Mechanisms Prducing Metallic Bnds in Cld Welding Scanning electrn micrscpe investigatin shws tw calescence mechanisms are invlved in the cld welding f aluminum t aluminum 2 i > LLI Q \ c LU BY N. BAY Q_ O ABSTRACT. In the cld welding f metals that are scratch-brushed befre jint plastic defrmatin, tw basic bnding r calescence mechanisms exist. One is fracture f the brittle cver layer frmed by scratch-brushing, extrusin f base material thrugh the cracks, and buildup f real cntact and calescing with base material f the ppsite surface. Fr AI-AI cld welding this mechanism is applicable t 60% f the weld area. In the remaining area n brittle cver layer is present, and welding is here established by fracture f the cntaminant film f xides and water vapr. A theretical mdel fr bth mechanisms shws the basic influence f bth surface expsure and nrmal pressure n weld strength. It is in gd agreement with experimental results. Experiments with milled surfaces were perfrmed t investigate the influence f the cntaminant film with n brittle cver layer being present. The results shw a threshld surface expsure independent f nrmal pressure in accrdance with the thery and an influence f nrmal pressure n weld strength that is much less prnunced. Intrductin Cld welding is a slid phase welding prcess characterized by the large number f pssible metal cmbinatins. Calescence is btained by jint plastic defrmatin f the tw metals. A basic variable gverning calescence is the degree f defrmatin expressed as the surface expansin r the surface exp- Paper presented under the title f "Bnding Mechanisms in Cld Pressure Welding" at the 63rd AWS Annual Meeting in Kansas City, Missuri, during April 26-30, 1982. N. BAY is Senir Lecturer- Dept. f Mechanical Technlgy, AMT, Technical University f Denmark, Lyngby, Denmark. sure Y f the weld interface surface: _ AT - A 0 A AT - AQ Y = At 0) where A 0 is the initial and A n the final area f the surface Table 1. In the case f rlling tw metal sheets tgether r cmpressing tw metal cylinders end t end, Y = R, the reductin. Figure 1 shws the weld strength in pure shear as a functin f the surface expsure fr a number f metal cmbinatins welded by rlling (Ref. 1, 2). Calescence is nt btained until a threshld surface expsure has been reached. Beynd this threshld value which is dependent n the metal cmbinatin, the weld strength increases rapidly with Y and reaches a steady prgress crrespnding t the strength f the weaker metal. This relatinship between weld strength and surface expsure is fundamental fr all cld welding prcesses. Hwever, the threshld surface expsure fr a given metal cmbinatin appears t be dependent n the type f prcess. Surface preparatin befre cld welding is anther variable that has a basic influence n the weld strength. Varius x P e> z tr t- c - I 3 200 ',60 120 80-40 (1) Cu-Cu (2) Cu-Fe (3) AI-AI (A) Cu-Al (5) Cu-Ni (6) Cu-Ag (7) Zn-Zn H 0.3 0A 0.5 0.6 0.7 0.8 Y- SURFACE EPOSURE Fig. 7 Weld strength in rlling as a functin f surface expsure (Ref. 1 and 2) frms f surface preparatins mechanical, thermal and chemical treatments have been investigated (Ref. 3, 4); these shw degreasing fllwed by scratchbrushing with a rtating steel brush t be the mst effective, giving the smallest threshld surface expsure. Experimental investigatins have als shwn the pressure distributin at the weld interface surface has an influence n weld strength (Ref. 1, 5-8). In this paper the calescence mechanisms are investigated, and a theretical mdel fr weld strength is develped, explaining the influence f the abvementined variables. Metallic Bnd Prducing Mechanisms. Brittle Cver Layer Earlier investigatins in cnventinal light micrscpe (Ref. 1) and in scanning Table 1 Symbls and Definitins Surface expansin f weld interface surface Y - Surface expsure f weld interface surface Y' Threshld surface expsure fr cntaminant film AT Final area f weld interface A 0 Initial area f weld interface A' Area crrespnding t Y = Y' r 0 Yield stress f basic material a B Weld strength CBC Weld strength, mechanism C r BF Weld strength, mechanism F i/v c Fractin f brittle cver layer t ttal area \pf Fractin f film layer t ttal area (S-fp PE Extrusin pressure p Nrmal pressure at base metal surfaces pc~nrmal pressure between end surfaces f extruded base metals R Reductin RE Extrusin reductin cc /) LU tx _J > c v> c z s c Crt ui c _l > C. \ c Ui tf) c WELDING RESEARCH SUPPLEMENT 1137-s
0.05 mm ' /, Fig. 2 Weld interface surface after fracture. Y = 0.375, p/r 0 = 1.9, TB/(T 0 = 0.02 electrn micrscpe (SEM) (Ref. 7-11) have indicated that a basic mechanism gverning calescence is fracture f the wrk-hardened surface layer prduced by scratch-brushing and expsure f the base material. Figures 2-4 shw SEM micrgraphs f the weld interface surface f Al Al (1100 ally) cld welds after fracture by tensin testing. At a surface expsure Y = 0.375 (Fig. 2), fracture in the brittle surface layer (characterized by hrizntal scratches frm brushing) has ccurred. The underlying base material is expsed and beginning t extrude up thrugh the cracks. The weld strength, hwever, is small, nly 0.02 times the yield stress 0 f the defrmed material, and n weld interface fractures are visible. At a little larger surface expsure Y = 0.383 (Fig. 3), base material has extruded thrugh the cracks. Real cntact and calescence with base metal f the ppsite surface has been established. Weld fracture is ductile and ccurs after lcal necking. Figure 4 crrespnding t Y = 0.72 shws extended areas f base material and calescence, unwelded regins f brittle surface layer being cnfined t small islated islands. Here weld strength is as = 0.71r. Figure 5 shws a micrgraph f a sectin thrugh the weld interface. The brittle surface layer is brken under 45 deg in a shear fracture. Measurement f the fragment length and the crack width shw that the fracture is brittle. The fragments are lying like pearls n a string. Figure 6 shws a SEM micrgraph f a scratch-brushed Al surface befre cld welding. The tpgraphy is characterized by an undulating cuntry with lng tngue-shaped hills. It is these "tngues" that are recgnized as being the brittle surface layer cracking when being defrmed Figs. 2-4. They are easy t distinguish by the regular pattern f hrizntal ridges frm the scratch-brushing whereas the remaining areas in Fig. 6 have a mre irregular tpgraphy. The width f each tngue is n the rder f 50-100 gm (2-4 10" 3 in.). The diameter f each wire in the steel brush is apprximately 0.3 mm (0.012 in.), indicating that each tngue might be made by a single wire. Fig. 4- Weld interface surface after fracture. Y = 0.72, p/ 0 = 1.8, trb/tr 0 = 0.71 Fig. 3 Weld interface surface after fracture. Y = 0.383, p/cr 0 = 2.0 a B /r 0 = 0.09 This seems supprted by the micrgraph in Fig. 7 shwing a sectin thrugh the scratch-brushed surface. Brushing is dne frm left t right. The tngue cnsists f a severely defrmed structure which, at the right end, is teared ver the riginal surface s that a crack is appearing between the tngue and the base material. This is generally seen, and smetimes the crack is mre than half the ttal tngue length. Micrhardness measurements were made bth n sectined specimens after welding and directly n tp f the scratch-brushed surface. They indicated a hardness frm 178 t 193 kp/mm 2 f the tngues and 38 t 55 kp/mm 2 (Vickers, lad 15 g) f the base material (in initial r defrmed state). Bth methds resulted in a hardness rati between tngues and basic metal f 3:7. This explains the brittle behavir f the tngues. They prbably cnsisted f defrmed material mixed with xides, and it is believed that they might have been frmed by Al particles that were picked up by brushing, adhered t the wire ends f the brush, and later were rejined t the surface by cld welding. This wuld explain the rather rugh surface tpgraphy fund next t the tngues Fig. 6. Frm the abve bservatins it fllws that the rle f scratch-brushing is nt just t clean the surfaces. It prduces a 138-s MAY 1983
'.'V O.aRmyi Fig. 5 Sectin thrugh Al Al cld weld hard brittle surface film; this will crack when being defrmed and thereby uncver underlying base material kept clean after expsure due t high vacuum in the cracks. The high nrmal pressure between the cntacting surfaces will allw n admissin f air, thereby keeping the uncvered surfaces clean. The surface film n tp f the tngues cnsists f an xide layer n the rder f 100 A thick and f a cntaminant film cnsisting f water vapr, grease and gases that is abut 30 A thick. This film is very thin cmpared t the thickness f the tngues (1-5 10 5 A). This means that the cntaminant film shuld nt be able t get in tuch with the base material in the crack. This als means that the uncvered material, when getting int cntact with uncvered material frm the ppsing metal, will bnd immediately as bserved in Fig. 3. In additin, weld strength will increase with increasing nrmal pressure building up the cntact area by plastic defrmatin f the surface asperities f the tw rugh surfaces. The brittle surface layer f tngues is, therefre, here termed the cver layer. Cntaminant Film As seen in Figs. 6 and 7 the brittle tngues d nt cver the whle surface. Measurements n larger areas f scratchbrushed surfaces shwed the brittle cver layer t be present as a fractin \p c = 0.355 f the ttal area. In the remaining area where I/'F = 1 \pc = 0.645, calescence is hindered by the cntaminant film. Placing tw scratchbrushed surfaces against each ther, the areas f cver layer "C" and f film layer "F" f the tw surfaces can be assumed t be cmbined randmly as C C, C F, F C and F F. In the areas where ne r tw brittle cver layers are present it can be assumed that the mechanism f metallic bnding will be as described abve, i.e., extrusin f base material thrugh cracks f the cver layer. Vaidyanath, et al. (Ref. 1), have shwn that tw ppsing cver layers will break up as ne single layer. In the remaining area, F F calescence is als btained and can be seen in Figs. 4 and 8. Figure 8 shws a surface with Y = 0.45, a surface expsure large enugh t btain clearly visible bnds. The areas f calescence are seen in the cracks f the brittle cver layer. Hwever, they are extended uninterrupted in a band frm the crack int the F zne. This phenmenn is als nticed in Fig. 4 where Y = 0.72 and brad bands f calescence have been established. Bands f calescence extending int the F znes frm the C znes are generally fund. This means that they are als established in the F F znes. The reasn fr this is assumed t be that the large lcal surface expsure due t cracking f the cver layer will cntinue in an extended band int the F zne. This is because it will be less energy-cnsuming than when establishing unifrm defrmatin in the F zne; the latter wuld require a large velcity discntinuity in the defrmatin pattern at the bundary f the tngue. This leads t the cnclusin that the tw mechanisms schematically utlined in Fig. 9 are applicable (Ref. 12): Mechanism C: Fracture f brittle cver layer, extrusin f base material thrugh the cracks and establishment f real cntact and calescence between base materials. Mechanism F: Fracture f cntaminant film and establishment f real cntact and calescence between base materials. Mechanism F is nly functining in the cntact znes F F which are cnfined t: 0 = ia F 2 = 0.645 2 = 0.4, r 40% f the ttal area. Fig. 7-Sectin thrugh scratch-brushed Al surface WELDING RESEARCH SUPPLEMENT 1139-s
a /Cver layer,- ^T^jfTxssa B a a a a a g B ^ g e S g i g B g cntaminant film nset f extrusin lcal thinning f cntaminant film Fig. 8-Weld interface surface after fracture. Y = 0.45, p/ 0 = 1.8, rb7r 0 = 0.24 Fig. 9 Schematic utline f calescence mechanisms Theretical Mdel fr Weld Strength A theretical mdel fr the weld strength based upn the first calescence mechanism namely, fracture f brittle cver layer has been described in the literature (Ref. 7, 8). A mdel taking bth mechanisms int accunt was later suggested (Ref. 12). It was develped by cnsidering each mechanism separately as if it were applicable t the whle surface, calculating the weld strength in each case, and then cmbining the tw expressins btained fr the weld strength. The analysis is briefly shwn belw. A mre cmprehensive descriptin is given in the literature (Ref. 12). Mechanism C At a threshld surface expsure dependent n the nrmal pressure p, extrusin f base material thrugh the cracks f the cver layer is initiated. The pressure PE necessary t start the extrusin is estimated using Jhnsn's (Ref. 13) and Hill's (Ref. 14) slipline analysis f plane strain extrusin thrugh square dies. Applying their results t the present extrusin mdel and intrducing the surface expsure Y instead f the extrusin reductin RE = 1 Y yields the relatinship between p E and Y shwn in Fig. 10. When the nrmal pressure p reaches the required pressure p E fr extrusin, the base material is frced thrugh the cracks in the cver layer. The extruded material f the tw metals meet, and real cntact is established. If the pressure p is further increased, the nrmal pressure pc and thereby the real calescence area between the rugh end surfaces increases; pc is given by: Pc PE (2) In rder t determine the nminal bnd strength r B c (crrespnding t the final area AT), it is necessary t estimate the true strength f the welds. The SEM micrgraphs shwed that the weld fracture in tensin testing was ductile. Cnrad and Rice (Ref. 15) have investigated the adhesin between clean surfaces f Ag, Al, Cu and Ni, using the technique f cld welding specimens previusly fractured in an ultrahigh vacuum. They fund that the weld strength btained was almst equal t the cmpressin lad applied in cld welding. Similar results were fund by Upit and Manik (Ref. 16). It is reasnable t assume (Ref. 7, 8, 12) that the weld fracture is established by plastic defrmatin in the welds and that the stress necessary t break the welds crrespnds t the nrmal pressure nec- 0.2 CK 0.6 0.8 Y -SURFACE EPOSURE Fig. 10-Extrusin pressure as a functin f surface expsure 1.0 essary t further increase f the calescence r weld area (by plastic defrmatin f surface asperities). This nrmal pressure equals the maximum applied cmpressive stress pc n the base metal surfaces. Frce equilibrium gives the relatinship between the weld strength r B c crrespnding t the nminal area AT and the true weld strength p c crrespnding t the true virgin area: O-BC A, = p c (AT - A 0 ) r by inserting (1) and (2): «5c = YP_ PE (3) As shwn earlier (Ref. 8, 12), tw maximum limits might be applied t the expressin abve: a /fraav -,_ ' 'max 1 = V3 telling that the maximum strength btained is equal t the yield strength f the calescence bridges frmed in the cracks between the cver layer. And: C >max2-1 telling that the maximum strength equals the yield stress f the defrmed metal. The crrect slutin is believed t lie in between the tw as discussed in the literature (Ref. 8 and 12). Mechanism F In the areas F F where n cver layer is present, calescence is btained when 140-s MAY 1983
a V 0 1.0 t- O 08 H 06 _ THEORY EPERIMENTS Al Al t /} = 0.415 V'= 0 35 p / 1.85 04 0.1 0.2 0.3 0.4 0.5 0.6 0.7 06 09 1.0 y - SURFACE EPOSURE Fig. 11 Theretical weld strength as a functin f surface expsure and nrmal pressure Al - Al 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Y- SURFACE EPOSURE Fig. 12 - Weld strength as a functin f surface expsure. Al - Al, scratched-brushed, p/tr 0 = 1.85 a threshld surface expansin Y' necessary t break the cntaminant film has been reached. This has been estimated t 0.30 Y' 0.35 in case f Al - Al cld welding as discussed later. Assuming n cver layer but nly a cntaminant film t be present the actual area f base metal being decvered will be Ai A' where A' is the area crrespnding t Y = Y'. Similar t the calculatin abve fr mechanism C, a frce equilibrium gives the relatinship between the weld strength cr BF crrespnding t the nminal area AT and the true weld strength crrespnding t the area f base material surfaces AT A'. The true weld strength equals p, the nrmal pressure in the base material surfaces accrding t the same arguments as abve based upn previusly reprted results (Ref. 15, 16): r B F AT = p(ai - A') (4) Frm equatin (1) it is seen that: 1 = A " 1 - Y' Tgether with (1) and (4) this gives: O-BF Y - Y' p (5) 1-Y'a 0 Cmbined Thery fr Mechanism C and F As mentined earlier, the cntaminant film fracture mechanism is acting in a fractin 13 = 0.4 f the ttal area whereas the cver layer fracture mechanism is acting in the remaining area. This puts frward the fllwing suggestin fr the cmbined thery fr the weld strength see equatins (3) and (5): ^= ( i-^)y p^b +/3 i^r.ii a 0 a 0 1 Y a a (6) In case f a nrmal pressure p PE, set p = p E, and incase f Y Y' set Y = Y'. It fllws that the first term in equatin (6) crrespnds t mechanism C which cmes int actin when p > pi. The secnd calescence mechanism is functining when the surface expsure Y exceeds the threshld value Y'. Figure 11 shws the theretical mdel fr Al Al cld welding when Y' = 0.35 and 8 = 0.4. At lw nrmal pressures, p/-q 2.0, welding is initiated when Y>Y' by fracture f the cntaminant film. Fr p/a 0 = 1.5, there is a bend n the curve at Y = 0.49 crrespnding t the starting pint f extrusin thrugh the cracks and weld frmatin between the segments f cver layer. This is in accrdance with Fig. 10 shwing the necessary extrusin pressure. Fr p/r 0 > 2.0 welding is initiated by fracture f the cver layer, extrusin thrugh the cracks and weld frmatin in between the segments f cver layer. Mechanism F, the fracture f the cntaminant film later cmes int actin, namely at Y = Y'; this explains the bend at Y = 0.35 f the curves fr p/a 0 > 2.0. Experimental Results and Discussin In rder t test the theretical mdel fr weld strength, experiments were perfrmed in equipment that allwed independent variatin f surface expsure and nrmal pressure when cld welding Al Al. The experiments as described in the literature (Ref. 7, 8) were dne by plane strain cmpressin in a channel frming tl. Weld strength was estimated by a specially develped tensin testing technique that allwed six small test pieces t be cut frm each specimen. The specimen material was annealed 1100 ally. Tw series f experiments were perfrmed. One with scratch-brushed surfaces allwed bth calescence mechanisms t functin. Fr the ther series f experiments, surfaces were milled withut a lubricant; this prduced a surface with a cntaminant film f xides and water vapr (frmed during the time I «U 04 - aa} 8 I 00' 9-& 0.2 03 04 0.5 0.6 0.7 Y-SURFACE EPOSURE THEORY ft t EPERIME >ITS Al - Al 0.415 0.30, 0.35 P/0=5.1 Fig. 13 - Weld strength as a functin f surface expsure. Al - Al, scratch-brushed, p/tr 0 =5.1 1- n 0 6 -i 5 0.4 0.2 9l O0=3,85 5 milled p/0=l.33 + milled p/0a3.85 scratch-brushed p/0 = 1.45 _ thery /3-l,V'=0.3 0 0.1 0.2 0.3 0.4 05 06 0.7 08 0.9 1.0 Y- SURFACE EPOSURE Fig. 14 - Weld strength as a functin f surface expsure and nrmal pressure, Al Al. Cmparisn between thery and experiments with milled surfaces, p/a 0 = 1.33 and 3.85. Experiments with scratchbrushed surfaces with p/r 0 = 1.45 WELDING RESEARCH SUPPLEMENT 1141-s
between milling and welding) but with n brittle cver layer, i.e., nly allwing the secnd calescence mechanism t functin. Welding in bth series was perfrmed after 5 minutes (min) expsure t the atmsphere. Figures 12 and 13 shw a cmparisn f the theretical calculated weld strength with experiments with scratchbrushed surfaces. Surface preparatin cnsisted f degreasing with tluene fllwed by scratch-brushing with a rtating steel brush; the brush had a 180 mm (7 in.) uter diameter, 0.3 mm (12 10~ 3 in.) wire diameters, a 1450 rpm running speed and 0.2 mm/s (8 10~ 3 in./s) feed. Figure 12 crrespnds t a mean nrmal pressure p/a Q = 185 and Fig. 13 t p/a 0 = 5.1. Althugh the scatter in the experimental results is quite large, it seems in reasnable accrdance with the thery; als, the sudden rise in weld strength with surface expsure in Fig. 13 is fund experimentally at abut Y = 0.35. Figure 14 shws the results frm the series with milled surfaces. After degreasing, a 0.1 mm (4 10-3 in.) chip was cut frm each f the bnding surfaces 5 min befre cld welding. Experiments were cnducted at tw different mean nrmal pressures p/cr 0 = 1.33 and 3.85. Cntrl experiments were cnducted with scratch-brushed specimens at p/ a 0 = 1.45 fr cmparisn with the earlier scratch-brushing results (Ref. 7, 8). Despite the lwer nrmal pressure btained with the cntrl experiments (Fig. 12), the results lie in the same range. The lwer nrmal pressure is due t mre careful lubricatin in these experiments. The reasn that the weld strengths btained are still in the same range as fr p/r 0 = 1.85 (in Fig. 12) has yet t be explained. It might be due t the fact that the relative humidity was very lw (dwn t 27%) when the last experiments ( in Fig. 14) were carried ut. Frm Fig. 14 it appears that milling gives larger weld strengths than scratchbrushing. Increasing nrmal pressure will increase the weld strength als in case f milled surfaces; hwever, the tendency is much less prnunced than thery indicates. The threshld surface expsure fr film break dwn in case f milled surfaces seems t be abut Y' = 0.3 independent f nrmal pressure. This independence was als adapted in the thery. The threshld value seems smewhat smaller than fr the F F znes f scratchbrushed surfaces (Fig. 13) where Y' = 0.35 seems a better fit. Cnclusins The SEM investigatins f the weld interface surface after fracture f Al Al cld welds shw tw calescence mechanisms t be active: 1. Fracture f the brittle cver layer frmed by scratch-brushing, extrusin f base material thrugh the cracks f the cver layer and establishment f real cntact between base materials. Calculatins shw that this mechanism is active ver 60% f the ttal area. In the remaining 40% f the area, n cver layer is present and welding is here btained by fracture f the cntaminant film f xides and water vapr and establishment f real cntact and bnding between expsed base material; in case f Al Al cld welding, this will ccur when the surface expsure Y exceeds Y' = 0.35. 2. Based upn the bserved calescence mechanisms a theretical mdel fr weld strength as a functin f the surface expsure and nrmal pressure has been develped. The mdel als takes int cnsideratin the variatin f the threshld surface expsure befre break dwn f the cntaminant film Y' and the fractin 8 f the weld area with n cver layer. 3. Experiments with cnventinal scratch-brushed surfaces at varying surface expsures and nrmal pressures shw gd accrdance with the theretical mdel. They cnfirm earlier experimental bservatins that nt nly the surface expsure but als the nrmal pressure is a basic variable gverning calescence; this explains the rather unnticed fact that the threshld surface expsure t btain calescence can vary frm prcess t prcess. The variatin f the threshld with the metal cmbinatin can als be explained since Y' and 8 will vary with the metal cmbinatin and since the ductility f the cver layer may depend n the base metal. Experiments with milled surfaces shw larger weld strengths than with scratchbrushed. This cntradicts results in the literature (Ref. 3). The threshld surface expsure fr breakdwn f the cntaminant film appears t be independent f nrmal pressure in accrdance with the theretical cnsideratins. Nrmal pressure has a less prnunced influence n calescence in case f milled surfaces than in case f scratch-brushed, and the effect is much smaller than thery states. Acknwledgment The authr wishes t thank Mr. O. Jagert fr helping with the experimental investigatins. References 1. Vaidyanath, L. R., Nichlas, M. C, and Milner, D. R. 1959. Pressure welding by rlling. Brit. Weld. J. 6:13-28. 2. McEwan, K. J. B., and Milner, D. R. 1962. Pressure welding f dissimilar metals. Brit. Weld. I 9:406-420. 3. Vaidyanath, L. R., and Milner, D. R. 1960. Significance f surface preparatin in cld pressure welding. Brit. Weld. J. 7:1-6. 4. Wdara, J. 1963. Einfluss der berflachenvrbereitung auf die Kaltpressschweissbarkeit vn Metallen. Schweisstechn. 13(12):548-552. 5. Gumm, P. 1964. Kmbinatin vn Umfrmung und Kaltpressschweissen beim Fliesspressen und Rhrziehen. Dissertatin. Technische Hchschule Carl-Wilhelmina, Braunschweig, GFR. 6. Vrm, T., Bay, N., and Wanheim, T. 1973. Influence f the hydrstatic stress cmpnent n critical surface expansin in frging cmpund prducts. Weld. Res. Int. 3(3):16-25. 7. Bay, N. 1976. Metallisk friktin g kldtryksvejsning. Ph.D.-thesis, AMT, Technical University f Denmark. 8. Bay, N. 1979. Cld pressure welding. The mechanisms gverning bnding. Trans. ASME, I Engn. Ind. 101(2):121-127. 9. Eggers, H., Krause, E., and Ruge, J. 1970. Zum Mechanismus des Kaltpressschweissens- Bruchflachenuntersuchungen mit dem Rasterelektrnenmikrskp. Schweiss. Schneid. 22:241-244. 10. Cave, J. A., and Williams, J. D. 1973. The mechanism f cld pressure welding by rlling. / Inst. Met. 101:203-207. 11. Wright, P. K., Snw, D. A., and Tay, C. K. 1978. Interfacial cnditins and bnd strength in cld pressure welding by rlling. Met. Techn.: 24-31. 12. Bay, N. 1981. Cld pressure welding. A theretical mdel fr the bnd strength. Prc. Instn. Metall. Cnf.: jining f Metals-Practice and perfrmance. Univ. Warwick, U.K.: 1-16. 13. Jhnsn, W. 1954-55. Extrusin thrugh wedgeshaped dies, Part I, /. Mech. Phys. Sl. 3:218. 14. Hill, R. 1950. The mathematical thery f plasticity: 185. Oxfrd: Clarendn Press. 15. Cnrad, H., and Rice, L. 1970. The chesin f previusly fractured FCC metals in ultrahigh vacuum. Metall. Trans. 1:3019-3029. 16. Upit, C. P., and Manik, J. J. 1969. The effect f lad n the chesive strength f virgin surfaces f plastic metals. Wear 13:77-84. 142-s MAY 1983