Parameters Evaluated in Long Cycle Aluminum Vacuum Brazing

Transcription

1 Parameters Evaluated in Long Cycle Aluminum Vacuum Brazing The special design requirements of aircraft heat exchangers imposed restrictions that were met successfully after a close study of all brazing variables BY E. B. GEMPLER ABSTRACT. The paper describes a development program undertaken at United Aircraft Products to establish the parameters involved in a long time-temperature cycle for vacuum brazing of aluminum alloys. The objectives were to evaluate the brazing capabilities of the various aluminum alloy combinations used inhouse, to evaluate the newer alloys developed especially for vacuum brazing, to determine the optimum parameters for use of the alloy systems and to compile this data into a form usable by the corporation. The parameters considered were variations in brazing sheet thickness, aluminum alloy combinations, fin shape and height, side piece (nosepiece) width and height, rate of temperature drive, brazing temperature and holding time, fixture pressure, magnesium catalyst and vacuum requirements. Results of the alloy systems brazed are discussed in detail. Particularly significant was the discovery of radical differences in behavior and microstructure of similar Mg-containing vacuum brazing filler metal alloys (Aluminum Association X7 designation) from two different sources. Subsequent work revealed these differences in various lots of the brazing sheet received from the same source. Corrective measures were found, and controls are reviewed. E. B. GEMPLER is Staff Metallurgist with United Aircraft Products, Inc., P.O. Box 37, Forest, Ohio. Paper was presented at the 7th AWS International Brazing Conference held during the 57th A WS Annual Meeting at St. Louis, Missouri, May 10-14, Introduction United Aircraft Products has fluxless brazed aluminum heat exchangers since the middle of Prior to the acquisition of a vacuum furnace, fluxless brazing was done by either of two methods. In-house brazing was done in a retort that was evacuated and then backfilled with dry argon. The retort was placed into a furnace for its heat source. Successful brazing was done by this technique for many years using magnesium chips as a catalyst for the process. In addition, successful brazements were made in vacuum by qualified outside brazing contractors for several years using UAP prescribed brazing cycles. A program was initiated to evaluate the various parameters affecting aluminum vacuum brazing in general, and specifically to determine those operating characteristics and parameters peculiar to our newly purchased vacuum furnace. The objectives were to evaluate the brazing capabilities of the various aluminum alloy combinations used inhouse, to evaluate the newer alloys developed especially for vacuum brazing, to determine the optimum parameters for use of the alloy systems, and to compile this data into a form usable by the corporation. Aircraft heat exchangers are designed to a variety of requirements. These designs many times result in the requirement to braze very heavy and very light sections. Additionally, the complexity of the design coupled with high pressure requirements often necessitate intricate and heavy braze fixtures which become a major part of the furnace load. UAP's furnace design had to consider the variety of sizes and shapes of furnace loads, and hence the time required to make a successful brazement. It was realized that a long brazing cycle could result. Consequently, both metallurgically and productionwise, the control of the brazing parameters and variables becomes increasingly important to minimize the problems of excessive diffusion, alloying, erosion, isolated fin melting and grain growth. Our past experience had showed relatively long brazing cycles of two hours or more compared to the short cycles of 15 to 30 minutes total heating time reported for brazing atypical automotive core in a high production furnace. Where this type of operation was done in a furnace designed specifically for a particular shaped family of cores, either self-fixtured or lightly fixtured, our vacuum brazing required a furnace designed to accept many sizes and shapes of heat exchanger cores. Figures 1 and 2 compare one usual difference between a typical automotive heat exchanger core design (Fig. 1) and those made for the higher pressure requirements of aircraft use (Fig. 2). Where the automotive core will be a construction in which the side enclosures are formed into the divider sheets (tube sheets) from thin clad braze material, the aircraft core with its more stringent design requirements may have rather thick, solid side pieces (we use the nomenclature "nosepieces") brazed to a thin tube sheet. The core with all thin materials will heat up faster and more evenly throughout than one with a combination of very thick and thin sections. The construction with two pieces of braze clad sheet in contact has double the amount of alloy available for large fillets and leaktight WELDING RESEARCH SUPPLEMENT I 293-s

2 LARGE FILLET 15 /ocladding,v.mn' I 15% I Ai11 :N, MATERIAL: ' '-... -:. fl-s«t SMALL FILLETS -I5%CLADDING AGAINST BARE METAL MATERIAL: X/ BRAZE SHEE! 020 r Fig. 1 Aluminum vacuum brazed core typical of automotive designs NOTE: WITH -HE SAME THICKNESS NO'S II.I2,21, BRAZE SHEET CLADDING IS IO% Fig. 2 Aluminum vacuum brazed core typical of the higher pressure design for aircraft use Table 1 Aluminum Alloy Data Brazing sheet Alloy No. 11 No. 21 No. 23 X7 Other alloys Alloy Sides clad Core alloy Nominal cladd ng comp osition AI-7.5%Si Nominal composition, % AI-7.5%Si AI-10%Si AI-9.7%Si-1 5%Mg AI-1.2Mn AI-1.0Mg-0.6Si-0.25Cr-0.25 Cu AI-0.7Mg-0.4Si AI-0.65Mg-0.35Si-0.25Cu joints, whereas that with one braze clad side contacting an unclad nosepiece lends itself to smaller fillets, less flow and half the alloy to produce leaktight joints. Experimental Procedure Materials Aluminum alloy combinations evaluated were the following: No. 23 braze sheet, No nosepieces, and No ruffled and lanced fins. No. 21 braze sheet, No nosepieces, and No ruffled fins. No. 21 braze sheet, No nosepieces, and No ruffled fins. No. 11 braze sheet, No nosepieces, and No ruffled fins. No. X7 braze sheet (from two sources), No nosepieces, and No ruffled fins. No. X7 braze sheet (from two sources), No nosepieces, and No ruffled fins. Table 1 shows the nominal compositions and temperature ranges for these alloys. Other Variables Nominal clad thickness 10%<.063 5%> %< %>.064 Brazing range, F Approximate melting range F Other parameters evaluated were brazing temperature, time at temperature, vacuum requirements, furnace drive, pre-braze cleaning, amount of magnesium, nosepiece width, fin shape and height, braze sheet thickness, total cycle time, fixture pressure and gap sealing ability. No effort was made on this program to encase the load in a retort which in effect would be a controlled volume for the magnesium vapor atmosphere. Equipment All samples were brazed in a Sunbeam horizontal front loading cold wall vacuum furnace with work zone dimensions of 18 x 30 by 84 in. long. The furnace was rated for 5 x 10""" torr at 2200 F (1204 C) and 5 X IO" 5 torr at 1200 F (649 C). Heating was by molybdenum elements in which the two ends, the sides and the top and bottom heaters could be separately regulated. Furnace temperature control was by a platinum/platinum-10 rhodium Type S thermocouple. Rated pumpdown time to 5 X 10" 5 torr, cold, was 20 minutes. Cooling was by inert gas recirculated over a water cooled, stainless steel finned tube heat exchanger. Twelve internal jacks were provided for Type K chromel/alumel thermocouples with a twelve point recorder for monitoring load temperatures. Vacuum was monitored by a cold cathode ionization gage, a hot cathode ionization gage and two thermocouple gages. Furnace operation could be either manual or automatic. Manual control was used for all development work. Braze Samples To simplify the test runs, a standard braze sample was used. All braze sheets were 3.5 in. (88.9 mm) OD discs. The nosepiece was cut from 3.5 in. (88.9 mm) OD, in. (5.51 mm) wall tubing, and the fins were Eloxed to 3.0 in. (76.2 mm) OD discs. The four pieces were stacked together as shown in Fig. 3A to form a hollow disc with internal fins. By machining the outside diameter of the nosepiece ring, the wall thickness was varied. Rings of three different compositions were used to vary the nosepiece alloy. The outer disc (core sheet) composition and thickness were varied during the program. Fin configuration was controlled by Eloxing the discs from several standard production fin configurations used in house. When comparing similar alloys from different sources the samples were assembled with one outer disc of brazing sheet from each of the two sources, both of similar thicknesses. The in. (0.5 mm) braze sheet was used as the standard thickness because much of it was used inhouse. The quality of all brazed joints was compared to those obtained with the standard thickness. Brazing Fixture The braze sample was placed in a simple but effective fixture as shown in Fig. 3D. This consisted of two sections made from a steel channel beam ground flat and parallel on both the fixture surface and bottoms of the legs. The pressure on the braze sample was controlled by varying the num- 294-s I OCTOBER 1976

3 .064" c. nos Lns. i at F 10 NOTE - All ^ Fig. 3A through 3G Various steps in preparing the braze samples for brazing, testing and micrographic examination WELDING RESEARCH SUPPLEMENTI 295-s

4 c. Run 31 <ie i. Fin - Core Sheet Joint Run No. 7 No. 23 core sheet, 6061 nosepiece 3003 fins F 5 minutes FIGURE 4 Pin - Core Sheet Joints Run N No. 23 core sheets, 6061 oosepi F FIGURE 5 FIGURE 6 a..020" No. 11 core sheet «. Run lloo'f 10 minutes * Run F 10 minutes ^^^^^_^^^^^_^^^ t>. Run F 15 minutes K r>cn" M n r.h, No. 21 core sheet, 6061 nosepiece and b - Run 43 nl " H20 F 10 minutes o.. 05U No. 11 core sheet ~ n n^ r fins No. 21 core sheet, 3003 fins and nosepiece Temperature F 10 minutes Vacations in tossing temperature. Variations in brazing temperature FIGURE 7 FIGURE 8 FIGURE 9 Figures 4 through 9 showing brazed core sheet joints resulting from different brazing sheets, temperatures and cleaning methods 296-s OCTOBER 1976

5 ber and size of the weights on the top half of the fixture. Samples were brazed in groups of three or four as shown in Fig. 3E. Temperature control was maintained by thermocouples placed in each sample and in the fixtures (Fig. 3E). Fixtures were centrally located in the furnace chamber and thermocouples attached to the 12 point recorder jacks. Sample Evaluation The brazed samples were gas tungsten-arc welded to a tubular stem and tested as follows: Visual Inspection Samples were examined for braze flow, braze voids, and general appearance. Leak Test Samples were attached to a pressure line and immersed in a lighted tank of water. The pressure was increased to 200 psi (1.379 MPa) for the samples with thin core sheets, and 300 psi (2.069 MPa) for those with thick core sheets. Any bubble formation indicated a leak. If the leaks were small, the sample was weld repaired for the subsequent burst test. Burst Test Both the leak tight and weld repaired samples were hydraulically tested to failure. Using a light hydraulic fluid, the pressure was slowly increased until the sample ruptured and an oil leak was observed. (Note core sheet deformation and burst crack in Fig. 3B). The ruptured sample was cut and examined visually for failure mode. Figure 3C shows a ruptured sample with an excellent braze in which all fins failed in tension. Metallurgical Examination Polished and etched sections were examined for braze quality, such as filler metal flow, fillets, diffusion, alloying, grain size, erosion, excess brazing filler metal and voids. When heating a core with wide variation in metal thicknesses, the thin fins get hotter faster and reach brazing temperature earlier than the heavy sections. In long cycle brazing the effects of erosion, diffusion and fin alloying, excessive brazing filler metal, etc., will be most noticeable at the fin brazed joints, thus the photomicrographs shown throughout this paper are those of the fin to core sheet joints, where these effects are most detrimental. Figure 3F is a photomicrograph at 75X magnification of an ideal braze joint in one of our recent heat exchanger cores with time to temperature of slightly over two hours. Note the good fillet and little to no alloying into the in. (0.127 mm) fin on the top. The particular braze sheet is X7 purchased from source 2. Figure 3G is a photomicrograph at 75X of X7 braze sheet from source 1. Total time to temperature was 18 minutes. Both samples were heated to 1105 F (540 C) and then cooled. Vacuum and Braze Cycle Requirements Early information indicated that a vacuum of 3 X 10~ 4 torr would give satisfactory brazements. Our tests showed that brazed joints made at approximately 1100 F (538 C) temperature in this vacuum range were borderline, and that consistently good vacuum brazes with all the alloys evaluated should be made in the 10~ 5 torr range. Initial runs were made by stabilizing all thermocouples within 25 F (13.8 C) just below 950 F (510 C), stabilizing to within 20 F (11 C) below 1030 F (554 C) and then increasing to brazing temperature as quickly as possible but keeping all thermocouples to within a total spread of 10 F (5.6 C), preferably 5 F (2.8 C). Later experiments entailed stabilizing all thermocouples to within 15 F (8.3 C) at 1000 F (538 C), then driving hard with the requirement that all thermocouples be within the 5 to 10 F (2.8 to 5.6 C) range at brazing temperature. Prebraze Cleaning Variations in cleaning requirements of all alloys were tried. Results indicated that all parts from alloys other than the X7 brazing sheet should be cleaned chemically prior to brazing. A proprietary cleaning procedure was used. In essence, the use of a caustic cleaning followed by cleaning with HN0 3 -HF and HN0 3 acids, deionized water rinsing, and drying proved satisfactory. The X7 brazing sheets from both sources were vapor degreased only. Discussion Series 1 Initial samples were run using in. (2.28 mm) thick No. 23 brazing sheet with No nosepiece rings and 3003 fins. This was done to optimize parameters for a production item using these alloys. Pressure was varied from 2.45 psi ( kpa) to 7.98 psi ( kpa). Where psi ( kpa) gave consistently leaktight joints for the in. (10.5 mm) high ruffled fins, a pressure of 7.98 psi was needed to reproduce these results using in. (2.54 mm) high lanced fins. Brazing temperatures and times were varied from 1065 to 1070 F (573 to 577 C) at 30 minutes to 1100 to 1105 F (593 to 596 C) at 30 minutes with many combinations in between. Figures 4 and 5 show variations in time and temperature from 1075 F to 1095 F (579 to 590 C) for five and ten minutes respectively. Note the increased alloying into the fin metal as the time and temperature are increased. The upper photograph of Fig. 5 shows a fin that failed in tension during the burst test. Series 2 A second alloy series evaluated was that using the No. 11 brazing sheet with all remaining parts being No alloy. The suggested brazing range for No. 11 braze sheet is from 1100 to 1140 F (593 to 615 C). Initial time and temperature variations ranged from 1120 F to 1145 F (604 to 618 C) for times ranging from just taking the part to temperature to holding for ten minutes. Results were borderline in these ranges, and it was only when temperatures were taken to between 1150 and 1160 F (621 and 626 C) that consistently good leaktight brazes were made (see Fig. 7). With the thickness of the brazing sheet decreased to in. (0.5 mm), in this series the pressure requirements for successful brazements were decreased to as low as 1.48 psi ( kpa). Figure 6 shows the results of an experiment in which prebraze cleaning of parts was varied to compare chemical cleaning versus vapor degreasing. Several runs were made and consistently the parts that were degreased only gave extremely poor to no bonding. Series 3 The results with this series using No. 21 brazing sheet with No rings and fins were more consistent with recommended brazing temperatures than those using the No. 11 braze sheet. Temperatures and times were varied from 1100 to 1140 F (593 to 615 C) for 10 to 20 minutes. The best overall results were obtained in the middle of this temperature range, approximately 1120 F (604 C). Figure 9 shows variations in brazing temperatures. As the upper temperature limits are approached, there is excessive fin alloying, large grain size and intergranular penetration as shown in the top photograph. Series 4 The alloy system using No. 21 brazing sheet with No nosepiece rings behaved somewhat differently. Temperatures were varied from 1075 Fto 1120 F (579 to 604 C) for 10 to 15 minutes knowing full well that the 1120 F (604 C) was good for the No. 21 braze sheet but above the eutectic WELDING RESEARCH SUPPLEMENT! 297-s

6 m i. Source 2-021" V. braze.; *W4 Ult.* l l - FIGURE 10 FIGURE 11 a. Source 1 X-7 b. Source 2 X7 2 X7 Run No. 16 Run No. 17 ^"»- * Temperature F 10 minutes Temperature F 20 FIGURE 12 FIGURE 13 FIGURE 14 Figures 10 through 14 showing the effects of diffusion and variations of time and temperature 298-s OCTOBER 1976

7 point of the 6061 alloy where localized melting would take place. Excellent brazes were observed from 1075 to 1100 F (579 to 593 C) as shown in Fig. 8. All of the aforementioned brazing sheets were developed for flux brazing processes. The vacuum brazing alloy with magnesium in the cladding does several things, namely, presents the magnesium at the joint interface where it is needed, holds the magnesium intact until approximately 1000 F (538 C) before vaporization of any amount takes place, and provides for less expensive and quicker prebraze cleaning. Series 5 and 6 These series used the Aluminum Association designation X7 braze sheet from two different sources with Nos and 6063 nosepiece rings. The X7 alloy supplied by two different companies has a published brazing range of 1080 to 1120 F (582 to 604 C) over which our evaluation was made. Time at temperature was varied from no holding time to 20 minutes at temperature. For direct comparison each sample was assembled with a core sheet from each source on it. The standard braze cycle used for comparison of the brazing variations was heating to 1105 F (596 C) and immediately cooling. Brazing sheet thicknesses evaluated were from to.064 in. (0.31 to 1.63 mm) with the standard being in. (0.5 mm). Pressure variations ranged from 1.26 psi (8.688 kpa) to 2.45 psi ( kpa). Metallurgical examination of this alloy brazed at 1105 F (596 C) showed a great difference in appearance between the brazing sheet from the two sources. Figure 10b shows the source 1 brazing sheet with large grains in the fillet, more diffusion into the core metal and less diffusion into the fin. The significance of this was not realized until later in the experiment when higher temperatures and longer times were used during brazing. Note the to in. (0.18 to 0.2 mm) undiffused core metal in source 1 brazing sheet, Fig. 10b, as compared to the to.013 in. (0.31 to 0.33 mm) undiffused core metal in source 2 brazing sheet, Fig. 10a. Figures 11, 12 and 13 are variations of time and temperature from 1080 F to 1120 F (582 to 604 C). As the brazing temperature was increased, the physical appearance of the source 1 core sheet sagged as shown in Fig. 11a whereas that of source 2 core sheet remained flat in all cases. Figure 14 shows the appearance of both sources of brazing sheet on the same sample brazed at 1115 to 1120 F (601 to 604 C) for 20 minutes. In the higher temperature ranges the degree of core sheet sagging became progressively worse as the time at temperature was increased (source 1 alloy only). A series of runs were made at the maximum temperature range of 1115 to 1120 F (601 to 604 C) from no holding time at temperature to 20 minutes at temperature. The width of the unalloyed core metal, in. (0.38 mm) initially, was measured and plotted against time at brazing temperature. The resulting plot, Fig. 19, shows the source 2 alloy being relatively stable up to 20 minutes at temperature, whereas the source 1 alloy progressively deteriorated with time. This was attributed to the rapid diffusion into the core metal causing subsequent weakening, sagging and porosity. Figure 15 shows the effects of brazing sheet thickness at the standard 1105 F (596 C) temperature. Note that the in. (0.31 mm) thick source 1 brazing sheet (Fig. 15a) sagged and became very porous. Figure 16 is the outer appearance of this sample. Note the many encircled 'holes through the braze sheet observed during the leak test operation. Figure 15b shows excessive fin alloying while Fig. 15c shows fin alloying, excessively large fillets and braze overrun with the in. (1.63 mm) thick braze sheet. Figure 17 shows the effects of ten minutes time in the range 1105 to 1120 F (596 to 604 C) with in. (1.27 mm) and in. (1.63 mm) braze sheet. The in. (0.19 mm) and in. (0.163 mm) thick cladding on the brazing sheets used in this run completely alloyed through the in. (0.127 mm) No fin metal. Based on these tests, recommendations were made to use only the source 2 alloy in thin sheets for long cycle brazing. Subsequent to this decision, it was discovered that this same radical difference, occurred between different lots of the X7 brazing sheet from the same source. Figure 18a shows the microstructure of source 2 alloy from a coil as compared to that in Fig. 18b from flat stock. Both photographs were from the same heat exchanger core. Corrective action was taken by the supplier to remedy the situation, and a rather simple brazing test was used to verify acceptability of the material. Figure 18c shows the material of Fig. 18a after reworking and brazing at 1110 F (543 C) for 20 minutes to determine if the core structure was acceptable. Clearance-Bridging Ability The last series of tests in this program was to evaluate the clearancebridging ability of the various thicknesses of the X7 braze sheet to the contact surface of the nosepiece. The nosepiece rings were modified by two techniques. The contact surfaces were tapered from the ID to the OD in one case giving a brazing sheet clearance from approximately in. (0.41 mm) at the ID to a in. (0.38 mm) clearance at the OD. The second technique incorporated rings that had a known clearance machined on the outside half of the contact surface. Results indicated that the in. (0.5 mm) brazing sheet would bridge a to.001 in. (0.013 to mm) clearance. The in. (0.76 mm) brazing sheet would bridge a to.0015 in. (0.025 to 0.04 mm) clearance and the in. (1.63 mm) brazing sheet would bridge a in. (0.05 mm) clearance. Magnesium Very little has been said about the amount of magnesium chips used with the silicon-bearing braze sheets. One reason for this is that our initial approach was based somewhat on the state-of-the-art rather than on a strictly statistical approach. In general, the recommended amount of magnesium depends on factors such as the furnace size, the load size, the heating time to brazing temperatures, available vacuum, the amount of magnesium vapor suppression employed, among others. In all cases magnesium must be available at the brazing temperature. Too little magnesium will be depleted prematurely. Excessive magnesium will cause localized burning (melting) of very thin fins. We have an in-house program underway to statistically determine the magnesium parameters. The aforementioned factors will be rigidly controlled. Our goal is to establish a simple formula into which the parameters can be plugged for the correct amount of magnesium to be used. Fixture Pressure Fixture pressure must be adequate to keep parts in contact throughout the brazing cycle. Often when excess pressure is required, it is needed more to flaten uneven and nonuniform parts, or to press down very strong fin shapes to the height of the nosepieces rather than just hold parts in contact. This was especially noticeable when the core sheet thickness was decreased from in. (2.36 mm) to in. (0.5 mm), and WELDING RESEARCH SUPPLEMENT! 299-s

8 oorce I r» ^.1 3 ^ L M FIGURE 16 / J I^B?J -W."'. b. X? FIGURE ui^ t.»fc*» LO llob'v and ; ' - - ' *, " < & «_ ' ^ T, /, I 2 064" X7 Core Sheet FIGURE 18 :JKIGINAL THICKNESS MATERIAL: BRAZE SHEET.021 THK MATI 15% CLAD TWO <;idfs * " : : AMPtEJ SAMPLE A SOURCE l USCE2 Cox< iheet Run No. 34 Result* of excess btaze VI toy on heavy core sheets and excess tine and temperature TIME A T TEMPERATURE TEMF FIGURE 17 FIGURE 19 Figures 15 through 19 showing various problems encountered in arriving at the final parameters as discussed in the text 300-s I OCTOBER 1976

9 where the fins were changed from the in. (10.5 mm) high ruffled to the in. (2.5 mm) high lanced structure. Pressure requirements can be minimized by presizing parts as well as the stacked heat exchanger core. Conclusions 1. Successful and practical long cycle aluminum vacuum brazing can be done with the Si-AI clad braze sheets with the aid of magnesium as a catalyst. 2. Successful and practical long cycle aluminum vacuum brazing can be done with the X7 (9.7 Si, 1.5 Mg bearing) clad brazing sheets providing the microstructure lends itself to a long hold time in the recommended maximum temperature range without sagging of the thinner sheets. 3. In long cycle brazing the best braze results were obtained at 1150 to 1160 F (621 to 626 C) with the No. 11 braze sheet and No alloy; at 1070 to 110OF (577 to 593 C) with the No. 23 braze sheet and No alloy; at 1100 to 1130 F (593 to 610 C) with the No. 21 braze sheet and No alloy; and at 1075 to 1100 F (579 to 593 C) with the No. 21 braze sheet and No alloy. Good braze results were obtained over the range of 1080 to 1120 F (582 to 604 C) with the X7 braze sheet and No alloy, however, the higher temperatures caused excessive fin alloying and braze overflow with the thicker brazing sheet. Results using this brazing sheet with the No nosepiece rings were inconclusive. 4. There are basic differences in the microstructure and brazing response between different lots of the X7 brazing sheet depending on the prior fabrication. More controls on the specification of this alloy should be established. 5. The most reproducible braze results were obtained in a vacuum range of 10-5 torr and with chemical cleaning of all alloys other than the X7, which can be degreased only. 6. Fixture design adequate to keep parts in contact throughout the brazing cycle may vary based on material, material thickness and material shape. Acknowledgment Special thanks goes to Mr. Gary Cobb, Engineering Test Technician, for his photographic and metallographic work during the program, and for the photography in this paper. WRC BULLETIN 207 JULY 1975 Joining of Metal-Matrix Fiber-Reinforced Composite Materials by G. E. Metzger The metal-matrix fiber-reinforced composites, their fabrication methods, properties, and some of the applications are described. Virtually all composite applications have been for demonstration or test structures, and most of these have been in the aerospace industry. The available literature on composite joining has been summarized and evaluated, mainly on joining as a secondary fabrication process; not for primary fabrication of the base material. The joining processes discussed include diffusion welding, fusion welding, resistance welding, brazing, soldering, mechanical fasteners, and adhesive bonding. Materials joined include Al/B, Al/graphite, Al/stainless steel, Al/Be, Ti/B, Ti/W, and Ti/graphite composites, although most work by far has been with Al/B materials. The greatest effort has been expended on brazing, resistance welding, and mechanical fasteners, with considerable work also on soldering. The tensile strength joint efficiency of joints by these processes shows a marked decrease as the base material tensile strength increases. For brazing and resistance welding, the joint efficiency is about 60/7 for low strength composites of 600 MPa tensile strength and about 30'V for the highest strength composites of 1400 MPa. The corresponding joint efficiencies for mechanical fasteners are about 40 and 20' 7. Efficient joining of high-strength composites by diffusion welding, with the exception of titanium-matrix and ductile-fiber composites, or by adhesive bonding, appears to be of low probability. The great difficulty of fusion welding composites has resulted in a minimum of work by these processes, but their potential is believed to be good. The publication of this report was sponsored by the interpretive Reports Committee of the Welding Research Council. The price of WRC Bulletin 207 is $6.50. Orders should be sent with payment to the Welding Research Council, United Engineering Center, 345 East 47 Street, New York, N.Y WELDING RESEARCH SUPPLEMENT I 301-s