TASK 2. MATERIALS TECHNOLOGY

This section summarizes the work carried out during this reporting period and in addition includes a brief summary of the current status and material problems to be resolved for the major converter components. Pertinent references are given in order to provide some degree of continuity for future workers in this field.

The development of the thermionic fuel-cladding system constitutes a major portion of the efforts of both the AEC and the NASA thermionic programs at Gulf General Atomic. Details have been described in various reports (see Bibliography in Appendix C). The recent status of thermionic fuel-cladding development can be found in a paper presented at the Third International Conference on Thermionic Electrical Power Generation (Ref. 1) and in another paper presented at the 1972 Intersociety Energy Conversion Conference (Ref. 2). More technical details are contained in the record of the First Thermionic Project Office Material Committee Meeting, which was held in September 1972*.

__________

*Copies of this record were distributed by D. Beard and L. Price of the USAEC to various committee members.

2-1

On the basis of the in-pile test results obtained to date at GGA, it appears that the life of the carbide-tungsten thermionic fuel cladding system is limited to about several thousand hours at a cladding temperature of 1800° to 1850°K. The major problems for the carbide-tungsten system lie in the degradation of thermionic performance due to fuel diffusion through the cladding and the deformation and cracking of the cladding by stress generated through fuel swelling. The carbide fuel strengthens and also embrittles the tungsten cladding so that the cladding cannot take much deformation without developing cracks. The stress caused by fuel swelling probably also enhances fuel diffusion through the cladding and accelerates thermionic performance degradation. Solutions proposed for resolving these problems consist of: (1) to incorporate more surface area into the fuel body to facilitate fission gas release and thus to minimize fuel swelling, (2) to soften the fuel body by lowering the concentration of the tungsten additive or eliminating it so that less stress is imposed on the cladding when the fuel body swells, (3) to improve the fuel body design so that less stress produced by fuel swelling is transmitted to the cladding, and (4) to use chloride-arc-cast tungsten cladding to reduce fuel transport rates and thus to minimize thermionic performance degradation. All these have been implemented into carbide converter design and fabrication development but completion of fabrication for in-pile evaluation had not been accomplished at the termination of the program. These items should be considered as top priorities for any further carbide thermionic fuel development program.

The in-pile test results on the oxide-tungsten system obtained to date at GGA indicate that the life of the oxide fueled tungsten emitter of reference design (40 mil cladding) is about 8000 hours at a cladding temperature of 1800°K, the major problem being the swelling of the emitter which leads to inter-electrode short. It appears that the life of the emitter could be increased to greater than 20,000 hours if the cladding thickness is increased to 80 mils or if the emitter diameter is reduced from the 1.1 inch prototypical size to 5/8 inch. Unlike the carbide fueled converters, the oxide fueled converters do not exhibit significant performance

2-2

degradation during long-term in-pile operation. Approaches for resolving the emitter swelling problem involve: (1) using lower fuel loading (72 vol. % instead of 85 vol. %) in the emitter cavity, (2) improving the fuel design so that very little of the stress generated by the fission gas bubble growth in the fuel body is transmitted to the cladding, (3) using high strength cladding material such as thoriated tungsten, and (4) lowering emitter operating temperature incorporating oxygen additive (e.g., oxygenated molybdenum collector) into the converter. To evaluate these approaches, one oxide-fueled capsule was under irradiation, while another oxide fuel capsule and an oxide fueled 6F TFE were being fabricated at the time the program was terminated. These approaches for reducing the swelling of the oxide-fueled emitter under irradiation should receive immediate attention in the event the thermionic reactor program is reactivated. Consideration should also be given to evening out the radial fission power distribution in the fuel during the irradiation test. This can be achieved by using low fuel enrichment or by irradiation in a fast reactor.

The development of the ceramic components for the thermionic fuel element, such as the insulator seal, the trilayer and the electrical insulation, was carried out under the AEC thermionic program. Detail information on work concerning these components can be located through the Bibliography in Appendix C. The conference records for the Thermionic Conversion Specialist Conferences show the annual progress by GGA and other laboratories, TECO, ORNL, and LASL. The development of the insulator seal for thermionic fuel element application was described in two topical reports (Refs. 3, 4).

The recent status for the insulator seal was given in a paper presented at the Third International Conference on Thermionic Electrical Power Generation (Ref. 5) and in another paper presented at the 1972 Intersociety Energy Conversion Conference (Ref. 2). The seal developed has operated successfully in converters and in thermionic fuel elements tested in thermal reactors under the AEC program at GGA. A cooperative effort with LASL for irradiating prototypical seals in EBR II was under planning. The development of Y₂O₃ seal for better resistance to fast neutron irradiation was initiated in 1972.

2-3

The development of the Nb-A1₂O₃ graded trilayer used in thermionic fuel elements was described in three topical reports (Refs. 6, 7 and 8). A number of problems remain to be resolved with regard to the fabrication and the in-pile operation of the Nb-A1₂O₃ graded trilayer. First, the niobium sheaths of the trilayers fabricated were often contaminated by oxygen and became embrittled, and the Al₂O₃ layers exhibited radial cracks. Secondly, there were indications that the electrical resistivity degraded when the trilayer was operating under an electrical potential in the presence of cesium vapor, both out-of-pile and in-pile even in a thermal reactor. Thirdly, cracks were observed in the ceramic layer when trilayer samples were irradiated in EER II in a fast neutron environment. The first problem can be resolved by better control of the impurities in the gaseous environment and the schedule during gas pressure bonding and plasma-spraying. The second problem needs more detailed investigations on the mechanism for the electrical resistivity degradation. If this problem cannot be resolved, then all the TFE concepts involving the contact of cesium vapor with trilayer become infeasible. Solutions to the third problem require improved trilayer design so that the niobium will move with the ceramic when the latter swells under irradiation, and the use of ceramics of better resistance to fast neutron damage, such as Y₂O₃. A joint program with LASL for irradiating prototype trilayers in EBR II and a developmental program for Y₂O₃ trilayers were in progress during 1972. The solutions to these trilayer problems are crucial to the success of the operation of a thermionic fuel element.

The developmental status of the electrical insulation for the intercell regions in a thermionic fuel element can be found in the AEC contractual reports. Plasma-sprayed Al₂O₃ has been the reference insulation layer used in the thermionic fuel element. There is concern over the contamination and embrittlement of the thin convolution in the insulator seal skirt during the plasma-spraying of the Al₂O₃ insulation layer. It has also been observed that the insulation layer always develops cracks at the bottom of the convolution, probably caused by thermal cycling. Thus there is considerable room for improvement

2-4

in the Al₂O₃ insulation layer for the intercell region. Sputtered Al₂O₃ coating has been investigated for such application but more work is needed before it can be considered for thermionic fuel element application.

In addition to the material problem areas for the converter components, there exist a number of crucial material problem areas for other thermionic fuel element components. First, to prevent the reaction between the fission products released from the fuel body and the Al₂O₃ insulation coating around the fission product venting holes, and to prevent the plugging of fission product venting holes, traps containing Al₂O₃ were installed into the emitter transition regions of a carbide fueled 6F TFE to trap the active ingredients such as Sr and Ba. With the disposal of all the irradiated capsules and thermionic fuel elements (except 6F1 and 6F2) without any hot cell examination, the effectiveness of such traps during in-pile operation cannot be evaluated. For oxide fueled devices, the out-of-pile study of fission product trapping (such as for tellurium) was about to begin when the thermionic program was terminated. Such out-of-pile and in-pile studies of fission product trapping are of utmost importance to the long-term operation of a thermionic fuel element. Secondly, most of the Al₂O₃ support discs for the Bellvelle washers in the two 6F TFEs (6F1 and 6F2) examined were found to be disintegrated into individual grains. It appears that these discs were subjected to high temperatures and stress. A white phase was observed in the tantalum holder for these discs at the tantalum-Al₂O₃ interface. Thirdly, excessive grain growth occurred in the thin convolutions of the niobium skirts of the insulator seal in the intercell region. The intercell region contains a large number of refractory metal joints located close to one another. Combined with large temperature and stress gradients and possible contaminations by fission products and fuel components, this region offers a number of potential leaks between the fission gas space and the cesium space. In fact, there was evidence that such leakage did occur in the majority of the thermionic fuel elements tested.

2-5

Any future thermionic program planning should include a redesign of this crucial region.

Another major problem in the thermionic fuel element developmental program is how to reduce cladding temperature in an operating thermionic fuel element and how to minimize the temperature non-uniformity in a single emitter as well as among all six emitters. Although this is not a material problem, the life and performance of all the materials in a thermionic fuel element, are strongly dependent upon the solution to this problem. Research and design efforts should be directed toward this area in any future thermionic program so that the material components will not be overburdened because of temperature uncertainty and penalty to performance caused by temperature non-uniformity. An additional area in need of attention is to study means for improving thermionic performance at low emitter temperature. This will help to lower the emitter operating temperature and thus to lessen the burden on emitter as well as other thermionic fuel element component materials.

A considerable amount of progress has been achieved during the last twelve years in developing thermionic fuel element component materials and in understanding the critical problem areas associated with long-term operation of the thermionic fuel element. It must be remembered that any material has its operating limits with respect to temperature and radiation damage. The development of a reliable system requires a thorough understanding of these limits and the designing of the system within these limits. The above described approaches raise these limits through the development of better materials (e.g., high strength cladding and better radiation resistant ceramics) and also provide wider safety margins below these limits through better fuel and TFE design. Such cooperative efforts in any future thermionic program could conceivably increase the thermionic fuel element life to beyond 20,000 hours for both space and underwater applications.

2-6

The work carried out during this reporting period under various subtasks is described below.

2.1 FUEL TECHNOLOGY

The objective of this subtask is to study the physical, chemical, and mechanical properties of carbide and oxide fuel materials for nuclear thermionic application and to develop new fuel materials and fabrication processes for reducing cost and improving the life and performance of thermionic fuel elements.

2.1.1 Carbide Fuel Phase Relationship (R. G. Hudson, L. Yang)

This work concerns two specific aspects of the phase relationship in carbide fuels. First, it is intended to find out the solubility of tungsten in UC and 90UC-10 ZrC as a function of temperature and C/U atom ratio. Second, it is intended to identify the nature of the metallurgical phases present when the solubility limit of tungsten is exceeded. The information is needed for the selection of fuel composition which is metallurgically stable toward the tungsten cladding. Due to the early termination of this work, studies were limited to an annealing temperature of 2073°K and UC specimens of C/U = 0.98, 1.01, 1.03 and 1.05 and tungsten concentrations equal to 0, 0.5, 1.0, 1.5, 2.0, and 4.0 wt. %. For specimens of C/U = 1.03, the tungsten concentrations were also extended to 6 and 8%.

The specimens were prepared by arc-melting of weighed amounts of uranium, tungsten and carbon. The arc-melted buttons were homogenized at 2523° K for 1 hour and then annealed at 2073° K for 16 hours before they were quenched in helium at a rate of 500° per minute to 800° K and then cooled to room temperature in helium. X-ray analysis of these speciments yielded the results shown in Table 2.1. The essential findings are summarized as follows:

2-7

1. There were five different phases identified by X-rays. In addition to the monocarbide matrix containing dissolved tungsten, the dicarbide phase, and the tungsten phase, two UWC2 phases were found. These were the orthorhombic phase (Ref. 9) and the h -phase (Ref. 10).

2. When the C/U atom ratio of the carbide was less than unity only the monocarbide matrix and the tungsten phase were found. The tungsten phase appeared when the tungsten concentration was between 1.5 and 2 wt. %.

3. When the C/U atom ratio was higher than unity, the dicarbide phase was observed at low tungsten concentration. When the tungsten concentration was increased, the UWC₂ phases and the W phase appeared. For a C/U atomic ratio of 1.01, this happened at a tungsten concentration between 1.5 and 2 wt. % and the UWC₂ present consisted of a mixture of the h -phase and the orthorhombic phase. With further increase in tungsten concentration, the orthorhombic UWC₂ phase disappeared but the h -UWC₂ phase remained. For a c/u atom ratio of 1.03, the UWC₂ and W phases appeared at a tungsten concentration between 2 and 4 wt. %, and the UWC₂ present consisted of a mixture of the h -phase and the orthorhombic phase. With further increase of tungsten concentration to 6 and 8 wt. %, the orthorhombic UWC₂ phase disappeared while the h -UWC₂ phase remained. When the C/U atom ratio was increased to 1.05, no W phase was observed even at a tungsten concentration of 4 wt. %; the only dispersion present was the orthorhombic UWC₂ phase.

2-9

These results showed that for hyperstoichiometric carbide fuel the tungsten concentration at which the W phase appeared depended upon the C/U atom ratio of the carbide before the tungsten was added, and that the sequence of the appearance of the various dispersion phases with the increase of the tungsten concentration was: orthorhombic UWC₂, h - UWC₂ + orthorhombic UWC₂ + w, h - UWC₂ + W. It appears that for the hyperstoichiometric specimens studied, higher carbon activity favors the formation of the orthorhombic UWC2 phase, while lower carbon activity favors the formation of the h -UWC₂ phase, and that the h -UWC₂ phase and the W phase always coexisted. For the hypostoichometric specimens, no UWC₂ phase was observed and the only dispersion present was the W phase.

The above described studies covered only a small part of the UC-W phase relationship. More extensive work is needed if the UC-W system is selected as the fuel material for the thermionic fuel element.

2.1.2 Compatibility and Trapping of Fission Products (G. Buzzelli)

The objective of this work is to evaluate the reaction between Al₂O₃ and the alkali earth fission products and the effectiveness of the trapping systems installed in TFE 6F5.

The experimental arrangement was described in the last thirdly report. Two tests were carried out previously, using tantalum capsules containing molybdenum discs coated with plasma-sprayed Al₂O₃ or Al₂O₃ chips as the trap and high purity strontium metal as the strontium source. The first test was carried out with a plasma-sprayed Al₂O₃ trap at 1350°K and a strontium pressure of 4 x 10^-2 torr for 16 hours. The second test was performed with a trap at 1310°K and a strontium pressure of 4.5 x 10^-3 torr for 32 hours. In each case, about 88% of the strontium atoms entering the trap was caught by the Al₂O₃ and

2-10

the tantalum housing of the trap SrO× Al₂O₃ was identified as the reaction product.

During this reporting period, another test was carried out with a plasma-sprayed Al₂O₃ trap at 1385° K and a strontium pressure of 1.5 x 10^-2 torr for 101.5 hours (Sr source temperature = 810°K). For a 6F TFE operating at a fission power density of 60-70 W/cc, the rate of formation of Sr is estimated to be about 10¹³ atoms per second per emitter. Assuming complete release from the fuel body, the Sr flux at the 0.4 mm diameter entrance to the trap is 8 x 10¹⁵ atom/cm sec. During steady state operation, the Sr pressure in the fuel chamber for maintaining such a Sr flux is about 10^-4 torr. The Sr pressure in the third test capsule is thus about 150 times higher than that in the fuel chamber of an operating emitter, and the total number of strontium atoms entering the trap in 101.5 hours corresponds to that for about 15,000 hour operation of the emitter in a TFE.

After the completion of the test, the test capsule assembly was dismantled and the components were analyzed for strontium distribution by X-ray fluorescence spectroscopy and electron microprobe techniques.

Table 2-2 summarizes the strontium analysis results. It can be seen that the baffles absorbed about 95% of the strontium atoms entering the trap, while about 2% of these strontium atoms leaked through the trap. This compares to a 40% absorption by the bafflesandl2% leakage through the trap for the first capsule test. The tantalum housing retained about 2% of these strontium atoms, compared to 47% for the first capsule test.

The individual baffle analysis indicates maximum absorption on the first baffle and decreasing quantities for subsequent baffles. This is in contrast to the first baffle experiment where there was no difference in the absorbed strontium levels for the various baffles. In the first test,

2-11

the strontium pressure was higher than this test by a factor of four, although the testing time was 16 hours, as compared to the 101.5 hours for the present test. It would appear that the absorbing capacity of the baffles was reached during the first test when the strontium flux was much higher, possibly because the arrival rate outweighed the rate of absorption into the Al₂O₃ by diffusion. This may explain the lower trapping efficiency of the first test. In the present test, the amount of strontium absorbed by the Al₂O₃ baffle decreased as the baffle moved away from the strontium source. This indicates that these baffles still possessed more capacity for absorbing strontium at the end of this experiment.

Another interesting observation concerns the absorption of strontium by the tantalum housing of the trap. It was observed in the first test that about 47% of the strontium atoms entering the trap was absorbed by the tantalum housing. In the present test, however, only 2% of these strontium atoms was absorbed by the tantalum housing. It was suspected that this difference might be related to the oxygen content of the tantalum, with tantalum of higher oxygen content trapping larger amounts of strontium. Analytical results showed that the tantalum for the housing of the trap used in the first experiment contained 78 ppm of oxygen after the test. On the other hand, the tantalum for the housing of the trap used in the present experiment contained 435 ppm of oxygen after the test. This is the reverse or what would be expected. Since there is a shortage of strontium inventory of about 4 mg in the present experiment, it is possible that these 4 mg of strontium may have diffused into the tantalum housing wall and were not removed by the acid leaching used for recovering the strontium from the tantalum housing. The tantalum housing for the present experiment will be sectioned and analyzed for strontium in the metal.

2-13

2.1.3 TFE Port Investigations (J. Kemme)

The objective of this task was to investigate the flow behavior of the present TFE port with respect to fission gas and cesium. This port, a 6 mil I.D. tube containing a 1 inch length of 4 mil diameter wire, forms an annular flow restriction in the fission gas venting path. In the event that leakage develops between this path and the inter-electrode space of a TFE, the port is designed to limit the loss rate of cesium to about 1g/10,000 hours. Because the port dimensions are small, there is some concern that it may become plugged with cesium or cesium compounds, a condition that would cause buildup of fission gas pressure and impair TFE performance. The approach used to study the behavior of the TFE port involved cold flow experiments with gas only and flow experiments with cesium only.

The block diagram for the experimental apparatus and its description can be found in the last thirdly report. Five cold flow tests were completed. Measurements were made with a precision glass tube with an internal diameter of 0.076 mm and a length of 120 mm (0.003 inch I.D. by 4.7 inches long) and with a TFE port of the type described above (designated as port No. 1).

The glass tube was sized so that it would have flow characteristics equivalent to that desired in the TFE (equivalent to a 0.051 mm I.D. by 25 mm long tube). The TFE port that was tested was known to have a flow rate higher than desired in the TFE and was found to behave like a 0.060 - 0.076 mm I.D. by 25 mm long tube.

The results of the tests are shown in Figures 2-1 to 2-4. Figure 2-1 shows the results of measurements made with helium in the viscous flow range (near atmospheric pressure), where the flow rate in torr cc/mm is plotted

2-14

versus the pressure drop across the port. One set of measurements was made with the glass tube and another with the TFE port. The glass tube data was in good agreement with viscous flow calculations; the TFE port data was in good agreement with that predicted for a 0.063 mm I.D. by 25 mm long tube. The viscous flow equations were taken from Reference 11.

Figure 2-2 shows results obtained with the glass tube with helium at low pressures. As indicated, the flow was slightly less than that predicted by the molecular flow and basic port flow equations. The molecular flow equation was taken from Reference 11. The basic port flow equation was taken from Reference 12 and is a semi-emperical equation which is in agreement with experimental data over the entire range of Knudsen numbers for flow of a single gas species. This equation is:

Ñ = D P ((2p )^1/2/6) (d³/l) (1/(mkT)^1/2) [ (3p /128) (1/K) + (p /4) (1/(1 + K)) + (K/ (1 + K))]

Ñ = flow rate, molecules/sec

D P = pressure differential across the tube, torr

d = hydraulic diameter of tube, cm

l = length of tube, cm

m = mass of the molecule, g/molecule

k = Boltzman constant

T = absolute temperature, ° K

K = l /d = Knudsen number

where l = 8.589 (h /P) (T/M)^1/2 cm (from Ref .12)

and

h = viscosity, poise

M = molecular weight, grams/mole

2-19

When the port was tested at low gas pressures, the flow was greater than expected. The results obtained with helium are compared in Figure 2-3 with curves calculated from the molecular flow and basic port flow equations for a tube, 0.076 mm I.D. by 25 mm long. Similar information for tests made with xenon is shown in Figure 2-4.

The results obtained with the glass tube indicated that the cold flow apparatus was functioning properly, but that additional tests would be required to characterize the flow in TFE ports.

The apparatus used for the cesium flow test is described in the last thirdly report. The primary objective of this test was to measure cesium loss rates through a TFE port at cesium pressures present in the TFE. In Figure 2-5, some typical test measurements are compared with a curve calculated from the basic equation for port flow (for port #2). Although the loss rates at lower cesium pressures were higher than predicted, this was expected because the technique for checking production ports indicated that the test port was slightly less restrictive than those accepted for the TFE. When it was found that the measured loss rates were lower at higher cesium pressures, it was postulated that at these pressures enough cesium was collecting at the port exit to cause refluxing, a condition that would create a back pressure and thus restrict port flow. However, as will be described later, the cube attached to the port exit was examined after the test and very little cesium was found in the region above the port.

A second objective of the test was to determine the temperature conditions required to form a condensation plug in a TFE port. It was found that plugging did not occur when the port temperature was a degree or two above the cesium reservoir temperature. However, a plug formed within a few hours when the port temperature was a degree or two below the reservoir temperature. Also, once plug formed, full flow was not restored for 120 hours, although the port

2-20

temperature was gradually increased during this time until it was 50 hotter than the reservoir.

The test was terminated after 720 hours. It was estimated that less than 5 mg of cesium transferred through the port while it was partially plugged. A more exact estimate, based on flow measurements and system pressure conditions, indicated that about 38 mg of cesium transferred during the remaining 600 hours. After the test, this estimate was checked by radiochemical analysis which showed that 46 mg of cesium collected in the tube that was attached to the port exit. This tube was designed to simulate the vent line leading to the charcoal trap in a TFE. The tube was analyzed in sections to determine the cesium distribution. The sections are identified in Figure 2-6. The amount of cesium in each section is given on the figure.

Because the tube at the top of the bend (see Figure 2-6) was below 390° K during the test, it was thought that condensation above the port would cause refluxing. Therefore, the location of essentially all of the cesium at the water-cooled end of the tube was not expected.

Based on the test results, the present TFE port should provide the desired cesium loss rate in the TFE, but the port should be kept above the cesium reservoir temperature to ensure that plugging does not occur. Questions concerning cesium refluxing in the vent line leading to the charcoal trap were not resolved.

2-22

2.2 CLADDING TECHNOLOGY

The objective of this subtask is to study the physical, chemical, and mechanical properties of chemically vapor deposited tungsten for thermionic emitter application, to establish techniques for quality control and to develop high performance and high strength cladding materials for improving the life and performance of thermionic fuel elements.

2.2.1 Study of Creep Behavior of Prototype Tungsten Emitter (L. Yang, H. Johnson, D. Allen)

The objective of this work is to study the creep behaviors of both unfueled and fueled prototype tungsten emitters as a function of stress, fluorine impurity concentrations, and nature of the fuel materials. The experimental results generated will be compared with those predicted on the basis of uniaxial creep data by the viscoelastic computer program under Task 4.

During the previous reporting periods, the deformation of a prototype fluoride tungsten emitter containing 5 ppm fluorine was studied at 2000° K and under helium pressures of 60 and 120 psi. The test data indicated the absence of primary creep and that the exponential dependence between stress and creep rate at low stress level differs from that at high stress level. The stress-strain rate curve defined by the test data was used to calculate the deformation of the emitter. The calculated results for the emitter deformation under 60 psi helium pressure were in good agreement with the observed values. It was also shown that the growth of voids at the grain boundaries of the cladding was accelerated by the stress and that a higher concentration of voids was present at locations where the tensile stress is the greatest. In addition to unfueled emitter, the study was also extended to a carbide fueled prototype fluoride tungsten emitter of 40 mil cladding thickness. The emitter contains 6 ppm of fluorine and the bottom 2.5 cm

2-24

was fueled with two 90 UC-10 ZrC (C/U = 1.01, W 4 wt. %) pellets. To insure a good contact between the fuel and the cladding, the cold gap was sized for a 2 mil diametric interference fit at 2000° K. The emitter was brought from room temperature to 2000° K in vacuum over a 48 hour period of time and was maintained at 2000° K at 24 hours. Upon cooling to room temperature, it was found that the emitter had expanded 2 mil diametrically in the fueled region. This is in good agreement with that predicted on the basis of the differential thermal expansion coefficient between the fuel and the cladding at 2000°K. The fueled emitter was then tested at 2000° K under 60 psi helium pressure for 1000 hours. The fueled region expanded diametrically about 1.4 mils, in agreement with the results obtained on an unfueled emitter, but the unfueled region and the emitter bottom showed no significant deformation.

During this reporting period, the emitter was tested for another 1000 hours at 2000° K and 60 psi helium pressure. This was followed by another 1000 hour test at 2000° K and 120 psi helium pressure. In both cases, no deformation was observed in the emitter structure, while previous tests of the unfueled emitter under similar conditions yielded significant deformation in both emitter diameter and emitter bottom. It appears that the tungsten cladding was considerably strengthened by the presence of the carbide fuel. Metallographic examinations showed the presence of a thin (~3 microns) UWC₂ layer at the fuel-cladding interface (Figure 2-7(a)) and porosities at the grain boundaries of the tungsten cladding (Figure 2-7(b)). Analysis of the cladding after the removal of the UWC₂ layer by polishing yielded a carbon concentration of 49 ppm, while the pre-test carbon concentration was only 8 ppm. The observed increase in the creep strength of the tungsten cladding is probably associated with the diffusion of carbon into the cladding. Such increase in creep strength, coupled with the formation of stress-induced porosities at the grain boundaries may lead to grain boundary cracking at low cladding deformation. This has been observed during the hot cell examinations

2-25

of irradiated carbide-fueled tungsten emitters (See Section 2.7). It is unknown whether the presence of oxide fuel will cause similar increase in the creep strength of the tungsten cladding. This is an area in need of more extensive work in any future thermionic fuel element development program.

2.2.2. Preparation of Uniaxial Creep Samples (J. Chin, R. G. Hudson)

Fluoride-chloride duplex tungsten samples are prepared for the study of their uniaxial creep rates at ORNL and TRW at temperatures and stress levels of thermionic interest. The data generated are used in the visco-elastic analysis of the dimensional stability of fueled thermionic emitters under irradiation. Deposits are made on square molybdenum mandrels of 6 inch length, from each of which four test samples can be obtained. Two low fluorine deposits (~5 ppm) were delivered to ORNL during this reporting period.

2-26

2.2.3 CVD Tungsten Development (J. Chin)

This work concerns the development of an infrared analytical technique for examining WF₆ used in the chemical vapor deposition of tungsten. This analysis was complicated by the corrosive interaction between and conventional materials used in analytical chemistry instrumentation.

Special sampling cells, Fig. 2-8, were constructed from Teflon and stainless steel and fitted with germanium CaF₂, NaCl, or BaF₂ windows. Sample cells were evacuated through a liquid nitrogen trapped diffusion pumping system, Fig. 2-9.

Infrared absorption was monitored with a Beckman TR-4 instrument capable of examining wave lengths of 1.0 to 15 microns. Impurities examined in were HF, HCl, and SiF₄. Wave lengths required for these examinations were 2.0-10 microns. Known quantities of these materials were added to for cell calibration. Infrared absorption versus concentration curves were then prepared for comparison with unknown samples. A typical infrared absorption curve is shown in Fig. 2-10.

Impurity contents found in eight vendor supplied bottles of WF₆ are shown in Table 2-3. The unexpected high concentration of HCl in several of the bottles were qualitatively related to difficulties experienced in depositing tungsten with controlled fluorine levels. The source of the HCl is thought to be WCl_xF_y or RCl_xF_y. Whichever is the source, these materials will interfere with the chemisorption, reduction and deposition reactions of WF₆ and H₂ necessary for the controlled fluorine deposition of tungsten. Fluoroform (HCF₃) was found in bottles from ORNL only.

When the bottles of WF₆ are nearly empty the impurity composition changes. These changes can lead to undesirable effects during chemical vapor deposition.

2-28

2.3 SEAL, TRILAYER, AND ELECTRICAL INSULATION TECHNOLOGY (J. Chin, C. Messick)

The objective of this subtask is to develop seals, trilayers, and electrical insulation which are stable under the thermal and irradiation environment of a thermionic fuel element.

2.3.1 Development of Y₂O₃ Seal

The reference seal for the converter in a thermionic fuel element consists of Lucalox ring, metallized with tungsten-2 wt% Y₂O₃ and brazed to niobium skirts with 60V-40Nb alloy. It provides electrical insulation between the emitter and the collector, and containment for the cesium vapor in the inter-electrode space. It is preferable that the seal remains leak tight during the designed life of the thermionic fuel element, although the thermionic fuel element design allows continuous operation even if the seal leaks. Such a reference seal operated successfully for 8000 hours in a prototypical converter. LASL EBR II irradiation results (Ref.13) indicated that Lucalox remained leak tight to helium after being subjected to a fast neutron exposure of 4.8 x 10²¹ nvt (> 0.1 MeV) which is equivalent to that for 20,000 hour operation of the 120 kWe thermionic reactor under consideration for nuclear electrical propulsion. To provide a safety margin for longer reactor life, seals containing ceramic materials of better resistance to fast neutron damage are required. Y₂O₃ which has a cubic lattice was shown by LASL EBR II irradiation results to be a promising candidate. The specific objective for this work is to develop a seal containing Y₂O₃ ceramic for fast thermionic reactor application.

Properties of yttria and rare earth doped yttria ceramics reported in previous GGA thermionic summaries brought out the following features:

1. Non-uniformity is hard to avoid in fabricating thermionic fuel element seal size ceramic bodies.

2-33

2. Yttria is mechanically weaker and more thermal shock sensitive than alumina.

3. The electrical resistivity of yttria is more erratic at 1073° K than that of alumina.

Attempts were made in this reporting period to overcome yttria nonuniformity and mechanical weakness problems by either additives to the ceramic or seal design modifications.

The results of the addition of two rare earth additives can be seen in the scanning electron photomicrographs (Figs. 2-11, 2-12). Schneider et al. (Refs. 14, 15, 16) theorize that the Y₂O₃ La₂O₃ mixed oxide should be monoclinic and the Y₂O₃-Ho₂O₃ mixed oxide should be cubic. The Y₂O₃-La₂O₃ mixed oxide structure in Fig. 2-11 is ill-defined. The Y₂O₃-Ho₂O₃ mixed oxide structure, Fig. 2-12, has many large tetrakaidecahedron grains. The addition of rare earths changed the microstructure of the yttria body. The effect of these changes on the properties of the yttria has not been determined.

Niobium and tungsten were added to yttria to form cermets. These metals were selected as yttria crack stoppers for improved mechanical properties. There are some deleterious reactions from these metal additions. Mixed WO₃-Y₂O₃ oxide formation is known to cause some lowering in electrical resistivity (Ref. 17) but this is not considered serious because of the limited amount of WO₃ present. Formation of Nb₂O₅ mixed oxides should also lower yttria electrical resistivity but this has not been verified. In practice, niobium cermets are strong and contain a minimum number of pores and cracks. These structures have worked well in graded yttria trilayer ceramic-to-metal transition zones. All ceramic cracking and tensile test failures occurred outside of this zone. The amount of niobium added to the yttria to form an insulating cermet is limited by the electrical conduction from touching niobium particles. This limit appears to be less than 5 weight per cent niobium for the materials and techniques now used for trilayer fabrication.

2-34

Recent yttria-based ceramic components were not completed in time for testing prior to program phase-out. One of these was a promising design for a ceramic-to-metal seal. The seal was designed with multiple, alternate layers of cermet and yttria as shown in Figs. 2-13 and 2-14. The intent of these multiple layers is to inhibit crack failure in the ceramic and enhance oxygen diffusion in yttria for higher density sintering.

2.3.2 Trilayer Development

The trilayer consists of concentric niobium cylinders bonded to an insulation layer between them. It provides electrical isolation between the collector and the liquid metal coolant and a thermal conductance path for heat rejection from the collector. A graded Nb-Al₂O₃-Nb trilayer was developed and was tested for thermal stability, property reproducibility, and irradiation stability. Metals and metal oxides were added to Al₂O₃ to stabilize its fine grain structure in an effort to improve its resistance to fast neutron damage. Cubic ceramics, such as Y₂O₃, were examined for the improvement of the resistance to fast neutron damage.

The objective of adding metals and metal oxides to stabilize the fine grain structure of alumina is to minimize the effect of fast neutron induced anisotropic expansion in hexagonal Al₂O₃. Select trilayer samples were prepared with tungsten, molybdenum, Pr₂O₃ and La₂O₃ added to the alumina. Fractured ceramic segments were examined before and after two hour, thermal treatments at 1973° K. The intent of this effort was to determine the effectiveness of the additives in inhibiting thermally induced grain growth in alumina, and thus the stability of the fine grain structure.

The cermets A1₂O₃-W and Al₂O₃-Mo were the most effective in inhibiting alumina grain growth (Table 2-4). Electrical resistivity decreased in all cermets but was not zero even in the 70 weight per cent tungsten sample. All samples were prepared by plasma spraying the metal-Al₂O₃ mixture and autoclaving the sprayed cermet at 1773°K and 10,000 psi for three hours.

2-37

Molybdenum like niobium smeared during this plasma spraying. Thus Al₂O₃ with 10 weight per cent molybdenum and greater were electrically shorted. The tungsten did not smear and the cermet was not shorted at higher metal / Al₂O₃ ratios. The 1973°K, 2 hour thermal treatment did not visibly influence the uniformity of the cermet grain structure, Fig. 2-15.

The rare earth additives were effective in inhibiting grain growth. Al₂O₃ with either 10% La₂O₃ or 12% Pr₂O₃ exhibited reduced grain growth in Al₂O₃. Lanthania and praseodymia pinned grains were twice the size of the Al₂O₃-W cermet grains but were equal to the molybdenum pinned alumina grains.

Needle grain growth was prominent in the Al₂O₃-Pr₂O₃ samples but was not evident in any other samples (Fig. 2-16). Komeya and Inoue (Ref. 18) reported that needle structures in AlN-Y₂O₃ increased the ceramic flexural strength. The importance of these needle structures to trilayer ceramics in a thermionic reactor cannot be judged without further data.

2.3.3 Thermal Stability of Al₂O₃ and Y₂O₃ Trilayers

The specific objective for this effort is to evaluate the long-term stability of the electrical resistivities of Al₂O₃ and Y₂O₃ trilayers at high temperatures under a continuously imposed electrical potential. The results should indicate whether these trilayers can maintain stable electrical performance in an operating thermionic fuel element and whether significant amounts of ionic transport occur in the Al₂O₃ and at the Nb-Al₂O₃ interface which may cause degradation of the structural integrity of the Al₂O₃ and the Nb to Al₂O₃ bond.

A recent thermal stability test of Nb-Al₂O₃-Nb and Nb-Y₂O₃-Nb trilayer samples at 1073°K under 10 VDC potential was terminated after a test time of 1524 hours. These electrically loaded trilayer segments were:

1. Nb-Al₂O₃-Nb graded trilayer

2. Nb-Y₂O₃-Nb graded trilayer

3. Nb-Al₂O₃-Nb TECO cermet

2-41

Fig. 2-17 shows the electrical resistivities of trilayer segments as a function of test time. Measured resistivities for the Nb-Al₂O₃-Nb, Nb-Y₂O₃-Nb graded trilayers were nearly constant for the l524 hour test. Resistivities measured across the TECO cermet were erratic when first measured but the overall trend was for a gradually increasing electrical resistivity. There were no detectable differences in the microstructural appearance of samples electrically loaded from those not loaded (Figs. 2-18 to 2-20). What appeared to be chains of touching spheres could be seen throughout the TECO sample. There was no evidence that this caused electrical shorting paths across the cermet. The Nb-Y₂O₃-Nb trilayer was more stable at 1073°K to a 10 VDC potential in this test than in any previous test.

2.3.4. Electrical Insulation Coating Development

The specific objective for this work is to develop electrical insulation coatings for intercell regions to prevent cesium vapor arcing and for other TFE components, such as bus bar, thermocouple sheath, and cesium reservoir, to prevent electrical shorting. Plasma-sprayed Al₂O₃ coating is currently being used in thermionic fuel elements under test. Alternate methods for preparing insulation coatings, such as sputtering and ion-plating, were also evaluated.

A newly developed technique for plasma spraying alumina was investigated. Plasmadyne, the developer, sent samples of Al₂O₃ sprayed on 410 stainless steel of 10 mil thickness. These newer coatings were applied in vacuo at estimated gas velocities of mach 3. Older TFE coatings were applied in air with maximum gas velocities estimated to be mach 2. Bright field and polarized light views of the new coating can be seen in Figs. 2-21a and 2-21b. The outer layer with a different polarized light appearance was softer (DPH 365) than the dull appearing inner layer (DPH 500). Both coatings are much softer than fully dense Al₂O₃ (DPH ~3000). The newer plasma-sprayed alumina looks interesting but because of its softness and non-uniformity, should be investigated further before being used in place of the older mach 2 coatings.

2-44

Evaluation of sputter coating as a means of’ electrical isolation continued this last reporting period. Plasma-sprayed Al₂O₃ over niobium was sputter coated in two trials to seal the coatings. Scanning electron photos of these surfaces showed either very little sealing or washing out of the plasma-sprayed structure by sputtered alumina coating (Fig. 2-22). The appearance of the surface was that of the sputtered coating only. This duplex coating has an advantage over straight plasma spraying in areas where the coating must be leak tight, where the thickness of the plasma sprayed coating is needed and where substrate bending is not a problem.

Sputter alumina was applied to niobium and NaCl substrates to determine the effect of substrate temperatures on the properties of the sputtered alumina. Fig. 2-23 (a), (b) shows scanning electron micrographs of coatings applied at 425 and 750°K. The higher temperature substrates appear to promote the formation of larger grains and continuous nucleation. Grain growth on the cooler substrate appears to be at sites where nucleation has already occurred.

The as-deposited alumina on NaCl was amorphous according to electron diffraction examination, Table 2-4. These coatings changed to q -Al₂O₃ plus some (Al₂O₃)5H after 2 hour, 1473°K thermal treatment, Fig. 2-24(a), (b). This work agrees with the findings of Dragoo and Diamond (Ref. 19) on structural changes in evaporative alumina coatings.

One ion-plated alumina sample on niobium was prepared. The appearance of the coating was similar to sputtered coatings. The properties of ion plated alumina coating and the conditions of deposition are not known in enough detail to compare it with sputter coatings.

It appears from the results of all physical vapor deposited alumina coatings tried in this program that sputter coatings can be applied to thermionic fuel element components as amorphous or g alumina coatings. Conversion of these coatings takes place in a predictable manner with thermal treatment. Coating adhesion is good and can be applied to full scale ceramic-to-metal components. Compression stress strain tests of two of these seals

2-50

before and after sputter alumina coatings were applied (Fig. 2-25) did not adversely affect the properties of the substrate and did not spall off the surface. This means these coatings could be applied to critical parts of the thermionic fuel element at intermediate fabrication stages without a detrimental effect on the coating or subassembly.

2-55

2.4 JOINT TECHNOLOGY (J. Grebetz)

The objective of this subtask is to develop joining techniques that would improve the structural integrity of thermionic fuel element components and reduce cost. Earlier work in this program defined electron-beam welding parameters for effecting joints in refractory metal TFE components.

2.4.1 Diffusion Bonding Study

The object of this program is to refine the diffusion bonding of the tantalum transition to F-series tungsten emitters so that a joint of highest integrity can be produced with minimum cost.

Earlier work in this area provided lapping/chemical cleaning procedures that improved surface finish. (See Historical Review below.) Scanning electron microscopy was used to determine the most effective post-lapping cleaning treatment that would ensure the removal of lapping compounds without increasing surface roughness to courser finishes than those created by the lapping process. As a result of this work, new lapping equipment was obtained to remove manual work from the lapping process and reduce its cost.

Work conducted during the past reporting period was to evaluate the effects of time, temperature and pressure with the overall objective of reducing fabrication costs by shortening the bonding cycle. Impetus was given to this work by observation of joint porosity in bonds examined during post-irradiation hot cell examination of fueled emitters. These bonds showed what appeared to be diffusion induced void formation on sections of bond perpendicular to the joint interface and a curious absence of such porosity on the portions of the emitter-transition piece joints that are on a 45° angle. The history of the development of the bond will serve to introduce the subject of current work in proper perspective.

Historical Review: The present method of diffusion bonding tungsten

2-57

to tantalum was initiated in 1963 with the fabrication of OC-4, and a series of improvements have been applied to some parts of the process since that time. The first study of diffusion bonding of these materials for thermionic applications was conducted by Elsner, Horner and Weinberg. The parameters devised were:

Joint temperature - 2073° K

Joint pressure - 2000 psi

Cycle –

1. Heat to 2073° K @ 500 psi

2. Raise pressure to 1800 psi for 1 hour

3. Raise pressure to 2000 psi for 1 hour

4. Cool to room temperature at 500 psi in ~ 40 minutes

System pressure: < 5 x 10^-5 Torr

A joint temperature of 2073° K represents close to a practical limit since the midpoint of the C Series and F Series emitter tubes was raised to ~ 2273° - 2173° K with the joint at 2073° K. The method of joint preparation was lapping with 600 grit silicon carbide abrasive.

During this period, configurations for lapping thin walled cylinders of tungsten to tantalum were studied. Various emitter tapers from 45° I.D. cones to 45° O.D. cones were studied. The best configuration was a flat (90° to the cylinder axis) surface with a trace (10 mils wide) conical chamfer at 45° on the inside of the cup. The chamfer allowed the machinist to lap the emitter in a true circular pattern.

In 1964 the lapping compound was changed to 15 micron diamond powder to ensure that any residual contamination from lapping would be carbon and a post-machining degassing treatment at 2273° K for 6 hours was added.

In 1969-70 the bonding chamber system pressure specification was revised to a value of < 5 x 10^-6 Torr and bonds have typically been made at pressures of 2 x 10^-6 to 8 x 10^-8 Torr. During this period a 1200° C stress relief anneal was added at the end of the bonding run to accommodate the differential thermal stresses that occur upon cooling. In addition a more thorough cleaning

2-58

specification was applied to the tantalum component, and the lapping procedure was revised. The lapping compound was changed to six micron diamond paste (natural diamond) to give a smoother surface finish, and lapped surfaces were inspected at magnifications from 10 to 60 diameters. The post bonding degassing treatment at 2273°K was eliminated and a 14 hour stress relief anneal at 1223°K was added.

Failures of diffusion bonds have been traced to several causes:

A. Microcracks in the Tungsten at the Tungsten-Tantalum Interface.

The microcracks are generally caused by the O.D. and end grinding sequences; for example, when the open end of the emitter is ground prior to the finish grinding of the tungsten cup O.D. (as it should be) and the last grinding pass is taken in a direction to allow the grinding wheel to pass from the flouride tungsten substrate at the open end (as it should not be), damage to the grain-to-grain bonding results. If such a structure is lapped to a transition piece and bonded, the structure generally cracks along a grain boundary. Evidence of this weakness is always found in the form of grain boundary pull-out at the lapped tungsten surface near the O.D. of the part. Parts that exhibit this behavior are now lapped upon a flat plate until all pull-out is removed, then lapped to a transition piece for bonding and re-inspected.

One example of an apparent inherent weakness in a flouride tungsten billet was observed during 1970. This billet was lapped and bonded twice--leaks occurred each time at the bond interface. Subsequent lapping followed by examination of the lapped tungsten surface at 20X revealed that grain boundary pull-out continued to occur even though several mils of material were removed during each lapping sequence.

B. Excessive Surface Roughness

It is believed that a rough surface texture results in "laps" in the surface of the tungsten that are not completely filled by tantalum as it creeps at the interface during the bonding step. It is highly probable that a rough surface is also one that is contaminated by embedded lapping materials (oil and diamond particles) that had not been removed by the chemical cleaning

2-59

process employed when coarser lapping compounds were employed.

C. Misalignment of Components

Component misalignment has caused bond failures to occur. Present machined accuracy of components and set-up techniques have eliminated this problem from our recent experience.

D. Temperature-Joint Loading Calibration

Some failures of bonds have been traced to pyrometer calibration or calibration of the force application system that yielded lower temperatures or lower pressures than those specified. This type of failure is accidental or related to operator error.

Current Work: A series of bonding study specimens were prepared and bonded at three different temperatures, 1673° , 2073° and 2194°K, with the overall objective of reducing fabrication costs by shortening the bonding cycle. Samples were bonded for three different times at temperature, 1 minute, 10 minutes and 100 minutes. A simple cylindrical sample configuration, described in Reference 19, was used because of its machining ease and easy characterization. Bonded samples were evaluated by metallographic examination, electron beam microanalysis and scanning electron microscopy. A sample of an F Series emitter tungsten-to-tantalum diffusion bond, prepared by the latest fabrication methods for surface preparation, was evaluated as a comparison sample by the same techniques. The results of these studies are summarized below:

A. Current practice in bonding F Series emitters results in an interface that is essentially void free. This result is a direct consequence of the improvements made in the lapping and post-lapping cleaning processes.

B. For cylindrical bond specimens, a gradient of void exists:

(1) Void near the O.D. of specimens was practically eliminated for samples bonded for 2 hours at 2073°K.

2-60

(2) Void present near the centers of specimens was determined to be related to the original surface roughness of the tantalum member by scanning electron microscopy. Void in these regions was regularly distributed with a spacing characteristic of the spacing of machining and lapping grooves, and many voids showed internal surfaces characteristic of being abrasively machined rather than being formed by a diffusional process.

Because of the void gradient, and because void density was unrelated to concentration gradients observed by electron beam microanalysis, it was concluded that void density was not related to volume diffusion effects.

2.4.2 Evaluation of Work

It was concluded that the presence of void could be related only to the original surface roughness of the samples that were bonded. Variations in void density between bonding temperatures and times for different samples were related to creep behavior and not to diffusional processes. Variations in void density along a sample diameter were a function of the varying creep rate (volume flow) of material along the diameter (increasing creep rates near the O.D. of the samples where volume restraint or sample edge strength was low, decreasing to low values at the centers of test samples where bulk flow was restrained by the surrounding sample volume). Direct parallels to these effects exist for hot forging operations in steel working where grain refinement may be poor in the centers of large forgings and adequate near the edges of the same forgings.

It was further concluded that the current bonding techniques for tungsten-tantalum joints for F and LE Series emitters produce an optimum joint.

2-61

2.5 SEAL AND TRILAYER IRRADIATION (J. Chin, C. W. Messick)

The objective of this subtask is to study the effect of fast neutron irradiation on the dimensions, structures and properties of seals and trilayers for thermionic fuel element application. The samples fabricated by Gulf General Atomic are irradiated in EBR II in heat pipe operated capsules designed and built by LASL. Post-irradiation examinations are carried out in the LASL hot cell.

2.5.1 Capsule K-3, 4

Irradiation of capsule K-3, 4 in EBR II has been completed. Post-irradiation examinations have been initiated in the LASL hot cell. The GGA samples irradiated in capsule K-3, 4 consist of a number of trilayers of the LASL size, a list of which was given in a previous thirdly report (Ref. 20). Most GGA samples were irradiated to a fast neutron fluence ( > 0.1 MeV) of 2.0-2.6 x 10²¹ nvt at 973°K. Out-of-pile ceramic impurity contents and microhardness measurements are shown in Tables 2-5 and 2-6. Examination results of irradiated K-3, 4 samples will be reported in a LASL topical report. A few remarks here summarize some of the post-irradiation examination results and what these results suggest in terms of trilayer stability to fast neutrons.

Irradiated alumina in all trilayer samples expanded at a rate predicted from measurements made on irradiated free standing alumina bodies. Niobium and Nb-1Zr trilayer skin swelling is estimated to be less than one-third of the alumina swelling. Trilayer samples made with high modulus Nb-1Zr either failed at the ceramic metal interface or caused cracks to form in the ceramic. Trilayers made with thin walled niobium skins followed the alumina swelling and did not crack. Where thick walled Nb skins were used, as in the case of the GGA graded trilayer (12 mil graded cermet, 26 mil collector and 28 mil thick inner sheath), the graded ceramic-to-metal joint was intact but the 5 mil thick alumina layer cracked.

2-62

Thus, in all cases when the trilayer skins could not follow ceramic movement, ceramic failure occurred. The trilayer wall should thus be made of a thin low-modulus material with a good thermal expansion match to the ceramic, which will still allow passage of the assigned DC currents with minimum Joule loss and yet can follow dimensional changes of the ceramic due to fast neutron irradiation.

Trilayers made with cubic ceramics such as Nb-Y₂O₃-Nb graded trilayers showed a slight amount of swelling which was approximately what one would expect from niobium swelling only. The cubic ceramic trilayers showed less evidence of neutron damage than the Al₂O₃ ceramic trilayers but swelling was slightly greater than free standing bodies.

2-65

2.6 FUEL-CLAD IRRADIATION (M. Yates, E. Thacher, D. Allen, A. Marshall, L. Yang)

The objective of this subtask was to study the dimensional stability and compatibility of thermionic fuel-clad systems as a function of design and operating parameters of thermionic fuel elements. To allow more flexibility in the incorporation of test variables and more reliability in the instrumentation for measuring the temperature of the test sample, a material irradiation capsule was selected as the test vehicle. Capsule irradiation of carbide fuel-clad systems will be carried out under a NASA-sponsored program. Capsule irradiation of oxide fuel-clad systems is included under this subtask.

Previously, two capsules, FC-1 and FC-2, were fabricated for the evaluation of the effect of cladding temperature and thickness on the dimensional stability of oxide fueled tungsten emitters (Ref. 21). The operational histories and test summaries are given in previous thirdly reports. Significant amounts of emitter distortion were observed in most of the FC-1 and FC-2 emitters. However, these results were not consistent with the results of the recent TFE tests, 6F3 and 1F1,as described in Section 3 of this report. One possible reason for the discrepancy is the difference in the pressure of the cover gas over the oxide fuel in these devices. The helium cover gas pressure was about one atmosphere over the fuel in the FC-1 and FC-2 capsules, as compared to the pressure of fission gases of 5-20 torr over the fuels in 6F3 and 1F1.

To continue the evaluation of the irradiation behaviors of oxide fueled emitters, two capsules, FC-3 and FC-4, were planned for the FY-73 program. The status of these capsules is summarized below.

2.6.1 FC-3 Oxide Fuel Capsule

Even though the observed dimensional changes in 6F3 and 1F1 are below those seen in FC-1 and FC-2, further improvements were desired. For this reason, an oxide fueled capsule designated as FC-3 was designed. This capsule contained four emitters; the major characteristics and schematic layout of the capsule and the emitters were given in prior thirdly reports (Refs. 22, 23). The capsule assembly

2-66

was completed in November 1972, and the capsule was put on test on the 17th of November. A total of 1379 hours was accumulated by the time the reactor was shut down on January 22, 1973. The test history and the results of the neutron radiographs taken after shutdown are given in Section 3.5.

Thermal analysis of the FC-3 capsule was performed with varying input power and helium and neon in the space between primary and secondary containments.

Figure 2-26 illustrates the calculational model. The principal heat transfer (radial, cold) gap (emitter to heat rejector) is 2.11 mm (0.083 inch). The heat rejector to stainless containment gap is 0.13 nun (0.005 inch), and the gap from emitter transition to heat rejector is 0.13 mm (0.005 inch). The emitters incorporate a 0.51 mm (0.020 inch) thick stem. Clearance at the bottom of the emitter is 3.81 mm (0.15 inch).

The average emitter temperature, maximum emitter temperature and maximum heat rejector temperature as a function of emitter heat flux are presented in Figure 2-27. Data points represent four computer runs at varying input power using the two gases, helium and neon, in the secondary containment.

Figure 2-28 illustrates the detailed temperature map for the case of the emitter average temperature at 1822° K. Note that average temperatures, average gap thicknesses at temperature and emitter collector heat fluxes are given. The radial (I) coordinates (1 through 13) and the axial (J) coordinates (18 through 44) are the same as those indicated on the scale illustration in Figure 2-26.

2-67

2.6.2 FC-4 Oxide-Fuel Capsule

In addition to FC-3, another oxide-fueled capsule, designated as FC-4, was planned for the evaluation of other fuel forms and thoriated tungsten cladding. The basic ground rules adopted for this capsule were based on the FC-3 ground rules, which were discussed in Reference 20.

Several changes in mechanical detail from the FC-3 configuration were incorporated in FC-4. These are documented on drawing 278-SK-2079E and its supporting subassemblies and details. The cell fuel assembly and emitter assembly designs are documented on the drawings tabulated below in descending sequence from the capsule top. All cells were fueled with UO₂.

Cell No. Assembly Drawing No.
1 Emitter 278-SK-2053-28
Fuel 278-SK-2151-1

2 Emitter 278-SK-2053-32
Fuel 278-SK-2166-l

3 Emitter 278-SK-2053-30
Fuel 278-SK-2169-l
4 Emitter 278-SK-2053-26
Fuel 278-SK-2050-l

Cell 1 fuel is insulated from the clad and on each end by 0.030 in. of ThO₂. The intent of this is to create a minimum temperature difference of 100°K between the fuel and the clad. Thus, the more plastic, weaker fuel would swell into the fuel central cavity rather than distort the clad. It also seemed possible that reduction or elimination of fuel ratcheting might occur because of the physical isolation of the fuel from the clad. Thoria was chosen as the insulator because:

a. Its thermal conductivity is low enough to allow a thin sleeve.

b. Its vapor pressure at the temperatures involved is low. Therefore, it would not redeposit significantly.

2-71

c. UO₂ diffusion through the ThO₂ was estimated to be small.

d. Thin ThO₂ sleeves are fabricable and relatively cheap.

The ThO₂ insulators on the fuel ends are required to reduce an anticipated increase in axial heat conduction in the fuel caused by the ThO₂ sleeve. The cell 2 fuel design is illustrated in Figure 2-29.

The fuel design for cell 2 (Figure 2-30) has six wedge-shaped slots equally spaced around its periphery. The purpose of the slots is to provide a volume at the clad-fuel boundary into which the fuel may swell. The total slot volume is enough to accept a maximum anticipated fuel volume increase of about 15%. At a constant fuel volume fraction of 0.72, the number of slots is a trade off between the optimum slot shape (small included angle, small radius at vertex), a fabricable slot shape (larger included angle, appreciable vertex radius), and the need to have a fuel central cavity diameter greater than 0.040 inch.

To prevent fuel redeposition inside the slots, each was sized so that it could be lined with 0.005 inch thick W-25Re foil. The foils are installed so that they are always in contact with the clad, but can be easily pushed back into the fuel slots, flush with the fuel surface, as the fuel expands during start-up. This should prevent the slot edges from placing concentrated loads on the clad inner surface.

The fuel for cell 3 (Figure 2-31) is a stack of ten fuel pellets and nine equally spaced W-25Re disc fins. The fins are 0.020 inch thick and have a 0.050 inch diameter central hole to allow fission gas passage. Each pellet is 0.182 inch thick and has a central hole. The hole diameter depends on the percent of theoretical density of the fuel. The fuel volume fraction of this configuration is also 0.72.

2-72

The disc fins take advantage of the redeposition characteristics of the fuel. They are sized to minimize the redeposited fuel thickness on the clad. Since there is then less fission-gas-producing material next to the clad, less clad deformation is expected. The fins run colder than the fuel, so the fuel tends to redeposit on them as well as on the clad.

The clearance between the disc fin outer diameter and the clad inside diameter was made equal to the differential expansion between the fins and the clad. This prevents loading of the clad by the fins.

The last cell in the fuel-clad capsule contains a conventional UO₂ fuel configuration sized for 0.85 FVF. However, the fuel cladding is thoriated tungsten coated with chloride tungsten instead of fluoride-chloride duplex tungsten.

The thoriated tungsten provides higher creep strength than fluoride tungsten, while the chloride tungsten layer provides the high vacuum work function needed for improved thermionic performance.

2-76

2.7 HOT CELL EXAMINATIONS (H. Johnson, G. Buzzelli, L. Yang)

The objective of this subtask is to study the dimensions, macroscopic appearances, microstructures, fission product distribution, and joint integrity of the components of capsules, converters and thermionic fuel elements after in-pile operation. The information is fed back into the material and thermionic fuel element development programs for the improvement of the life and performance of thermionic space power systems.

During this reporting period, the main effort was devoted to the examinations of the 90 UC-10 ZrC-4 wt % W fueled TFE 6F1 and TFE 6F2. The results are described as follows.

2.7.1 Examination of 90 UC-10 ZrC-4 wt % W Fueled Thermionic Fuel Element 6F1

The essential components and operating history for 6F1 were described in Table 2-6 of the last thirdly report. The fission product release and distribution in 6F1 were studied during the last reporting period and the results obtained were also included in the last thirdly report. Macroscopic and metallographic examinations of various 6F1 components were carried out during this reporting period.

2.7.1.1 Emitter deformation. The maximum diametric expansions for the top five emitters are as follows: No. 1 (top emitter) 23 mils, 1/2 inch from emitter bottom; No. 2, 17 mils, 1/2 inch from emitter bottom; No. 3, 5 mils, 1/4 inch from emitter bottom; No. 4, 17 mils, 1/2 inch from emitter bottom; No. 5, 18 mils, 1 inch from emitter bottom. These variations of emitter deformation probably reflect the temperature profiles of these emitters. Cracks were visible in the claddings of all these five emitters, mostly along grain boundaries (Figure 2-32).

2-77

2.7.1.2 Macroscopic Appearance of Emitter Cross Section. Typical examples are shown in Figures 2-33 and 2-34 for emitter No. 1 and emitter No. 3, respectively. The top fuel pellet of emitter No. 1 was loose and fell out of the emitter cavity during handling; therefore, only seven fuel pellets were shown in Figure 2-33. It can be seen that the fuel pellets showed little or no swelling near the top of emitter No. 1. On the other hand, the fuel pellets near the bottom of emitter No. 1 exhibited such a large amount of swelling that the fuel central cavity disappeared. This indicates the presence of a very large axial temperature gradient in emitter No. 1. The appearance of the cross section of emitter No. 3 (Figure 2-34) implies a much lower axial temperature gradient in this emitter and the fuel central cavity showed the typical hourglass shape.

2.7.1.3 Fuel-Cladding Interaction. Figures 2-35 and 2-36 illustrate the fuel-cladding interaction layer. In each emitter the interaction layer becomes thinner near the emitter stem where the temperature is lower. The maximum thickness for the four emitters examined are: No. 1, 8 mils; No. 2, 12 mils; No. 3, 8 mils; No. 4, 7 mils. There does not appear to be a relationship between fuel enrichment and the thickness of the fuel-cladding interaction layer.

2.7.1.4 Microstructures of Fuel. The fuel body in all the emitters examined remained highly porous, although the fuel density appeared to be slightly higher in the fuel body near the cladding (Figures 2-37 (a), (b) and (c)). The lack of densification in spite of the high operating temperature of the fuel is probably caused by the presence of closed pores containing fission gases. The slightly higher fuel density near the cladding may be caused by the vapor transport of fuel material toward the cooler region. There is no evidence of segregation of tungsten-containing phases at the surfaces between fuel pellets and at the fuel-cladding interfaces (Figure 2-38) as was observed in the UC fueled 1F2 emitter (see last thirdly report).

2-79

2.7.1.5 Microstructures of Cladding. The cladding in general exhibited columnar grain structures for the fluoride tungsten and equiaxial grain structure for the chloride tungsten (see Figure 2-36). Grain growth in fluoride tungsten was observed in regions where large cladding deformation was observed. In these cases, the boundary between the chloride tungsten and the fluoride tungsten vanished and porosities appeared at the grain boundaries, which eventually led to grain boundary cracking. (See Figures 2-39 and 2-40.)

2.7.1.6 Bellvelle Washer Support. For all the emitters examined, at least part of the Al₂O₃ disc supporting the Bellvelle washer disintegrated into individual grains (Figure 2-41). In addition, the tantalum support reacted with the Al₂O₃ disc to form a white interaction layer (Figure 2-42). It appears that the Al₂O₃ disc was subjected to temperatures much higher than the design value.

2.7.1.7 Collector-Trilayer. Gaps were found between the ceramic layer and the niobium outer sheath of each of the collector-trilayer structures examined. Typical examples are shown in Figure 2-43. In some cases, gaps were also observed between the ceramic layer and the niobium collector. These gaps were probably formed by the sintering of the plasma-sprayed Al₂O₃ during the in-pile operation of the TFE. The presence of these gaps impeded the heat transfer between the collector and the niobium outer sheath and caused the collector to operate at temperatures much higher than the design value. The microhardness of the niobium at the I.D. of the ceramic (collector side) was higher than that at the O.D. of the ceramic (outer sheath side) (Knoop hardness = 200-230 versus 140-150), while the microhardness of the niobium collector and the niobium outer sheath (Knoop hardness = 100-120) was lower than these values. This indicates that contamination of the niobium surrounding the Al₂O₃ layer occurred during the operation of the TFE and that the collector side suffered more contamination because of its higher operating temperature.

2-86

2.7.1.8 Intercell Region and Fission Product Venting Orifices. The plasma-sprayed Al₂O₃ insulation layer remained white except at locations surrounding the fission product venting orifices where a dark region existed (Figure 2-44). This dark region was believed to be caused by the interaction of fission products, such as Sr and Ba, with the A₂O₃. Similar interactions between Al₂O₃ and Sr and Ba fission products were observed in the two cell E-series TFE 2E3. The reaction shown by 6F1 was less severe than that by 2E3, probably because the fission product release was less from the 6F1 fuel. The low fission product release from 6F1 fuel was attributed to the fact that most of its open porosity was lost when being operated at very high temperatures because of poor heat transfer across the fuel-cladding gap formed by fuel sintering. Figure 2-44 also shows the presence of cracks in the Al₂O₃ layer at the bottom of the convolution of the insulator skirt. These cracks were probably formed by stress induced through thermal cycling. Their presence may impair the ability of the Al₂O₃ insulation layer to prevent cesium arcing between the outer sheath and the insulator skirt. Figure 2-45 shows the surfaces of the Nb-1 Zr sheath around the fission product venting orifices in emitter No. 2. There was no heavy deposit on these surfaces, although the areas directly facing the orifices showed a darker color. The lack of heavy deposit is consistent with the observed low fission product release from 6F1.

To insure that the low fission product release was not due to the plugging of the Al₂O₃ venting tubes, each intercell region was mounted, sectioned and polished to reveal the cross section of these venting tubes. Figure 2-46 is a typical example of such cross sections. No significant plugging was found in all the samples examined.

2-90

2.7.1.9 Joints. No major defects were observed in all the joints examined. A white phase, however, was found in the weld joints between niobium and tantalum (see Figures 2-47 and 2-48) in the intercell regions above emitters No. 2, 3, 4 and 5 (Emitter No. 6 was not examined). A similar white phase was also present in the niobium convolutions of the insulator skirts of these emitters (see Figure 2-49), which grew from the side in contact with the cesium vapor toward the side coated with the plasma-sprayed Al₂O₃ insulation layer. This white phase was probably formed by some contaminants, such as oxygen, in the cesium vapor and its formation was sensitive to temperature. The niobium-to-tantalum welds above emitter No. 1 did not contain this white phase, probably because the temperature there was cooler than that above the other emitters because of heat loss through the electrode lead. Samples were sent to LASL hot cell for the determination of the composition of this white phase and the nature of the contaminant, but the work was not completed due to the termination of the thermionic program.

Figure 2-50 shows the tantalum-to-tungsten diffusion bond in emitter No. 2. Some Kirkendall voids were found on the tantalum side in the flat portion of the bond but the tapered portion of the bond contained fewer Kirkendall voids. The bond in general appeared to be in good condition. Examinations of the tantalum-tungsten diffusion bond in emitters No. 1, 3, 4 and 5 yielded the same results. These observations are in general agreement with the results obtained in previous hot cell examinations made on other thermionic devices. The tantalum-to-tungsten diffusion bond appears to be a promising joint for long-term application in a thermionic fuel element.

2.7.2 Examination of 90 UC-10ZrC-4 wt% W Fueled Thermionic Fuel Element 6F2

Due to the termination of the thermionic program early in January 1973, hot cell examination of 6F2 was limited to macroscopic examinations of the emitters, the Al₂O₃ insulation layers in the intercell regions, and the Al₂O₃ support for the Bellvelle washer, and the determination of the fission gas Kr-85 released into the charcoal trap.

The essential components and operating history of 6F2 are summarized in Table 2-7.

2-93

2.7.2.1 Determination of Fission Gas Release into Charcoal Trap. A total of 79.0 millicuries of Kr-85 was collected from the charcoal trap after bakeout at 250°C. This corresponded to 3.5% of the Kr-85 formed during the irradiation. This was much higher than that present in the charcoal trap of 6F1 (2 x 10^-5%). The release, however, was much less than that observed for carbide fuels irradiated in high pressure (~ 1 atm) inert gas environment (~ 40-60%). This is probably caused by the loss of open porosity in the fuel body upon irradiation in low pressure (< 20 torr) inert gas when the fuel temperatures reached very high values because of poor heat transfer across the fuel cladding gap.

2.7.2.2 Macroscopic Examination of the Emitters. All six emitters had cracks in the cladding. A typical example is shown in Figure 2-51. The cracks appear to follow the grain boundaries.

2.7.2.3 Intercell Region Al₂O₃ Insulation Layer. The Al₂O₃ insulation layer of the intercell region was darkened around each fission product venting hole. The area of the darkened region was much bigger than that in 6F1. This is consistent with the observed larger release fraction (3.5% versus 2 x 10^-5%) of fission gas (presumably also other fission products) through these venting orifices into the charcoal trap. The dark layer is believed to be formed by the interaction between Al₂O₃ and the fission products Sr and Ba. A typical example is shown in Figure 2-52. The Nb-1 Zr sheath surface facing the intercell region in Figure 2-52 is shown in Figure 2-53. It has a general dark appearance, probably due to the condensation of some of the vented fission products such as Cs, Sr and Ba.

2.7.2.4 Al₂O₃ support for Bellvelle Washer. All the Al₂O₃ supports for the Bellvelle washers in 6F2 were found to be cracked, presumably by the stress imposed when the emitter structure expanded during heating. A typical example is shown in Figure 2-54. This is an area in need of better design in any future TFE developmental program.

2-97

1. Yang, L., Hudson, R.G., Johnson, H., Horner, H., and Allen, D.T., "Material Development for Thermionic Fuel-Cladding Systems". Presented at the 3rd International Conference on Thermionic Electric Power Generation, Julich, Federal Republic of Germany, June 5-9, 1972.

2. Yang, L., and Chin, J., "Development Status of Thermionic Materials", Presented at the 1972 IECEC Conference, San Diego, California, September 25, 1972.

3. Chin, J., and Messick, C. W., "High Temperature Ceramic-to-Metal Seal Development". GA-7752, June 1968.

4. Chin, J., and Messick, C. W., "High Temperature Ceramic-to-Metal Seal Development". GA-9118, February, 1969.

5. Chin, J., and Messick, C. W., "Ceramic-to-Metal Component Development for Thermionic Fuel Elements". Presented at the 3rd International Conference on Thermionic Electric Power Generator, Julich, Federal Republic of Germany, June 5-9, 1972.

6. Chin, J., and Messick, C. W., "Sheath Insulator Investigations". GA-8951, October 1968.

7. Chin, J., and Messick, C. W., "Thermal Stability Testing of an Electrically Loaded Sheath Insulator". GA-9465, October 1969.

8. Chin, J., and Messick, C. W., "Sheath Insulator Development", GA-9590, October 1970.

9. Nowotny, H., et.al. , Monatschaft Chem. 89, 692, 1958.

10. Ugajin, M., and Takahashi, I., "The Phase Reaction in the UC-W System". Journal of Nuclear Materials 37, 30, 1970.

11. Dushman, S., "Scientific Foundations of Vacuum Technique". Chapter 2, August, 1962.

12. Weber, S., "Uber den Zusammenhang Zwischen der Lammaren Stromung der reinen Gase durch Rohre und dem Selbstdiffusion-Kueffizienten". Danske Matematisk-fy stske Meddelelser 28, No. 2 1954

2-102

13. Reichelt, W. H., Backus, C. E., Patrick, A.J., and Ranken, W. A., "Effect of Fast Neutron Irradiation on Ceramics and Ceramic-Metal Seals". Conference Record of 1971 Thermionic Conversion Specialist Conference, October 4-6, 1971. p.128-p.132.

14. Roth, R. S., and Schneider, S. J., "Phase Equilibra in Systems Involving the Rare Earth Oxides. Part I. Polymorphism of the Oxides of the Trivalent Rare Earth Ions". J. of Research of the National Bureau of Standards, 64A(4), 309-316, 1960.

15. Roth, R. S., and Schneider, S. J., "Phase Equilibra in Systems Involving the Rare Earth Oxides. Part II. Solid State Reactions in Trivalent Rare Earth Oxide Systems". J. of Research of the National Bureau of Standards, 64A(4), 317-332, 1960.

16. Roth, R. S., Schneider, S. J., and Waring, J. L., "Solid State Reactions Involving Oxides of Trivalent Cations", J. of Research of the National Bureau of Standards, 65A(4), 345-374, 1961.

17. Borchardt, Hans J., "Yttrium-Tungsten Oxides", Inorganic Chemistry, 2 (1), 170-173, 1963.

18. Komeya, K., and Inoue, H., "The Influence of Fibrous Aluminum Nitride on the Strength of Sintered AlN-Y₂O₃", Trans. British Ceramic Soc. 70 (3), 107-113, 1971.

19. Dragoo, N., and Diamond, J. J., "Transitions in Vapor-deposited alumina from 300-1200°C". J. Amer. Ceramic Soc. 50 (11), 568-574, 1967.

20. "Development of a Thermionic Reactor Space Power System", Summary Report for Period July 1, 1972 through October 31, 1972. GA-A12397.

21. "Development of a Thermionic Reactor Space Power System", Summary Report for Period July 1, 1970 through October 31, 1970. GA-11009.

22. "Development of a Thermionic Reactor Space Power System", Summary Report for Period November 1, 1971 through February 29, 1972, GA-A11073.

23. "Development of a Thermionic Reactor Space Power System". Summary Report for Period March 1, 1972 through June 30, 1973, GA-A12251.

2-103