TASK 3. THERMIONIC FUEL ELEMENT DEVELOPMENT

The thermionic fuel element (TFE) development task goal is to provide a thermionic fuel element for use in thermionic reactor power systems to produce electrical power for space and terrestrial applications. The application presently most influencing its development objectives is space nuclear electric propulsion. Detailed information on TFE planning and test schedules to meet these development objectives is given in Ref. 1.

The expected power output from a TFE for long-term applications is 800 watts to 1 kWe. Specific lifetime goals in the development program were operation of a six-cell TFE (6F) for 10,000 hours by July 1974, and 20,000 hours by January 1977. Present lifetime on a 6F has surpassed 8,000 hours, and 8,000 hours has been achieved on a single-cell F-series. 10,000 hours has been achieved on experimental in-core tests and on a 2-cell TFE of smaller diameter.

The principal development areas are emitter dimensional stability and performance stability. The primary objectives of nearly all tests being operated or designed is to provide data relating to one or both of these areas. Each test also provides data on other development areas including fission product venting characteristics, insulator stability, envelope integrity, etc. Development has been underway on TFE design and fabrication procedure modifications aimed at reducing TFE cost. A summary of development work alone in each major development area and present status is given in Section 5 of the TFE Plan, Ref. 1.

The thermionic fuel element development task encompasses the design, engineering analysis, fabrication, encapsulation, and testing of TFEs and components. Fission heated TFEs are tested in the core of the Thermionic Test Reactor (TITR) at Gulf General Atomic. Data summarizing the current status of fuel elements is given in Sections 3.2 and 3.5 and in Appendix A.

3-1

3.1 ENGINEERING AND HARDWARE DESIGN (E.F. Thacher)

The objective of this subtask is to produce drawings and specifications (except materials and process specifications) and engineering analyses in support of the fabrication of development hardware. This includes design of single-cell, two-cell, and full-length elements and of fuel-clad capsules for testing in the TITR.

The approach taken is to prepare designs, to develop mathematical models of the designs and their environment, and then calculate heating rates, temperatures, stresses, strains, and electrical performance. Design versions axe based on the analytical results and test data.

The tables in Appendix A summarize past effort on Subtask 3.1. Major work completed in this reporting period was: (1) conceptual design of TFE 2F4, design of fuel-clad capsule FC-4 (Section 2.6.2), analysis of TFE 6F5, design of the UO₂ fuel press, preliminary design of TFE 2LF2, modification of BRITL code to analyze thermionic emitter distortion, and laboratory F-cell design.

3.1.1 F-Series TFE Design Analysis (E.F. Thacher, D.T. Allen, D.J. Heisser).

3.1.1.1 TFE 2F4

Conceptual design of the fuel for F-series TFE 2F4 was completed. This carbide-fueled device was to be used to test means for reducing emitter deformation. The methods employed were increasing fuel surface-to-volume ratio, improving the vented surface distribution in the fuel, and reducing fuel tungsten content. TFE details, other than for the fuel, were standard for the 2F series.

Figure 3-1 depicts the fuel concepts selected for TFE 2F4. No detailed design work on these concepts was done. The advantages of

3-2

the platelet concept shown in Fig. 3-1 are high surface-to-volume ratio, better distribution of vented surface in the gas-producing volume and a potential for using pellets of differing enrichment to produce more axially uniform volumetric heat generation. The standard geometry fuel was intended to act as a control to help judge the effectiveness of the platelet concept.

Both fuels in Fig. 3-1 contain the same low percentage of tungsten, perhaps £ 1%. The exact percentage had not been selected when work was stopped. The expected effect of lowering the tungsten content below 4% (which was standard) was reduction of the fuel creep strength. Measured creep data indicated that reduction by a factor of 100 might be achieved. A weaker fuel body would accommodate fuel swelling by allowing the fuel to move inward into the central cavity rather than moving outward to increase the cladding diameter.

The dimensions shown in Fig. 3-1 for the platelet concept yield a vented surface-to-volume ratio of 9.9 in.^-1. This can be compared to a surface-to-volume ratio of 1.5 in.^-1 for the standard fuel shape. The pellet thickness shown for the platelet design is about the smallest considered fabricable.

3.1.1.2 TFE 6F5

The relative effects of fuel configuration on stress and deformation at startup were determined from elastic calculations for the "shaped" and the reference fuel forms of TFE 6F5. All of the 6F5 fuel forms were designed with 0 to 0.003 in. interference between fuel and cladding at the design temperature. The analysis assumed an interference of 0.003 in. at 1900°K. The fuel shapes in the 6F5 converter are described in Ref. 2.

Cladding stresses appear to be significantly reduced by the slots in the fuel, but the peak stresses in the fuel are increased. Fuel effective stresses at points

3-4

of concentration are no higher than 494 MN/m² (71,700 psi). The F37 configuration has lower calculated peak cladding stress than the F39 configuration, but the latter has 19% less elastic cladding expansion, indicating that the average hoop stress is lower.

An elastic thermal stress analysis in the plane normal to the axis was performed for both the F37 and F39 fuel configurations. The F37 fuel has a cross-shaped void in the center. The F39 fuel has a small central hole with radial slots from the outside inward. The calculations were made with the computer program FEATS (Ref. 3) and modeled as a case of generalized plane strain. The initial cold gap between fuel and cladding was accounted for by adjusting thermal expansion expressions to zero at the calculated isothermal temperature at which the gap closed.

For comparison of the FEATS results with the stress condition in the axisyimmetric fuel configuration used generally before 6F5, a calculation was made for the axisyimmetric fuel form with the same power and fuel volume fraction and the identical initial radial gap and fuel properties. Resulting temperatures and deformation of the emitter are summarized in Table 3-1.

In comparison to the axisyimmetric fuel form, the D T across the fuel and cladding is 49% greater for F37 and 60% greater for F39 configurations. The fuel-weakening configurations for F37 and F39 result in 23% and 38%, respectively, less emitter expansion than the axisymmetric standard fuel shape.

The worst stress condition for F39 appears in the emitter-cladding interface at the outside corner of the radial slot. This appears to be an artificial result of the computer program’s requirement for continuity from cladding to fuel. The high tensile and shear stresses indicate that, in fact, the fuel and cladding will slide on each other as the TFE is started up.

3-5

The value for maximum cladding stress in the F39 configuration is probably closely coupled to this unreal "locking" of fuel and cladding. Relief of the shear stress along the fuel-cladding interface would result in lower stresses in the cladding. Unlike the axisymmetric configuration or F37, the F39 fuel will be forced to move along the fuel-cladding interface during thermal cycling. This motion will serve to reduce stresses in the cladding below the values calculated.

For both shaped configurations, the maximum fuel stress is located in the notch of the radial cut. Table 3-2 summarizes the maximum calculated stresses in fuel and cladding for the three configurations studied.

The lower cladding stresses in F37 and F39 indicate the effectiveness of weakening the fuel. The higher fuel stresses in F37 and F39 result from stress concentrations caused by the grooves.

For the F37 configuration maximum effective stresses increase 131% in the fuel and decrease 69% in the cladding compared to the axisymmetric configuration. The F39 fuel as modeled has 56% higher maximum fuel stress and 61% lower clad stress than the axisymmetric form.

5.1.2 TFE Cost Reduction

3.1.2.1 UO₂ Fuel Press

Design of a UO₂ fuel press which will reduce the UO₂ fuel fabrication cost was completed. Figure 3-2 is a sketch of the fuel press.

The knock-down design of the fuel press die allows a quicker removal of the completed pellet from the press than does the conventional solid die configuration, thereby reducing fuel fabrication time.

3-7

The center pin is used with the die to form a central hole in the fuel and reduce subsequent machining operations. The center pin is made more easily removable by hardening it so that it deforms elastically less than the fuel. Fuel "spring back" releases pressure on the pin when the die segments are knocked away.

5.1.2.2 TFE 2LF2

The proposed design for TFE 2LF2 differs from that of 2LFl. The stem assembly of 2LFl is essentially a modified F-series design; the stem assembly of 2LF2 has been redesigned completely to reduce its cost and to reduce the amount of heater power necessary to control the cesium reservoir temperature. The insulator seal has also been redesigned for lower cost and the lead adapter trilayer section has been eliminated. Figure 3-5 illustrates all of these changes.

In order to simplify the cesium reservoir design and eliminate as much machining as possible, the reservoir is made of a piece of niobium tubing which is pinched off and welded at its upper end after cesium loading. The cesium vapor transfer passage is a tantalum tube pinched off at its upper end. This tube is welded to the reservoir bottom plug and the resulting assembly is welded into the reservoir tube. When the reservoir assembly is EB welded to the manifold block, the lower reservoir wall forms a vacuum wall insulator to keep the transfer tube hot and prevent cesium condensation inside it. The reservoir assembly is designed to maintain the cesium liquid vapor interface as the coldest surface in the TFE.

Preliminary calculations indicated that reservoir heater power consumption will be about 25 watts. The total heat leakage into the stem assembly from the TFE will be about 103 watts, (about double the leakage of the short stem F-series TFEs) with 17 watts of this total going up the reservoir stem and 86 watts up the positive current lead. Significant reduction in total leakage can still be made. Our means for doing this would be reducing the positive current lead thickness.

3-10

The fission product port is heated by conduction from the manifold block through the helium gas. The port and fission product exhaust tube are configured so that the port entrance temperature is 50 - 100° C hotter than the surface of the liquid cesium, while the port exit is at least as hot as this surface. The design objective is to prevent blockage of the port by condensing volatile fission products on the inside surface of the section of fission product tube upstream of the port and to keep cesium from condensing inside the port itself.

The positive current lead, no longer integral with the reservoir as in the F-series TFE, has about 40% of the electrical resistance of the F-series lead over its length with the dimensions shown on Figure 5-3. The temperature of the Cu/Nb joint is about 600°K, with 86 watts flowing into the positive current lead.

The 2LF2 insulator seal was redesigned to reduce cost and to keep the stress in the flexing member below yield. Lowering seal cost was accomplished by replacing the machined convolution with a commercially fabricated bellows and by using a simply configured brazed seal. This concept is illustrated in Figure 3-4. A savings of about $1,000,000 over the standard F-series seal for 1000 units was estimated, based on a bellows unit cost of $75,00. The low bid for the bellows manufacture was $18/unit for lots of 100, indicating that the savings may be even larger. Owing to project termination, further investigation of the low bidder’s understanding of bellows fabrication with niobium was not initiated.

5.1.3 Modification and Use of BRITL Computer Code to Analyze Thermionic Emitter Distortion

To expand and update previous work on fuel-clad interaction and clad distortion (Reference 4), a one-dimensional (radial) analysis of TFE fuel and emitter was performed including the following phenomena:

3-12

Fuel: Restructuring due to vaporization and condensation and surface diffusion, volume swelling, and both thermal and radiation induced creep.

Clad: Volume swelling and elastic-plastic-creep deformation due to fuel contact pressure.

The calculations were performed with the help of the BRITL computer program (Reference 5), which had been developed to analyze UO₂ fuel pins in the gas-cooled fast reactor. The program was modified to incorporate the latest creep equation for tungsten and a fuel swelling model believed to be more accurate in the temperature range of interest for thermionic emitters.

For fuel swelling, the model proposed by Warner and Nichols (Reference 6) was used. Since the burnup levels were less than 4 x 10²⁰ fissions/cm³, only the initial swelling rate, which is independent of burnup, was used. The dependence of swelling rate on temperature was calculated by interpolation between the initial slopes of the five curves in Reference 6.

The clad creep model used was

(-13.071 + 1.185 lns ) (-120500/1.986 T),
? = l.955 x 10²² s e

where ? = creep rate (in./in./hr),

s = equivalent stress (psi), and

T = temperature (°K).

This model is the best fit of ORNL and TRW data and the results of the biaxial creep test (Reference 2). Since the model deviated considerably at stresses under 400 psi, the creep rate below 400 psi was assumed linearly related to stress, equal to zero at zero stress.

3-14

Initial calculations were performed for a reference case identical in dimensions, fission power distribution, and clad temperature to those estimated for TFE 1F1. The deformation rate calculated under these conditions was about 1% of that measured during irradiation of 1F1. It does not appear that such a large discrepancy can be explained by errors in the model or in estimating the operating conditions. The most likely explanation is that small thermal cycles (±10°K) in the course of testing increase hoop stresses in the clad periodically. Because of the highly non-linear dependence of creep strain on stress, even short periods of time at high stress levels can produce a distortion of the clad very much larger than that which would result from the continuous slow swelling of the fuel due to fission gas.

To evaluate the effect of thermal cycles, the swelling rate of the fuel was increased by factors of 100 and 1000. These rates of fuel swelling correspond to rates of temperature rise of 7 and 70° K per hour.* The results of these calculations are given in Table 3-3.

While these results are not rigorous (they are based on equilibrium swelling and deformation rates and do not include the buildup in stress levels when the temperature begins to rise), they do indicate the potential for a large deformation of the emitter due to a repetition of small thermal cycles. The effects of these thermal cycles are expected to be cumulative, since the fissioning fuel cannot support significant tensile stress when the temperature is reduced without forming voids. While the fuel temperature is near the lower extreme, the voids will be closed by vaporization and condensation and surface diffusion so that dense fuel will be present to stretch the clad when the temperature is again raised.

___________

*Since the fuel has a higher thermal expansion coefficient than the clad, there is a net growth in the dimensions of the fuel relative to those of the cladding when the temperature is increased. In terms of stress and deformation of the cladding, the effects of fuel growth from differential expansion are expected to be indistinguishable from the effects of the same growth due to swelling.

3-15

Several cases were run with the BRITL program varying clad thickness, emitter temperature, distribution of power density in the fuel, and the diameter of the central hole. They are compared with the reference case, TFE 1F1, in Table 3-4. The parameters of 1F1 were 1 mm clad thickness, 1780° K emitter temperature, a power distribution peaked at the outer edge, and a central hole diameter of 3.05 mm. While the distortion predicted for 1F1 is only 1% of that observed, the relative effects may be indicative of the importance of the parameters varied.

The effect of thickening the cladding is generally as expected although the magnitude of the reduction in deformation is somewhat larger than observed in 6F3. The results emphasize the important effect of emitter temperature on distortion. Modifying the power density through the fuel to a uniform profile (simulating fast flux operation) resulted in no change in clad deformation. It had been expected that the reduction in fission rate next to the clad would reduce deformation somewhat.

The unexpected increase in clad deformation from increasing the central hole diameter may possibly be due to the difference in temperature sensitivity between fuel swelling and creep models. As the hole diameter increases the fuel mean temperature decreases (for constant heat generation). This lowers both the creep and swelling rates. If the creep rate decreases significantly while the swelling rate does not, the clad will experience a larger contact pressure causing increased deformation.

3.1.4 Laboratory F-Cell Design (E.J. Steeger)

The "Laboratory F-Cell" is an F-series thermionic diode with its emitter heated by an electron bombardment heater. The purpose of the cell is to investigate the thermionic performance of the F-series design

3-17

in a radiation-free environment and without the complicating effects of fuel swelling, fission production evolution, etc. The cell was designed to operate over a wide range of temperature by controlling collector temperature with a heater and cooled niobium collector heat sink.

The GGA thermal analysis computer program, TAC2D, was used to determine the thermal performance and the computer program TACTIC was used to determine the performance with electron cooling. The axial temperature gradient along the cell was varied to simulate the gradient along a fueled cell assembly by introducing a stepped gas gap between the collector and the collector heat sink.

A thermal stress analysis was made with particular emphasis on the end closures of the collector heat sink to allow for thermal expansion between the collector and heat sink without over-stressing the closures.

3-19

3.1.5 Evaluation of Work

Efforts to reduce the cost of TFEs through better tooling (UO₂ fuel press) and improved design (TFE 2LF2) are important to increase the applicability of thermionic fuel elements. Cost is a major factor for both aerospace and undersea applications. Significant progress in this area was made during the last four months.

The work in fuel-clad modeling indicated that thermal ratcheting coupled with fuel swelling may be more significant than fuel swelling alone in causing distortion of UO₂ fueled emitters. Further work in modeling the thermal cycles typical of those observed during testing is necessary to evaluate this mechanism.

3-20

3.2 TFE FABRICATION (H. Horner, J. Kay)

Thermionic fuel elements which incorporate the technology required to support long-life testing in TRIGA are fabricated under this task. The fuel elements are prototypic of those planned for use in thermionic reactor applications.

The technical approach used applies developed technology and technology being developed under Task 3.3 to the fabrication of thermionic fuel elements incorporating one, two, or six thermionic converters assembled in series. The F-series converter is the unit cell for these fuel elements and is produced at a constant rate. Fuel elements are completed as the cells become available. Materials and process specifications are prepared within this task.

Prior work in fuel element fabrication resulted in completion of 50 F-series cells made into 12 F-series TFEs. Three two-celled E-series fuel elements were also fabricated. The technology used for fabrication of F-series and E-series is an extension of technology used earlier to fabricate C-series converters for the unit cell reactor concept and single cell test converters, called Mark VI cells, that were built and tested to demonstrate the feasibility of long-term operation in a reactor environment. This work is summarized in References 7, 8, and 9. The total cells and TFEs fabricated arc summarized n Tables 3-5, 3-6, and 3-7.

3.2.1 Recent Fabrication Accomplishments

Since the last thirdly reporting period, TFEs 6F6, 1F5 and 2LF1 were completed bringing the total number of F-series TFEs fabricated under this contract to 15. The sixteenth TFE, 6F7, was assembled into a sheath tube and readied for final sheath tube bonding, degassing and cesium loading.

3-21

Four emitters were loaded with UO₂ fuel forms of differing geometries to satisfy requirements for fuel-clad capsule FC-3. Work on emitters for fuel-clad capsule FC-4 was initiated.

An emitter and cell components were prepared for the F-series laboratory cell. All components were machined except for finish machining of heater wire grooves in the collector, heat sink body. All components for this device were boxed separately so that easy continuation of work would be possible.

Ceramic-to-metal seals were prepared for TFE 2LF2 and the emitters were approximately 80% completed.

Emitter fabrication for NASA clad-capsule CC-l and for NASA Exhibit B, D and D/2 sized capsules was initiated by preparation of molybdenum mandrels required for deposition of tungsten. Twenty mandrels were fabricated and processed, and deposits were prepared for six D sized fuel cups.

During this work period, eight emitters for F-series cells were completed bringing the total F-series emitter fabrication during this program to 58 excluding the four fabricated for fuel-clad capsule FC-3. An additional eighteen F-series emitters were in work at various stages at the time of program cancellation.

Seven trilayers were completed during this period and an additional eight were in various stages of work at program cancellation.

Eleven ceramic-to-metal seals were completed for device application. Of these, four were of the configuration required for the LF-series TFE. An additional batch of eight seals was processed to the point of readiness for assembly brazing.

Program close-out work in fabrication (other than the technical work described above) involved the disposition of TFE components and material,

3-25

disposition of records (history packages), disposition of control paper work, and disposition of AEC Equipment.

3.2.2 Evaluation of Work

Efforts in thermionic fuel element fabrication over the past several years have resulted in the development of a controlled system capable of repetitive fabrication of TFEs of the F-series design with complete end item documentation. This controlled system is expandable to handle the numbers of fuel elements required for a thermionic reactor experiment and other thermionic reactor fuel element fabrication tasks. The features of this controlled system are described in the Thermionic Reactor Quality Assurance Plan, Appendix B, and the Q.A. Plan provides a time period for implementation of Q.A. functions as a Thermionic Reactor Experiment program would evolve into test and ground prototype reactor construction.

During the past year, several component fabrication areas that caused schedule and cost problems were subjected to close scrutiny and process or equipment changes were implemented or initiated. Some of these efforts were related to development of components of modified design for the LF series TFE and are discussed in Section 3.3, TFE Fabrication Development. Others are discussed below.

A severe trilayer contamination problem caused levels of oxygen in niobium components of the trilayers to be in the range of 300 to 600 ppm after gas pressure bonding. Yields of only 50% of acceptable parts were suffered because of this problem. Emphasis was placed upon protection of the components during gas pressure bonding and upon an improved plasma spraying technique involving automatic part rotation and gun traverse motions and use of water cooling inside the sprayed substrate. Substrates were increased in length by a factor of 2.5 to improve spraying and gas pressure bonding economies. Swaging of a niobium sheath over the sprayed

3-26

trilayer substrate prior to gas pressure bonding was eliminated (to avoid gas entrapment) and replaced by an assembly technique using ring inserts for welding at the ends of the trilayer blank. These changes in processing were not fully evaluated; however, the yield of trilayers had increased to about 75% acceptable for TFE use at the time of contract termination.

Maintaining schedule on chemical vapor deposition of tungsten emitter blanks (H₂ reduction of WF₆) has been demanding with a single emitter deposition rig. A CVD tungsten deposition rig capable of producing three emitters in one deposition run was made fully operational during the year. The production runs on the rig resulted in components containing 3-10 ppm flourine and deposition efficiency increased from approximately 35% to approximately 60% utilization of WF₆. Approximately one-half as many man hours were used for each billet with the three-emitter production rig as with a single emitter deposition rig.

Schedule delays and high unit cost problems were eased for ceramic-to-metal seal fabrication by installation of seal production equipment in the TFE fabrication facility. Major pieces of equipment consisted of a tungsten mesh element hydrogen metallizing furnace capable of metallizing batches of 20 ceramic rings in wet hydrogen at 1850° C, a 1200° C air firing furnace facility for metallizing preparation, application and seal mechanical assembly, and a seal brazing furnace. The brazing furnace was capable of automatically programming batches of six seals through the complex brazing cycle (Ref. 10) required for vanadium-niobium seal brazing at 850° C without human intervention or control. Brazing cycles were completely reproducible and two brazing cycles could be made per shift. Seal production capability per man hour for brazing and metallizing operations was increased by a factor of about 10 by installation of the equipment and the possible production rate was increased by factors of from 12 to 20.

3-27

A controlled plating system designed for the specific geometry of F-series sheath tubes was placed into service during the year. The facility has provisions for repetitive nickel plating of sheath tubes (using a process covered by a written specification) so that a precise quantity of nickel can be applied to the tube. The facility and procedure have reduced plating rejections and increased plating quality with respect to adherance, thickness, appearance and thickness uniformity.

Efforts in the process modifications described had reduced the work content of affected operations and increased product quality and reproducibility. Process specifications for seal fabrication, sheath tube nickel plating and tungsten fabrication using the three-emitter rig are a part of the records prepared for storage as part of the AEC contract termination.

Fabrication of TFE 2LF1 has demonstrated the effectiveness of the low cost features of the design, and the fabricability of this TFE concept.

3-28

3.3 TEE Fabrication Development (J. Grebetz)

The objective of this task is to develop the fabrication technology required to construct thermionic fuel elements. The approach is to apply the most recent advances in manufacturing technology in a manner that will lead to the efficient and economical fabrication of thermionic devices, while maintaining their quality and reliability.

Past efforts in this task have included the development of all fabrication techniques and processes and the development of quality control procedures necessary for fabrication of thermionic fuel elements. The work has included: 1) development of thermionic fuel element assembly techniques by well characterized electron beam welding procedures that specified welding fixtures and parameters; 2) experimental characterization of component abilities to withstand the TFE environment in thermal cycling and thermal stability testing; 3) development of machining procedures for TFE components including tungsten emitters, ceramic-to-metal seals, trilayers and thin-walled, close-tolerance sheath tubing; 4) ultrasonic testing procedures to evaluate sheath tube-to-trilayer and trilayer ceramic-to-metal bonding; and 5) TFE final assembly and processing techniques including evaluation methods to ensure that final processing does not degrade trilayer ceramic-to-metal bonding.

Work during the current reporting period included final assembly of 2LF1, refinement of a low cost brazing process for ceramic-to-metal seals, and completion of a parametric study of the tungsten/tantalum diffusion bonding process. Work on 2LF1 is covered in a general discussion of low cost component development (3.3.1).

3.3.1 Low Cost Component Development

Techniques for the assembly of the low cost F-series TFEs were defined during the assembly of TFE 2LF1. Work performed this period included definition of the process for preparing low cost copper-nickel brazed ceramic-to

3-29

metal seals, definition of welding parameters and fixtures for "slip-over ring" welding, and TFE final assembly and degassing procedures.

Low Cost Ceramic-to-Metal Seals. Seals were prepared by brazing simple grooved rings to metallized ceramic rings of rectangular cross-section using copper-10wt-% nickel braze foil. The configuration for brazing is described in Ref. 11. Warpage of the ceramic rings during metallizing caused braze gaps greater than 0.007 inch in order to assemble components for initial brazing experiments. Thermal cycling of these components with large gaps after brazing and finish machining caused leakage through ceramic-metal interfaces. Use of lower metallizing temperatures (a reduction of approximately 300°C) and better metallizing fixturing reduced the braze gaps to approximately .001 inch and yielded seals capable of withstanding 5 cycles to 800°C without loss of vacuum or structural integrity. A thermal cycling experiment was performed to determine if the stress lowering convolution machined integrally with the low cost seal would be subject to low cycle fatigue failure as indicated by analytical studies. Fifteen thermal cycles to 850°C were performed upon a convolution operating in a "real life" cyclic stress environment without leakage failure. Eight ceramic-to-metal seals were prepared for 2LF1 and 2LF2 applications.

Slip Over Ring Electron Beam Welding Parameters. Slip ring welding parameters were defined using samples that mocked up the trilayer of the LF series TFE and the thin wall (.010 inch thick) CB 1% Zr slip over ring. Welding was accomplished using a niobium holding fixture to compress the slip over ring uniformly onto the trilayer. Weld parameters were devised that did not affect the metallurgical integrity of the niobium-alumina bond.

TFE 2LF1. Final assembly of 2LF1 was accomplished by a straight-forward application of welding parameters and techniques devised with experimental samples. Assembly consisted of welding units consisting of a lower trilayer and an upper trilayer with an emitter, slip over ring and an insulator sandwiched in between. The weld fixture needed no fixturing other than standard collets to hold the 1.314 inch O.D. trilayers in proper concentric relative position and alignment. Five welds were made for each setup compared to the

3-30

one or two welds made for each setup for F series TFEs.

Low Cost Emitter Preparation using the "double diffusion bonding" technique described in Ref. 11. By using a flat surface rather than one with a tapered joint, a reduction of a factor of six in lapping time was realized in joint preparation and bonding time was cut by one-half because two bonds were prepared during the same bonding run. Work performed on tungsten-tantalum diffusion bonding parameters and reported in Section 2.4 showed that no differences in joint integrity should be expected by employing this new process.

3.3.2 Evaluation of Results

The new development processes refined during the current reporting period proved successful in providing low cost fabrication techniques for the manufacture of thermionic fuel elements. The LF series seal fabrication was simplified, compared to F series, by reducing the complexity of the ceramic-metal joint and by limiting accurate machining to a past brazing operation. The number of setup operations required for TFE assembly was reduced by a factor of three for final welding of the LF TFE compared to the F series TFE.

3-31

3.4 INSTRUMENTATION AND ENCAPSULATION (D. Schwarzer)

The objective of this task is to instrument the TFE with thermocouples, voltage probes, and the high current bus bars. The TFE is then outgassed under high vacuum and encapsulated for testing in the TITR core.

Pre-assembly of the primary containment feedthrough seal and the secondary containment was initiated. The pre-assembly of the required instrumentation through the primary and secondary containment seals permits these units to be assembled and stored, reducing the elapsed time from receipt of the TFE until it is ready for testing in core. In conjunction with the pre-assembly, modified outgassing and encapsulation methods were developed. These new methods were initiated to reduce set up times while maintaining high quality.

During this report period, the instrumentation and encapsulation of TFE 6F5 and capsule FC-3 were completed. Instrumentation for TFE 6F6 was initiated.

3-32

3.5 IN-CORE TESTING (G. O. Fitzpatrick, M. K. Yates, D. E. Schwarzer)

Thermionic devices and fuel-clad capsules were tested in the Thermionic Test Reactor (TITR) at Gulf General Atomic. The objectives of the in-core tests were to demonstrate improved lifetimes through upgrading components and to determine the combined effects on performance of fuel and fission product diffusion and the nuclear environment. Testing included device startup, performance mapping, periodic logging of data, diagnostic measurements, neutron radiographs, and analysis and interpretation of the data obtained.

Thirty-eight thermionic devices were life tested in-core between 1962 and 1973. Table 3-8 summarizes the test hours, longest tests, and failure modes of these devices. The devices were fueled with either carbide or oxide fuel. Considerable progress was made in improving in-core converter lifetimes with the longest tests increasing from 1000-2000 hours in 1965 to 10,000 hours by 1970. The number of devices tested stayed at about the three or four per year level until mid 1971, when it was increased to 10-15 per year. This increase was allowed by the installation of a new core in TITR (coincident with the startup of the first full-length TFE, 6F1) which allowed 10-15 simultaneous tests. This increase in the number of tests resulted in an increase in the information available to the development program and shortened the feedback time. The test operational hours for the period from July 1971 to January 1973 are plotted versus year in Figure 3.5.

The lifetimes and performance of all the prototype F-series TFEs are given in Table 3-9. Table 3-10 summarizes the design and operating parameters and lifetimes of recent test devices and gives the latest information from the neutron radiographs. Complete summaries of these tests as well as of all in-core and out-of-core tests since 1969 are given in Appendix A. The tables given in this Appendix summarize all pertinent operating parameters as well as the current neutron radiograph results.

3-33

Twelve devices were tested in the reactor during this reporting period. They include:

3 Full-length TFEs (six cell), 6F3, 6F4, and 6F5

4 Two-cell TFEs, 2F1, 2F2, 2F3, and 2E2

1 One-cell TFE, 1F4

1 Prototype cell, Mark VIIA IC-D3

1 Experimental cell, Mark VI IC-I6

1 Fuel-Clad Capsule, FC-3

Tests of two new devices, TFE 6F5 and the fuel-clad capsule FC-3, were initiated during this reporting period, and operating and diagnostic correlations were derived for each. Two tests were terminated, TFE 2F1 and Mark VIIA IC-D3. Nine devices were on test prior to the shutdown on January 22, 1973; all eight thermionic devices were producing power at the electrodes.

The test hours achieved with 6F TFEs include 8062 hours on TFE 6F3 at an average power output of 647 watts, 2956 hours on TFE 6F4 at an average power output of 664 watts, and 1379 hours on TFE 6F5 at an average power output of 825 watts. Tests of partial-length TFEs reached 11084 hours on TFE 2E2, 4928 hours on TFE 1F4 with an average 6F equivalent lead power of 821 watts, 8021 hours on TFE 2F1 with an average equivalent 6F lead power of 679 watts, 5574 hours on TFE 2F2 with an average equivalent 6F lead power of 895 watts, and 2956 hours on TFE 2F3 with an average equivalent 6F lead power of 894 watts.

Off-design testing of two devices was conducted during this period. The emitter temperature of IC-I6 was raised from a normal temperature of 1800°K to a value of 1900° K for a period of ~5 hours once each week in order to ascertain the effect of this type of cycle (thermal ratching) on clad swelling. The result was swelling rate increase equivalent to 1/2 mil (0.12 mm) per cycle. IC-D3 was taken open circuit at 280 hours. An emitter temperature of 2300°K resulted. At 330 hours the emitter shorted to the collector, resulting in an emitter temperature drop to 1200°K. Testing was terminated at 1577 hours.

Figures 3-6 and 3-7 show the relative power histories of all current tests. Details of each test history are included in Appendix A.

3-38

Included in this section is a brief summary of results which have been achieved in the areas of performance, emitter dimensional stability, and parasitic discharges. Where applicable, oxide and carbide fueled tests are discussed separately. A summary of in-pile electrical discharge results is also included.

3.5.1 Thermionic Performance

There are two important aspects of thermionic performance which can be examined. These are the initial levels of performance and performance stability of a device.

In general, oxide fueled tests have exhibited consistently good initial thermionic performance and have maintained this performance throughout their tests. Carbide fueled tests have tended to have lower and less consistent initial performance and have exhibited less stable performance during testing. The longest UO₂ fueled tests, IC-I4 (9754 hours), 2E2 (11084 hours), and 2E1 (12535 hours) had stable performance within + 10% (in the case of 2E1 until a short at 7035 hours). The performance of the longest UC-ZrC fueled test, IC-C11 (7881 hours), decreased ~15% over its lifetime. This difference in performance trends is ascribed to the different effects of the fuel constituents which diffuse through the tungsten cladding.

The two most recent carbide fueled tests, TFE 6F4 (2956 hours) and TFE 6F5 (1379 hours), incorporated design changes intended to reduce fuel operating temperature. Both devices performed stably near design levels until test termination.

3.5.1.1. Initial Performance

The initial performance level of a device is based on its electrode power output at the time of the initial performance mapping. The mapping was usually done one to two weeks after test startup when its performance n stabilized. The initial thermionic performance of most of the tests operated

3-41

to date is plotted against emitter temperature at a current density of 7 Amps/cm² in Figure 3-8. The electrode materials, spacing, performance relative to CPOP, and method of determining emitter temperature are given in Table 3-11. TFE 2F1 and 6F2 have not been included since they apparently had special initial condition problems which were not considered representative. In order to provide a better comparison, data acquired for each test is corrected to show the device’s performance with its collector at its optimum temperature. Out-of-pile cell (OC-A4) mapping data was used to make this correction. No correction was made to account for differences in spacing and this factor should be considered when comparing individual tests.

The curve labeled CPOP is a calculated curve which is based on out-of-pile W(110)/Nb electrode data (Ref. 12). The CPOP results were those used in systems and reactor calculations. The most significant source of differences in performance in in-pile tests revealed by this data is the emitter material. Devices with W(WCl₆) emitters performed with a relative power 8% better than those with W(WF₆) emitters. There was also an indication that oxide fueled devices performed ~ 4% better than carbide fueled devices, and cells with molybdenum collectors produced ~ 7% more power than those with niobium collectors. The variation in performance which cannot be attributed to these differences amounts to ± 4%, the equivalent of emitter temperature estimate errors of ± 25°K. This is consistent with the errors expected and indicates the initial performance of the devices is reproducible within measurement errors if certain devices are not included (TFE 2F1, 6F2).

3.5.1.2 Performance Stability

Changes in the performance of the devices tested may be attributed to one or a combination of the following factors:

1. Leaks between the cesium or fission gas spaces and the containment gas (envelope failures).

3-42

2. A leak between fission gas and cesium spaces resulting in fission gas and fission products in the inter-electrode space.

3. Failure of the bonds in the collector-sheath tube structure resulting in high collector temperatures.

4. Contamination of the electrodes during fabrication.

5. Electrode contamination due to fuel component diffusion through the emitter clad.

6. Inter-electrode spacing changes.

Envelope failures were a frequent problem in early Mark VI and Mark VII tests, accounting for fifteen out of twenty failures. They were almost always associated with a braze joint, and consequently, an all-welded design was used in the fabrication of TFEs. The new design solved the problem as indicated by the fact that only one out of eight failures, TFE 2E3, has been the result of an envelope leak.

Leaks between fission gas and cesium spaces were common, having occurred in TFEs 6F1, 6F2, 6F3, 2F2, and 2F3. The sources of the leaks were not ascertained in all cases, although emitter cracking is one observed source. The primary result of a leak of this type is a loss in thermal efficiency. Particularly in the case of oxide fueled devices, such as 6F3 and 2F3, thermionic performance is maintained. In the case of the carbide fueled devices, 6F2 and 2F2, there is evidence that electrode contamination may have been associated with such a gas leak, possibly due to fuel migration through clad cracks (discussed in more detail in a following section). In the oxide devices, where contamination associated with leaks has not been a problem, thermionic performance losses due to gas pressure buildup in the inter-electrode space may be minimized by venting such gases to larger traps through larger ports. To prevent cesium loss, larger reservoirs may also be needed.

3-45

Poor bonds in the collector-sheath tube structure have been definitely identified as a performance change mechanism in TFE 6F1. An early style trilayer employing a pure alumina layer between niobium sleeves was used in that device. Later TFEs incorporated a graded niobium/alumina cermet trilayer structure, which appears to have solved the problem.

Contamination of electrode surfaces during fabrication is a problem that may have been present in two forms. Most TFE tests have initially performed at a higher level than predicted and have suffered an initial performance loss resulting in stabilization with performance near that predicted. The performance loss occurs over a period of one to several hundred hours. This improved performance has been attributed to the presence of oxygen sources within the TFE, possibly due to tungsten oxide residue on the emitter as the result of electropolishing the emitter. Tests of TFEs 1F2, 2F1, and 6F2 had low initial performance which could be improved by short periods of operation with high emitter temperature. This suggested the possibility that the electrodes may have been contaminated during the fabrication process. Work was conducted to ascertain the existence and extent of any such contamination. The possibility of contamination due to handling or the environment (oil, etc.) was eliminated. The possibility of metal spray contamination during the welding process was not eliminated.

Electrode contamination due to fuel component diffusion has been visible in the carbide fueled devices. Much effort has been concentrated on the solution of this problem and a generally satisfactory one has been found in the use of duplex tungsten clads and fuels with closely controlled stoichiometry. The most recent carbide tests, TFE 2F2 and TFE 1F4, indicate a new mechanism for fuel transport to the emitter surface may be important. Thermionic performance remained high and stable past 4,000 hours in the case of TFE 2F2 and 3500 hours in the case of TFE 1F4 but began to fall as the result of contamination after these points. Fuel sintering was observed prior to these times in each test, but swelling to full diameter occurred during the same period as the performance loss. This association between performance loss and fuel swelling confirms observations made with earlier F series carbide fueled tests (Ref. 2). Out-of-pile tests (Ref. 2) indicate the clad stresses resulting from fuel swelling promotes grain boundary void information, which would result

3-46

in an increased fuel transport rate. In order to prevent sintering, TFEs 6F4 and 6F5 were designed to minimize fuel temperatures, in the case of TFE 6F4 by back filling the fuel spaces with a heat conducting gas (40 torr argon) and in the case of TFE 6F5 by providing an interference fit between fuel and clad at operating temperatures. Some sintering has been observed in TFE 6F4 at 2956 hours indicating that gas over pressure may indeed be an effective means of preventing sintering.

Inter-electrode spacing changes appear to have a minimal effect on thermionic performance prior to inter-electrode shorting. A small loss in performance in TFE 2E2 (~10% at 11084 hours) was attributed to spacing changes and there are indications that a similar loss may have occurred in IC-I6. The primary problem associated with spacing changes is the attendant errors in ignition current based emitter temperature estimates. Further discussion of the emitter swelling problem is included below.

3-47

3.5.2 Emitter Dimensional Stability

3.5.2.1 Carbide Fueled Emitters

Fuel Behavior: Fuel sintering has been observed in almost all of the F series TFEs fueled with carbide fuel except in TFE 6F4 and in some emitters of TFE 6F5. During this reporting period:

1. The fuel in the 6F4 emitters showed no evidence of sintering as of 2956 hours.

2. The fuel in emitters 2 and 3 of 6F5 sintered ~ 1/2% by 1379 hours. No sintering was observed in the other emitters.

3. The fuel in 2F2 was sintered ~ 1.0% (on the diameter) at 4195 hours. It had returned to full diameter by 5574 hours and was swelling into the central cavity and stem area.

4. The fuel in 1F4 had sintered ~ 1% on the diameter at 3549 hours. At 4928 hours the top two pellets remained sintered ~ 1% but the bottom two had returned to full diameter.

The observed sintering may be due to an initial fuel-clad gap and the lack of any conducting gas in the fuel cavity (see Ref. 2). Under these conditions heat is transferred from fuel to clad primarily by radiation with a fuel-to-clad temperature differential as high as 500°K. The sintering history data on 6F2 (Appendix A) indicates that there is an enrichment effect on the sintering behavior. The fuel in the lowest enrichment emitters sintered the most and remained sintered the longest and that in the highest enrichment emitters sintered the least.

3-48

From the data on emitter dimensional stability which is discussed in the next section it appears that the sintering of the carbide fuel is detrimental to dimensional stability. Therefore, several approaches are being followed to reduce this problem. These include:

1. Changing the fuel-clad gap to insure an interference fit at temperature (adopted for 6F5 and later TFEs). The presence of sintering in two emitters in 6F5 at 1379 hours indicates that some sintering may still result.

2. Backfilling the fission gas space with a small amount of gas. This technique requires that the gas pressure be low enough to prevent the loss of a significant amount of power if the gas should leak into the TFE, but high enough for good thermal bonding of fuel and clad (adopted for 6F4 and 6F6). Forty torr of argon was used as a backfill gas in TFE 6F4. No sintering was observed as of 2956 hours, indicating this may be an effective technique.

3. Pre-sintering the fuel (planned for a fuel-clad capsule experiment).

Fuel-clad reaction layers were observed in most of the emitters of carbide fueled TFEs where the fuel enrichment is higher than 40%. In general these layers were present at the time of the first radiograph (1200-1300 hours) and did not change significantly after that time. Usually 0.13 mm (5 mils) of clad reacted. Reaction layers were generally not observed in emitters whose enrichment is less than or equal to 20% at times around 5000 hours (£ 0.08 mm of clad reacted). This dependence of the reaction layers upon the fuel enrichment apparently results from the fact that the fission rate near the clad increases significantly as the enrichment goes up. In an actual

3-49

thermionic reactor, the fission power density is uniform over the volume of the fuel pellet. Comparable fission distributions can be obtained in TITR by using low enrichments. As a result of this fact, and the fact that the reaction layer does not include more than about 0.13 mm of clad, this reaction layer is not felt to be a major problem in the TFE development program.

Emitter Dimensional Changes: A comparison of all of the early GGA data on carbide fueled emitters is given in Ref. 2. Data from most of the F series tests is given in Fig. 3-9 as the maximum diametral increase plotted against burnup. The average emitter temperature values refer to the average values at the time of the last set of neutron radiographs. The dimensional changes which have been observed in most of the long-term carbide fueled tests (6F1, 6F2, 2F1, and 1F2) appear to be associated with the return of the fuel pellets to full diameter following sintering. It is postulated that the as-built open porosity of the fuel disappears or becomes closed porosity during the sintering process. Later the fission gas pressure in the fuel builds up since the gas cannot be released, and the fuel begins to swell. The fuel is running at high temperature during this period since it is not generally in contact with the clad. This results in a rapid fuel swelling rate (rates of 1-2% per 1000 hours swelling on the diameter have been observed), and the stresses on the cladding are high enough to cause clad swelling and rupture. Internal gas pressures in the emitters due to plugging of the fission gas vent tubes may be contributing to the swelling problem.

Work is proceeding to solve both the fuel sintering and gas pressure problems. The fuel sintering work has been discussed under Fuel Behavior and is directed at preventing the fuel from sintering into a dense body with little open porosity. Laboratory tests have shown (Ref. 11) that baffles in the fission gas space can trap fission products which might plug the vent tubes or channels. This design change was incorporated in TFE 6F5.

The data from the last two 6F TFEs (TFE 6F4, 2956 hours; TEE 6F5, 1379 hours) appear encouraging. No dimensional changes

3-50

were observed in TFE 6F4 (£ 0.18% on the diameter), and the dimensional changes observed in TFE 6F5 (from £ 0.18% to 0.44%) are attributed to the initial fuel-clad interference fit. These data points were deleted from Fig. 3-9 because of the low values of burnup that were achieved. However, no fuel sintering was observed in TFE 6F4, and, therefore, high fission gas release and improved dimensional stability should result.

No fuel sintering was observed in four of the 6F5 emitters and the same conclusions can be drawn. However, further irradiation time is necessary to confirm these predictions. Two emitters in 6F5 contained fuel slotted in manners intended to relieve stress (Ref. 2). No clad swelling was observed in one of these and £ 2 mils (0.05 mm) in the other, although the other four emitters exhibited significant swelling resulting from the interference fit. Such stress relief slots may serve to prevent clad swelling after sintering has occurred.

3.5.2.2 Oxide Fueled Emitters

Fuel Behavior: The fuel in the oxide fueled TFEs has usually redeposited and a central cavity formed by the time of the first neutron radiograph (1000-2000 hours). After this time very little change in fuel geometry is seen (TFEs 2E1, 2E2, 1F1, 6F3). Cracks are sometimes observed in the fuel, but these are expected to be due to differential thermal expansion on cooling.

Emitter Dimensional Changes: All of the early GGA data on UO₂ fueled emitter dimensional stability was summarized in Ref. 2. Data from most of the current tests is given in Fig. 3-10 as the maximum diametral increase in per cent plotted against burnup. The average emitter temperature values refer to the average values at the time of the last set of neutron radiographs.

The last neutron radiographs of TFE 2E2 were taken at 11084 hours. The difference in the swelling rates between the two emitters in 2E2 is ascribed to a 100°K temperature difference between them. The observed changes are very similar as those observed in TFE 2E1 (Ref. 2), a TFE of the same design as 2E2. No shorting has been observed in this TFE even though the swelling of the bottom emitter has closed 72% of the initial gap (assuming circumferential uniformity).

3-52

The longest test to date of an F-series 1 mm clad emitter is that of 1F1, which was operated 8560 hours. Partial shorting occurred (a 25% loss in electrode power density) at the neutron radiography shutdown at 8228 hours, and the test was removed from core shortly thereafter. The partial short occurred when the gap was 56% closed, apparently because the swelling was eccentric. A plot of the diametral distortion as a function of time and axial location is shown in Fig. 3-11. The point of’ maximum swelling is at or somewhat below the midplane of the emitter, where the temperature is the highest.

The last set of neutron radiographs of TFE 6F3 were taken at 8062 hours. The swelling of the six emitters varied from 0.5 - 0.9% at that time. If the swelling remained uniform, partial shorting would not be expected before 20,000 hours for the most rapidly swelling emitter and before 40,000 hours for the least rapidly swelling emitter. The only quantity which varies significantly from emitter to emitter is the enrichment; however, plotting swelling rates versus enrichment shows no obvious correlation. Differences in emitter fluorine content or fuel stoichiometry are not significant in this TFE and, therefore, cannot be used to correlate the data. In earlier data it appeared that the relative position in the core might correlate the data. This no longer appears to be the case. Even though the calculated emitter temperatures cannot be used to correlate the data, it is possible that the data could be correlated by emitter temperature because of relative emitter-to-emitter temperature uncertainties (± 50°K).

Plots of diametral distortion as a function of time and axial position are given in Fig. 3-12 for TFE 6F3. Figure 3-13 shows the axial profile of each emitter at the 8062 hour radiograph. No significant differences are noted from emitter to emitter or between these emitters and the TFE 1F1 emitter. This lack of difference from emitter to emitter in 6F3 is surprising because of the differences in the fuel shape inside the emitters (see Appendix A). The TITR flux shape effects the shape and the distribution of the isothermal central cavities in UO₂ fueled devices. This results in an increase in the amount of fuel in the lower half of the emitters in the bottom cells of the TFE and a decrease in the amount of fuel in the lower half of the emitters in the top cells of the TFE. Since the maximum

3-54

swelling occurs in the lower half of the emitter, this increase in the amount of fuel present there could increase the swelling rate. This might be the reason for the increased swelling in emitter number five, but the pattern is not uniformly followed. It is possible that the problem could be resolved and a correlation determined if two dimensional flux, temperature, and stress profiles were calculated for this TFE.

The clad thickness appears to have a definite effect on swelling rate. The 1 mm clad emitter in 1F1 distorted at a rate between 1.5 to 2.5 times as fast as the 2 mm clad 6F3 emitters. The 1 mm clad emitter in 2F3 was swollen 1% at a burnup of 3.2 x 10¹⁹ f/cc whereas no swelling was detected on the 2 mm clad emitter (£ 0.18%), even though these emitters should have been operating at the same temperature.

3-58

3.5.3 Electrical Discharge Testing

In-pile electrical discharge tests have been conducted with the 1F TFEs 1, 2, 3, and 4. (See Ref. 2 and Appendix A for test details).

Testing with TFE 1F1 was concluded at 8560 hours. At that time the TFE was stably supporting a one ampere discharge current with an emitter to sheath potential difference of 9 volts. The sources of the discharge current were not known.

Testing of TFE 1F2 was conducted between 2120 and 5618 hours with an emitter to sheath tube potential difference of 5.0 volts. The resulting current grew slowly from 12 milliamps to ~ 330 milliamps. Following a shutdown at 5618 hours the emitter and sheath tube were found to be shorted, as were the emitter and collector. Although no correlation between the two shorts other than time has been found it is possible that the high collector temperatures which would be produced where the emitter touched the collector may have contributed to a trilayer short.

Discharge testing of TFE 1F3 was terminated after 551 hours of testing with an applied voltage of 5.0 volts, following a collector-sheath tube short. It should be noted that the cesium and fission gas spaces in this cell were deliberately connected.

TFE 1F4 stabily maintained a current of 1 ma at 5 volts for 1507 hours, followed by 280 hours at 10 volts with stable current of 4 ma. At that time the polarity was reversed and the current rose slowly to 300 ma. At 2320 hours a bistable discharge mode developed with currents in the high current mode passing through the collector sheath insulator. Testing was terminated at 2400 hours.

The sheath insulator in TFE 1F1 performed well. It was of an early design and was not graded. The insulators in TFEs 1F2 and 1F3 were of the GGA graded cermet type. It was not clear in these tests whether the observed shorts.

3-59

were through these insulators or through plasma-sprayed insulating coatings. As noted in Section 3.7 the graded sheath insulator tested in cesium out-of-pile performed well for 1921 hours, and the evidence indicates the current flow in that test may have been through the insulating coating. The peculiar switching mode behavior observed in TFE 1F4’s sheath insulator, a TECO cermet, is similar to that observed in out-of-pile tests of GGA insulators which had been contaminated (Ref. 13). The nature of this behavior is not understood. It should be noted that the resistances of the two TECO trilayers tested out-of-pile quickly dropped to and stabilized at resistances near 0.2

In summary, in-pile and out-of-pile tests have demonstrated the feasibility of the trilayer insulators and alumina insulating coatings at up to 10 volt levels for up to one year. There is evidence that breakdown can occur particularly in cesium, however, and further work is needed to understand the mechanisms and prove the techniques.

3-60

3.5.4 Neutron Radiography

Work with the neutron radiography unit has shown that dysprosium foils provide superior resolution to indium foils, whereas indium foils show fuel cavities and cracks more clearly. The use of both during radiography is thus desirable. The work has also shown that:

1. For dysprosium pictures, foil order, backing shields and position in the radiation fixture has no significant effect on sharpness. Foil exposure time has a significant effect. An optimum exposure for the existing system was obtained.

2. For indium pictures foil order, cadmium backing, and exposure time have no significant effects. Indium backing is beneficial in thicknesses to 1/16 inch. Dysprosium also has a beneficial effect, but to a smaller degree.

A topical report (Ref. 14) has been written which summarizes the design and use of the neutron radiograph equipment and the techniques used in data acquisition and reduction.

3-61

3.5.5 Evaluation of Work

Two of the primary problem areas associated with TFE development are performance stability and emitter dimensional stability. The initial performance of most devices tested has agreed with predictions within the errors in estimates of the device’s operating condition. Prior to inter electrode shorting the thermionic performance of oxide fueled devices has proven stable to within the measurement errors of ± 10%, in the case of TFE 2E2 reaching 11084 hours. Carbide fueled tests generally have proven less stable, a fact attributed to fuel component diffusion through the tungsten emitter clad. Much of the fuel diffusion problem appears to have been solved by control of fuel stoichiometry and cladding structure as demonstrated by TFE 2F2, which performed stably past 4000 hours. In that case a performance loss was experienced only after the carbide fuel had sintered, returned to full diameter, and had swelled and possibly cracked one emitter clad.

Present system designs require that the emitter dimensional strains do not lead to electrical shorting at burns of 1.3 - 1.6 x 10²⁰ fission’s/cc (20,000 hours, 1 mm clad thickness). The data from the oxide fueled test 1F1 (0.25 mm spacing) indicates that further improvements in dimensional stability are necessary to meet the 20,000 hour goal with 1 mm clad thickness. The data from 6F3 (0.36 mm spacing) indicates that the 20,000 hour goal can be met or exceeded, but the effect on the system design of using the thicker 2 mm clad with its lower fuel inventory must be considered. TFE 2F3 has confirmed the effectiveness of a 2 mm clad in reducing the swelling rate when compared to a 1 mm clad. Data acquired with IC-I6 demonstrated the fact that temperature cycles can induce clad swelling. Operation in a mode which minimizes such cycles may prove effective in reducing the swelling rate. The use of different fuel designs, such as those incorporated in the oxide fuel clad capsules FC-3 and FC-4, may also prove useful in reducing swelling.

The data from the carbide fueled tests (6F4, 6F5, 2F1, 2F2, 1F4) indicates that the 20,000 hour goal may be feasible if the fuel sintering problem can be solved. The results obtained with TFE 6F4 (which has a 40 torr argon fuel cover gas) are encouraging since no sintering, swelling or performance loss was observed in 2956 hours of testing.

3-62

3.6 TRIGA OPERATION (D. Schwarzer, G. Fitzpatrick, M. Yates)

The objective of this task was to supply a neutron environment for the testing of thermionic fuel elements and fuel-clad capsules and to provide a source for neutron radiography. A total of eleven devices were tested in TITR, nine of them simultaneously.

During the November 13-17 period, the reactor was shut down for neutron radiography of in-core tests, normal reactor maintenance, and thermionic instrumentation maintenance and upgrading. Also tests Mark VIIA IC-D3 and TFE 2F1 were removed from core and TFE 6F5 and capsule FC-3 installed during this period.

A reactor scram caused by a thermionic low current instrumentation trip on November 22 kept the reactor down until November 29. The long down period was necessary due to the holidays and the diagnostic investigation required to discover where the malfunction occurred. The malfunction was found to be an intermittent shorting of switches in the cross-bar section of the data acquisition system causing spurious signals. The cross-bar was replaced and the malfunctioning unit sent out for repair.

The reactor was shut down on December 7 for relocation in-core of TFEs 6F5, 2F2, and 2F3.

Due to cancellation of the thermionic program, the reactor was shut down on January 22, 1973. All nine in-core tests were neutron radiographed and disassembly of the instrumentation initiated.

The reactor was operated for a total of 1700 hours out of a possible 1982 hours giving a duty cycle of 86%. During the calendar year of 1972, the reactor was operated for 7983 hours out of a total of 8784 hours with a reactor duty cycle of 91%.

3-63

3.7 TFE DEVELOPMENT TEST SUPPORT

The objective of this task was to conduct parasitic electrical discharge tests and to provide test support as needed for TFE development.

The electrical discharge tests were designed to study possible parasitic losses within the TFE, particularly in the event of a cesium leak from the inter-electrode to the fission gas spaces. The approach used in the investigation was to:

1. Determine what conditions are necessary to prevent electrical breakdown.

2. Determine any adverse effects of discharges on the outer sheath or the converters as a function of voltage and cesium vapor pressure.

3. Provide data, both through the literature and by experiment, on breakdown voltages and discharge current-voltage characteristics.

The work has shown that an unprotected TFE may be effectively shorted by an electrical discharge in the event of a cesium leak to the fission gas spaces; however, the use of alumina insulating coatings on exposed surfaces and alumina tubes in the fission gas vent holes will reduce electrical losses to an acceptable level. Tests of a TFE intercell region with such protection conducted at normal TFE operating sheath temperatures and cesium pressure showed no damage following sustained application of TFE open circuit voltages (14 volts) for over one hundred hours. A more detailed summary of the tests results are given in Ref. 11.

Three life test specimens were tested during this work period, LT-1A, and LT-2A, and LT-2B. LT-1A uses a GA graded cermet trilayer (Ref. 15). LT-2A andLT-2B use TECO cermet trilayers (Ref. 16).

3-64

Testing was conducted with the samples at 1075°K with a cesium pressure of 5 torr and an applied voltage of five volts. LT-2A and 2B were run with opposite polarities and LT-1A was operated with the insulating coating at a positive potential. The histories of these tests are shown in Figs. 3-14 and 3—15.

The resistance of LT-2A and LT-2B fell during the first one hundred hours of testing, stabilizing near 0.2 ohms in each case. The IV characteristics of these devices were linear, indicating a resistive short. Both devices were operated with currents near 6-1/2 amperes until test termination at 1921 hours.

The resistance of LT-1A dropped slowly during the test, stabilizing near 40 ohms at 1400 hours. At 1627 hours the voltage was raised to 10 volts, resulting in an ignition and an accompanying resistance drop to approximately 15 ohms. It remained stable at this level until test termination at 1921 hours. A representative IV curve taken at the 1920 hours is shown in Fig. 3-16.

The post-operational appearance of the insulating coatings and trilayer insulators are shown in Figs. 3-17, 3-18, and 3-19. The chip in the insulating coating visible near the edges occurred during sample disassembly. No shorts were visible in the trilayer sections examined. One or two trilayer cracks were visible in each case, but these may have resulted from sample preparation. The sheath insulation in LT-2A and 2B was leak tight and LT-1A had a helium leak rate of 1.2 x 10^-7 std cc/sec. A microprobe analysis of the sheath insulators, alumina insulating coatings, and a weld exposed to cesium revealed no cesium at or above the 1/2% level.

The three sample trilayers were divided into quarters and halves and their resistances measured.

The results were:

LT-1A LT-2A LT-2B

Half 10^7W Half 0.4W Half 0.6W

Half 10^7W Quarter 14.4W Quarter 1.0W

3-65

These indicate the shorts observed in LT-2A and LT-2B were distributed to some degree. The resistance of LT-1A after testing was quite high. This, in conjunction with the form of the TV curve, suggests the discharge current observed in that case occurred through the alumina insulating coating.

The LT-1A test confirmed earlier tests which demonstrated the feasibility of graded sheath insulators and alumina insulating coatings to reduce discharge currents to acceptable levels. In conjunction with in-pile tests, they also suggest the need for further development, both of the graded sheath insulators and the alumina insulating coating. The TECO cermet insulators in particular need further effort to determine the source of and solution to their low resistances in the presence of cesium.

3-72

1. Roth, R. S., and Schneider, S. J., "Phase Equilibra in Systems Involving the Rare Earthy Oxides. Part I. Polymorphism of the Oxides of the Trevalent Rare Earth Ions." 64A, (14) 309-316 (1960).

2. "Development of a Thermionic Reactor Space Power System," Summary Report for Period March 1, 1972 through June 30, 1972. USAEC Report GA-A12251, Gulf General Atomic.

3. "FEATS - A Computer Program for the Finite Element Thermal Stress Analysis of Plane or Axisymmetric Solids," Westinghouse Astronuclear Laboratory.

4. "A Statistical Fuel Swelling and Fission Gas Release Model," Nuclear Applications and Technology, Volume 9, August, 1970.

5. Thomson, W. I., "BRITL, A Fuel Rod Analytical Model for Thermal or Fast Flux Irradiation, April 1, 1970.

6. Warner, H. R., and Nichols, F. A. "A Statistical Fuel Swelling and Fission Gas Release Model," Nuclear Application and Technology, Volume 9, August 1970.

7. Holland, J. W. "Thermionic Fuel Element Development Status Summary", Presented at the 3^rd International Conference on Thermionic Electrical Power Generation., AEC Report GA-A12067, Gulf General Atomic, June 1972.

8. Ream, J. T., Weinberg, A. F., and M. H. Horner, "Development of a Unit Cell Thermionic Reactor Fuel Element," Presented at the Thermionic Conversion Specialists Conference, AEC Report GA-7321, Gulf General Atomic, November 2, 1966.

9. Weinberg, A. F., et. al. "Fission-Heated Thermionic Cells for Application to Nuclear Space Power Systems," Presented at the Conference on Applied Thermionic Technology, May 13-14, 1965.

10. Chin, J. and Messick, C. W., "High Temperature Ceramic-to-Metal Seal Development," USAEC Report GA-9118, Gulf General Atomic February 10, 1969.

11. Yates, M., et. al., "Development of a Thermionic Reactor Space Power System, USAEC Report GA-A12397, Gulf General Atomic, December 1, 1972.

3-73

12. Gietzen, A., "CPOP. A Program for Thermionic Converter Performance Analysis and Optimization," USAEC Report GA-8873, Gulf General Atomic, April 30, 1969.

13. "Development of a Thermionic Reactor Space Power System, Summary Report for Period March 1, 1971 through June 30, 1971. USAEC Report GA-B11043, Gulf General Atomic.

14. Schwarzer, D. E., and Fitzpatrick, G. O., "Equipment and Techniques for the Utilization of Neutron Radiographs with Thermionic Fuel Elements", USAEC Report GA-A12521 (To be published)

15. Chin, J., Messick, C. "Sheath Insulator Development," Proceedings of the Thermionic Conversion Specialists Conference, WASH-1166.

16. R. Breitwieser, private communication.

3-74