Electron Beam Welding of Dissimilar Metals
Using electron beam welding and braze welding tech niques, thirty-five dissimilar metal joints were made and rated on the basis of visual, metallographic, tensile and bend testing
BY G. METZGER AND R. LISON
ABSTRACT. At the outset of this study Cu/Fe18Cr8Ni, Cu/V10Ti, In order to take full advantage of on the weldability of dissimilar metal Cu20Ni/Fe18Cr8Ni and Ti/Zr2Sn the properties of various materials, it joints, some preliminary information welds. The Cu/Ni weld had deep un is necessary to produce high quality on the weld metal properties of dercut, but was in other respects ex joints between them. Only in this way various joint combinations was ob cellent. The mechanical properties of can the designer use the most suit tained from arc melted buttons. When the Ag/Fe18Cr8Ni weld were poor, able materials for each part of a given the information proved to be of min but the defect could probably be cor structure. The growing availability of imum value in predicting the weld rected. Difficulty with cracking was new materials and the higher require ability of the electron beam welds un experienced with the Al/Ni and ments being placed on materials cre der study, this approach was dis Fe18Cr8Ni/V welds, but sound welds ates a greater need for joints of dis continued. had excellent mechanical properties. similar metals. In those applications The remaining welds Al-Cu, where adhesive bonding or mechan Thirty-three two-member combi AI/Cu20Ni, AI/Fe18Cr8Ni, ical fasteners are not acceptable, nations of dissimilar metals were AI/Ni15Cr7Fe, Cu20Ni/V, welded or brazed joints must be used. electron beam welded as square- groove butt joints in 0.08 and 0.12 in. Cu20Ni/V10Ti, Cb/Zr2Sn, Ni/Ti, The electron beam welding process sheet material. Many joints were Ni15Cr7Fe/V, Ni15Cr7Fe/V10Ti, and has several advantages for welding "braze welded" by offsetting the elec Ti/V were unsuccessful, due to brittle dissimilar metals. The ability to tron beam about 0.02 in. from the butt phases, primarily at the weld metal- precisely locate the weld and the joint to achieve fusion of the lower base metal interface. favorable shape of the fusion zone melting point metal, but no signif In addition to the two-member permit good control of the relative icant fusion of the other member of specimens, several joints were made amounts of the two metals in the weld the pair. The welds were evaluated by by buttering. Longitudinal weld spec metal. When a given combination visual and metallographic examina imens of the three-member combi cannot be welded directly, interme tion, transverse tensile tests, and nation AI/Ni/Fe18Cr8Ni and the five diate metals may be used. The very bend tests. member combination narrow fusion zone of the electron Fe18Cr8Ni/V/Cb/Ti/Zr2Sn showed beam weld permits a greater number The welds Ag/AI, Ag/Ni15Cr7Fe, good tensile strength and satisfac of intermediate metals to be inserted Cu/NM5Cr7Fe, Cu/V, tory elongation. without the total width of the welded Cu20Ni/Ni15Cr7Fe, Fe18Cr8Ni/Ni, joint becoming too large. Fe18Cr8Ni/NM5Cr7Fe, Cb/Ti, Cb/V, Introduction The present paper gives the results Ni/NM5Cr7Fe and Cb/V10Ti were of an investigation, which was intend readily welded and weld properties The joining of dissimilar metals ed to further the development of were excellent. Others which had only often presents a difficult technolog welded joints of austenitic stainless minor defects included the ical problem, and in only a few in steel/aluminum and austenitic stain Ag/Cu20Ni, Ag/Ti, Ag/V, stances has this problem been satis less steel/zirconium alloy. In the factorily solved. In many cases, where course of the investigation other G. METZGER, formerly a research asso a welded or brazed joint of dissimilar pairs, which could possibly be ap ciate at the Institute ot Reactor Materials, metals would be advantageous, either plied as intermediate materials for the Atomic Research Center, Juelich, Ger the technique to produce the joint is above two pairs, were also welded. many, is now a Welding Engineer, Air not available or the properties of the The work was carried out in 1965 Force Materials Laboratory, Wright-Pat joint are not acceptable. In other through 1967 at the Atomic Research terson Air Force Base, Ohio. R. LISON is a cases mechanical fasteners or Center in Juelich, Germany and the research associate at the Atomic Re adhesive bonding is used, even search Center. complete results are given in a report Paper was presented at 49th AWS An though these methods do not pro (Ref. 1) published in German, which nual Meeting held at Chicago, Illinois, dur duce joints which are completely was reduced by about 70% for the ing 1-5 April 1968. satisfactory. present paper.
230-s | AUGUST 1976 The alloys are identified according Base Metals Copper Shield to their chemical composition; the -Copper Hold-Down Clamps B0-»- base metal is given first, followed by the alloying elements with the weight percent, to the nearest whole per Electron Beam cent, preceding each chemical sym Beam Angle- A bol. Broken Corner n of Copper Shield
Arc Melting Experiments In a preliminary experiment, small buttons were arc melted, using as alloy components various commer cial grade metals, which represented the base metals of welded dissimilar metal pairs. The buttons were exam ined by metallographic means and - copper Backup Bar hardness measurements to obtain in EB > ELECTRON BEAM CS • COPPER SHIELD formation on the weld metal prop Fig. 1 — Welding fixture BO < BEAM OFFSET BA = BEAM ANGLE erties of the corresponding dis similar metal welded joints. Fig. 2 — Weld cross-sections with various In some cases the weldability of a geometric relationships between the joint given pair could be estimated from and the electron beam the button properties; with other of the sheet at the joint, as shown in pairs, the welding results contradict Fig. 2a. ed the melting experiment results. It For dissimilar metals, which pro was concluded that, in general, the duce brittle, crack prone, or other value of melting experiments did not wise defective welds, when both justify the expended effort. The condi metals are fused, the electron beam tions that make melted buttons a poor was offset from the joint to limit fusion substitute for welded joints are the largely to the metal B of Fig. 2b. If the presence of diffusion layers, often fused upper corner of metal A caused brittle, at the interface of weld metal poor weld properties, its fusion was and base metal in a welded joint and, almost completely eliminated by the of lesser significance, the much faster use of a copper shield, as shown in cooling rate of a weld than of an arc Fig. 2c. In some cases, the weld root melted button. showed incomplete fusion, as illus trated in Fig. 2d. This flaw could usually be prevented by changing the Electron Beam Welding angle between the electron beam and Material* the sheet surface, as shown in Fig. 2e. In most cases, the electron beam The base metals were commercial was offset in the direction of the metal grade sheet of 2 or 3 mm (0.08 or 0.12 with the lower melting point. The elec in.) thickness in the annealed condi tron beam was perpendicular to the tion. The sheet specimens were pre plane of the sheet, unless otherwise pared for welding by milling a straight noted. edge perpendicular to the plane of the sheet, followed by pickling in a suit Each pair was first welded with a able acid solution. No filler metals number of short joints of 10 to 50 mm Fig. 3 — Ag/AI weld were used. (0.4 to 2 in.) length, in order to deter mine an approximation of suitable Welding Procedure welding conditions. Visual inspec tion, and in some cases metallo Welds were made in a vacuum of graphic examination also, was used to an average of 35% and the beam about 6.66 mPa (5 X 10"s torr) with an decide on further welding experi offset by an average of 22%. One pair, electron beam welding machine hav ments. Most of those welds which did Ni15Cr7Fe/V, was welded only in 2 ing a maximum power rating of 3 kW not appear to be virtually impossible mm (0.08 in.) sheet. at a beam potential of 150 kV and a of success, were welded with longer beam current of 20 mA. The machine joints in sufficient quantity to arrive at Visual Observations and welding conditions which yielded was equipped with a device for oscil Metallographic Results lation of the beam parallel or perpen satisfactory joints, as judged by visual dicular to the welding direction, but examination. These welding condi One, two, or three specimens, beam power pulsation was not avail tions were then used to produce depending on weld length, were able. welds for mechanical testing and removed, transverse to the weld, for All welds were square-groove butt detailed metallographic examination. metallographic examination. A sum joints, made by a single pass from The welding parameters for each mary of some of the metallographic one side in the flat position. The dissimilar weld, with a sheet thick results is presented in Table 2. Addi general arrangement for welding is ness of 3 mm (0.12 in.), are present tional metallographic information and shown in Fig. 1. If the metals of a ed in Table 1. Welds in 2 mm (0.08 in.) the results of visual observations are given dissimilar weld were metallurg sheet were made at the same weld given in the following discussion. ically compatible, the copper shield of ing speed and electron beam poten Unless otherwise noted, a satis Fig. 1 was not used and the electron tial as for the 3 mm (0.12 in.) welds, factory weld was produced with no beam was perpendicular to the plane but the beam current was reduced by significant flaws.
WELDING RESEARCH SUPPLEMENT! 231-8 It is to be noted that the photo AI/Fe18Cr8Ni weld failed. These in weld metal porosity were often ob graphs of the following discussion are cluded beam oscillation with offset to served. An example of severe under not representative of the weld quality limit fusion of steel and a beam angle fill is shown in Fig. 8. The obtained in the investigation. Most of of 15 deg, again with beam offset. Cu/Fe18Cr8Ni weld also contained the photographs were selected to il Most welds failed in the intermetallic grain boundary microcracks at the lustrate weld defects or irregularities compounds (640 to 1060 DPH) steel interface due to Cu penetration. which were observed. Two examples formed at the steel interface, Fig. 6. The combination of ferromagnetic Ni of many good quality welds are shown The W. M. had a hardness of only with diamagnetic Cu caused an in Figs. 3 and 4; the first (Ag/AI) weld about 75 DPH. erratic deflection of the electron beam ed with the beam offset to fuse only No difficulty was experienced in in the direction of the Cu, which was the Al, and the second producing excellent welded spec counteracted by welding with the (Fe18Cr8Ni/Ni15Cr7Fe) with the imens for mechanical testing of the beam offset in Ni and with oscillation beam on the joint. binary Al/Ni weld. The usual Ni/W. M. transverse to the joint. Severe under Molten Ag penetrated the Cu20Ni interface is shown in Fig. 7, although cut in Cu was observed with the Cu/Ni B. M. (base metal) of the Ag/Cu20Ni occasional localized inclusions were weld. The results for the Cu/V10Ti weld at a few locations up to 50 (im observed. Later when welding spec weld were similar to those of the Cu/V depth. In the Ag/Fe18Cr8Ni weld, a imens for the ternary weld, except severe grain boundary layer of about 20 ^m thick of steel AI/Ni/Fe18Cr8Ni weld, a dark grey, penetration of the Cu W. M. into the dendrites, interspersed with Ag, soot-like coating was often deposited V10Ti B. M. was present. formed in the steel B. M. at the W. M. during welding on the faying surface Numerous grain boundary micro (weld metal) interface. of the Ni, which caused poor wetting. cracks, due to W. M. penetration, A metallographic specimen of the Welding with a beam angle of 5 deg were present in the steel at its inter Al/Cu weld failed in the mounting improved the results, but the wetting face with the Cu20Ni W. M. in the press at the interface between Cu and flaws could not be consistently pre Cu20Ni/Fe18Cr8Ni weld. The results hard diffusion layer, Fig. 5. The hard vented. The AI/Ni15Cr7Fe weld con for the Cu20Ni/V10Ti weld were sim ness of the wide dark layer was 250 tained numerous cracks, after cool ilar to the Cu/V weld, except for the DPH. The behavior and microstruc ing from welding temperature, due to formation of a wide multiple-phase ture of the AI/Cu20Ni weld was sim brittle weld metal. zone at the V10Ti/W. M. interface, ilar to that of the Al/Cu weld. Welds, in which Cu was fused, had where light bending caused failure in All attempts to produce a good a tendency to underfill. Undercut and a brittle manner. For the reason de-
Table 1 — Welding Parameters. Sheet thickness: 3 mm (0.12 in.). Welding speed: 10 mm/s (23.6 in./min)
Base metal Side Side Side Electron beam offset (a) A B shielded kv mA,b| in mm in.
Ag Al Ag 130 14 Al 0.6 0.024 Ag Cu20Ni none 120 12 Ag 0.3 0.012 Ag Fe18Cr8Ni Fe18Cr8Ni 120 13 Ag 0.5 0.020 Ag Ni15Cr7Fe none 120 13 Ag 0.5 0.020 Ag Ti Ti 120 13 Ag 0.6 0.024 Ag V none 130 14 Ag 0.5 0.020 Al Cu Cu 120 15 Al 0.6 0.024 Al Cu20Ni Cu20Ni 120 13 Al 0.6 0.024 Al Fe18Cr8Ni see text Al Ni Ni 130 12 Al 0.6 0.024 Al Ni15Cr7Fe Ni15Cr7Fe 120 11 Al 0.7 0.028
Cu Fe18Cr8Ni Fe18Cr8Ni 120 13 Cu 0.5 0.020 Cu Ni none 130 12 Cu 0.3 0.012 Cu Ni15Cr7Fe none 120 15 Cu 0.3 0.012 Cu V V 140 18 Cu 0.8 0.032 Cu V10Ti V10Ti 140 16 Cu 0.5 0.020
Cu20Ni Fe18Cr8Ni Fe18Cr8Ni 130 10 Cu20Ni 0.6 0.024 Cu20Ni Ni15Cr7Fe none 130 9 — 0 0 Cu20Ni V V 130 16 Cu20Ni 0.6 0.024 Cu20Ni V10Ti V10Ti 140 16 Cu20Ni 0.5 0.020
Fe18Cr8Ni Ni none 130 8 Ni 0.3 0.012 Fe18Cr8Ni Ni15Cr7Fe none 130 10 — 0 0 Fe18Cr8Ni V V 120 9 Fe18Cr8Ni 0.4 0.016
Cb Ti none 130 13 Cb 0.4 0.016 Cb V none 120 12 — 0 0 Cb V10Ti none 130 14 — 0 0 Cb Zr2Sn none 130 12 — 0 0 Cb Zr2Sn Cb 130 12 Zr2Sn 0.5 0.020
Ni Ni15Cr7Fe none 130 8 Ni 0.3 0.012 Ni15Cr7Fe V10Ti V10Ti 130 13 Ni15Cr7Fe 0.4 0.020 Ti V none 120 9 V 0.3 0.012 Ti Zr2Sn none 120 10 — 0 0
(a) beam potential (Ul beam uunenl
232-S I AUGUST 1976 scribed with the Cu/Ni weld, the beam with a beam angle of 12 deg. A few a location with many cracks, polished was offset in Ni, in order to obtain a welds contained transverse cracks in on a section parallel to the sheet sur weld at the joint of the Fe18Cr8Ni/Ni the W. M., which were more numer face, is shown in Fig. 9. One weld weld. ous and wider on the steel side than cracked longitudinally in the W. M. The Fe18Cr8Ni/V pair was welded on the V side. A specimen taken from for a short distance, when a trans verse cut with a power shear was made. The Cb/Zr2Sn welds, which were welded with the beam on the joint, formed a diffusion layer with cracks at the Zr2Sn/W. M. interface, as shown in Fig. 10.The hardness of the diffu sion layer was 490 DPH and of the W. M. 200 DPH. Welds with the beam offset in Zr2Sn also had a diffusion layer with cracks, but it formed at the Cb/W. M. interface (Fig. 11). The hardness of the diffusion layer was 440 DPH and the W. M. 300 DPH. In an attempt to produce a Ni/Ti Nil5Cr7Fe weld with W. M. of the ductile NiTi in termetallic compound, welds were made with varying beam offset on either side of the joint. All the welds contained phases of very high hard i • ness from 530 to 1410 DPH and numerous cracks. w A double diffusion layer at the w ~ 500 pm Ti/W. M. interface was observed with 1 •—1 the Ti/V weld (see Fig. 12) with a typ WT ical crack at the interface of the two Fig. 4 — Fe18CrNi/NI15Cr7Fe weld diffusion layers. The hardness of the layer on the left was 230 DPH, the layer on the right 430 DPH, and in the W. M. 260 DPH. A weld broke open at the Ti/W. M. interface when a trans Table 2 — Metallographic Results verse cut was made with a power shear. B/W.M.(a) diffusion Mechanical Properties layer The transverse tensile strength of Base metal pair thick DPH (t>; Under Side A SideB lire DPH W.M. B.M fill Porosity binary welds was determined with the two types of tensile specimens of Ag Al 4 — 73 22 B none none Figs. 13a and 13b to determine the Ag Fe18Cr8Ni none 45 36 A moderate none effect of weld bead surface removal. Ni15Cr7Fe none 41 29 A none none Ag The results are presented in Table 3, Ag Ti none 74 36 A severe none where each average value repre Ag V none 39 29 A moderate severe sents three specimens. Al Cu 5 550 158 28 A none slight Bend specimens (Fig. 13c) were Al Cu20Ni 28 483 48 28 A none none made from welds in 3 mm (0.12 in.) 22 A Al Ni 1 — 47 none none sheet by machining 0.5 mm (0.02 in.) Cu Fe18Cr8Ni none 93 60 A severe slight from each side. These specimens Cu Ni none 100 60 A slight moderate were tested by first bending into a vee- Cu Ni15Cr7Fe none 160 60 A slight none block with an angle of 90 deg by a Cu V none 70 60 A slight none plunger of 50 mm (2 in.) radius, fol Cu20Ni Fe18Cr8Ni none 130 80 A none none lowed by plungers of decreasing radii Cu20Ni Ni15Cr7Fe none 145 80 A none none until the first crack appeared. The Cu20Ni V 40 300 115 105 A slight none results of the bend test are given in Table 4. Fe18Cr8Ni Ni none — 100 B none slight Fe18Cr8Ni Ni15Cr7Fe none 145 140 A none none Combinations of More Than Fe18Cr8Ni V none 305 140 A none none Two Materials Cb Ti 50 330 — 80 A none none Cb V none — 80 A none slight The two material combinations Cb V10Ti none — 80 A none none AI/Fe18Cr8Ni and Zr2Sn/Fe18Cr8Ni Ni Ni15Cr7Fe none — 100 A none none were welded in 3 mm (0.12 in.) sheet, Ni15Cr7Fe V 30 450 150 145 B none none with suitable intermediate metals. The joint preparation and the welding con Ni15Cr7Fe V10Ti 30 630 165 145 B none none ditions were as previously described Ti Zr2Sn 8 250 320 145 B moderate none for the binary combinations. After welding each joint of the multiple (a) Diffusion layer formed at the interface of the weld metal and base metal B. An AI/W.M diffusion layer formed only with the Ag/AI weld. The thickness was 10 *im and the hardness was 185 DPH. material combination, the inter (b) Base metal of lower hardness. mediate material was milled to a width
WELDING RESEARCH SUPPLEMENT) 233-8 of 2 mm (0.08 in.), before the next shown in Fig. 14. The results are given to successfully complete the quinary joint was welded. in Table 5. combination, as shown in a cross The welds were examined by section in Fig. 15. The results are metallographic methods and the ten given in Table 5. sile properties were determined. The Fe18Cr8Ni/V/Cb/Ti/Zr2Sn tensile specimens of Fig. 13b were The welding sequence for this Discussion used, except that the welds were quinary combination was as follows: parallel to the longitudinal axis and lo Welds with Silver Zr2Sn/Ti joint, Ti/Cb joint, cated in the center of the specimen. V/Fe18Cr8Ni joint, and the Cb/V joint Silver proved to be the best mate to complete the combination. While rial for use in dissimilar welds. All six AI/Ni/Fe18Cr8Ni welding the last joint, the previous welds with Ag, including those which The Al/Ni joint was welded first, V/Fe18Cr8Ni welds cracked longi form intermetallic compounds, Ag/AI and the ternary combination was tudinally in the W. M. along their en and Ag/Ti, were successfully welded. completed by welding the tire length. After removal of the re The greatest difficulty with Ag was ex Ni/Fe18Cr8Ni joint. A macrophoto- maining W. M. from the V by milling, cessive underfill, which appeared to graph of the weld cross section is a new steel sheet was welded to the V, be caused by violent expulsion of Ag particles from the molten weld metal. The most probable cause of this "spit ting" is the Ag vapor pressure at its melting point, which is about two orders of magnitude greater than the pressure in the welding chamber. The Ag vapor formed in the molten Ag could escape from the melt at high velocity, thus carrying with it particles of molten Ag. This action could also be the cause of porosity, if Ag vapor was entrapped by solidifying Ag. Other investigators were able to suc cessfully electron beam weld 3 mm (0.12 in.) sheet of the Ag/Ni pair without underfill (Ref. 2). Despite the presence of brittle in termetallic compounds in the Ag-AI phase diagram, the Ag/AI welds did not exhibit brittle behavior. The usual underfill, when welding pairs with Ag, was not present with the Ag/AI weld, since the welding technique resulted in weld metal with a very low Ag con tent. The bending behavior of the Fig. 5 — Cu/W. M. interface of broken weld, when welding with the electron Al/Cu weld beam on the joint, appears to be less Fig. 7 — Ni/W. M. interface ot Al/Ni weld favorable (Ref. 3) than with the beam located in the Al. This would be due to the greater amount of Ag-AI inter metallic compounds in the W. M. Although there are no intermetal- lics in the phase diagrams of Ag and the three elements of Fe18Cr8Ni, the Ag/Fe18Cr8Ni welds formed cracks at a large radius in the bend test. It is probable that the poor bend prop erties are due to the microstructure of interdendritic Ag in a steel matrix at the Fe18Cr8Ni/W. M. interface. The low strength of Ag would result, dur ing bending, in a concentration of plastic deformation in the Ag. Since the Ag is present as thin layers between steel dendrites, its deforma tion limit would be reached with a low total deformation of the bend spec imen. The tensile strength was also in fluenced by the microstructure at the Fe18Cr8Ni/W. M. interface. The specimens from the thicker sheet, with the described Ag — steel den drite microstructure, had a lower strength than specimens from the thinner sheet, which did not have the Fig. 6 — Fe18Cr8Ni/W. M. interface ot Fig. Cu/Fe18Cr8Ni weld unfavorable microstructure. broken AI/Fe18Cr8Ni weld Other investigators (Ref. 4) con-
234-s I AUGUST 1976 eluded that the ductility of Ag/Ti gas Several attempts with various tech tungsten-arc welds was determined niques were made to weld the by the Ti content of the W. M. More AI/Fe18Cr8Ni pair, but all of the welds favorable welding conditions, than broke easily at the Fe18Cr8Ni/W.M. were used in the present work, may interface m a reaction layer of inter have reduced the Ti content of the metallic compounds. Another inves W. M., with an improvement of the tigator has reported success with this weld bend properties. combination (Ref. 13). Welding with a The Ag/V welds demonstrated ex pulsed electron beam produced cellent tensile strength, but a defi welds which could be bent 180 deg, ciency at the V/W. M. interface was providing the deformation took place revealed by the bend tests. However, mostly in the Al member (Ref. 12). the worst specimen passed a bend Acceptable welds were made by the radius of 20 mm (0.7 in.), corre gas tungsten-arc process, after first sponding to an elongation of 5% in the Fig. 9 — Fe18Cr8Ni/V weld outer fiber, and one specimen showed no cracking at a 2 mm (0.08 in.) radius. Further work would be necessary to establish more precise ly the conditions for consistently pro ducing welds with the ductility of the latter specimen. Table 3 — Transverse Weld Tensile Strength
Tensile strength I Spec Welds with Aluminum Base metal pair imen Joint(i 3) Side A SideB Fig- MPa ksi Eff. Failure location Aluminum proved to be one of the most difficult metals to weld in dis Ag Al 13a 77 11.2 100 Al similar combinations. Of the six pairs Ag Al 13b 96 13.9 100 Al Fe18Cr8Ni 13a 23.6 98 Underfill with Al which were investigated, four Ag 163 Ag Fe18Cr8Ni 13b 131 19.0 79 Fe18Cr8Ni/W.M. were unsuccessful because of brittle interface welds. Ag Ti 13a 177 25.7 100 Underfill The Al/Cu combination was readily Ag Ti 13b 176 25.5 100 Ag welded with a good visual appear Ag V 13a 168 24.4 100 Underfill ance, but when lightly stressed during Ag V 13b 162 23.5 97 W.M. preparation of metallographic spec Al Ni 13a 76 11.0 100 Al imens, the welds failed in a brittle Al Ni 13b 90 13.1 100 Al layer at the Cu/W. M. interface. Two other attempts (Refs. 2, 3) to Cu Fe18Cr8Ni 13a 225 32.6 95 Underfill electron beam weld the Al/Cu pair Cu Fe18Cr8Ni 13b 228 33.1 100 Cu were also unsuccessful. The welds Cu Ni 13a 171 24.8 57 Undercut Cu Ni 13b 221 32.1 100 Cu broke either in the weld fixture upon Cu Ni15Cr7Fe 13a 228 33.1 97 Underfill cooling or from light bending Cu Ni15Cr7Fe 13b 228 33.1 100 Cu stresses. Gas tungsten-arc and gas Cu V 13a 216 31.3 91 Underfill metal-arc welds are normally not pro Cu V 13b 229 33.2 100 Cu duced by directly joining Al to Cu, but by using intermediate metals such as Cu20Ni Fe18Cr8Ni 13a 314 45.5 100 Cu20Ni Cu20Ni Fe18Cr8Ni 13b 326 47.3 100 Cu20Ni Zn (Refs. 5, 6), Sn (Ref. 7), or a Cu20Ni Ni15Cr7Fe 13a 316 45.8 100 Cu20Ni Ag16Cu16Zn18Cd alloy (Ref. 8). Cu20Ni Ni15Cr7Fe 13b 319 46.3 100 Cu20Ni,W. M. It has been demonstrated (Ref. 9) Cu20Ni V 13b 51 7.4 16 V/W. M. interf. by the annealing of pressure welded Al/Cu specimens that welds of good Fe18Cr8Ni Ni 13a 444 64.4 100 Ni strength and formability are attained Fe18Cr8Ni Ni 13b 469 68.0 100 Ni Fe18Cr8Ni Ni15Cr7Fe 13a 568 82.4 100 Fe18Cr8Ni when the intermetallic layer is limited Fe18Cr8Ni Ni15Cr7Fe 13b 579 84.0 100 W.M. to no more than 5 to 10 fim. This may Fe18Cr8Ni V 13a 404 58.6 100 V explain the excellent mechanical Fe18Cr8Ni V 13b 427 61.9 100 V properties of welds between Al and Cu tubes used for many years in the Cb Ti 13a 298 43.2 100 Cb refrigeration industry produced by Cb Ti 13b 208 30.2 100 Cb V 294 42.6 100 flash welding, when the welding con Cb 13a Cb Cb V 13b 189 27.4 100 Cb ditions are carefully controlled (Refs. Cb ZrSn(b) 13a 257 37.3 93 Cb,Zr2Sn/W.M. 10, 11). Another report (Ref. 12) of interface good ductility in electron beam welds Cb Zr2Sn(b) 13b 157 22.8 76 Zr2Sn/W.M. made with a pulsed beam is ques interface tionable, since the applied bend test Cb Zr2Sn(c> 13a 240 34.8 87 Cb/W.M. interf. was probably not suitable for a con Cb Zr2Sn
WELDING RESEARCH SUPPLEMENT] 235-8 coating the steel with Ai by a hot dip Welds with Copper and Cu20N i ed (Refs. 19, 21, 22). In the present ping process (Refs. 14-16). work the mechanical properties of the The Al-Ni phase diagram contains Most of the welds of dissimilar Cu/Fe18Cr8Ni or the a number of intermetallic com metals with Cu or Cu20Ni were of Cu20Ni/Fe18Cr8Ni welds were not pounds, with very high hardnesses of good to excellent quality. The greatest affected by the Cu intergranular 720 and 1000 DPH (Ref. 17). Based on difficulty was a tendency to underfill penetration. this, the excellent weld properties at and, to a lesser extent, undercut and Five of the nine weld combinations tained with the Al/Ni combination porosity. Others (Ref. 2) have elec with Cu or Cu20Ni were without in could not have been predicted. The tron beam welded 3 mm (0.12 in.) dications of brittleness, the joint effi dark colored deposit, which formed sheet of the Cu/Ni pair, apparently ciency of tensile specimens with the on the Ni faying surface and some without serious underfill or undercut. weld bead surfaces removed by times prevented good wetting by the In the present work, the severe under machining was 100%, and no cracks Al weld metal was apparently caused cut in Cu with the Cu/Ni weld was appeared at a bend radius of 1 mm by the vaporization of some constitu largely due to the transverse beam (0.04 in.). ent from the weld pool and its subse oscillation, necessitated by the The tensile strength and bend duc quent deposition on the Ni ahead of magnetic influence of Ni on the beam. tility of the Cu20Ni/V weld were very the weld pool. The addition of Cr and This resulted in a concentration of poor due to premature failure in a Fe to Ni drastically reduced the com heat at the edge of the W. M. where brittle reaction layer formed at the patibility of Ni with Al so that the the transverse velocity of the beam V/W. M. interface. Although the AI/Ni15Cr7Fe welds were very poor was zero, before reversing its direc mechanical properties of the because of brittle weld metal. tion, causing overheating and metal expulsion at this location. The deflec Cu20Ni/V10Ti weld were not deter tion and instability of the electron mined, other indications lead to the beam found in this investigation when conclusion that the properties are welding the Cu/Ni pair, was not ex similar to those of the Cu20Ni/V weld. perienced by others (Ref. 2). The ex The reaction layer of the latter two planation for this discrepancy could welds may be assumed to consist of be a residual magnetism in the Ni or Ni-V compounds, since the layer did an unknown external electromag not form with either the Cu/V or the netic effect. Cu/V10Ti welds.
The technical literature contains a Welds with Fe18Cr8Ni number of references to the inter granular penetration of steels by At no time were cracks observed in molten Cu or Cu alloys during welding the V/W. M. interface of the (Refs. 18-22). The use of an inter Fe18Cr8Ni/V weld and no reaction mediate Ni material to prevent this layer was detected. However, a type of defect has been recommend tendency for weld metal cracking was
Table 4 — Weld Bend Test Results
Radius1 Single Base metal pair values Average SideA SideB mm mm in. Crack location
Fig. 10 — Zr2Sn/W. M. interface of Ag Al 8,8,8,6 8 0.24 Cracks in W.M. Cb/Zr2Sn weld. Welded with the electron Ag Fe18Cr8Ni 35,35,35,35 35 1.38 Cracks, Fe18Cr8Ni/W.M. beam on the joint interface Ag Ti 1,25, 10, 10 12 0.47 Cracks, W.M. and Ti/W.M. interface Ag V 14, 18, 1 11 0.43 Cracks, V/W.M. interf. Al Ni 2,2,2,4 3 0.12 Cracks in W.M.
Cu Fe18Cr8Ni None Cu Ni None Cu Ni15Cr7Fe None Cu V None
Cu20Ni Fe18Cr8Ni None Cu20Ni Ni15Cr7Fe None Cu20Ni V 25, 25, 25 25 0.98 Cracks, V/W.M. interf.
Fe18Cr8Ni Ni None Fe18Cr8Ni Ni15Cr7Fe None Fe18Cr8Ni V None
Cb Ti None Cb V None Cb'b) Zr2Sn 1, 1,35, 50 22 0.82 Cracks, W.M. and Cb Cb/W.M. interface Ti V 4, 1, 1 2 0.08 Transv. cracks, W.M. SO Mm Ti Zr2Sn 16, 16, 16 16 0.063 Transv. cracks. W.M. I 1
(a) Radius of plunger at which the first crack appeared. Where no values are given, three specimens of each combina Fig. 11 — Cb/W. M. interface of Cb/Zr2Sn tion were bent with a 1 mm (0.04 in.) radius plunger without the appearance of cracks. Plungers were at intervals of 5 mm weld. Welded with the electron beam offset from 50 to 20 mm radius. 2 mm from 20 to 6 mm radius, and 1 mm from 6 to 1 mm radius. in 7r?Sn (b) Electron beam offset in Zr2Sn
236-s I AUGUST 1976 present. There were two types of the $ to a transformation; e.g., the co - made with shielded Cb and the elec cracking. The first was hot cracking phase. The excellent weldability of tron beam in Zr2Sn, the very low Cb transverse to the weld bead, which is combinations of Cb and Cb alloys content of the weld metal would have believed to be due to the difference in with Ti and Ti alloys has been con shifted the compound layer to the Cb the thermal coefficient of expansion firmed by other investigators (Refs. interface (Fig. 11); but there would be of the two materials; that of Fe18Cr8Ni 23, 25-27). sufficient Cb in the weld metal to in is about twice that of V. Upon cooling No difficulty was experienced in crease its hardness to 300 DPH, from from welding, this could cause tensile producing welds of excellent quality the 150 DPH of the Zr2Sn base metal. stresses to develop on the steel side with the Cb/V pair, except for a slight of the weld metal, as demonstrated by tendency to porosity. Limited experi Fig. 9. There is a possibility that these ments with the Cb/V10Ti weld indi microcracks would be eliminated by cate that its weldability is com using an austenitic steel containing parable to the Cb/V weld. Welds with more ferrite in the weld microstruc the Cb/V pair, using the gas tung ture, which is known to reduce the sten-arc process exhibited good bend tendency of austenitic weld metal to ductility (Ref. 23). form hot cracks. Other welding condi The gas tungsten-arc welding of tions, to decrease the V content of the the Cb/Zr and Cb/Zr2Sn pairs with no weld metal, might also decrease the reference to embrittlement, has been hot cracking tendency. reported (Refs. 23, 28). Other inves The second type of cracking en tigators (Refs. 29, 30), however, have countered is the brittle failure, longi observed brittle behavior in gas tung tudinally through the weld metal, of sten-arc and electron beam welds of one weld when a transverse cut was Cb alloy/Zr alloy joints, which indi made with a power shear. A higher V cated that the embrittlement was due content in the weld metal of this weld, to the formation in reaction layers of as compared to the remaining welds the intermediate co -phase during of this combination, was indicated by transformation of the metastable /?- a higher hardness. An electron beam phase. The hard, brittle reaction angle of less than the 12 deg which layers of the Cb/Zr2Sn weld, pro was used, should decrease the V con duced in the present investigation, Fig. 12 — Ti/W. M. interface of Ti/V weld tent of the weld metal. cannot be explained by the presence Gas tungsten-arc welding of the of the co -phase. The hardness of 440- Fe18Cr8Ni/V pair produced brittle 490 DPH, measured 7 to 8 weeks after welds (Ref. 23) believed to be caused welding, was much lower than it by Ni-V intermetallic compounds. should have been, if a significant Attempts to join V foil to thicker aus amount of the co -phase had been tenitic Cr-Ni steel by diffusion, gas formed. Also, welds produced with tungsten-arc, and electron beam the electron beam on the joint, which 7O »r~ welding were unsuccessful due to were annealed for one hour at 600, cracking and embrittlement (Ref. 24). 800 and 1000 C showed an increase 75(2 95)- However, welds in 0.5 mm (0.02 in.) in the Zr2Sn/W. M. reaction layer I85 (7.28) - sheet of the Fe18Cr8Ni/V pair pro hardness, rather than the decrease duced by pulsed electron beam that would be expected if the co - a Tensile specimen with weld bead in as - welded condition welding were reported to be of good phase were present. weld- quality (Ref. 12). The significance of Although it has not been con this latter result is questionable, as firmed, indications of an intermetallic discussed for the Al/Cu weld. compound at 3% Cb in the Cb-Zr phase diagram has been reported I85 (7.28 ) =- -75(2.95)—| ~| Welds with Columbium (Ref. 31). This would explain the ob servations made in the welds of the Excellent mechanical properties present work. A weld metal of high Cb b. Tensile specimen with weld bead surfaces removed were achieved with Cb/Ti welds, content, as was the case when weld Weld - despite the formation of a reaction ing with the electron beam on the is layer of high hardness at the Ti/W. M. O to joint, would have caused the bulk of : interface. There are no intermetallic the intermetallic compound to form at ~f" compounds in the Cb-Ti phase dia theZr2Sn interface (Fig. 10), since the _i gram and the hardness of 330 DPH in Cb content of the weld metal would is6 the reaction layer is assumed to be have been too high to form much c. Bend Specimen "•" caused by an intermediate phase of compound. When the welds were Fig. 13 — Mechanical property specimens
Table 5 — Tensile Properties of Welds With More Than Two Materials
Tensile strength Elong.10' Base metal combination Average Calculated(a) Avg., A B C D E MPa ksi MPa ksi % Al Ni Fe18Cr8Ni 318 46.1 344 49.9 24 Zr2Sn Ti Cb V Fe18Cr8Ni 487 70.6 420 60.9 6
(a) Based on the proportional cross-section of each material in the combination and its strength (b) Gage length — 62 mm (2.44 in.)
WELDING RESEARCH SUPPLEMENT! 237-8 Fig. 14 — AI/Ni/Fe18Cr8Ni weld Fig. 15 — Fe18Cr8Ni/V/Cb/Ti/Zr2Sn weld with the Fe18Cr8Ni/V joint welded last
Table 6 — Summary of Results 2! C z O o 5 OO CM Visual Metallo Transv. Bend 3 3 cn Overall > > is U 2 o
WELDING RESEARCH SUPPLEMENT! 239-8 9. Voropai, N. M. and Shinyaev, A. Ya., "Mechanical Properties of Welded Joints Report BMI 1619, pp. X1-X2. "Influence of Heating on Diffusion Pro Between Steel and Aluminum," Weld. 25. Gorin, I. G., "Welding Titanium cesses and Properties of Bimetallic Joints Prod.. No. 1 (1966) pp. 23-27. Alloys to Nickel-Base Alloys," Weld. Prod., of Aluminum and Copper," Metallove- 17. Petty. E. R., "Hot Hardness and No. 12 (1964) pp. 46-53. denie i Termicheskaya Obrakotka Metal- Other Properties of Some Binary Inter 26. Shchetanov, D. P., "A Method of lov, No. 12 (1967) pp. 55-57. metallic Compounds of Aluminum," J. Inst. Welding VN1 and VN2 Niobiums to the 10. Schneidmadl, E., "The Production Metals, Vol. 89 (1960-61) pp. 343-349. VT1 and OT4 Alloys," Weld. Prod., No. 7 of High Quality Joints Between Copper 18. Peshekhonov, V. D., Kobylyanskii, I. (1971) pp. 31-32. and Aluminum by the Resistance Flash F. and Dubitskii, A. K., "Welding Joints 27. Novosadov, V. S.. "The Argon TIG- Welding Process," Schw. u. Schn., Vol. 9 Between Sheets of Copper and KM8N10T Welding of Titanium to Niobium," Auto, (1957) pp. 59-64. Steel," Weld. Prod., No. 1 (1966) pp. 27-31. Weld., No. 10 (1970) pp. 67-68. 11. Haessly, W. F., "Flash Welding Alu 19. Koecher, R., "The Joining of Copper 28. Evans, R. M., "Dissimilar Metals minum to Copper Tubing," Welding Jour and Copper Alloys with Steel — The Joining," Reactor Materials, Vol. 9, No. 2 nal, Vol. 33 (1954) pp. 1162-1170. Welding of Copper-Clad Steels," Metall, (1966) pp. 146-147. 12. Scheibe, W., The Welding ot Vol. 15 (1961) pp. 1107-1111. 29. Novosadov, V. S., "The Weldability Dissimilar Materials Which Have Inter 20. Moore, P. T. and Metcalf, J., of Zirconium Alloys with Niobium Alloys," metallic Compounds in Their Phase "Fabrication of Pressure Vessels in Explo Weld. Prod., No. 8 (1969) pp. 40-44. Diagrams. Deutscher Verlag f. Schweiss sively Bonded Copper-Clad Austenitic 30. Novosadov, V. S., "Effects of Heat technik, Krefeld. Strahltechnik II, DVS-Ber. Steel," Metal Constr., Vol. 1, No. 12s (1969) Treatment on the Properties of Joints Be 4, 1968, pp. 17-41. pp. 36-40. tween a Zirconium Alloy Containing 2.5% 13. Scheibe, W., The Influence of 21. Schultz, H., "Joining of Dissimilar Nb and Niobium Alloys," Auto. Weld., No. Intermetallic Phases on Electron Beam Metals by Inert Gas and Electron Beam 1 (1969) pp. 27-32. Welds in Thicker Sections of Dissimilar Welding," Schw. u. Schn.. Vol. 17 (1965) 31. Shunk, F. A., Constitution of Binary Metals, Deutscher Verlag f. Schweiss pp. 288-296. Alloys, Second Suppl. New York, McGraw- technik, Duesseldorf. Strahltechnik III, 22. Bott, H. B., "How to Weld Copper to Hill Book Co., 1969. DVS-Ber. 5, 1969, pp. 51-56. Stainless Steel and Mild Steel," Welding 32. McBee, F. W., Henson, J. and Ben 14. Miller, M. A. and Mason, E. W., Journal, Vol. 38 (1959) pp. 236-238. son, L. R.. "Problems Involved in Spot "Properties of Arc Welded Joints Between 23. Strizhevskaya, L. G. and Starova, L. Welding Titanium to Other Metals," Aluminum and Stainless Steels," Welding L., "Fusion Welding of Certain Dissimilar Welding Journal, Vol. 21 (1956) pp. 481-s- Journal, Vol. 35 (1956) pp. 323-S-328-S. Metals," Weld. Prod.. No. 1 (1966) pp. 8- 487-s. 15. Gorin, I. G., "Aluminum Alloys 13. 33. Novosadov, V. S., "Properties of Welded to Stainless Steels," Auto. Weld., 24. Hodge, E. S. et al., "Joining of Welded Joints Between Zirconium Alloy No. 1 (1966) pp. 16-21. Regenerative Cell Components." Battelle with 2.5% Nb and Titanium Alloys," Weld. 16. Ryabov, V. R. and Yumatova, V. I., Memorial Institute, Columbus. Ohio, 1963. Prod., No. 11 (1969) pp. 41-44.
WRC Bulletin 215 May 1976
I "Development of Design Rules For Dished Pressure Vessel Heads"
by E. P. Esztergar
The present study is the continuation of an extensive investigation aimed at the development of rational design rules for formed pressure vessel heads. This problem was one of the first research pro jects the Design Division of PVRC selected for study and by maintaining its interest stimulated an in creasing amount of work on the diverse behavior of seemingly similar head shapes. In the course of this work, not only new analysis techniques were developed but also a number of new failure modes unique to these deceivingly simple looking geometrical shapes were uncovered.
II "The Effect of Geometrical Variations on the Limit Pressures For 2:1 Ellipsoidal Head Vessels Under Internal Pressure"
by J. C. Gerdeen
This theoretical study has been conducted for the Subcommittee on Shells of the PVRC Design Division of the Welding Research Council. The impetus for this general study grew out of another spe cific study of the analysis of limit pressures of actual formed heads. Since it was found that actual formed heads are not uniform in thickness and that the thickness variations do affect the limit pressure, the thickness variation was studied in detail using a computer analysis that included bending theory. Publication of these papers was sponsored by the Pressure Vessel Research Committee of the Welding Research Council. The price of WRC Bulletin 215 is $8.00 per copy. Orders should be sent with payment to the Weld ing Research Council, United Engineering Center, 345 East 47th Street, New York, N.Y. 10017.
240-8 I AUGUST 1976