Tungsten and its alloys can be successfully joined by gas tungsten-arc welding,
gas tungsten-arc braze welding, electron beam welding and by chemical vapor deposition.
The weldability of tungsten and a number of its alloys consolidated by arc casting, powder metallurgy, or chemical-vapor deposition (CVD) techniques was evaluated. Most of the materials used were nominally 0.060 in. thick sheet. The joining processes employed were (1) gas tungsten-arc welding, (2) gas tungsten-arc braze welding, (3) electron beam welding and (4) joining by CVD.
Tungsten was successfully welded by all of these methods but the soundness of the welds was greatly influenced by the types of base and filler metals (i.e. powder or arc-cast products). For example, welds in arc-cast material were comparatively free of porosity whereas welds in powder metallurgy products were usually porous, particularly along the fusion line. For gas tungsten-arc (GTA) welds in 1/ 1r, in. unalloyed tungsten sheet, a minimum preheat of 150° C (which was found to be the ductileto-brittle transition temperature of the base metal) produced welds free of cracks. As base metals, tungsten-rhenium alloys were weldable without preheat, but porosity was also a problem with tungsten alloy powder products. Preheating appeared not to affect weld porosity which was primarily a function of the type of base metal.
The ductile-to-brittle transition ternperatures (DBIT) for gas tungsten-arc welds in different types of powder metallurgy tungsten were 325 to 475° C, as compared to 150。 C for the base metal and that of 425° C for electron beamwelded arc-cast tungsten.
Braze welding of tungsten with dissimilar filler metals apparently did not produce better joint properties than did other joining methods. We used Nb, Ta, W-26% Re, Mo and Re as filler metals in the braze welds. The Nb and Mo caused severe cracking.
Joining by CVD at 510 to 560° C
eliminated all but a small amount of porosity and also eliminated the problems associated with the high temperatures necessary for welding (such as large grains in the weld and heat-affected zones).
Introduction
Tungsten and tungsten-base alloys are being considered for a number of advanced nuclear and space applications including thermionic conversion devices, reentry vehicles, high temperature fuel elements and other reactor components. Advantages of these materials are their combinations of very high melting temperatures, good strengths at elevated temperatures, high thermal and electrical conductivities and adequate resistance to corrosion in certain environments. Since brittleness limits their fabricability, the usefulness of these materials in structural components under rigorous service conditions depends greatly upon the development of welding procedures to provide joints that are comparable in properties to the base metal. Therefore, the objectives of these studies were to (1) determine the mechanical properties of joints produced by different joining methods in several types of unalloyed and alloyed tungsten; (2) evaluate the effects of various modifications in heat treatments and joining technique; and (3) demonstrate the feasibility of fabricating test components suitable for specific applications.
Materials
Unalloyed tungsten m叮10 m. thick sheets was the material of most interest. The unalloyed tungsten in this study was produced by powder metallurgy, arc casting and chemical-vapor deposition techniques. Table 1 shows the impurity levels of the powder metallurgy, CVD and arc-cast tungsten products as received. Most fall within the ranges nominally found in tungsten
but it should be noted that the CVD material contained more than the norma] amounts of fluorine.
Various sizes and shapes of tungsten and tungsten alloys were joined for comparison. Most of them were powder metallurgy products although some arc-cast materials were also welded. Specific configurations were used to determine the feasibility of building structures and components. All matenals were received in a fully cold worked condition with the exception of the CVD tungsten, which was received as-deposited. Because of the increased brittleness of recrystallized and large-grained tungsten the material was welded in the worked condition to minimize grain growth in the heataffected zone. B e c a u s e o f th e high cost of the material and the relatively small amounts available, we designed test specimens that used the minimum amount of material consistent with obtaining the desired information.
Procedure
Since the ductile-to-brittle transition temperature (DBTT) of tungsten is above room temperature, special care must be used in handling and machining to avoid cracking1. Shearing causes edge cracking and we have found that grinding and electrodischarge machining leave heat checks on the surface. Unless they are removed by lapping, these cracks may propagate during welding and subsequent use.
Tungsten, like all refractory metals, must be welded in a very pure atmosphere of either inert gas (gas tungsten-arc process) or vacuum (electron beam pro:::ess)2 to avoid contamination of the weld by interstitials. Since tungsten has the highest melting point of all metals (3410° C), welding equipment must be capable of withstanding the high service temperatures.
Table 1
Three different welding processes were used: gas tungsten-arc welding, gas tungsten-arc braze welding and electron beam welding. Welding conditions necessary for complete pcnetration at a minimum energy input were determined for each material. Before welding, sheet material was machined into囚in. wide blanks and degreased with ethyl alcohol. The joint design was a square groove with no root opening.
Gas Tungsten-Arc Welding
All automatie and manual gas tungsten-arc welds were made in a ehamher that was maintained below 5 x I o-r. torr for about 1 hr and then backfilled with very pure argon. As shown in Fig. lA, the chamber was fitted with a traversing mechanism and torch head for automatic welding. The workpiece was held in a copper fixture provided with tungsten inserts at all points of contact to prevent it from being brazed to the work by the welding beat. The base of this fixture housed the electric cartridge heaters that preheated the work to the desired temperature, Fig. 1 B. All welds were made at a travel speed off 10 ipm, a eurrent of about 350 amp and a voltage of 10 to 15 v.
Gas Tungsten-A『c Braze Welding
Gas tungsten-are braze welds were made in a ehamber with an inert atmosphere by techniques similar to
those described above. The bead-onplate braze welds made with tungsten and W—26% Re filler metal were made manually; however, the butt braze welds were welded automatically after the filler metal was placed in the butt joint.
Electron Beam Welding
The eleetron beam welds were made in a 150-kV 20-mA machine. A vacuum of about 5 x I o-6 torr was maintained during welding. Electron beam welding results in a very high ratio of depth to width and a narrow heat-affected zone.
』oining by Chemical Vapor Disposition
Tungsten joints were made by depositing unalloyed tungsten filler metal via the chemical vapor deposition process3. Tungsten was deposited by hydrogen reduction of tungsten hexafluoride according to the reaction-t
heat
WFs(g) + 3H,(g)一–+W(s) + 6HF(g).
The use of this technique for joining required only minor changes in fixtures and reactant flow distribution. The primary advantage of this process over more conventional methods of joining is that, since the low temperatures employed (510 to 650 ° C) are much lower than the melting point of
tungsten (3410 ° C), recrystallization and possible further cmbrittlement of the wrought tungsten base metal by impurities or grain growth are minimized.
Several joint designs including butt and tube-end closures were fabricated. Deposition was performed with the aid of a copper mandrel which was used as a fixture, alignment piece and substrate. After deposition was completed, the eopper mandrel was removed by etching. Since other work” has shown that CVD tungsten possesses complex residual stresses as deposited, these joints were stress relicvcd I hr at 1000 ° to 1600 ° C before machining or testing.
Inspection and Testing
Joints were inspected visually and by liquid penetrant and radiography before they were tested. Typical welds were chemically analyzed for oxygen and nitrogen (Table 2) and extensive metallographic examinations were performed throughout the study.
Because of its inherent simplicity and adaptability to small specimens, the bend test was used as the primary criterion for joint integrity and eomparison of the processes. Ductile-tobrittle transition temperatures were determined with a three-point bending apparatus for joints both as-welded and after aging. The basic specimen for the bend tests was the longitudinal
face bend, 24t long by 12t wide, where t is the specimen thickness. Specimens were supported on a 15t span and bent with a plunger of radius 4t at a rate of 0.5 ipm. This geometry tended to normalize data obtained on various thicknesses of materials. Specimens were usually bent transverse to the weld seam (longitudinal bend specimen) to provide uniform deformation of the weld, heat-affected zone and base metal; however, a few specimens were bent along the weld seam (transverse bend specimen) for comparison. Face bends were used in the initial portions of the investigation; however, because of the slight notch found on the faees of most welds due to the weight of the molten metal, root bends were substituted in later tests. The recommendations of the Materials Advisory Board6 concerned with bend testing of sheet specimens were followed as closely as possible. Because of limited material, the smallest advisable specimens were selected.
To determine the bend transition temperature, the bending apparatus was enclosed in a furnace capable of quickly raising the temperature to 500 ° C. A bend of 90 to 105 deg was eonsidered a full bend. The DBTT was defined as the lowest temperature at which the speeimen bent fully without craeking. Although the tests were conducted in air, discoloration of the specimens was not evident until test temperatures reached 400 ° C.
Figure 1
Results for Unalloyed Tungsten
General Weldability
Gas Turzgstea-Arc Welding—In gas tungsten-arc welding of 1乍in. thick unalloyed sheet, the work must be substantially preheated to prevent brittle failure under stress induced by thermal shock. Figure 2 shows a typic·al fracture produced by welding without proper preheating. The large grain size and shape of the weld and heataffected zone are evident in the fracture. Investigation of preheating ternperatures from room temperature to 540°C showed that preheating to a minimum of 150°C was necessary for consistent production of one-pass butt welds that were free of cracks. This temperature corresponds to the DBTI of the base metal. Preheating to higher temperatures did not appear to be necessary in these tests but material with a higher DBTI, or configurations that involve more severe stress concentrations or more massive parts, may require preheating to higher ternperatures.
The quality of a weldment depends greatly upon the procedures used in fabricating the base metals. Autogenous welds in arc-cast tungsten are essentially free from porosity, Fig.
3A, but welds in powder metallurgy tungsten are characterized by gross porosity, Fig. 3 (b), particularly along the fusion line. The amount of this porosity, Fig. 3B, particularly along 3C, in welds made in a proprietary, low porosity product (GE-15 produced by General Electric Co., Cleveland).
Gas tungsten-arc welds in CVD tungsten have unusual heat-affected zones due to the grain structure 0£the base metaF. Figure 4 shows the face and corresponding cross section of such a gas tungsten-arc butt weld. Note that the fine grains at the substrate surface have grown due to the heat of welding. Also evident is the lack of growth of the large columnar
grains. The columnar grains have gas
bubb_les at grain boundaries caused by fluorme impurities8. Consequently, if
the fine grain substrate surface is removed before welding, the weldment does not contain a metallographically detectable heat-affected zone. Of course, in worked CVD material (such as extruded or drawn tubing) the heat-affected zone of the weld has the normal recrystallized grain structure.
Cracks were found in the columnar grain boundaries in the RAZ of several welds in CVD tungsten. This cracking, shown in Fig. 5, was caused by rapid formation and growth of bubbles in the grain boundaries at hightemperatures9. At the high temperatures involved in welding, the bubbles were able to consume much of the grain boundary area; this, combined with the stress produced during cooling, pulled the grain boundaries apart to form a crack. A study of bubble formation in tungsten and other metal deposits during heat treatment shows that bubbles occur in metals deposited below 0.3 Tm (the homologous melting temperature). This observation suggests that gas bubbles form by coalescence of entrapped vacancies and gases during annealing. In the case of CVD tungsten, the gas is probably fluorine or a fluoride compound
Electron Beam Welding—Unalloyed tungsten was electron beam welded with and without preheating. The need for preheat varied with the specimen. To ensure a weld free of cracks, preheating at least to the DBTT of the base metal is recommended. Electron beam welds in powder metallurgy products also have the weld porosity mentioned previously.
Gas Tungsten-Arc Braze Welding一In an effort to establish whether braze welding could be used to advantage, we experimented with the gas tungstenarc process for making braze welds on powder metallurgy tungsten sheet、 The braze welds were made by preplacing the filler metal along the butt joint before welding. Braze welds were produced with unalloyed Nb, Ta, Mo, Re, and W-26% Re as filler metals. As expected, there was porosity at the fusion line in metallographic sections of all joints (Fig. 6) since the base metals were powder metallurgy products. Welds made with niobium and molybdenum filler metals cracked.
The hardnesses of welds and braze welds were compared by means of a study of bead-on-plate welds made with unalloyed tungsten and W一26% Re as filler metals. The gas tungstenarc welds and braze welds were made manually on unalloyed tungsten powder metallurgy products (the low porosity, proprietary (GE-15) grade and a typical commercial grade). Welds and braze welds in each material were aged at 900, 1200, 1600 and 2000°C for l, 10, 100 and 1000 hr. The specimens were examinedmetallographically, and hardness traverses were taken across the weld, heataffected zone, and base metal both as-welded and after heat treatment.
Table 2
Figure2
Since the materials used in this study were powder metallurgy products, varying amounts of porosity were present in the weld and braze weld deposits. Again, the joints made with typical powder metallurgy tungsten base metal had more porosity than those made with the low porosity, proprietary tungsten. The braze welds made with W—26% Re filler metal had less porosity than the welds made with the unalloyed tungsten filler metal.
No effect of time or temperature was discerned on the hardness of the welds made with unalloyed tungsten as filler metal. As welded, the hardness measurements of the weld and base metals were essentially constant and did not change after aging. However, the braze welds made with W—26% Re filler metal were considerably harder as produced than the base metal (Fig. 7). Probably the higher hardness of the W-Re br立e weld deposit was due to solid solution hardening and/ or the presence of er phase finely distributed in the solidified structure. The tungstenrhenium phase diagram11 shows that localized areas of high rhenium content could occur during rapid cooling and result in the formation of the hard, brittle er phase in the highly segregated substructure. Possibly the er phase was finely dispersed in the grains or grain boundaries, although none was large enough to be identified by either metallographic examination or X-ray diffraction.
Hardness is plotted as a function of distance from the braze-weld center line for different aging temperatures in Fig. 7A. Note the abrupt change
in hardness at the fusion line. With increasing aging temperature, the hardness of the braze weld decreased until, after 100 hr at J 600° C, the hardness was the same as that of the unalloyed tungsten base metal. This trend of decreasing hardness with increasing temperature held true for all aging times. Increasing time at a constant temperature also caused a simiJar decrease in hardness, as shown for an aging temperature of 1200° C in Fig. 7B.
Joining by Chemical Vapor Deposition—Joining of tungsten by CVD techniques was investigated as a method for producing welds in various specimen designs. By use of appropriate fixtures and masks to limit deposition to the desired areas, CVD and powder metallurgy tungsten sheets were joined and end closures on tubing were produced. Deposition into a bevel with an included angle of about 90 deg produced cracking, Fig. 8A, at the intersections of columnar grains growing from one face of the bevel and the substrate (which was etched away). However, high integrity joints without cracking or gross buildup of impurities were obtained, Fig. 8B, when the joint configuration was changed by grinding the face of the base metal to a radius of飞in. tangent to the root of the weld. To demonstrate a typical application of this process in fabrication of fuel elements, a few end closures were made in tungsten tubes. These joints were leak-tight when tested with a helium mass spectrorr:eter leak detector.
Figure 3
Figure 4
Figure 5
Mechanical Properties
Bend Tests of Fusion Welds一Ductile-to-brittle transition curves were determined for various joints in unalloyed tungsten. The curves in Fig. 9 shows that the DBTT of two powder metallurgy base metals was about I 50° C. Typically, the DBTT (the lowest temperature at which a 90 to 105 deg bend could be made) of both materials increased greatly after welding. The transition temperatures increased about 175° C to a value of 325° C for typical powder metallurgy tungsten and increased about 235° C to a value of 385° C for the low porosity, proprietary material. The difference in the DBTTs of welded and unwelded material was attributed to the large grain size and possible redistribution of impurities of the welds and heat-affected zones. The test results show that the DBTT of typical powder metallurgy tungsten welds was lower than that of the proprietary material, even though the latter had less porosity. The higher DBTT of the weld in the low porosity tungsten may have been due to its slightly larger grain size, Fig. 3A and 3C.
The results of investigations to determine DBTT’s for a number of joints in unalloyed tungsten are summarized in Table 3. The bend tests were quite sensitive to changes in testing procedure. Root bends appeared to be more ductile than face bends. A properly selected stress relief after welding appeared to lower the DBTT substantially. The CVD tungsten had, as welded, the highest DBTT (560℃);yet when it was given a 1 hr stress relief of 1000℃ after welding, its DBTT dropped to 350℃. stress relief of 1000° C after welding, its DBTT dropped to 350° C. Stress relief of arc welded powder metallurgy tungsten for 1 hr at 18000 C reduced the DBTT of this material by about 100° C from the value determined for it as-welded. A stress relief of 1 hr at 1000° C on a joint made by CVD methods produced the lowest DBTT (200° C). It should be noted that, while this transition ternperature was considerably lower than any other transition temperature determined in this study, the improvement was probably influenced by the lower strain rate (0.1 vs 0.5 ipm) used in tests on CVD joints.
Bend Test of braze welds-gas tungsten-arc braze welds made with Nb. Ta, Mo, Re, and W-26% Re as filler metals were also bend tested and the results anre summarized in table 4. the most ductility was obtained with a rhenium braze weld.
Although the results of this cursory study indicate that a dissimilar filler metal may produce joints with mechaniccal properties interior to thouse of homogeneous welds in tungsten,some of these filler metals may be useful in practice.
Results for Tungsten Alloys.