WELDING RESEARCH

NOVEMBER 2006-s252

ABSTRACT. An explosive welding proce-dure for joining 9.5-mm-thick niobiumplate to 203-mm-thick 6061-T651 Al platehas been developed in order to maximizethe weld tensile and impact strengths andthe amount of welded material across thesurface of the plate. This procedure im-proves upon previous efforts, in which the9.5-mm-thick niobium plate is welded di-rectly to 6061-T4 Al plate. In this im-proved procedure, thin Nb and Al inter-layers are explosively clad between thethicker niobium and aluminum plates.Welds produced using these optimized pa-rameters display a tensile strength of ap-proximately 255 MPa and an impactstrength per unit area of approximately0.148 J/mm2. Specialized mechanical test-ing geometries and procedures are re-quired to measure these weld propertiesbecause of the unique weld geometry. Inorder to ensure that differences in thethermal expansion coefficients of alu-minum and niobium do not adversely af-fect the weld strength, the effects of ther-mal cycling at temperatures between –22°and 45°C on the mechanical properties ofthese welds have also been investigated bytesting samples in both the as-receivedand thermal cycled conditions. Based onthe results obtained from this series ofme-chanical tests, thermal cycling is shown tohave no adverse effect on the resultingtensile and impact strengths of the weldsproduced using the optimized weldingparameters.

Introduction

A preliminary investigation of the fea-sibility of explosively welding niobium toaluminum has been previously published

(Ref. 1). In that study, 9.5-mm-thick com-mercially pure Nb is welded to a 178-mm-thick 6061 Al plate in the T4 condition.These initial welds are produced in a con-ventional manner with the detonatorplaced at the edge of the plate. The alu-minum plate is required to be in the soft-ened (T4) heat treated condition to facili-tate the explosive welding process. Inorder to transform the 6061-T4 Al to thedesired T6 heat treated condition, a post-welding heat treatment is required. An ul-trasonic evaluation of these welded platesshows that areas of disbonding are presenton the side of the plate directly oppositefrom the detonator.

Radially symmetric welding across theweld interface can be achieved by movingthe detonator to the center of the plate. Adivot, surrounded by a small area of non-welded material, is produced at the deto-nator location in the center of the plate.Additional areas of disbonding, as con-firmed by ultrasonic scans, tend to be iso-lated near the edges of the plate, far fromthe locations where well-welded materialis required. A ring of well-welded mater-ial extends radially outwardmore than 430mm from the nonwelded region in the cen-ter of the plate.

In order to eliminate the postweld heattreatment and to directly weld the nio-bium to the 6061 Al plate in the hardened

(T6) condition, explosively welded nio-bium and aluminum interlayers are intro-duced between the aluminum and nio-biumplates. Smaller explosive charges canbe used to weld these interlayers to thethick aluminum plate, thus making itmuch easier to work with the aluminumplate in the hardened condition. The ef-fects of the introduction of these interlay-ers on the resulting mechanical propertiesof the explosive weld are also investigatedin this study.

The tensile properties of the initialwelds produced with the detonator lo-cated at the edge of the plate were inves-tigated using conventional tensile testswith a dog-bone geometry (Ref. 1). Inorder for the samples to have a sufficientlength to place the Al-Nb weld interfacein the center of the gauge length of thetensile bar, niobium extension bars wereelectron beamwelded to the niobium-cladplate. During testing, these samples failedin the explosively clad niobium at loca-tions near the weld interface, but not atthe weld interface, showing that thestrength of the weld exceeds that of theniobium. Additional mechanical tests,using sample geometries optimized fortesting the tensile, shear, and impactstrengths of the welds, have also been per-formed on samples taken from both theedge-detonated and center-detonatedwelds. These procedures are discussedfurther in a following section.

Table 1 provides a summary of the me-chanical properties of the edge-detonated(Ref. 1) and center-detonated explosivewelds. For both sets of welds, failure oc-curs at the Al-Nb weld interface, and thenotched (268 ± 14.5 MPa and 287 ± 35.4MPa, respectively) and unnotched (351 ±17 and 268 ± 9 MPa, respectively) tensilestrengths compare favorably with the ulti-mate tensile strengths of both niobium(310 MPa) and 6061-T6 Al (312 MPa)(Ref. 1). The measured shear strengths

Development of an ExplosiveWelding Process for ProducingHigh-Strength Welds between

Niobium and 6061-T651 Aluminum

Thin interlayers of Nb and Al were used to improvethe joining of thicker plates

BY T. A. PALMER, J. W. ELMER, D. BRASHER, D. BUTLER, and R. RIDDLE

KEYWORDS

AluminumInitiation PointExplosive WeldingMechanical PropertiesNiobiumSolid StateDissimilar Metal JoiningImpact Strength

T. A. PALMER, J. W. ELMER, and R. RIDDLEare with Lawrence Livermore NationalLaboratory, Livermore, Calif. D. BRASHER andD. BUTLER are with High Energy Metals, Inc.,Sequim, Wash.

WELDING RESEARCH

-s253WELDING JOURNAL

(224 ± 7 MPa and 180 ± 5, respectively)are approximately 72% and 58%, respec-tively, of the measured tensile strength ofthe 6061-T6 Al. These measured shearstrengths are consistent with the vonMises ductile failure criteria, which pre-dict that the shear strength should be ap-proximately 58% of the ultimate tensilestrength (Ref. 2). It is also in line with thereported shear strength of 6061-T6 Alplate (207 MPa).

On the other hand, there are signifi-cant concerns about the impact strength ofthe resulting welds. As shown in Table 1,the measured impact strengths of theedge- and center-detonated weld samplesare low (≥ 0.02 J/mm2). All of the samplesfailed along the Al-Nb weld interface, andthe majority of the fracture surface dis-plays a rather flat appearance, indicativeof a brittle fracture mode. Figure 1A andB shows micrographs taken of typicalareas on a fracture surface obtained froman edge-detonated impact sample. Whilethemajority of the fracture surface is char-acteristic of a brittle failure mode, thereare small areas interspersed on the frac-ture surface displaying features similar tothose more commonly associated with amore desirable ductile failure mode.

These low impact strengths and thebrittle failure mode can be correlated with

characteristics of the weld interface mor-phology. An examination of the weld crosssection, which is shown in Figs. 4 and 5 inRef. 1, shows regions of intermixing of theAl and Nb (Ref. 1). These regions consistprimarily of submicron-sized Nb-rich par-ticulates mixed with larger fragments ofNb in an Al-rich matrix. This behavior isconsistent with the formation of Al-Nb in-termetallic phases at the weld interfacecaused by the melting of the Al during theexplosive welding process. It appears thatthese regions of Al and Nb intermixing re-sult in a brittle weld being formed over alarge area of the interface between the Aland Nb plates.

Changes to the welding process arethus required in order to produce weldswith the desired tensile and impactstrengths. In the current study, a three-step explosive welding process is devel-oped to join Grade 1 (Reactor) niobiumdirectly to 6061-T651 Al plate, thus re-moving the follow-on heat treatment.These welds must meet minimum criteriafor tensile, shear, and impact strengths,which are measured using specialized me-chanical testing geometries and proce-dures. In addition to meeting these me-chanical property requirements, theamount of well-weldedmaterial across thesurface of the thick Al plate must be max-

imized. In order to meet these require-ments, a procedure involving the explosivewelding of thin sheets of Al and Nb be-tween the thicker Al and Nb plates, hasbeen developed. The introduction of thesethin interlayers not only produces weldswith the required tensile, shear, and im-pact strengths, but also allows the thick Alplate to be welded in the high-strength (T-651) condition, thus removing the re-quirement for a postweld heat treatmentof the clad material.

Experimental

Explosive Welding

The explosive welding (EXW) processis a solid-state joining process used to joina wide variety of materials that cannot bejoined using traditional fusion weldingprocesses (Ref. 3). A schematic diagramshowing the basic components of thisprocess (the explosive charge, a basemetal, and a clad metal) is shown in Fig. 2.Prior to welding, the clad metal is offset agiven distance above the base metal, and apredetermined amount of explosive isplaced on top of the clad metal. A con-trolled explosive detonation is used to ac-celerate this cladmetal into the basemetalat a sufficient velocity that the collision

Fig. 1 — SEM micrographs of the fracture surface on one of the edge-detonated Izod samples.

Table 1 — Summary of Mechanical Properties of Preliminary Single-Welded Al-Nb Clad Plates

Edge-Detonated Plates Center-Detonated PlatesNumber of Tests Average Standard Deviation Number of Tests Average Standard Deviation

Tensile Strength (MPa)Unnotched 6 350.6 17.0 5 266.8 10.07Notched 3 265.6 14.5 3 287.3 35.4

Shear Strength (MPa) 4 224.3 6.76 3 180.4 5.38Impact Strength (J) 2/7 0.8/1.5(a) 0.1/0.1(a) 12 1(b) 0.2(b)

Impact Energy/Unit Area 2/7 0.020/0.023 0.003/0.002 12 0.015 0.003(J/mm2)

(a) Impact strength sample geometries are varied from a cross section of 3.2 x 10.2 mm to one of 6.4 x 10.2 mm.(b) Impact strength sample geometry has a cross section of 12.7 x 10.2 mm.

WELDING RESEARCH

NOVEMBER 2006-s254

causes the two metals to fuse together.The force of the explosion sets up an an-gular collision, which produces a plasmaphase, which is ejected ahead of the lead-ing edge of the weld interface. Since theplasma jet is located in front of the colli-sion point, it is inferred that melting is notnecessarily part of the welding occurringbehind it. This plasma jet removes impu-rities from the surfaces of both the baseand clad plates, thus leaving clean metalsurfaces for joining. The pressures at thecollision point (690 to 4137 MPa) areenough to cause the metals to behave likeviscous fluids. This behavior is responsiblefor the wave-like weld pattern produced atthe interface of the two materials andshown in the figure (Refs. 3, 4).

The joining of niobium to aluminum ishindered by the potential formation ofseveral intermetallic phases, as shown intheAl-Nb phase diagram in Fig. 3 (Ref. 5).The presence of these intermetallicphases can lead to significant decreases inthe weld strength, thus making the use offusion welding techniques to join thesetwo materials impractical. Therefore, a

solid-state joining process, where nomelt-ing or significant intermixing of theAl andNb occurs, is required. Explosive weldingis preferable to other solid-state joiningprocesses in this case because of the thick-ness of the niobiumplate (9.5mm) and thelarge surface area, approximately 508 ×508 mm, to be welded.

In the development of this explosiveweld, a number of different materials areused. These materials include a 203-mm-thick 6061-T651 Al plate, a thin (0.8/1.0mm) 6061-O Al sheet, a thin (0.33-mm)Nb sheet, and a thick (9.5-mm) Nb plate.The 6061-T651 Al base plate meets thechemistry and mechanical property re-quirements in ASTM B209M-02a (Ref. 6)and AMS-QQ-A-250/11 (Ref. 7), whilethe Nb sheet and plate materials meet therequirements for R04200-Type 1 ReactorGrade material, which appears in ASTMB393-03 (Ref. 8).

Because of the nature of the weld de-velopment process, which evolved over anextended period of time, a number of dif-ferent heats have been used for each of thematerials defined in Table 2. Changes in

material heats are made between Welds#1 and #2, which encompass the initialdevelopment work on single welds. Thefirstmultiple welds with a 0.8-mm-thickAlinterlayer inWelds #3a through#3c use asingle heat of each material. Differentheats of each material are used when the0.8-mm Al interlayer is replaced by a 1.0-mm-thick Al interlayer in Welds #4a and#4b. Compositional variations betweenthe different heats are not considered sig-nificant since the weld parameters play amuch more dominant role than the mate-rial chemistry in producing an explosiveweld with suitable properties.

The explosive welding operations de-scribed here have been performed byHighEnergy Metals, Inc. All welds are madeusing an ammonium nitrate-fuel oil(ANFO)-based explosive mixture. In theexplosive welding process, there are sev-eral essential parameters that must betightly controlled andmonitored as part ofthe welding process. These parameters in-clude the surface finish and cleanliness ofthe welded face of each material, theplacement of the detonator, and the ex-

Fig. 2 — Schematic figure showing the basic features of the explosion weldingprocess and the properties of the resulting weld.

Fig. 3 — Plot showing the Al-Nb phase diagram (Ref. 5).

Table 2 — Summary of Explosion Welding Parameters Used during the Development of the Al-Nb Weld

Material Thickness (mm) Explosive Energy (MJ)Welding DetonatorParameter Location Al Plate Al Interlayer Nb Interlayer Nb Plate Al-Al Bond Al-Nb Bond Nb-Nb BondIdentification

Single Welds1(a) Edge 203 — — 9.5 — 3.1 —2(a) Center 203 — — 9.5 — 3.1 —

Multiple Welds0.8-mm Al Interlayer

3a Center 203 0.8 0.33 9.5 0.9 1.0 3.083b Center 203 0.8 0.33 9.5 1.01 0.95 3.763c Center 203 0.8 0.33 9.5 0.82 1.0 3.08

1.0-mm Al Interlayer4a Center 203 1.0 0.33 9.5 1.37 1.54 3.084b Center 203 1.0 0.33 9.5 1.37 1.54 3.08

(a) A postweld heat treatment is required to convert the 6061 Al plate from the T4 heat treatment condition to the T6 heat treatment condition.

WELDING RESEARCH

-s255WELDING JOURNAL

plosive energy. The explosive energy is de-fined as the kinetic energy of the impact-ing plate for a given explosive detonationvelocity, which is controlled by mixing theANFO-based explosive with fine silica,and stand-off distance between the twoplates. The explosive velocity of a smallportion of the explosive mixture is mea-sured before thewelding operation by tim-ing the speed of the explosive wave frontover a distance of approximately 305 mm.

The surfaces to be welded are groundto a surface finish of 3.2 µm root meansquare (RMS) or better and cleaned inorder to remove any oxide, dirt, or otherforeign objects. Each of the plates, whichare all 508 × 508 mm in size, must alsomeet a flatness tolerance, of at least 0.2mm over its length, in order to maintain auniform stand-off distance between theclad and base metals. The development ofa set of suitable welding parameters hasinvolved a number of iterations. Table 2

provides a summary ofthe various welding pa-rameters for both thepreliminary singlewelds, as well as themul-tiple welds. The locationof the detonator and theexplosive energy are var-ied in an attempt to pro-duce a weld with accept-able properties. Thestand-off distances usedin each of the welds isapproximately equal tothe thickness of thesheet or plate beingwelded. Smaller-capac-ity detonators are usedto weld the thin interlay-ers, which require lowerexplosive energies (ap-

proximately 1 MJ) during the welding op-eration. A higher-capacity detonator isused to weld the 9.5-mm-thick Nb plate,which requires a higher explosive energy(> 3 MJ) during welding.

Nondestructive and MetallographicEvaluation of Weld Integrity

The quality of each explosive weld isexamined in the as-welded condition usingan ultrasonic evaluation technique to in-spect the integrity of the Al-Nb weld re-gion. The ultrasonic evaluation of each ex-plosive weld is performed by immersingthe welded plates into a water tank andusing a pulse echo ultrasonic examinationtechnique (Refs. 9–11) to detect indica-tions of voids, defects, and areas of dis-bonding in the explosive weld. A 5-MHz,12.7-mm-diameter, 32-mm sphericalfocus transducer, which has a theoreticalspot size of 0.76 mm, is used. During each

scan, the transducer is automaticallyrastered over the surface of the plate witha maximum step size of 0.5 mm in order toprovide a complete picture of the weld re-gion across the entire plate. The results ofeach scan are used to judge the extent ofwell-welded material on each plate and asa means of determining whether any de-fects are present in the region where me-chanical testing samples are removed.

An ultrasonic calibration standard,fabricated from a piece of explosively cladAl-Nb weld, is used to calibrate the equip-ment prior to each scan and to ensure thatthe system is able to detect defects of agiven size. Known reflectors, in the formof flat bottomed holes and rectangles witha minimum area of 0.79 mm2, are placedin the standard at the Al-Nb weld inter-face and are used to establish the primaryreference responses of the equipment.The sizes of the individual defects in thecalibration standard are chosen to ensurethat the system is able to detect defectssmaller than the maximum allowable de-fect size in the final part.

Metallographic examination of thecross section of the explosively weldedmaterial is typically performed in the as-polished condition. In this condition, thedifferent weld regions are visible. Moredetailed information about the weld re-gions can be obtained by etching the sam-ple in a chemical bath containing 20 mLof glycerol, 10 mL of hydrofluoric acid,and 10 mL of nitric acid. Because the alu-minum etches at a more rapid rate thanthe niobium, care must be taken to ensurethat the aluminum is not overetched.With etching, the microstructure of theweld region is better revealed, making itmuch easier to identify areas of intermix-ing between the dissimilar metals. Thisexamination of the weld cross section al-

Fig. 4 — Schematic diagrams. A — The tensile testing setup used tomeasure the tensile strength of the Al-Nb explosion weld; B — the ramtensile specimen.

Fig. 5 — Typical stress-strain curve obtained during the tensile testing of anexplosively welded sample in the unnotched and notched configurations.

A

B

WELDING RESEARCH

NOVEMBER 2006-s256

lows the weld wave pattern to be evalu-ated and determine if any changes to thewelding parameters are required in orderto produce the most desirable weld pat-tern morphology.

Mechanical Testing

The mechanical properties of the ex-plosive welds produced during each weld-ing iteration are measured to ensure thatthe welds produced during the explosivewelding process have the maximum possi-ble strength. Mechanical test samples areremoved from specific locations in eachwelded plate. The samples taken fromWelds #1 and #3a are removed from thecenter region of thewelded plate. Samplestaken fromWeld #2 are removed from lo-cations at the corners of the plate. Those

samples used totest the mechanicalproperties ofWelds#3b, #3c, and#4(a and b) arealso taken from thecenter region of thewelded plate out-side the center de-fect area. In theseplates, though, thesamples are re-moved from betterdefined radial loca-tions, 152 mm fromthe center of theplate. The loca-tions where the-samples are re-moved in eachplate have been ul-

trasonically scanned to ensure that onlyproperly welded material is used in themechanical testing samples. The resultsreported for Welds #3a–c and #4a in-clude results from only a single weldedplate, while those reported for Weld #4bare an average of results taken from threedifferent Al-Nb clad plates welded usingidentical procedures.

Modifications to traditional mechani-cal test geometries and procedures are re-quired to test the strength of the weld.Typical tensile and Charpy impact speci-men geometries would require an exten-sion be welded onto the rather thin nio-bium clad layer, as done in the previousstudy (Ref. 1). Because the addition ofthese welded extensions introduces an ad-ditional level of complexity to the testing,a set of modified mechanical testing

geometries and procedures, based on ex-isting standard techniques, are employedto test the tensile, shear, and impactstrength of the Al-Nb weld (Refs. 12–14).

The tensile strength of the weld is mea-sured using the tensile testing setup andcorresponding ram tensile specimenschematically shown in Fig. 4 A and B(Ref. 12). This test specimen, shown inFig. 4B, can be divided into two compo-nents. The first component encompassesthe 6061-T651 Al side of the weld and hasan outer diameter of 33.3mm and a heightof 25.4 mm. An internal hole with a diam-eter of 19.1 mm is machined through theheight of the aluminum and into theniobium-clad layer. The second compo-nent of the ram tensile specimen consistsof the niobium- clad layer, which is ma-chined to a diameter of 25.4 mm. In thedesign of these specimens, the niobiumportion of the specimen in contact withthe plunger is made thick enough (8.89mm) so that the niobium does not yieldduring testing. Samples with and withouta notch machined at the Al-Nb weld in-terface are tested. In the notched samples,a 60-deg ± 2 deg notch with a depth of0.635 ± 0.127 mm is used. During testing,a plunger is inserted into the center hole,and a compressive force is exerted whilethe Al portion of the sample is held inplace. As a result, the strength of the weldis measured in tension.

All of these ram tensile tests are per-formed on an Instron electromechanicaltest machine, equipped with an 89-kNload cell. The load (N)-displacement(mm) behavior of the tensile sample ismonitored at a constant crosshead speedof 0.5mm/min during each test. After test-

Fig. 6 — Schematic diagrams. A — The shear testing setup used to measure the shear strength of the Al-Nb explosion weld; B — the shear strength specimen.

Fig. 7 — Typical stress-strain curve obtained during the shear testing of anexplosion welded sample.

A B

WELDING RESEARCH

-s257WELDING JOURNAL

ing, the load is then converted to stress bydividing the measured load by the initialannular cross-sectional area at the weldinterface. Figure 5 shows typical stress-displacement curves for tensile tests per-formed on explosively welded samples inboth the unnotched and notched configu-rations. The tensile tests continue until thesample fails, and the tensile strength ofeach weld is equivalent to the maximumstress measured during each test.

The shear strength of the weld is mea-sured using the testing setup and shearstrength sample shown in Fig. 6A and B(Ref. 12). In order to test the shearstrength of the weld, a rectangular sampleis fabricated from the explosively clad Al-Nb weld material by removing a majorityof the niobium-clad layer, leaving only asmall nub of niobium measuring 6.35 mmwide. The aluminum portion of the sam-ple is 63.5 mm in length by 12.7 mm inwidth and 25.4 mm high. When the sam-ple is placed in the fixture, it is supportedby this niobium nub, which is located19.05 mm from the end of the sample andextends across the 12.7 mm width of thesample.

With the niobiumnub restrained by thefixture, a compressive force is applied tothe top of the sample placing the Al-Nbexplosive weld region in shear. The test

continues, measuring load vs. sample dis-placement, until the sample fails. Theload-displacement behavior of the shearsample is monitored at a crosshead speedof 0.5 mm/min on the Model 4400R1225Instron electromechanical test machine.After testing, the load is converted tostress by dividing themeasured load by thecross-sectional area at the weld interface,as defined by the weld interfacial area be-tween the aluminum and the niobium nubon the shear specimen. Figure 7 shows atypical stress-strain curve for a shearstrength test performed on an explosivelywelded sample. The resulting shearstrength of the weld corresponds to themaximum shear stress measured duringthis test.

In order to measure the impactstrength of the Al-Nb explosive weld, amodified Izod impact testing procedureand sample, which are shown in Fig. 8Aand B have been developed (Refs. 13, 14).The holding fixture of the Izod tester hasbeen modified to grasp the undersizedniobium-clad layer, which is 8.89 mmthick. The 6061-T651 aluminumportion ofthe sample is 31.75mm in length, which al-lows the hammer to strike the sample inthe appropriate location, as defined by thegoverning standards (Refs. 13, 14). A 45-deg ± 5 deg angled notch with a depth of

2.54 mm ± 0.05 mm is machined into theIzod sample at the Al-Nb weld interfaceusing a slitting saw to minimize potentialmachining damage.

During testing, the notch is positionedso that it points in the direction of hammerimpact, thus placing the weld in tensionduring testing and subsequent failure ofthe sample. Different sample sizes, rang-ing from 3.2 × 12.7 mm to 12.7 × 12.7 mm,are tested. All of the Izod impact tests areperformed on a Tinus Olsen Charpy-Izodimpact test machine, Model 66, with amaximum capacity of 22.6 J. The resultingfracture surfaces of selected samples havebeen examined using an FEIModel XL30S FEG scanning electron microscope inorder to identify the prevailing failuremodes and mechanisms.

Thermal Cycling of Al-NbExplosion Welds

A comparison between several perti-nent mechanical properties of niobiumand 6061 Al is shown in Table 3. (Refs. 15,16). Since niobium and aluminum havesignificantly different thermal expansionproperties, as shown by the differences intheir respective coefficients of thermal ex-pansion (7.20 × 10–6 K–1 for niobium and2.36 x 10–5 K–1 for aluminum), thermal

Fig. 8 — Schematic diagrams. A — The Izod impact testing setup used to measure the impact strength of the Al-Nb explosion weld; B — the modified Izodspecimen.

Table 3 — Summary of the Base Metal Room-Temperature Mechanical Properties for Niobium and Aluminum Demonstrating Their Differences inMechanical Properties (Refs. 15–16)

Youngs’ Modulus Poisson’s Ratio Coefficient of Yield Strength Ultimate Tensile(GPa) Thermal Expansion (K–1) (MPa) Strength (MPa)

Niobium 104.9 0.397 7.20 x 10–6 207 5856061-T6 Al 70.6 0.345 2.36 x 10–5 280 3106061-O Al 70.6 0.345 2.36 x 10–5 48.3 117

A B

WELDING RESEARCH

NOVEMBER 2006-s258

stresses can develop across the weld inter-face with changes in temperature. As a re-sult, residual stresses of a magnitude suf-ficient enough to affect the long-termweld integrity can build between the Aland Nb. In order to examine the effects ofchanges in temperature on the explosionweld properties, selected ram tensile andIzod specimens are subjected to a series often thermal cycles.

In each thermal cycle, the samples arefirst placed in a bath heated to a tempera-ture of 49°C. The temperature of the sam-ples is monitored by a thermocouple at-tached to one of the samples. After thesamples are immersed in the bath, thesamples are allowed five minutes to reachthe desired temperature. At the comple-tion of this time period, the samples arethen held in the bath for a period of 15minutes. After this 15-minute hold is com-plete, the samples are removed from thebath and immediately placed in a second

bath cooled to a temperature of –22°C.The sample temperature is again allowedto equilibrate for five minutes, after whichthe samples are held in the bath for 15minutes. This process is repeated until atotal of ten hot/cold cycles are completedon each sample. After the completion ofthe thermal cycling of these samples, thetensile and impact strengths of the weldsare then tested in the same manner asthose samples in the as-receivedcondition.

Results and Discussion

Development of Multiple-Layer Welds

An explosive charge of 3.1 MJ is origi-nally used to clad the 9.5-mm-thick Nbplate directly to the 203-mm-thick 6061-T4Al plate in the initial weld developmentstudy discussed previously (Ref. 1). Withsuch a large explosive charge, tempera-

tures at the weld interface can becomehigh enough to melt the aluminum, whichthen reacts with the niobium to form brit-tle intermetallic phases. Visual confirma-tion for intermixing of the aluminum andniobium at the weld interface has beenpreviously discussed (Ref. 1). Since theimpact strength of the weld producedusing these procedures is not acceptable,improvements to the process are requiredin order to significantly increase the im-pact strength of the welded material whilemaintaining the tensile and shearstrengths at or near their current levels. Inorder to achieve this goal, a means foravoiding melting and intermetallic forma-tion during welding must be developed.

Figure 9 provides an overview of theensuing weld development process under-taken in an attempt to produce explosionwelds with the desiredmechanical proper-ties. In this figure, the evolution of thewelding process from the single weld pro-cedure described previously (Ref. 1) to theensuing multiple weld procedures is sum-marized. These multiple welds are pro-duced using a three-step welding processbased on the addition of thin sheets or in-terlayers of aluminum and niobium be-tween the 9.5-mm-thick niobium and the203-mm-thick aluminum plates. Thewelding of each thin interlayer requires asmaller explosive charge, thus decreasingthe likelihood for melting and intermetal-lic formation and allowing the explosionwelds to be made directly on the 6061-T651 Al plate, thus removing the finalheat treating step. A summary of the ex-plosion energies required to join these in-terlayers is provided in Table 2.

Welds #3a–c utilize a 0.8-mm-thick6061-O Al sheet, which is welded directlyto the 6061-T651 Al plate. Since the Alsheet is in an annealed condition, it ismuch more ductile than the Al plate, al-lowing it to be welded with a rather low ex-plosive energy (0.8 to 1.0 MJ). After the

Fig. 9 — Schematic diagram summarizing the evolution of the explosion weld development. The circleson the top view of the explosion weld schematics indicate the locations of the detonators. In the singlebonds shown in part A, the dotted lines indicate the circular shape of the Al plate during welding withthe detonator placed at the edge of the niobium plate.

Fig. 10 —Micrographs of the weld cross section taken at an orientation perpendicular to the explosive wave. A— The overall weld, including the Al-Al, Al-Nb,and Nb-Nb weld interfaces. B — a closer view of the Al-Nb and Nb-Nb weld interfaces.

A B

WELDING RESEARCH

-s259WELDING JOURNAL

welding of this thin Al interlayer, a 0.25-mm-thick niobium sheet is welded directlyon top of it. Since a low explosive energy(0.95 to 1.0 MJ) is used to make this weld,the possibility of intermixing between theAl and Nb is significantly decreased. As afinal step, the 9.5-mm-thick niobium plateis welded onto this thin niobium interlayerusing a much larger explosive energy (3.08to 3.76 MJ). By joining the thick niobiumplate (9.5 mm) to the thin niobium sheet,the possibility of creating conditionsunder which any melting can occur isnearly eliminated because the niobiumhas a high melting point (2468°C).

Figure 10A and B provides views of atypical weld cross section taken at an ori-entation perpendicular to the explosivewave front in Weld #3a. In Fig. 10A, theentire weld region, including the Al-Al,Al-Nb, and Nb-Nb interfaces, is shown. Inthis figure, the Al-Nb weld interface ap-pears as a dark line. This feature is a rem-nant of the metallographic preparation ofthe sample and results from the differencein the height of the niobium and 6061-T651 Al after polishing, caused by the dif-ference in hardness of the two materials.Each weld interface displays a wavy ap-pearance, which is typical of explosivewelds. Of the three weld interfaces shownin this figure, the Al-Al interface displaysthemost prevalent wavyweld interface ap-pearance. The Al-Nb weld interface,

which is highlighted inFig. 10B, is also muchwavier in appearancethan that observedwith the single-layerwelds. There is also novisual evidence ofmelting, intermetallicformation, or a mixedzone along the Al-Nbinterface, which is animprovement over theweld interface ob-served in the previouswork (Ref. 1).

Results from thetensile and shearstrength testing forWelds #3a–c are sum-marized in Table 4.Taking an average ofthe results from thesethree welds, average tensile strengths of243 ± 20 MPa and 231 ± 25 MPa are ob-served for the unnotched and notchedconditions, respectively. The tensilestrengths of the multiple welds in both theunnotched and notched conditions are ap-proximately 9% and 20%, respectively,lower than those observed in the single-welded plates. In addition, failure in theram tensile specimens in the multiplewelds (Welds #3a–c) occurs within thethin Al interlayer, as compared with the

Al-Nb weld interface in the single welds(Welds #1 and #2).

Welds #3a–c also display lower shearstrength values (112±22MPa) than thoseobserved in the edge- and center-deto-nated plates (224 ± 7 MPa and 180 ± 5MPa, respectively). Results of the shearstrength testing of these welds are alsosummarized in Table 4. This decrease inmeasured shear strength is due, in part, tothe addition of the interlayers between thealuminum and niobium plates. In the sin-

Fig. 11 — Summary of measured impact strengths for the multiple weldplate made using parameters #3a, showing the effects of differences in fail-ure location on the measured impact strength.

Table 5 — Summary of Impact Strengths Measured in Each Al-Nb Explosion Weld

Impact Energy (J) Impact Energy/Unit Area (J/mm2)Welding Number Average Standard Average StandardParameter of Tests Deviation DeviationIdentification

0.8-mm Al Interlayer3a(a) 11 6.8/19.2 2.6/1.7 0.052/0.148 0.020/0.0133b 10 16.8 3.6 0.130 0.0283c(a) 10 9.2/13.3 0.8/1.0 0.071/0.103 0.006/0.008

1.0-mm Al Interlayer4a 3 21.2 2.6 0.164 0.0204b 19 19.2 3.5 0.148 0.027

(a) Two distinct failure modes (Al-Al interface and Al interlayer) are observed.

Table 4 — Summary of Tensile and Shear Strengths Measured in Each Al-Nb Explosion Weld

Tensile Strength (MPa) Shear Strength (MPa)Welding Number Unnotched NotchedParameter of Tests Average Standard Average Standard Number Average StandardIdentification Deviation Deviation of Tests Deviation

0.8-mm Al Interlayer3a 3/3 264 9 246 26 3 88 393b 3/1 225 8 202 — 3 128 53c 3/3 240 52 244 52 3 12 1

1.0-mm Al Interlayer4a 3/— 251 21 — — 3 127 24b 6/6 255 13 284 25 — — —

WELDING RESEARCH

NOVEMBER 2006-s260

gle-weld plates, a distinct failure surface atthe Al-Nb interface is observed in the shearstrength samples after testing. Themultiplewelds display no such clear failure mode.Rather, during testing, the thinAl interlayeryields, causing the explosively clad niobiumlayer on the shear strength sample to onlybe displaced and not be removed from theAl plate during testing.

Even though the tensile and shearstrengths are lower than those observed inWelds #1 and #2, this decrease in weld

strength is not a concern because the ten-sile and shear strengths are still accept-able. Most importantly, though, thesemultiple welds display significant en-hancements in the measured impactstrength when compared with the edge-and center-detonated plates. As shown inTable 5, the impact strength of the multi-ple weld plates increases to a level of 19 J(0.147 J/mm2), which is much higher thanthat observed in the single weld plates(0.020 J/mm2). Based on these results, it is

clear that the use of a 0.8-mm-thick Al in-terlayer in the explosion welding processhas dramatically increased the impactstrength of the Al-Nb weld, while main-taining desirable levels of tensile andshear strength. However, there is a ratherwide range of impact strength values ob-served in the welds produced using the0.8-mm-thick Al interlayer. In Weld #3a,in particular, a bimodal distribution in im-pact strengths, as shown in Fig. 11, is ob-served. The average values of the two lev-

Fig. 12 — Micrographs. A — Showing the cross section; B and C — fracturesurface of an Izod impact sample, which fails at the Al-Al weld interface.

Fig. 13 — Micrographs. A — Showing the cross section; B and C — fracturesurface of an Izod impact sample, which fails in the Al interlayer.

A

B

C

A

B

C

WELDING RESEARCH

-s261WELDING JOURNAL

els vary by nearly a factor of three (0.052and 0.148 J/mm2).

The bimodal distribution of impactstrengths, which appears in Fig. 11, is at-tributed to the presence of two distinctfailure modes. In the first failure mode,which results in low weld impact strength,the weld failure occurs very close to theAl-Al weld interface. Figure 12A–C dis-plays the cross section and fracture sur-face of an Izod impact sample with low im-pact strength taken fromWeld#3a. In thisfigure, the Al-Nb weld interface appearsas a single dark line, as a result of the dif-ferences in the heights of the aluminumand niobium in the as-polished specimen,resulting from the metallographic prepa-ration. It is apparent in the cross sectionshown in Fig. 12A that the 0.8-mm Al in-terlayer remains nearly intact, and the re-sulting fracture surface is generally flat.Figure 12B and C shows images of thesame fracture surface at different magni-fications. In this figure, the fracture sur-face displays a mixed mode fracture ap-pearance with areas exhibiting both brittleand ductile fracture modes present.

In the second failure mode, which isobserved in the welds with the higher im-pact strength, failure occurs within the6061-O Al interlayer. As shown in thecross-section view in Fig. 13A, the region

of failure displays a tortuous morphologyacross the width of the Al interlayer. Mi-crographs of the fracture surface, shown inFig. 13B and C, display a dimpled mor-phology, indicative of a ductile fracturemode.No regions indicative of brittle frac-ture are observed, and the dimpled mor-phology covers the entire fracture surface.This failure mechanism is desirable.

A closer examination of the cross sec-tions of the two fracture surfaces in Figs.12A and 13A provides evidence that theexplosion welding pattern is playing a rolein the formation of these two distinct fail-ure mechanisms. In Fig. 13A, it appearsthat the tortuous fracture surface mor-phology correlates with the wavy weld pat-tern observed in the weld cross section inFig. 10A. On the other hand, the failuremechanism shown in Fig. 12A for the low-impact strength sample indicates that theexplosion welding pattern at this locationin the welded plate does not develop theclearly defined waves observed in Fig.10A. These differences in failure mecha-nism can thus be correlated with corre-sponding differences in weld morphology.

The formation of the characteristicwavy weld interface morphology in explo-sion welds is known to be controlled by theflyer plate collision velocity (Ref. 17). Inorder for this preferred weld interface

morphology to form, the interface kineticsgoverning the welding of the two platesmust be controlled by a turbulent flowmechanism. Sufficient collision velocitiesare required in order for this mechanismto dominate and for the preferred weldline morphology to be formed. If the col-lision velocity becomes excessive, the tur-bulence can become extreme, leading tovery large waves, which can result in amix-ing and melting of the constituent metalsat thewave crests and troughs, and the for-mation of either voids or brittle zones, re-sulting in a decrease in strength. This con-dition dominates in Welds #1 and #2,where evidence for melting at the Al-Nbweld interface is observed. On the otherhand, at collision velocities below thethreshold value where turbulent flowdominates, the resulting interface kineticsis controlled by laminar flow. When lami-nar flow dominates, the resulting weld in-terface morphology is devoid of the wavepatterns characteristic of turbulent flow.

In the case of Weld #3a, it can be as-sumed that the collision velocity, driven bythe explosive energy, is in a transition re-gion, where both laminar and turbulentflows are present. As a result, both flat andwavy weld interface morphologies are ob-served. The presence of this nonuniformwelding pattern across the surface of the

Fig. 14 — Micrographs. A — Showing the entire weld cross section; B — the Al-Nb and Nb-Nb welds taken at an orientation perpendicular to the explosivewave front for a multiple weld plate with a 1.0-mm Al interlayer.

Table 6 — Summary of Tensile and Impact Strengths Measured in Thermal Cycled Samples

Tensile Strength (MPa) Impact Strength (J)Welding Number Unnotched NotchedParameter of Tests Average Standard Average Standard Number Average StandardIdentification Deviation Deviation of Tests Deviation

0.8mm Al Interlayer3a(a) 4/4 175 50 256 65 9 4.3 1.6

1.0mm Al Interlayer4a(b) 3/— 243 17 — — 3 17.3 8.44b(b) 13/6 267 28 277 38 19 19.4 2.9

(a) Izod impact samples display failure at the Al-Al interface.(b) Izod impact samples display failures within the Al interlayer.

A B

WELDING RESEARCH

NOVEMBER 2006-s262

welded plate also creates a large degree ofuncertainty in the weld strength and sub-sequent performance. Therefore, changesin the explosive energies used to weld thethin Al interlayer to the thick Al plate aremade in an attempt to produce more uni-form weld properties and to promote thedesired failure mechanism within the Alinterlayer during impact testing.

There is an optimization point at whichthe turbulent flow is sufficient to producea wavy weld interfacemorphology withoutthe appearance of either a flattened weldinterface morphology or intermixed re-gions resulting from excessive turbulentflow. By increasing the explosive energy, acollision velocity can be attained wherethe entire welded area consists of a wavyinterface. Table 2 shows that Weld #3buses a higher explosive energy to weld the6061-OAl sheet to the 6061-T651 Al platethanWelds #3a and #3c. With this higherexplosive energy, the collision velocity isnow of a sufficient magnitude to produceturbulent flow across the entire weld in-terface. As a result, the bimodal distribu-tion of impact strengths and multiple fail-ure modes are not observed in this weld.

On the other hand, Weld #3b does notcontain an adequate region of acceptablewelding, thus prohibiting the use of thesewelding parameters. The increase in ex-plosive energy required to achieve thewavy interface when using the 0.8-mm6061-O sheet in Weld #3b not only de-creases the amount of well-welded mater-ial but also leaves a roughened top surfaceon the thin aluminum sheet. Because ofthe small thickness of the Al sheet (0.8mm), the associated surface roughnessnearly reaches the depth of the explosionwave pattern through the thickness of thesheet. Changes to the welding proceduresare therefore required in order to producewelds with the desired wave pattern weldinterface morphology and resulting highimpact strength levels.

In order to use higher explosive ener-gies to weld the 6061-O Al interlayer to

the 6061-T651 Al plate, a thicker Al inter-layer is used. The combination of thethicker Al interlayer and the increased ex-plosive energy is expected to produce astronger weld between the 6061-O and6061-T651 Al and allow for an enhancedamount of acceptable welding area acrossthe surface of the plate. Welds #4a and#4b, as summarized in Table 2, utilize a1.0-mm-thick Al interlayer, allowing anexplosive energy, on the order of nearly50% higher than those used in Welds #3athrough #3c, to be used for both the Al-Al and Al-Nb welds.

Figure 14A shows a micrograph of theweld cross section at an orientation per-pendicular to the explosion wave fronttaken from Weld #4a, which utilizes the1.0-mm-thick Al interlayer and higher ex-plosion welding energies. When com-pared with the weld formed using a thin-ner Al interlayer, which is shown in Fig.10A and B, the increased thickness of theAl interlayer is apparent, as is the desiredwavy weld pattern. In Fig. 14B, the Al-Nband Nb-Nb welds are shown at a highermagnification. Both welds display the de-sired wavy weld pattern, and the Al-Nbweld shows no evidence of a mixed regionor intermetallic formation.

The use of a 1-mm-thick 6061-O Al in-terlayer does not adversely affect eitherthe tensile or shear strength of the weld, assummarized inTable 4. In general, the ten-sile and shear strengths are equivalent tothose observed in the welds that use the0.8-mm-thick Al interlayer. During thetesting of the impact strengths of thesewelds, as summarized in Table 5, the weldswith the thicker Al interlayer display a sig-nificantly higher impact energy, ap-proaching 21 J for a 12.7 × 10.2 mm cross-sectional area inWeld#4a, than the weldsformed with the thinner Al interlay. Thisimpact energy corresponds to a value forthe impact energy per unit area of 0.119J/mm2. In addition, only a single failuremode, occurring in theAl interlayer, is ob-served in all of the tests. The resulting

fracture surfaces exhibit only features thatindicate a ductile fracturemode, similar tothat shown in Fig. 13C.

Even though the results of the me-chanical testing of Weld #4a are a signifi-cant improvement over those observed inWelds #3a–c, additional changes aremade to the welding procedures. Thesechanges are primarily related to how theAl andNb sheets andNb plate are fixturedduring the explosion welding process inorder to achieve more uniform weldingproperties across the surface of the Al-Nbclad plate. They are also meant to mini-mize any potential plate-to-plate varia-tions in the properties of the Al-Nb explo-sion welds during the manufacture ofmany clad plates. With these changes inplace, the tensile and impact strengths ofWeld #4b are similar to those observed inWeld #4a.

In summary, the development of an ex-plosion welding process for joining a 9.5-mm-thickNbplate to a 203-mm-thick 6061-T651 Al plate has been successful inproducing welds with high tensile and im-pact strengths. In this process, thin 6061-OAl and Nb sheets have been utilized to pro-duce aweld having a tensile strength (255±13MPa) approximately 80%of the 6061-T6Al basemetal ultimate tensile strength (312MPa).The impact strengthof theweldspro-duced using this process (19.2 ± 3.5 J or0.148 ± 0.0027 J/mm2) exceed the impactstrength of the 6061-T6 Al base metal (9.8± 0.5J or 0.076 ± 0.004 J/mm2) by nearly afactor of two. The explosive energies andfixturing of the individual layers during thewelding process have also been optimizedto maximize the area of acceptable weldingacross the surfaceof the203×203mm6061-T651 Al plate.

Effects of Thermal Cycling onWeld Properties

Thermal cycling tests have been per-formed in order to determine if the me-chanical properties of the welds are af-

Fig. 15 — Comparison between measured tensile strengths in the as-receivedand thermal cycled multiple welded plates.

Fig. 16 — Comparison between measured Izod impact strengths in the as-received and thermal cycled multiple welded plates.

WELDING RESEARCH

-s263WELDING JOURNAL

fected by exposures to extremes of tem-perature. Notched and unnotched ramtensile and Izod impact samples takenfrom multiple welds with both 0.8- and1.0-mm-thick Al interlayers have beentested. Table 6 provides a summary of thetensile and impact strengthmeasurementsmade on samples exposed to the thermalcycling sequence described above. Testshave been performed on samples takenfromWeld#3a, which has a 0.8-mm- thickAl interlayer, and Welds #4a and #4b,which both have a 1.0-mm-thick Al inter-layer. In general, the welds made with thethicker Al interlayer display significantlyhigher tensile and impact strengths afterthermal cycling than the weld with thethinner Al interlayer.

Comparisons of these results with thoseobtained from the testing of similar samplesin the as-received condition are given inFigs. 15 and 16, respectively, for the tensileand impact strengths of the welds. In thecase of the thinner Al interlayer, there is anevident decrease in the tensile (175.3±49.5MPa) and impact energy per unit area(0.027 J/mm2) after thermal cycling, withthe most prominent decrease occurring inthe impact strength.Thisdecrease in impactstrength can be traced to the appearance ofonly a single failure mode at the Al-Al in-terface for all of the samples tested in thethermal cycled condition.

On the other hand, the welds madeusing a thicker Al interlayer (1.0 mm) dis-play no such decrease in impact strengthafter thermal cycling. In both of the weldstested with the thicker Al interlayer, thetensile and impact strengths measuredafter thermal cycling basicallymatch thosemeasured in the as-received condition. Infact, Weld #4b shows the highest mechan-ical property values of all of the weldstested. Failure in each weld is observedwithin the Al interlayer, with the resultingfracture surfaces showing evidence of duc-tile failure, similar to that observed in theas-received samples.

Weld #4b displays unnotched andnotched tensile strengths of 267± 28MPaand 277 ± 38 MPa, respectively, in thethermal cycled condition. Both values fallwithin 6% of the as-received values. Themeasured impact strength for the thermalcycled samples taken from Weld #4b is19.4 ± 2.9 J (0.150 ± 0.22 J/mm2). Thisvalue varies from that measured in the as-received condition by only 1%. Based onthese results, it is thus apparent that theexplosive welds produced with the 1.0-mm-thick Al interlayer are unaffected bythe thermal cycling.

Summary and Conclusions

An explosion welding procedure wasdeveloped to clad 9.5-mm-thick niobiumplate to 203-mm-thick 6061-T651 Al plate

using a three-step procedure. This weldconsists of three separate explosively cladlayers: a 1.0-mm-thick 6061-O Al sheet, a0.33-mm-thick Nb sheet, and a 9.5-mm-thick Nb plate. The introduction of thethin 6061-O Al and niobium sheets allowsthe 9.5-mm-thick niobium plate to bewelded to the Al plate in the high-strength(T-651) condition, thus removing the needfor a postwelding heat treatment requiredin earlier welds. Metallographic examina-tion of themultiple weld region also showsno evidence of melting or intermetallicformation.

In the final welding process, the deto-nator is located in center of plate, and dif-ferent explosive energies are used to makethe three welds. For the weld between the6061-O Al sheet and the 6061-T651 Al-plate, an explosive energy of 1.37 MJ isused. An explosive energy of 1.54 MJ isused to weld theNb sheet to the 6061-OAlsheet, and an explosive energy of 3.08 MJis used to weld the thick Nb plate to thethin Nb sheet. The resulting weld displaysa tensile strength of approximately 255 ±13 MPa in the unnotched and 284 ± 25MPa in the notched condition. These val-ues are approximately 80% of the ultimatetensile strength of the 6061-T6 Al. The im-pact strength of the welds, which is 19.2 J± 3.5 J (0.148 ± 0.027 J/mm2) is also threetimes that of the high-strength aluminum.

Selected tensile and impact strengthsamples were also exposed to acceleratedthermal cycling treatments between –22°and 45°C to determine if the significantdifferences in the coefficients of thermalexpansion for the niobium and aluminumcause the weld to be weakened over time.After being exposed to these extremes oftemperature, the tensile and impactstrength samples were tested and com-pared with results from as-welded sam-ples. These thermally cycled samplesshowed no degradation in mechanicalproperties when compared with the as-re-ceived material. For example, the un-notched tensile strength of the thermal cy-cled samples is 267± 28MPa, the notchedtensile strength is 277 ± 38 MPa, and theimpact strength is 19.4 ± 2.9 J (0.150 ±0.22 J/mm2). In each case, the differencesbetween the tensile and impact strengthsof the as-welded and thermally cycledwelds are insignificant.

Acknowledgments

This work has been performed underthe auspices of the U.S. Department ofEnergy, by the University of California,Lawrence Livermore National Labora-tory under Contract No. W-7405-ENG-48. Additional thanks go to Dennis Free-man for his extensive support during themechanical testing, Edwin Sedillo

(LLNL) for providing the scanning elec-tron microscopy support, Robert Vallier(LLNL) and JacksonGo (LLNL) for theirmetallography support, Robert Huber(LLNL), Trung Le (LLNL), and PaulSouza for their ultrasonic evaluation sup-port, and Mark Gauthier (LLNL) for ma-terial handling and movement.

References

1. Elmer, J. W., Terrill P., Brasher, D., andButler, D. 2002. Joining depleted uranium tohigh-strength aluminum using an explosivelyclad niobium interlayer. Welding Journal 81(8):167-s to 173-s.

2. Juvinall, R. C. 1967. Engineering Consid-erations of Stress, Strain, and Strength. NewYork, N.Y.: McGraw Hill.

3. Welding Handbook, Vol. 2, WeldingProcesses. 1991. Miami, Fla.: American Weld-ing Society.

4. ASM Metals Handbook Vol. 6: WeldingBrazing, and Soldering. 1993. Materials Park,Ohio, ASM International.

5. Massalski, T. B. 1990. Binary Alloy PhaseDiagrams, Vol. 3, 2nd ed. Materials Park, Ohio:ASM International.

6. ASTM B209M-02a, Standard Specifica-tion for Aluminum and Aluminum-Alloy Sheetand Plate (Metric).

7. AMS-QQ-250/11, Aluminum Alloy 6061,Plate and Sheet.

8. ASTM B393-03, Standard Specificationfor Niobium Alloy Strip, Sheet, and Plate.

9. ASTMA578M-96, Standard Specificationfor Straight Beam Ultrasonic Examination ofPlain and Clad Steel Plates for Special Applica-tions.

10. ASTM E214-01, Standard Practice forImmersed Ultrasonic Examination by the Reflec-tion Method Using Pulsed Longitudinal Waves.

11. ASTM E1001-99a, Standard Practice forDetection and Evaluation of Discontinuities bythe Immersed Pulse-Echo Ultrasonic MethodUsing Longitudinal Waves.

12. ASTM B898-99, Standard Specificationfor Reactive and Refractory Metal Clad Plate.

13. ASTM D256-02, Standard Test Methodsfor Determining the Izod Pendulum Impact Re-sistance of Plastics.

14. ASTM D4812-99, Standard Test Methodfor Unnotched Cantilever Beam Impact Resis-tance of Plastics.

15. Brandes, E. A. ed. 1992. Smithells MetalsReference Book, 7th Edition. London, Butter-worth and Heinemann.

16. Metals Handbook, 1990. Vol. 2, Proper-ties and selection: Nonferrous alloys and spe-cial-purpose materials, 10th Ed. MaterialsPark, Ohio: ASM International.

17. Cowan G. R., Bergman O. R., andHoltzman A. H. 1971. Mechanism of weld zoneformation in explosion clad metals. Metallurgi-cal Transactions. 2 (11): 3145–3155.