) is the Frenkel-Halsey-Hill slab theory^^^ According to this theory, the coverage 6 (in molecular layers) is given by Jin £- = -Aee"'' ,^^ ^" PQ kT ' ^'^ v/here p is the pressure of the vapor, p is the vapor pressure corresponding to the surface temperature T, and Ae is the difference between the adsorption energies in the first layer and in the bulk material (6 ^- »). The para meter r is a number, probably betv;een 2 and 3, that corresponds to the exponent for the fall-off of the excess adsorption energy of the first layer. The energy difference-Ae for ceSium oxide is not known. However, the (31) dissociation energy.for Cs^O is 3.38 eV^ , and it is known that the first layer of Cs,,0 is bound to a tungsten surface with forces strong enough to (24) compete with the corresponding Cs^O molecular binding forces. ' Thus, the binding energy of this first layer could be the order of 2 eV or more. The adsorption energy for the bulk is 1.2 eV. For a crude estimate, therefore, we can assume Ae to be on the order of 1.0 eV. (32) The vapor pressure of cesium oxide is given by^ ' log pjCs^O) =-^+ 7.71, (2) or iln p(Cs.O) = -^^^f^+ 17.75. (3) 0 ^ ' -65- Thus, we can write 13834 |I_fl _ 1i 1I ^ _ Ae • e"'' (4) hi] kT If we assume T = 500°K Ae ^ 1 eV 11,600 °K/eV = 23.2 500°K e"'' = .037, we obtain T = 485°K. Thus, the temperature difference between the collector and the reservoir should be the order of 15 C in order to develop the low work function, multilayer surface. This corresponds to a T/Tp ~ 1.03. Some work function minima observed by Denisot were at T/Tp ~ 1.4 (see Fig. 38). Other data by Denisot indicated lower values. These results are consistent with the above estimate of T considering the uncertainties in r and Ae. o ^ Choosing the Operating Point The thermionic converter is a non-equilibrium device because of the net current and because of the emitter-collector temperature difference, but the beginning of an understanding for specifying a converter operating point can be the equilibrium system. Consider, therefore, the cesium- oxygen system in equilibrium. According to Taylor and Langmuir^ ^ the vapor pressure of cesium is given by log Pjj(Cs) = 11.0531 - 1.35 log T - ^Y^, (5) (32) and according to Touzain ' the vapor pressure of cesium oxide (Cs„0) is given by log Pjj(Cs20) = - ^+ 7.71, (6) where the pressure is in torr and the temperature is in degrees Kelvin. We can assume Roult'slaw for partial pressures over a mixed reservoir, although small deviations are recognized^ '. Thus, for the mixed system -66- 450 500 550 600 650 700 T (°K) Fig. 38 Cesium and Cesium Oxide Partial Pressures Versus Reservoir Temperature and Percentage Reservoir Oxidation -67- p(Cs20) * Po(Cs20) x(Cs20) (7) P(Cs) ^ p^(Cs) [1 - x(Cs20)], (8) where p(Cs20) and p(Cs) are the partial pressures in equilibrium with the reservoir and x(Cs20) is the mole percentage of Cs^O in the reservoir. With the help of equations 5-8, the plots of x vs T for constant partial pressures can be plotted, as in Fig*. 38. On this figure, the desired Cs-O partial pressure can be chosen to give the desired oxygen effect at the emitter, and the desired cesium partial pressure can be chosen to give the desired arc drop. This locates a point on the plot. The coordinates of this point give the desired degree of reservoir oxidation and the reservoir temperature. There are, of course, deviations from this plot in an actual converter. Cesium oxide from the collector, for example, will be dissociated at the emitter. To some extent the emitter material will be evaporating as metal-oxide and will be depositing on the collector. And there are deviations from Roult's law. Nevertheless, the above equilibrium situ ation should give an approximate operating point. As an example, suppose a pressure p(Cs) = 0.3 Torr is needed for minimum arc drop with the spacing chosen (perhaps d ~ 50 mil) giving a pressure-spacing product of 15 mil-Torr. Also, suppose an oxygen arrival rate corresponding to p(Cs20) = 10" Torr is desired (chosen be cause of the data in Fig. 36). This means that the reservoir should be about 15% oxidized and operated at about 500°K. This operating point is indicated in Fig. 38. Cesium Oxide Test Converter A simplified converter was designed to test the cesium oxide concept. It is shown schematically in Fig. 39. This converter uses a tungsten spiral filament for an emitter. Current to this filament is half-wave rectified and the I-V curve is studied during the off part of the cycle. The collector is simply the stainless steel vacuum wall of the apparatus. That wall also serves as the reservoir. The wall is thickened in these collector and reservoir regions to improve temperature uniformity and -68- TO VACCUM STATION GUARD RINGS RESERVOIR COLLECTOR EMITTER FILAMENT IONIZER FILAMENT ACCELERATION GRID Fig. 39 Simplified Converter for Testing the Cesium Oxide Concept V -59- control. The plasma for the converter is generated by electrons from the ionizer filament. A grid at the end of the spiral emitter accel erates the electrons into the center of the spiral to generate the plasma. Guard rings restrict the region of the collector to that immediately adjacent to the spiral emitter. The temperature of the collector and the oxidation level of the reservoir is chosen by the considerations of Fig. 38. The reservoir temperature is chosen relative to the collector temperature by the use of Eq. 4. Negative Ion Study (L. K. Hansen, H. Woo): During this period numerous attempts were made to see negative cesium ions. The preliminary data previously reported was obtained with a relatively crude apparatus. Now, with the apparatus redesigned and completely rebuilt we saw nothing. There were some start-up bugs to work out in the apparatus. These were overcome, but still no indication of a Cs~ signal was found. Work is continuing to determine the source of the current problems or to identify the cause of the Cs~ signals previously reported. 1.3.4 Thermionic Converter Data Review (L. K. Hansen) Introduction: Information on various aspects of converter physics or operation is widely dispersed in both published and unpublished form. Few single references exist which summarize this information. This task will begin the process of reviewing and summarizing this information in such a format that the result of previous work will be generally available. Current Effort: The effort to acquire electrcn-neutral cesium cross-section data was continued. TASK 2 - THX POWER MODULE EVALUATION 2.1.1 THX Test Module (L. L. Begg, B. Carlsmith) Introduction: The overall objective of the THX Test Module Task is to provide a proper evaluation of the THX power module concept prior to a decision to develop prototype devices. -70- This evaluation will be made by designing, fabricating, and testing a preprototypic device. The modular nature of the THX permits such a realistic evaluation of the technology at minimum cost. The preprototype THX will include all essential components, includ ing heat transfer system (heat pipe), power output system and representa tive THX cells (both the end cells and a central cell). Since the THX consists of two end cells and a variable number of central cells, the development of this three cell module will be representative of THX modules with more than one central cell. The THX Test Module will be sized so as to permit evaluation for both central station power plant and industrial cogeneration applications. During this current contract period effort will concentrate on designing the THX Test Module in sufficient detail to project probable performance, input power requirements, component operating temperatures and heat fluxes, materials, and required auxiliary test equipment. Supporting component tests, such as material strength or joint design, will be performed as needed to support the design effort. Current Effort: Effort on this task during this reporting period concentrated on obtaining a better method for predicting the long term creep properties of TZM. The creep strength of TZM is the structural criteria used to size the wall of the TZM heat pipe used in each THX, and this component has a major impact on THX cost and reliability. Previously a Larsen-Miller Parameter was used to determine stress to rupture at a given time and temperature. The parameter is calculated from: P = (20 + log t)(T + 460)(10"^) t = time, hrs. T = temperature, °F Having calculated P, the stress to rupture was found from the following chart obtained from Ref. 36. -71- P=(20 + log t) (T + 460) LARSON-MILLER PARAMETER The creep rate e, was calculated by the equation: t = 6/t^ t = creep rate, %/hr t = time to rupture, hr Recently discussion of this problem were held with Mr. William Klopp of NASA Lewis. Mr. Klopp has accumulated a data base and de veloped an equation to product creep rates of worked and recrystallized TZM which is dependent upon stress level and temperature. The creep rate may be predicted according to the relation: -72- t = K(sinhaa/E)"(e"^l/'^'^+ce"V'^'^) t = creep rate, sec K =: proportionally constant, sec a = constant, dimension!ess 0 = stress, ksi E = Young's modulus, ksi n =: stress factor exponent, dimensionless Qi = high temperature activation energy, cal/g-mol R = gas constant = 1.986 cal/deg-mol T = temperature, K c = constant, dimensionless «2 = low temperature activation energy, cal/g-mol Although the Klopp parameter must be used with caution, it indi cated that the allowable stress in the TZM wall are higher than those currently being used in the THX computer program. Interestingly, both Klopp's parameter and Ref. 37 indicate that, at the time and temperature experienced by a THX, recrystallized TZM has a lower creep rate than stress relieved TZM. Fig. 40 compares extrapolated stress levels which would limit strain to 0.5% for 263,000 hrs. (30 yrs.) using the Larson-Miller parameter for stress relieved TZM and the Klopp parameter for recrystal lized TZM. The possibility of using inertial welding techniques to fabricate a large heat pipe have been explored with Manufacturing Technology, owners of the process. In the inertia welding process one workpiece is fixed in a stationary holding device. A second workpiece is clamped in a spindle chuck, usually with an attached flywheel, and spun rapidly. At a predetermined speed, driving power is cut and the first part is thrust against the second piece. Friction between the parts decelerates the flywheel, converting stored energy to frictional heat-enough to soften, but not melt, faces of the parts. The energy level is adjusted to achieve foregoing temperatures. The rate of energy consumption is determined by -73- r ,LARSON-MILLER EXTRAPOLATION FOR STRESS RELIEVED TZM KLOPP EXTRAPOLATION FOR RECRYSTALLIZED TZM 1000 1100 1200 1300 1400 1500 1600 1700 TEMPERATURE (°K) Fig. 40 Projected Stresses for 0.5% Creep Strain in 263,000 Hours for TZM -74- the weld itself-when fully consumed, the weld is completed and flywheel rotation stops. Because the bond is a forgoing and no melting occurs, no recrystallization takes place. The bond is characterized by a fine grain structure and ductility is retained. Manufacturing technology has the capacity presently to weld pieces 8" in diameter and 40' long. They have had experience in welding molyb denum and TZM. 2.1.2 Heat Source (D. C. Johnson, L. L. Begg) Introduction: The objective of this task is to provide the furnace facility required for development and operation of the THX Test Module. Current Effort: Work was initiated during this reporting period. A ground rule established at the beginning was the requirement that the furnace should reach a sufficient temperature so as to radiate heat into the THX Test Module heat pipe at the desired heat flux. This is in contrast to a method of direct flame and/or gas impingement and more nearly represents the anticipated operating environment. The first approach explored used combustion on or close to the sur face of the furnace wall. The combustion products would heat the furnace wall sufficiently to radiate the heat pipe. Attempts were made to locate a commercially available furnace of this type and the requisite capacity. There are conventional furnaces built that reach the neces sary temperature and could be modified to accept the heat pipe but they are designed to operate without the thermal sink that the heat pipe represents. The result of this effort is the conclusion that combustion heat could not be transferred to the walls in a reasonably sized test furnace at a rate sufficient to maintain the required radiant heat flux to the heat pipe. -75- A porous wall furnace was identified as an early design candidate. An air/fuel mixture could travel through a porous refactory furnace wall material where it would combust on the surface of the furnace wall. Such furnaces are commercially available but have a temperature limita tion of ~2000°F. Above these temperatures the refactory material neces sary to withstand the higher temperatures becomes too dense to allow sufficient flow. . Refractory cloths were tested which could sustain the temperature required at the furnace wall (~3000°F). However all of these cloths become brittle after firing when the binders or stitching burn out. Thus they are unsuitable for use as a free-standing furnace wall. Several attempts were made to obtain the necessary wall temperature using a ceramic fixed-bed furnace. An air/fuel mixture was flowed through a ceramic pebble bed with combustion occurring on the surface of the bed. The surface temperature, however, did not reach a value high enough to radiate the required heat flux. Following these tests the design approach was changed. Instead of burning on the radiating surface a volume large enough to allow complete combustion was provided upstream of the surface. Following combustion, the hot products were forced to flow through the surface transferring heat by convection which in turn was radiated from the surface. A small test furnace was constructed to test the concept and various candidate materials for the porous wall. The furnace, shown in 2 Fig. 41, can test radiating areas up to 175 cm . The first attempt at this post-combustion, porous wall heat transfer was a checkerwork of refractory material. Air and fuel were injected into a volume where they were mixed and allowed to combust. After combustion, the hot products would leave the furnace through various networks and meshes of checkerboard refractory material. Figs. 42 and 43 show the configuration of the furnace and checkerwork refractory material. The finer the checkerwork evolved, the higher the temperature the wall achieved. Again, however, the wall didn't reach a sufficient temperature to radi ate the desired heat flux. -76- Fig. 41 Test Furnace ^^o-^^WW if£i«'^j*S^^ •77- I f (^ P.^-VNiUlkKr V* 'V- Fig. 42 Top View of Furnace with Fig. 43 Exit View of Furnace with Checkerwork Refractory Checkerwork Refractory -78- The next series of attempts replaced the checkerwork refractory with alumina tubes a shown in Fig. 44. This arrangement showed a marked increase in the temperature but it still fell short of the required flux. The most recent tests used walls made from cellular ceramic cubes. These cellular cubes have a high surface, area to volume ratio. This permits more heat to transfer to the ceramic cube in turn driving it to a higher temperature and radiation heat flux. Fig. 45 shows some of the various meshes and materials which were tested. Initial tests were made using a cellular block of cordierite. The cordierite exhibited good thermal shock resistance but melted at ~1400°C. Cellular mullite blocks were tested up to 1550°C. Fig. 46 shows the radiant surface of a cellular mullite mesh during testing. Cellular alumina blocks will be tested next in an effort to achieve a longer lifetime material at 1650°C. 2.1.3 Output Power Transfer System (R. S. Dick, E. J. Britt) Introduction: The objective of this task is to provide the design, fabrication, and performance data needed to ensure proper operation of the THX Test Module power transfer subsystem. Three possible subsystems are currently being considered: flux reset inductive power coupling, push-pull inductive power coupling, and series power coupling. The basic feasibility of inductive power coupling was demonstrated on a small scale early in this program, and series power coupling is the approach conventionally used. The effort during this contracting period will concentrate on selecting one of these approaches for use in the THX Test Module. Following selection of the approach, a full scale engi neering test will be accomplished at the currents and voltages typical of the THX Test Modules. The test will serve to confirm the early design methods and identify practical problem areas. Further tests will then be conducted to eliminate these problems. Finally a full scale test simulating actual THX power transfer conditions will be run in order to confirm the THX power system design and to obtain operating experience. -79- Fig. 44 Furnace Test Using Alumina Tubes •80- I I'lTi'l I'l'i'i'i'i'i'i'i': m|.|i|i|mpyinmi|i|Mi|i|i|i|t|i|i|imi|.ji|iii|i|i|i|ij iiliiiliiiiiiliiiiiliiitfiiiliiiiiliiii^^^ H mmmmMBm, rti «••• < ••••••••••iflaa^v --^••••••••••••••••••r.HavBBi-••••••••••••• kiB|SE"'5i.' —•••••••••siH ••••••••••.' r: iiiss::;!: •»« !!!i"S!i! Fig. 45 Samples of Cellular Materials Used in Experiments -81- r Cellular Mullite Blocks Under Test Fig. 46 Mullite Cellular Block Under Test -82- I Current Effort: The promising results discussed in the previous report showed that series connected TFE's could be used with acceptable current losses through the steam system. Work during this period con centrated on defining inverter subsystem sizes and costs and the possi bility of using high pressure insulating sections in the steam lines. Inverters used for power conditioning vary strongly in cost as a function of input voltage and current: as voltage goes down, cost goes up for the same current. Because the cost per kilowatt, weight, and size of the inverters drops so steeply with increasing voltage, as shown in Figs. 47, 48, and 49, placing the modules in series is desirable. Point 'a' in the figures is a recent estimate supplied by Udylite Corporation of the cost of a 10 vdc, 100 k A inverter. The price of $300/kW is for a single unit and does not include design engineering charges. The rest of the cost data in the figures is in 1966 dollars. By placing the modules in series and raising voltage output, the cost per kilowatt will be reduced. A limit is reached wherein placing more units in series, though continuing to reduce cost also reduces redundancy of outputs, and thereby reliability. Work is continuing on array arrangements necessary to minimize cost for a given redundancy. Because the modules have low voltage output, power conditioning equipment must be placed as close as possible to the furnace. The area covered by power conditioning equipment is of concern. As can be seen in Figs. 48, and 49 the area per kilowatt and weight per kilowatt drops steadily with increasing output voltage. The parametric system analysis being performed in Task 2.2 will take account of the further trade-off between THX system output voltage, structure and area necessary to support inverters. 2.1.4 Heat Transfer System (L. L. Begg) Introduction: The objective of this task is to provide the design, fabrication, and performance data needed to ensure proper operation of the THX Power Module heat transfer system. The heat pipe which trans fers thermal energy from the furnace to the THX module is a critical -83- (D' N^\ 7500 A ^10,000 A ^ _,—I—«- -H 1 1 1—H 10 20 40 60 80 100 200 400 400 600 800 DC Voltage Fig. 47 Power Conditioning Specific Cost Characteristics -84- 20 30 40 300 400 DC Voltage Fig. 48 Rectifier and Inverter Equipment *Footprint Versus Voltage 100 90 80 70 60 50 40 4 I 00 CT> I 5 30 20- 5000 A 8000 A 10,000 A 10 -(—(—•- 5 6 7 8 910 20 30 40 50 70 100 300 400 DC Voltage Fig.49 Rectifier Equipment Weight Versus Voltage element of the module design. It must survive the combustion environ ment, transfer heat at high fluxes and temperatures, mate reliably to the power module, and provide a stable precision surface as an emitter ^ substrate. This task will serve to establish the first two capabilities- combustion environment lifetimes and heat transfer capability—in heat pipe tests using the heat source facility. Supporting tests needed to establish the fabrication procedures, and define the heat pipe start up dynamics will be performed as necessary. The effort during the current contract period will concentrate on defining the best heat pipe configuration (wick, arteries, structural supports, etc.) for use under THX operating conditions. Following definition of these aspects of the heat pipe design and initial results from the Heat Transfer Technology subtask (Task 1.1.2) a reference heat pipe design will be developed and fabrication processes established. A THX-scale heat pipe will then be built and tested at the heat flux and temperatures representative of THX operating conditions. Based on the results of this test recommendations will be made for modifications to the Reference THX Power Module Design. Current Effort: No work was accomplished on this task during this reporting period. 2.2 Conceptual Plant Design (G. 0. Fitzpatrick, E. J. Britt) Introduction: The objective of this task is to realistically define the performance potential of a coal-fired thermionic central station power plant. Preliminary system studies reported earlier in (4) this program^ ' have shown that combined thermionic-steam central station power plants can be both highly efficient and economic. How ever, these studies have been made assuming either relatively conven tional power plant components and steam cycles, or advanced components designed for other applications (such as an MHD type high temperature air preheater). The resources have not been available to fully optimize a thermionic power station, as opposed to a simple thermionic topping system. As a result the full potential of thermionics in the central -87- station application has not been well defined. To provide such defi nition a more fully optimized design is being generated in three phases: • Phase One - Baseline system definition and parametric analysis t Phase Two - Conceptual plant design • Phase Three - Development plan definition In Phase One, which will be accomplished during the period of the current contract, a baseline coal-fired steam electric plant incorpor ating thermionic converters will be defined, and parametrically analyzed to determine, at a conceptual level of accuracy, the most economic process conditions and equipment designs. Data from this phase will provide the input for the conceptual design of a power plant equipped with thermionic converters in Phase Two. In addition it will provide initial estimates of system performance and major process component installed costs. Based on the results of Phases One and Two, Phase Three will com plete the study by reconmending development goals for commercially acceptable thermionic converter systems for steam electric power plant applications. The study is the joint effort of Rasor Associates, Inc., Bechtel National Corporation, and Foster Wheeler Development Corporation. The effort is divided between the participants as follows: • Rasor Associates will provide overall coordination and required information pertaining to the performance, design and cost of the THX (Thermionic Heat Exchanger) modules. • Foster Wheeler will provide required information pertaining to the performance, design and costs of coal-fired combustors equipped with thermionic converters. • Bechtel will provide required information pertaining to the performance, design, environmental intrusion, capital costs, construction schedule, and power generation economics of steam electric power plants equipped with the thermionic boilers as defined by Rasor Associates and Foster Wheeler. -88- Current Effort: Ground rules for the study were set during this quarter, parameters defined, their ranges set, and work was started by Bechtel National and Foster Wheeler. The form of output was decided, and a number of input values set. The following ground rules were established to ensure comparability with previous Energy Conversion Alternatives Study (ECAS). Ground Rules 1. POWER OUTPUT • Base loaded plant t Electrical power putput at 60 Hz, 3 phase, 500 kV at transformer output side 2. DESIGN GOALS t Plant useful life 30 years • Availability 90% 3. FUEL • Illinois number 6 coal • Higher heating value 10,788 Btu/lb • Coal composition (as received) Constituent % By Weight C 59.5 H 5.9 S 3.9 N 1.0 0 20.0 Ash 9.6 • Fuel cost $1.00/MBtu, in mid-1975, delivered • Fuel storage at site - 60 days dead, 3 days live -89- EFFICIENCIES • All efficienices basedon higher heating value of the fuel as delivered to the plant site • Thermodynamic efficiency _ Gross Electrical Energy Generated Thermal Input to the prime cycle • Power plant efficiency = Overall efficiency (coal-pit to busbar) _ Net Electrical Output HHV of the fuel supplied EMISSION STANDARDS t Design and Cost Basis: Standard Pollutant (LB/MBtu Heat Input) SO^ 1.2 NO^ 0.7 Particulates 0.1 • Discuss capability to meet current and future standards EXENT OF THE PLANT STUDIED • Coal handling equipment at the central station plant, including facilities for coal unloading from rail cars, storage facilities for a 60 day supply, and conveyor equipment (but not including coal transportation from the mine to the plant site) a Combustors and emission control equipment a Ash and other waste removal and disposal equipment within the plant site property limits a The energy conversion subsystems, including heat input heat exchangers, and electrical generators a All auxiliaries and balance of plant, including buildings, land, offices, shop facilities, special maintenance equip ment, water treatment equipment, protective devices, etc. -90- • Power and voltage control subsystems suitable for isolated operation or operation in parallel with existing generating units in a utility system a Heat rejection subsystem a Current inversion equipment for those conversion concepts generating direct current a Transformers to raise voltage to transmission line or dis tribution voltage but not the high voltage breaker and switch yard 7. UNIT RATING AND SIZING a Unit rating based on average daily ambient temperature 59F with a coincident relative humidity of 60% a Generator and fuel handling equipment sized for maximum power output at ambient air temperature of 20F with a coincident relative humidity of 60% • Heat rejection utilizes evaporative mechanical draft cool ing towers sized for 5% summer condition corresponding to an ambient wet bulb of 76F and a pressure of 1 atm (101 kPa) 8. PLANT SITE a "Middletown, USA" a Make-up water available from a nearby "North River" 9. ECONOMIC ASSUMPTIONS a Capital cost basis mid-1979 a Discount rate = Weighted Cost of Capital = 10% • Escalation = 6.5% a Fixed Charge Rate = 18% a Electrical-Mechanical field labor rate = $n.75/hr a Construction cash flow in accordance with the given "S-curve" a Capacity factor = 70% -91- Fig. 50 is a flow schematic diagram of the power system to be studied, additionaly one other flow system may be considered which is presented in Fig. 51. The second system will be considered if it is found that the increase in efficiency in the thermionics resulting from lowered collector temperatures offsets the losses to efficiency from bypassing more heat to the steam system around the THX modules. A total of nine plants will be examined. Three furnace capacities, 900, 1800, and 2700 MWth will be treated. Three air preheat tempera tures, 1500, 2000, and 2500°F will be used. Parametric performance and cost trends will be based on the more detailed analysis of these nine. Table 1 lists these and some of the other variables which will be examined. -92- Coal r 017,000 Ib/h Air Coal Primary Air 980.000 Ib/h 8.1305 X lO' ib/h 59*r Procesilng _2nrmi Sprond.Trv Air . 860*F /f\\ni > Fans A 300* F TilERMIOIlIC High Teap. Finishing Low leap. Preclpltatort Air Heaters superheater Air Heaters and Slack FUMIACE 2300*F H9A*F 930*F ID Fans DC/AC Inverter! Thenalonlc Converter* J633..5 Rclieater Steaa Generator 595.5 HJg 602*F lOOO'F ICOO'F " AC ST 4H ST -i- Generator •715 MU, 4.75 » 10^ Th/h Clrc. 1 r Wet Punp ^j_Z___L_ Cooling Feedwater r -i(r] ^A /Towers jiniL. Heaters Steaa > ri> Condensers ' Condensate Funp x/ Coolliig Water Kake-up Fig. 50 Flow Schematic Diagram I I -93- Coal f 817,000 Ib/h Air 8.1305 X lo' Ib/h Coal Primary Air 980.000 Ib/h 59*F Processing ?nnn'. rr ^prnndnrv Afr . 860* F n\\|F D Fans P\ / ' I 300*F THERMIOMIC High leap. • Finishing Loo«u Teap. _ZT*" Prcclpltacori Air Heaters luperheater Al r Heaters ^^ and STi»rk FURIIACE 2300*F 1494*F 930*F ID Fans ''.L / / DC/AC / y Inverters Thermionic Converters / «33..S Finishing Rcheater Steaa / Generator Reheater / 595.5 602T r-t 1000*F 1000*F " AC ST -i-l ST -Oh Generator •715 )«e 4.75 X 10* m/h Clrc. \—r Uec Pufflp I \ Cooling [ Towers 4 •J'"*'' Feedwater Heaters Steaa < -0" n> Condensers ' Condensate XJ Funp Cooling Vater >'.aie-up Fig. 51 Flow Schematic Diagram V -94- TABLE 1 THERMIONIC POWER PLANT STUDY List of variables and their range for parametric study Variable Name Min Base Max Furnace Capacity (MWth) 900 1800 2700 Air Preheat Temperatures (°F) 1500 2000 2500 Heat Pipe Temperature (°K) 1300 1600 1700 Feed Steam to FSH (°F) 720 820 Radial Heat Flux of Heat Pipe (W/cm^) -20 -48 <100 Thermionic barrier index (eV) 2.0 1.5 1.0 Cell Current Density (J, watts/cm ) 1.0 10 30 Voltage output of THX string 1[Volts ) 200 -95- PART II - NASA/JPL PROGRAM TASK 1 - CONVERTER FABRICATION AND TEST (G. L. Hatch, M. L. Manda) Introduction: The overall objective of this effort is to measure the operating efficiency of converters using the best available elec trodes and operating modes. In 1978 this effort concentrated on establish ing the design and fabrication procedures for a new cylindrical converter test vehicle specifically engineered to meet this need. Five such converters were built and tested, proving the effectiveness of the design. These five were used to study structured electrode effects. A two-level two-factor experimental design was used in order to obtain the most possible information with the limited number of converters. Fig. 52 shows the design used, called the "initial design." The first converter built used smooth molybdenum electrodes in order to simplify the fabri cation process. The four following converters use rhenium emitter instead of molybdenum in order to obtain higher performance. The two factors studied were emitter surface structure and collector surface structure. The test results indicated that structured surfaces could provide an output voltage increase of ~0.1 volt, with a concurrent improvement in efficiency below T^ ~1700°K. Above 1700°K efficiencies were not improved, probably because of increased radiation losses due to the higher emissivities of the structured surfaces. The best perfor mance demonstrated at the design temperature of 1700°K was an efficiency 2 of 12% with a lead power density of 2.6 W/cm . The converter providing this performance had a structured rhenium emitter and smooth molybdenum collector. For comparison Figs. 53 and 54 show the meastired electrode efficiency and average operating electrode power densities of the cylindrical Converters tested at Gulf General Atomic between 1965 and 1973. The operating range for the four NASA/JPL cylindrical converters are shown as cross hatched regions in both plots. The output power of the planar W/WO TECO converter with Vg -1.9 eV is also shown for comparison. As can be seen, steady progress is being made on all fronts, reducing emitter temperatures, increasing power density, and increasing efficiency. -96- KEY FACTORS: + A - Emitter Surface Smooth Structured B - Collector Surface Smooth Structured C - Collector Material Mo *xOy EXPERIMENT DESIGN: Converter Test Factor 1978 1979 1980 A JPL-2 1 + + JPL-3 RCC-6 2 + + JPL-4 3 + (JPL-5) RCC-7 4 + RCC-9 6 + RCC-8 5 + RCC-10 7 RCC-11 8 Initial first experimental design examined factor A and B Completion of eight tests permits examination of factors A, B, C and their interaction. Fig. 52 Converter Test Plan -97- 20 15.. *s ol0 UJ I—I <_> Ll- U_ UJ \ LU O 1978 O 1970-1972 JPL 2,3,4 5 .. X GA (Ref. 9) Rasor (Ref. 10) + 0 1 cs 6) ca s S ca s ca S s s ca s s in (O t^ GO O) ca t-t »-« r~t »-l T-t «-« CM Od EMITTER TEMPERATURE (°K) Fig. 53 Cylindrical Converter Efficiency Performance -98- o IS ta S ta ta ca s S csa (sa (a ca s in (D t^ GO O) ca T-l •r-l »-l •r-l «-• T-» CM CM EMITTER TEMPERATURE (°K) Fig. 54 Cylindrical Converter Output Power Performance -99- Beginning in March 1979, the specific objective of this task is to operate a converter with a lead efficiency of 15% and an output power density of 5 W/cm at an emitter temperature, nominally 1650°K, suitable for application to a space nuclear power system. To achieve this objec tive three cylindrical converters will be built and operated. The experimental design will be expanded to include the effect of a high performance collector, surface, molybdenum oxide, although insufficient resources exist to test all of the converters called for by the plan this year. Special tests may be made of particularly attractive con figurations, such as the hybrid mode. Current Effort; The first of the three converters to be built this year, RCC-6 will have a structured ("rough") chemically vapor deposited rhenium emitter (RReCVD) and a smooth molybdenum collector (SMo). Fabrication of RCC-6 continued during this reporting period. Almost all of the converter body parts were machined. Approval for fabrication of the electrode surfaces was received in August. The emitter blank was machined and sent to Ultramet for rhenium deposition. The collector was also machined. Cleaning and outgassing of parts was initiated and will be completed early in the next reporting period. Completion of RCC-6 is being delayed primarily due to delivery dates associated with vapor deposition. RCC-6 should, however, be completed and tested in the next reporting period. Modifications to the ceramic-metal seal were implemented to facili tate converter assembly. The inside diameter of the seal was increased to 1.42 cm (.560") to accommodate the emitter lead heat shields more easily. An order for these new seals was placed and delivery is expected in early November. The metal portions of these seals were machined from Kovar tubing. These will be brazed to ceramics by RW Products. The life test of JPL-5 (STR/STR) converter has continued to establish the integrity of the converter envelope. No failure in the converter com ponents or testing apparatus has been experienced. Over 3100 hours of operation had been accrued by the end of September. -100- TASK 2 - UNINSULATED HEAT PIPE THERMIONIC POWER ANALYSIS (E. J. Britt. M. D. Smith. R. S. Dick Introduction: The objective of this task is to evaluate the pos- siblity of using uninsulated converters in a nuclear space power system for application to electric propulsion. In the current system design a sheath insulator is used to electrically isolate the emitter of each converter in the power conversion array from the heat pipes which transport heat from the reactor core to the array. If the need for this insulator could be eliminated a difficult materials requirement would also be eliminated, permitting a broader choice of operating parameters (emitter temperature, etc.). During 1978 a review of a variety of techniques showed that the use of flux reset or push-pull power coupling was possible, although the latter would probably result in an exces sively heavy system. The most attractive approach appears to be one with series coupled converters utilizing the electrical resistance of the heat pipes between the converter and the reactor core to minimize electrical losses. During this contract period these systems will be examined paramet- rically following development of a converter performance code. The potential and advantages of the uninsulated system design will then be evaluated. Current Effort: Two areas were worked on during this period: (1) modification of the JPL NEP computer program and (2) parametric mapping of specific weight and efficiency. The JPL NEP computer program was modified to include calculation of the optimum electrical lead size for connection to the emitter. In optimizing the efficiency of a thermionic converter two major parameters must be considered; the heat loss by conductance through the load and the voltage loss by resistance in the leads to the load. Fig. 55 shows the heat flow and voltage drops in a typical thermionic converter. -101- EMITTER COLLECTOR MAIN HEAT FLOW "^\ AT = T^ - Tc } ° HEAT AV SHUNTED AROUND LEAD CONVERTER THROUGH THE K\ LOAD LOAD LOAD Fig. 55 Converter Energy Flow Diagram If the lead region is made large in order to minimize the voltage drop, then most of the heat input is shunted around the converter. If the lead region is made small in order to minimize the heat conduction, then the voltage loss in the lead is increased. In both cases the efficiency approaches zero. There must be some optimum lead design which produces maximum efficiency for any set of converter operating conditions. The key to lead optimization lies in the fact that most refractory metals obey the Lorentz relationship: Thermal Conductivity Electrical Conductivity = LT, (1) where L is the Lorentz coefficient (almost a constant), and T is the average of Tr and T^. -102- {39) Unfortunately L is not completely constant among various metals^ ' and -8 2 may range from 1.6 to 2.5 10~ x (volts/K) . For this study the theo retical value was used: PK/T - 7T^k^/3e^ = L = 2.44 x 10"^ (volts/K)^. (2) This value is typical for most refractory metals but the designer using this study should be careful that the L value of the specified lead material does not vary widely from the theoretical value. The methodology of this report is to (1) define the efficiency of a thermionic converter; (2) rewrite the thermal conduction terms of the efficiency equation into electrical resistance terms using the Lorentz relationship; (3) differentiate and set dri/dR = 0 to find optimum resis tance; (4) use the optimum resistance thus found to calculate the optimum efficiency; and (5) use an HP9825A computer to plot optimum lead design for various operating conditions. The efficiency of any thermal device can be defined by: _ Power Out '^ Power In In a thermionic converter the power out is simply the electric power at the load minus the power lost in the leads. In terms of current density J (A/cm^): IV Power Out _ o lAV ,., ,.., A^ A^ A^ " '^^o • ^^^' where V is the electrode voltage, AV is the voltage drop in the lead, Ap is the area of the emitter, and I the total lead current. The power input is determined by a first law analysis of an iso thermal control volume surrounding the emitter. In terms of heat flux q 2 (watts/cm ) an energy balance for one square centimeter of the emitter is: -193- ^in ~ ^electron cooling "^radiation ^lead conduction ^ cesium convection The electron cooling term in this case is: Electron cooling = J(t)r + J(2kTr/e), where (^^ is the emitter work function, k is the Boltzmann constant, T^ is the emitter temperature, and e is the electron charge. The radiative heat transferred is given by: q^ = a£(T—/ £ 4 - TQ4 )\ . cf is the Stefan-Boltzmann constant, e is the effective emissivity of the emitter and collector. In this study the threee convective and radiative terms are combined into the empirical equation: p ^= 1.636 X 10"^J T^ + 1.090 x 10"^^(T^^ - T^."^). (3) This equation is a reasonable fit to observed data and obviates the need to find -104- ^con = —^— 1/2 J R* where T is an average temperature of the lead, (Tn + Tp)/2, and AT the difference (Tr - Tp). The total energy required per converter, per unit area of the emitter is: p. ^ = 1.636 • 10"^ J T^ + 1.09 • 10"^^ {^^ - 1^) + LTAT J^R* [^* ~ 2 » 0'^ P P — 2 lin _ Icr . LTAT J^R* ^E A^ R* • 2 • The total output is: Pout = "o - V2 l2 R^. which becomes: p = JV„ - 1/2 I J R, , and -A^p 0 L substituting R. = R*/Ap this becomes: P 2 ^out _ ,., J^R* -j^ = JVQ - -y- The efficiency defined earlier is now: JV„ - J^R* 0 n = ^cr ^ LTAT J^R* Ag R* -105- The relationship for V represents an ideal diode modified by inclusion of the barrier index V. : VQ = kT^ Jin (AT^^/J) - V^,. (5) This substitution will not be made since it would unnecessarily com plicate the differentiation of Eq. (4). Setting the differential equal to zero will yield the maximum: A -J^(P.., - '^J^R* + LTAT/R*) - (JV„ - J^R*)(-^^J^ - LTAT/R*^) '^^* (P^^ - ^J^R* + LTAT/R*)^ Solving for R* yields the following quadratic equation: R* = 2JLTAT ± V(-2JLTAT)^ - 4(-JPc^r + jV)(0 V 0 LTAT) _ (gj °Pt 2(-JP^^ + J^V^) The optimum lead resistance is now uniquely determined by J, Tp, Tj,, and the barrier index implied in V . Substituting R* . back into Eq. (4) gives the maximum efficiency for a given set of operating condi tions. 0 opt /-7\ P^ + I'TAT/R*^p, - ^aj'R*opt ^E Two important design parameters can be obtained from the above equations: JR* Lead Voltage Loss _ opt ,o\ Electrode Voltage V„ ' ^^' ^ 0 Lead Conduction ^ K^ . ,,* ^ ^,. ,qx Total Heat In P.^ ' ^"^^^ ^ ^"^l" ^^' -106- To implement the optimization study a computer program was written for an HP9825A calculator and the 9872A plotter. The lead optimization routine was then entered into a converter characteristics calculation program which produced a plot of the efficiency and specific weight as a function of collector temperature. The computer program was employed to predict converter efficiency and output voltage. Those output voltages were adjusted to account for current losses down the length of the reactor heat pipes. This then determined the system output power and therefore an efficiency. Using that efficiency the mass of the system was adjusted to obtain the proper output power. The relative specific mass was then calculated. Both the system efficiency and relative specific mass are plotted in Figs. 56 and 575. Those plots lead to the following conclusions: (1) current tech nology (Vg=2.0 V) yields a 400 kWe system with a specific mass of 26 kg/kWe; (2) reduction of collector work function to 1.1 eV would yield a 400 kWe system with a 23 kg/kWe specific mass; (3) reduction of the arc drop to V . = 0.1 eV is predicted to yield a 400 kWe system with specific mass of 16 kg/kWe; and (4) the optimum specific mass minimum occurs at a collector temperature greater than the optimum collector temperature for system efficiency. Our next effort will be to incorporate the current loss calculation into the computer program along with an analytic equation for predicting the specific mass of a system. -107- NEP THERMIONIC REACTOR SYSTEM T^ = 1550°K J = 10 A/CM^ ^00 KWE ?. n I. 5.. RELATIVE i.n SPECIFIC MASS 1.0 = 18.8 kg/kWe \ \ \ " CURRENT THERMIONIC TECHNOLOGY- 0.5.. *C = ^-^ 1.6 eV, V 0.4 V \ \ \ \ ^ —h COLLECTOR TEMPERATURE (°K) Fig. 56 Variation of Efficiency and Specific Mass for the NEP Reactor System as a Function of Collector Temperature,Current Thermionic Technology and Two Systems with Advanced Thermionic Converter Performance NEP REACTOR SYSTEM WITH CURRENT TECHNOLOGY THERMIONICS Tr = 1650°K J = 10 A/CM^ m KWE 20 ?.n , 15 1.5.. (j)^ + V^ = 1 .5 + 0.4 •'.MJ , •';i'•'•~^'y•;i!•^i••'^V^•^•:'i•^•^;•i^^^••:-^: S CONVERTER LEAD RELATIVE f EFFICIENCY ^^ "i" • V\V'.'''''>!''N/'/l'i^."''i''''-''Vi''.''''''i''i'iViHi'-rtr ' •'^^V•'•••''M^^^v^'/^^•,.•VVA •^:^:\^;^i ^':•V•^.^/^^^''^i•';''^^•'^^^^V•^'•:VV^^'-^ I '-• ' - ' 1 .. .^1J ' > ' 1 •!. ^li;;^ 1.0 = 18.8 kg/kWe ({)p + V , = 1.6 + 0.4 0.5 0i J..flL in COLLECTOR TEMPERATURE (°K) o I Fig. 57 Efficiency and Relative Specific Mass of a NEP System I—' en with Current Technology Thermionics I (JO I REFERENCES "ZERO - The Worlds Largest Thermionic Energy Converter," L. L. Begg and G. 0. Fitzpatrick, presented at the 14th lECEC, August 1979. "Advanced Thermionic Energy Conversion," Progress Report 9/74-8/75, COO-2263-4, NSR 2-4. "Thermionic/Thermoelectric Energy Conversion Technology Development Program," Vol. 1 - Progress Report: October 1978 - January 1979 May 3, 1979, 730-49, Jet Propulsion Laboratory. "Thermionic Power Plant Design Point Selection: The Economic Impact," G. 0. Fitzpatrick and E. J. Britt, presented at the 13th lECEC, 1978. "Design Study of a Coal-Fired Thermionic-Topped Power Plant Using an Advanced Boiler," G. Miskolczy and A. E. Margulies, presented at the 14th lECEC, 1979. J. Morris, NASA Lewis Research Center, private communication. "Advanced Thermionic Converter Development," F. N. Huffman et.al, presented at the 11th lECEC, Stateline, Nevada, September 1976. "A Summary of USSR Thermionic Energy Conversion Activity," N. S. Rasor, presented at the lECEC, August 1978. "Development of a Thermionic Reactor Space Power System, Final Summary Report," Gulf-GA-A12608, UC-33 Propulsion Systems and Energy Conversion, June 30, 1973. "Advanced Thermionic Energy Conversion, Joint Highlights and Status Report," Rasor Associates, Inc., COO-2263-14, NSR 2-14, January- March 1979. "A Study of Cesium Oxygen Coadsorption on Metallic Substrates," J.-L. Desplat, NSR 6-2, NASA CR-152272, May 1979. J.-L. Desplat and C. A. Papageorgopoulos, "Interaction of cesium and oxygen on W(llO). I - Cesium adsorption on oxygenated and oxidized W(no)," NASA rep. CR-152272, May 1979. Accepted for publication in Surface Science. L. R. Danielson and L. W. Swanson, "High temperature coadsorption study of zirconium and oxygen on the W(IOO) crystal face," JPL contract 955156, Feb. 1979. Accepted for publication in Surface Science. C. A. Papageorgopoulos and J.-L. Desplat, "Interaction of cesium and oxygen on W(llO). II - Codeposition: cesium oxide formation," NASA rep. CR-152272, May 1979. Accepted for publication in Surface Science. -110- J. J. Uebbing and L. W. James, Behavior of cesium oxide as a low work function coating," J. of Appl. Phys., 41, 4505, 1970. Metals Handbook, Vol. 8, American Soc. for Metals, Metals Park, Ohio, 109, 1973. P. E. Gregory, et al, J. Appl. Phys., 46, 3525 (1979). J.-L. Denisot, CEA Report R-4508, Saclay (1973). T. Alleau and M. Bacal, Thermionic Conversion Specialist Conference, Miami (1970), p. 434. J.-L. Desplat and J.-L. Denisot, Proceedings Thermionic Conversion Specialists Meeting, Eindhoven (1975), p. 25. C. Warner, Thermionic Conversion Specialist Conference, San Diego (1971), p. 170. M. G. Burt and V. Heine, J. Phys. C: Solid State Phys. jl, 961 (1978). J.-L. Desplat, NASA Report CR-152272, Ames Research Center, (1979). To be published in Surface Science. D. Lieb and F. Rufeh, Proceedings 3rd International Conference on Thermionic Electrical Power Generation, Julich, (1972), p. 1091. D. Lieb, et al. Proceedings Thermionic Conversion Specialist Meeting, Eindhoven (1975), p. 11. B. Gunther, et al. Proceedings 3rd International Conference on Thermionic Electrical Power Generation, Julich, (1972), p. 1103. I. G. Gverdtsiteli, et al, Sov. Phys. - Tech. Phys. U, 76 (1972). I. G. Gverdtsiteli, et al, Sov. Phys. - Tech. Phys. 21, 1030 (1976). D. M. Young and A. D. Crowell, Physical Adsorption of Gases, Butterworths, Washington (1962). A. V. Gusarov, et al, High Temperature 5^, 584 (1967). Ph. Touzain, J. of Therm. Anal. 9, 441 (1976). J. B. Taylor and I. Langmuir, Phys. Rev. 51, 753 (1937). L. K. Tower, NASA Technical Memorandum NASA TM X-2157, Lewis Research Center, Cleveland (1971). M. N. Arnol'dov, et al, Izv. Akad. Nauk SSSR, Met. publ. 75, issue 4, p. 98. -Ill- Nuclear Space Power Systems, H. M. Dieckamp, Atomics International, Conoga Park, CA, September 1967. The Effects of Fabrication Variables on the Creep-Rupture Properties of Molybdenum-Base Alloys, R. L. Stevenson, Transactions of the Me- tallurgical Society of AIME, 1969. A. Schock, "Optimization of Emission-Limited Thermionic Generators," J. Appl. Phys., 32 (8): 1564-1570 (1961). R. A. Flinn and P. K. Trojan, Engineering Materials and Their Applications, (Boston: Houghton Mifflin Company, 1975). G. 0. Fitzpatrick and E. J. Britt, "Thermionics and its Application to the SPS," in Radiation Energy Conversion in Space, Progress in Astronautics and Aeronautics, Vol. 61, American Institute of Aeronautics and Astronautics, New York, 1978. •112- APPENDIX I ' THE IGNITED DIODE ANALYTICAL MODEL "IMD-4" NOMENCLATURE d = interelectrode gap width (cm) e = electron charge (conlomls) E = electric field (volt/cm) E(o) = electric field at the emitter surface " " Ep = electric field at the emitter sheath- " " plasma interface 2 Jp " Richardson current (Amps/cm ) Jp = net electron current emitted into the " " plasma from the emitter Jp = electron back emission from the collector " " J„r = ion emission from the emitter " " J P = net forward electron current at the emitter " " sheath-plasma interface (computed from boundary conditions) J P = net forward electron current at the collector " " sheath-plasma interface J.p = net forward ion current at the emitter " " sheath-plasma interface J.p = net forward ion current at the collector " " sheath-plasma interface Jp(o)= net forward electron current at the emitter " " sheath-plasma interface (computed from transport equations) J = net current through the diode " " k = Boltzmann's constant (erg/°K) kgT = electron thermal diffusivity -dimensionles k.jT = ion thermal diffusivity -dimensionles Kg = electron thermal conductivity (watt/cm °K) m = electron mass (gm) m. = cesium ion mass (gm) n = electron density (cm ) n. = ion density " n = charged particle density " np = charged particle density at the emitter " sheath-plasma interface Hp = charged particle density at the collector " sheath-plasma interface A-1 n^ = neutral atom density (cm ) P = cesium gas pressure (torr) (degrees/cm q = electron kinetic energy flux (watt/cm Q = total electron energy flux " " Q P = electron energy flux at the emitter " " sheath-plasma interface (calculated from the boundary conditions) Q P = electron energy flux at the collector " " sheath-plasma interface Qg(o)= electron energy flux at the emitter " " sheath-plasma interface (calculated from transport equations) Rio ~ ''0" ^'^='9 ^^^™ (volt/cm 3 S = ion source coefficient (cm /sec T = electron temperature (°K) T. = ion temperature " T = neutral gas temperature " Tgp = electron temperature at the emitter " sheath-plasma interface T p = electron temperature at collector " sheath-plasma interface T^ = emitter temperature " Tp = collector temperature " Tp = cesium reservior temperature " U^ = initial energy of ions entering the emitter (erg) sheath from the plasma U = initial energy of electrons entering the (erg) emitter sheath from the emitter V = electron thermal velocity (cm/sec) v'. = effective ion velocity at the " " 1 sheath-plasma interface V = electron potential (volts) Vp = emitter sheath height " Vf^ = collector sheath height " V = plasma drop " V. = arc drop " A-2 double sheath barrier height (volts) output voltage cesium ionization potential spatial coordinate in interelectrode gap (cm) recombination coefficient (cm /sec) 2 electron particle flux (= Jg/g) (1/cm -sec) ion particle flux (= J-/ ) 1 e electron particle flux at emitter sheath-plasma interface (= Jgc/g) electron particle flux at collector sheath-plasma interface (= Jgr/p^ ion particle flux at emitter sheath-plasma interface (= Ji^/e) ion particle flux at collector sheath-plasma interface (= Jic/g) emitter sheath attenuation -dimensionless- factor collector sheath attenuation -dimensionless- factor electron mobility (cm /volt-sec) ion mobility II II II field-free emitter work function (volts) effective emitter work function collector work function A-3 APPENDIX I THE IGNITED DIODE ANALYTICAL"MODEL "IMD-4" ALGORITHM The ignited mode thermionic converter is described by a set of five transport equations and their associated boundary conditions. The trans port equations are those derived by Wilkins and Gyftopoulous. Charge neutrality in the plasma is assumed, so that n = n.. The particle fluxes are ?. = -Uol^T^Vn + nt + (1 + kl)n^ V TJ (1) e ^e|_e e e e ej for the electrons, and r, = -yJ^T.Vn- nt + (1 + kLn^VT, - 1 1 (2) for the ions, where the ion drag term ^. is le ^•e " -Si^ATgVn + nt) + (g^ - g^)nkVTg. (3) Here g, and g^ are functions of the electron collision frequency and are evaluated by Wilkins and Gyftopolous. The kinetic energy flux q is given by q = t (5/2 + k"^)kT^ - K VT . (4) ^e e e' e e e The total energy flux is just where V is the electron potential. The ion source equation gives Vo f = V° r. = Sn n - an% (6) e 1 a where n is the neutral particle density and S and a are the ionization d and recombination coefficients, respectively. Completing the set of differential equations is Voq =-rot-V.Vo? + AE . + AE , . (7) ^e e 1 e rad el as A-4 Here AE . and AE , represent energy transfer by radiation loss and elastic collisions, respectively. These terms will be ignored in this analysis. f Combining Eqns. (7) and (5) gives Vo^g = (V - V.)Vo?^. (8) In this analysis, all equations will be solved in one dimension only. The boundary conditions at the emitter for an electron retaining sheath are Jc n V (0) / eV re.E = ^-^^^exp-T-—J ^^P-kT ^ , (9) eE, 3E_ / ^W np v^ ^•E = f-P -WT --^-^ ' no) and "eE ' T <2 "E * ^E * ^P' - (T - I-eEJl^ "^j * V, . Vp). (11) Here the zero of electron potential is taken at the collector edge, so that Vp represents the change in the electron potential in going from the collector to the emitter. The ion emission from the emitter, J p will be neglected. The ion velocity v.'(O) is the ion Bohm velocity evaluated at the emitter edge. The collector boundary conditions are _ n, Vg(d) / eV,X J, ^eC 4 ^^P (- kt;^ ) - T ' n^^ n^.'(dlip V.I ) r,ipC = -^2— , (13) and 5eC = (''eC^T)(2"eC*V-^<2^Tc*''c)- t^"' A-5 In order to treat either the Schottky effect or a double sheath barrier at the emitter, which both modify Jp, the electric field at the r emitter must be evaluated. Solving Laplace's equation in the emitter sheath results in the McKeowen equation: U 2 _ + E(0) Sire /? (-r^E) Mw^ nk T eV, eE 1 - exp (- ^^f (1 kTe E e ^Jm. where f(x) = /I + x - A". (1 The terms U and U_ represent the initial energies of ions arriving from the plasma and emitted electrons, respectively. The final equation necessary for the solution of the problem is the conservation of current e e (1 Preparing to integrate the equations in one dimension, Eqs. (1), (2), (4), (6), and (7) must be solved explicitly for dJ„ dT^ . ,,, dQ^ e e dn^ dV . ^e Eq. (6) gives dx ' dx ' dx' dx' ^"° dx dJ ^= e (Sn^n - an^. (1 Combining (4) and (5) gives T dT^ Q = r [(5/2 + k')kT + V] - K -r-^ ^e e e e e dx (1 A-6 ^^e Solving for -j-^ yields Jg [(5/2 + k;)kTg + V] - Qg (20) dx K. Eqs. (1) and (2) can be used to eliminate E, giving T. _ k dT. r^ r, (1 + k')nl^ -k dT (1 + k'.)n^-r^ _e + _L 4. e' e e ^ ^ i' e dx le dx dn ^e ^ dx (21) ^e ^(T e + T.1 ) Using Eq. (17) to eliminate r. and rewriting gives J dn ^/j_.X\.JJ_,(i,kT) n^^ dx e IPg \i. I e \i. ^ e' e dx T k ^T,- + (1 + k')n^-7-^ - R. /-(T + T.) (22) i' e dx le e^ e 1 Solving Eq. (1) for E yields rJ_ , kT (23) dx ^ e ]in e n dx ^ e' e dx Finally, (8) gives dQ, dJ^ (24) dx ^^ ^•^ dx The coefficients S and a in Eq. (18) are evaluated as functions of T and n based on the results of Norcross and Stone. The ion tempera ture T. is assumed to be equal to the neutral gas temperature, which varies linearly between the electrode temperatures. Thus, (25) 'i 'E ^'C 'E^ d ' and ^- VJE (26) dx d A-7 The neutral density n^ is given by the ideal gas law; a P = n kT^ + nkT. + nkT , (27) which can be rewritten as "a = kTT - " (l * TT)- (28) Here the relationship T. = T has been used. 1 a The ion drag term R. is found by using Eqs. (3) and (1) to give /j T I. dT\ . dT The electron transport coefficients u , k and k are evaluated e e e using the results of Stoenescu and Heinicke, for Nighan's best guess elastic collision cross section. The coefficients y., k., g,, and g^ are evaluated according to the formulation of Wilkins and Gyftopolous, -13 2 -14 2 using a cross section of 1.0 x 10 cm for the ions and 3.5 x 10 cm for the electrons. Eqs. (18), (20), (22), (23), and (24) are integrated using a standard fourth-order Runge-Kutta algorithm, using a step size of d/50. II. COMPUTATIONAL ALGORITHM The computational model operates by specifying a value for J, the output current density, which specifies a position on the volt-ampere characteristic. The output voltage, V , is the value to be calculated. Values are also specified for the emitter temperature Tp, the collector temperature T„, the reservoir temperature T^, the cesiated emitter work function ({jr, the collector work function (J)-, and the interelectrode spacing d. The saturation and back-emission current densities are calculated by the Richardson-Dushman equation: A-8 J,= 120T^2 exp(-^), (30) and p / *r \ r J, = 120Tc^ exp {-^J. (31) The gas pressure P is calculated from the cesium reservoir temperature by p = saiijo! exp (- ^Y (32) which is an empirical formula. The system of equations is solved using the shooting method. This method involves guessing those values which are unknown is the boundary conditions at one side, integrating over to the other side using a standard stepping technique, and iterating on the unknown values until the boundary conditions at the far side are satisfied within a specified degree of accuracy. This method is useful when the differential equations to be integrated are strongly nonlinear. In this model, integration is begun at the collector edge and proceeds toward the emitter. The values to be guessed are n- and T p. The iteration on T „ is rested within that for n-. To begin propagating the equations across the gap, values of parameters at the collector edge are calculated first. The ion and electron velocities at the collector edge are calculated: (33) and /8k(T + T ) ^' (^) V '^- • ^''^ A-9 The collector ion current is given by Eq. (13), enp v.' (d) J., = -^ . (35) The collector electron current is now found from Eq. (17), J^^ = J + J.(,. • (36) A value for the collector sheath barrier factor is also required. This is defined to be h = ^^P 1^- ^ 1' (37) and using Eqn. (12) gives 4 (Jg^ + J^) ^C - en^ Vg(d) (3^) Thus, the collector sheath is M^ = kTg^ ln(l/?^). (39) The energy flux carried by electrons is given by Eq. (14), which can be rewritten for the collector edge as QeC = ^eC^^^T^c" V^^C^^^T^C- V- ^''^ Values have now been either found or guessed for all parameters at the collector side. Values of n^, T p are guessed and J p and Q „ are calculated. The electron potential is taken to be zero at the collector edge. Numerical integration of the differential equations can now proceed. At the end of the integration, values are available for J„(0), T r. ir. V(0)(=Vp)(0)(=Vp), and QQg(0 (0) ca t emitter edge. The emitter boundary conditions are now solved. A-10 The ion and electron velocities at the emitter side are found by /8k(Tp + T^p) r V.' (0) = / ^ §i- , (41) 1 V irm. and V (0) =/ ^ . . (42) » e The ion flux is found using Eq. (10); en^ V.' (0) 0,,= -^i— . (43) Eqns. (17) gives J,E = J + J^E . (44) A value for J^ is now guessed. The value of Jp will be iterated in order to solve either the case of a double sheath barrier or Schottky modification of Jp. Defining % '= ^xp (- r^) • («> and using Eq. (9) gives 4(Jp - J-p) h =- en V (0 • (^^) e e The emitter sheath is given by VE = ^ 1" (^) • (") The electric field at the emitter, E(0), is evaluated using Eq. (15), with A-11 U^ = 3/2 k(T^ + Tg^) (48) and U_ = 3/2 kT^. (49) The value of Ep in Eq. (15) comes from evaluation of Eq. (23). All quantities in the right hand side of Eq. (15) are first converted to Gaussian units, which results in a value for E(0) in statvolts/cm. This value is then converted to the practical units of volt/cm. For the case of a double sheath barrier, Jp is iteratively adjusted by the secant method to make E^(0) = 0 in Eq. (48). The sheath barrier height Vg is found by (50) If the current J is too high for the existence of a double sheath barrier, this procedure will find a negative value for Vn. In this case, a procedure is used to evaluate the Schottky effect. This procedure is similar to that just described, except that a Jp is found using a false position method that will make AJj E Jj - J^ exp f^^MEI) = 0 . (51) where E(0) is again calculated with Eq. (15). The energy flux Q P is now found using Eq. (11). QeE = ^E 2^C^E - V "^eE (2k T^p . V^ + Vp). (52) Where J p was obtained from Eq. (44) and J (0) comes from the numerical integration of Eq. (18). A-12 Two functions are formed which are the bases of iteration on 1-r. and eC np. The first of these is •'(^eC^ = ^eE - %^^^' (53) where Q^p comes from Eq. (52) and Qg(0) is obtained from the numerical integration of Eq. (24). The second function is H\) = Jg£ - Jg(0). (54) where J^p was obtained from Eq. (44) and Jg(0) comes from the numerical integration of Eq. (18). The value of T p must be ajusted to make F(T p) small. In general, F(T p) will be positive for values of T p which are too large and negative for T p too small. This is complicated by the fact that for values of np and T p which are much larger than the converged values the integration of n(x) results in a negative value before reaching the emitter and therefore cannot proceed. This is handled by assigning F(T p) a positive default value when n(x) is less than zero. The value of T p is adjusted in order to find two non-default values for F(T p) which are of opposite sign. The solution of T p is then found by using the secant method. The iterative solution of np, which has the T p interation nested inside of it, proceeds in a much more straight forward way. The secant method is used to find the value of n- which will make F(np) zero. The parameter F(np) is always evaluated after T - has converged. This process will finally lead to a simultaneous solution for T p and np. It should be noted that a problem occurs if the order of the nested iterations is changed. For certain values of T p, no possible adjustment of np will result in F(np) = 0. A-13 Upon convergence, the output voltage is calculated. The effective emitter work function (J)p is given by r (J)p' = 4)^ + Vg (55) in the case of a double sheath, and by (|)p' = (t,p - kTp ln(Jp/J^) (56) in the Schottky region. The arc drop is given by Vd = Vp - V^ + Vp. (57) Finally the output voltage is calculated; \ = *E - *C - ^d- (58) The current J is changed, and the calculation process begins again. A-14