BRAZING & SOLDERING TODAY

MARCH 200762

TThere is considerable interest in join-ing dissimilar materials such as met-als to carbon-carbon (C/C) com-

posites for a variety of aerospace andland-based applications (Refs. 1–3). Inparticular, thermal management systemsare being developed for various space ex-ploration applications that require thejoining of metal tubes to C/C compositeface sheets (Ref. 4). For this reason, stud-ies are underway to determine the effec-tiveness of various braze and solder ap-proaches toward joining Ti tubes todifferent C/C face-sheet materials.

In an earlier study (Ref. 5), a tube-platetensile test was developed to determinethe tensile load-carrying ability of Ti tubesbrazed to C/C plates. The three differentbraze materials used in that study all wetthe Ti and the C/C composite and ap-peared to strongly bond to the Ti and theC/C substrates for simple flat-on-flatjoints (Ref. 6). The simple tube-plate tech-nique was very good at illuminating suchissues as how well the braze spreads to fillin the region between the tube and theplate, which dictates the bonded area andultimately the load carrying ability of thejoint. In an earlier study, higher tempera-ture brazes were used on a lower conduc-tivity C/C composite (woven T300 fibers;resin-derived carbon matrix). In thisstudy, lower temperature braze materialsand one solder were used to braze similar

tube-plate structures as well as more con-ventional butt-strap lap shear specimensfor Ti and a higher stiffness (woven P120fibers), higher conductivity CVI carbonmatrix composite.

Experiment Details

Two types of joint structures were usedin this study. For all the structures, com-mercially pure Ti (Grade 2, TIMET Inc.,

Mo.) was joined to a C/C composite con-sisting of stacked three-harness satin 2Dwoven carbon fabric (P120) and chemi-cally vapor infiltrated (CVI) carbon ma-trix composite (Goodrich Corp., Santa FeSprings, Calif.). The C/C thickness was ap-proximately 1.2 mm.

The composition and liquidus temper-atures of braze and solder materials usedin this study are listed in Table 1. The brazealloys were obtained from Morgan Ad-

Comparison of Braze and SolderMaterials for Joining Titanium

to CompositesStudies were conducted to determine the tensile load-carrying ability of joints

for a curved surface and the shear strength of joints in a butt-strap lap test

BY GREGORY N. MORSCHER, MRITYUNJAY SINGH, AND TARAH SHPARGEL

GREGORY N. MORSCHER (Gregory.N.Morscher@grc.nasa.gov) and MRITYUNJAY SINGH are with Ohio Aerospace Institute/NASAGlenn Research Center, Cleveland, Ohio. TARAH SHPARGEL is with ASRC Aerospace at NASA Glenn Research Center, Cleveland, Ohio.

Fig. 1 — Schematic representations of joint tests. A — Tube-plate tensile; B — butt-strap tensile.

A B

Morscher Layout:Layout 1 2/7/07 8:28 AM Page 62

BRAZING & SOLDERING TODAY

63WELDING JOURNAL

vanced Ceramics. Braze foils, ~ 50 μmthick, were used in all cases. For CuSilABA and TiCuSil, joints were fabricatedusing a paste form of the braze as well. Thebraze and solder materials’ processingtemperatures were always performed afew degrees higher than the liquidus tem-perature for each braze. Soldering was

performed by S-Bond Technologies, LLC,Landsdale, Pa. The soldering temperaturewas approximately 420°C (Ref. 7), signifi-cantly lower than the braze materials’ pro-cessing temperatures. Soldering also in-volved a proprietary metallizationtreatment to the C-C surfaces.

Two types of joint specimens were fab-

ricated: tube-plate tensile specimens andbutt-strap-tensile (BST) shear specimens— Fig. 1. Tube-plate tensile tests wereperformed on all the braze systems butnot on the solder system. Butt-strap-tensile tests were performed on most ofthe braze systems and on the solder sys-tem. The C/C panels were machined into2.54 × 1.26-cm plates that were joined to1.26-cm-diameter × 2.54-cm-long tubesfor the former and 1.26 × 75-mm lengthsof a Ti flat plate for the latter test config-uration, respectively.

Investigation Results

Tube-Plate Tensile Tests

The tube-plate tensile data are shownin Fig. 2 for four different braze systems.At least three specimens were tested foreach condition. The TiCuSil and CuSil-ABA braze compositions were tested inpaste as well as foil form. The CuSil-ABApaste system had the best load-carryingability of all the braze systems followed bythe TiCuSil paste and TiCuSil foil systems.All three systems showed excellent brazespreading and large metallic bondedareas. Figure 3 shows a typical fracturesurface of a tube-plate specimen wherethe braze spread and strongly bonded tothe outer C/C ply over a significant dis-tance of the plate. The CuSil-ABA foil sys-tem as well as the CuSin-1 ABA and In-CuSil ABA braze systems, which were thetwo lower-temperature systems, allshowed poorer load-carrying ability. Forthese systems, considerably poorer braze

Fig. 2 — Tube tensile failure loads for different braze approaches.

Fig. 3 — Tube and C/C plate fracture surfaces for CuSil-ABA pastebraze material showing the bonded area of the outer ply of C/Cbrazed to the Ti tube (left) and C/C plate (right).

Fai

lure

Lo

ad, N

TiC

uSil

Foi

l

TiC

uSil

Pas

te

Cus

il-A

BA

Foi

l

Cus

il-A

BA

P

aste

Cus

in-1

AB

A

Foi

l

Incu

sil A

BA

Foi

l

Table 1 — Braze and Solder Properties

Braze/Solder Composition Liquidus Temp., °C

TiCuSil Braze 68.8 Ag, 26.7 Cu, 4.5 Ti 900CuSil ABA Braze 63.0 Ag, 35.3 Cu, 1.8 Ti 815CuSin-1 ABA Braze 63.0 Ag, 34.3 Cu, 1.8 Ti, 1.0 Sn 806InCuSil ABA Braze 59.0 Ag, 27.3 Cu, 1.3 Ti, 12.5 In 715S-Bond 400 Alloy Solder Zn, Ag, Al, with trace amounts 410–430 (Ref. 7)

of Ga and Ce (Ref. 7)

Table 2 — Tube-Plate Tensile Results

Braze #Tested ⊥; // Failure Load, N ⊥; // Stress, MPa ⊥; //

TiCuSil Paste 3;0 41.1 ± 3.0(a) 0.45 ± 0.02

TiCuSil Foil 6;6 34.2 ± 10.0; 0.86 ± 0.44;34.3 ± 12.8 0.39 ± 0.14

CuSil ABA Paste 11;6 48.7 ± 9.3; 0.47 ± 0.09;47.8 ± 13.0 0.32 ± 0.18

CuSil ABA Foil 6;0 18.0 ± 9.9 0.55 ± 0.19

CuSin-1 ABA Foil 3;3 15.1 ± 5.7; 0.68 ± 0.44;5.8 ± 2.2 0.56 ± 0.06

InCuSil ABA Foil 5(b); 0 8.2 ± 4.1 NA

(a) Standard deviation.(b) Two specimens failed in handling for InCuSil foil. It was apparent that this was a poor braze for joining Tito this C/C composite. The bonded area was not continuous, so the determination of stress was not performed.

Morscher Layout:Layout 1 2/7/07 8:29 AM Page 63

BRAZING & SOLDERING TODAY

MARCH 200764

spreading and smaller bonded areas wereobserved — Fig. 4. Note that in Fig. 4 (In-CuSil ABA foil), metallic bonding of thebraze material to the C/C composite waspatchy, predominantly only the outer towsof the C/C surface ply were bonded. Thiswas one of the poorest braze spreading ex-amples. For the CuSil-ABA foil andCuSin-1ABA foil, the bonded area usuallyextended over the width of the C/C plate,but the distance the braze spread alongthe length of the plate was significantlyless compared to Fig. 3.

Using the size of the bonded areas, theapparent failure stress of the joints can bedetermined. Table 2 lists the average fail-ure load and failure stress for the differentbraze combinations. The data were sepa-rated for each braze system based on theorientation of the outer fibers on the sur-face of the C/C composite. The C/C com-posite is made up of stacked 2D harnesssatin woven cloth where the exposed sideof the composite is dominated by the fibertows either oriented perpendicular or par-allel to the axis of the tube (for example,most of the surface fiber tows in the har-ness satin weave are oriented perpendicu-lar to the tube axis in Figs. 3, 4). In the ear-lier study (Ref. 5), it was observed thatfiber tows orientated perpendicular to thetube axis had higher load-carrying abilitycompared to the case where the fiber towswere predominantly oriented parallel tothe tube axis. This effect was not as no-ticeable for the composite system tested inthis study. The composites used in thisstudy have a 3 harness satin weave (i.e.,each fiber tow crosses over two perpendi-cular tows and then under one perpendi-cular fiber tow) and a larger number

(2000) of fibers per towcompared to the compos-ites used in Ref. 5 (5 har-ness satin, 1000 fibers pertow). For the compositesof this study, the net effectis a lower aspect ratio ofsurface tow orientationsince the tow widths arewider and the outer towsonly cross over two per-pendicular tows com-pared to four perpendicu-lar tows in the 5 harnesssatin weave. Therefore, itis not surprising that thefailure loads for perpen-dicular and parallel ori-ented composites werenot significant in thisstudy (Fig. 2 combines allthe data of Table 2).

The TiCuSil foil was also used in Ref.5 with the T300 resin-derived carbon ma-trix composites. For that system, the fail-ure loads were less than 10 N compared to34 N for the P120 CVI carbon matrixplates used in this study and the failurestress was on the order of ~ 0.2 MPa com-pared to ~ 0.6 MPa for the P120 CVI car-bon matrix plates used in this study (Table2). This is commensurate with the differ-ence in interlaminar tensile (ILT)strengths between the two systems. Forthe T300 resin-derived carbon matrixcomposites, the ILT strength (ASTMC1468) was found to be 2 MPa, whereasthe ILT strength of the P120 CVI carbonmatrix composites of this study were 6.5MPa (Ref. 8).

One issue of concern with the braze

systems was the presence of cracks afterprocessing within the braze reaction layer— outer-ply of the C/C. Figure 5 shows acrack for a TiCuSil paste specimen. Thiswas most apparent for the TiCuSil speci-men, which would be expected due to thehigher processing temperature of theTiCuSil (see discussion below). Eventhough these cracks were more apparentand larger for the TiCuSil specimens, ten-sile load-carrying ability of the tube-plateconfiguration was primarily dependent onthe spreading of the braze, i.e., bondedarea of the C/C.

Butt-Strap Lap Tensile Shear Tests

Butt-strap lap tensile (BST) tests wereperformed as shown in Fig. 1B for the

Fig. 4 — Tube and C/C plate fracture surfaces for InCuSil ABA foilbraze material showing the distinct bonded areas of the outer ply ofC/C brazed to the Ti tube (left) and C/C plate (right).

Fig. 5 — TiCuSil paste brazed tube-plate specimen showing acrack after processing. A — Edge-on perpendicular to the tubeaxis; B — at an angle parallel to the tube axis.

A

B

Morscher Layout:Layout 1 2/7/07 8:29 AM Page 64

BRAZING & SOLDERING TODAY

65WELDING JOURNAL

TiCuSil and CuSil ABA braze composi-tions and one solder material (S-Bond). Inaddition, a BST test was performed for theC/C composite brazed to the C/C compos-ite using the CuSil ABA paste. The S-Bond is processed at a significantly lowertemperature (Table 1), which should min-imize the buildup of thermal stresses be-tween the Ti and the C/C. For the case ofC/C brazed to C/C, no thermal stresseswould be expected between the joinedplates since they are the same material.

Only thermalstresses that wouldoccur between thebraze and the C/Cwould develop,which are expectedto be minor due tothe ductility of thebraze.

Figure 6 shows theshear strength re-sults for the differentbraze and soldercombinations. Atleast three speci-mens were tested foreach combination. Asignificant increasein shear strength wasobserved for the S-Bond solder and C/Cbrazed to C/C. Itshould be noted thatall of the brazesspread and bondedthe entire area wherethe Ti and C/C plates

overlapped except for the CuSil ABA foilspecimens, which accounted for the poorshear strengths for that system.

Formation of Cracks

Cracks were observed within the outerply of the C/C composite-braze reactionlayer (Fig. 7), which were due to the ther-mal stresses caused on cool down due tothe difference in thermal expansion coef-

ficients between the Ti (8.6 × 10–6/°C) andthe C/C (–1 × 10–6/°C). The crack in Fig. 7extended several millimeters along thelength of the specimen. These preexistingcracks were the initial flaw sources forjoint failure of the Ti to C/C brazed joints.No cracks were observed in the S-Bondsolder joints or the C/C to C/C brazedjoints.

Since thermal-induced residualstresses appear to control joint perform-ance for these systems, a first-order ap-proximation of the thermal strains withinthe joint was estimated using the simplerelationship (assuming elastic behavior)ΔαΔT, where Δα is the difference in thethermal expansion coefficients betweenthe two plates of the joint and ΔT is theprocessing temperature of the braze orsolder (Table 1) subtracted by the test tem-perature (25°C). The estimated thermal-induced residual strain on shear strengthis plotted in Fig. 8. For the braze joints, thethermal-induced strains would be veryhigh and it is not surprising that they areaccommodated by cracking within theouter ply of the C/C. However, the forma-tion of the cracks significantly reduces theshear strengths of those joints. Apparentlythe buildup of strains on the order of 0.4%could be accommodated elastically ormore likely via plastic deformation withinthe S-Bond solder itself since cracking inthe C/C was not observed. Not enoughtests were performed to determine if thedifference between the S-Bond solderjoint of Ti to C/C was statistically signifi-cantly lower than the CuSil ABA paste

Sh

ear

Str

eng

th, M

Pa

BS

T S

hea

r S

tren

gth

, MP

a

TiC

uSil

AB

A F

oil

TiC

uSil

AB

A P

aste

CuS

il A

BA

Foi

l

CuS

il A

BA

Pas

te

S B

ond

Sol

der

C/C

to C

/C w

/CuS

ilA

BA

Pas

te

Fig. 6 — Shear strength of different braze and solder joints. Fig. 7 — Optical micrograph of a typical crack within the outer plyof the C/C for a joint region of a Ti plate joined to C/C compositewith CuSil ABA paste.

Fig. 8 — Effect of the estimated thermal-induced residual elasticstrain in the joint on shear strength. The CuSil ABA foil joint datawere not used because this braze material did not spread and bondthe lap section effectively.

Morscher Layout:Layout 1 2/7/07 8:29 AM Page 65

BRAZING & SOLDERING TODAY

MARCH 200766

joint of C/C to C/C. However, it is evidentthat the CuSil ABA paste provides a rea-sonable joint strength for C/C to C/C.

Conclusions

Two-dimensional woven C/C compos-ites with high-conductivity carbon fiberswere joined to Ti tubes and plates withboth braze and solder approaches. CuSilABA paste and TiCuSil braze materialsperformed the best of the braze conceptsevaluated due to their superior spreadingand resulting bonded joint area. However,due to the buildup of thermal stresses dur-ing cooling of the joined structure fromthe braze processing temperature, crackswithin the outer layer of the C/C compos-ites formed limiting the joint strength. Thelower-processing temperature solderproved to be a much stronger joint whentested in shear since the residual stresses

were sufficiently low to prohibit process-ing-induced cracks in the C/C composite.Relatively high joint shear strengths werealso measured for C/C plates joined toC/C plates with a CuSil ABA braze.♦

Acknowledgments

This work was supported by NASA’sPromethius Program. We would like tothank Dr. Michael Halbig for the review ofthe manuscript.

References

1. Schwartz, M. M. 1995. Joining of Com-posite Matrix Materials, ASM International, Ma-terials Park, Ohio.

2. Trester, P. W., Valentine, P. G., Johnson,W. R., Chin, E., Reis, E. E., and Colleraine, A.

P. 1995. Proceedings of the Seventh InternationalConference on Fusion Reactor Materials, Ob-ninsk, Russia.

3. Goodman, D. L., Birx, D. L., and Dave,V. R. 1995. Nuclear Instruments and Methods inPhysics Research B 99, pp. 775–779.

4. Mason, L. S. 2004. Power conversion con-cept for the Jupiter icy moons orbiter. Journalof Power and Propulsion, 20: 902–910.

5. Morscher, G. N., Singh, M., Shpargel, T.,and Asthana, R. A Simple test to determine theeffectiveness of different braze compositionsfor joining Ti-tubes to C/C composite plates.Mater. Sci. Eng. A, in print.

6. Singh, M., Shpargel, T. P., Morscher, G.N., and Asthana, R. 2005. Active metal brazingand characterization of brazed joints in tita-nium to carbon-carbon composites. Mat. Sci.Eng. A, 412: 123–128.

7. S-Bond 400 Alloy Property Bulletin, No.14.01.07, issue date 06/2004. www.s-bond.com

8. Morscher, G. N. unpublished data.

Circle No. 61 on Reader Info-Card

Morscher Layout:Layout 1 2/7/07 8:30 AM Page 66