Effects of Temporal Pulse Shaping On

WELDING RESEARCH

JANUARY 2008, VOL. 87-s18

ABSTRACT. Pulsed laser beam welds of6061-T6 aluminum alloy hermetic pack-ages for encapsulation of sensitive elec-tronic and optoelectronic componentstypically exhibit severe solidificationcracking when a conventional rectangularor on/off laser pulse is employed. In thisstudy, the effects of temporal pulse shap-ing on the solidification cracking suscepti-bility of Nd:YAG pulsed laser welds madein 6061-T6 aluminum were investigated.The results showed that it was possible toeliminate solidification cracking in 6061-T6 pulsed Nd:YAG laser seam welds bydecreasing the ramp-down gradient of thelaser pulse power after the main weldingpulse sector. Crack-free welds were pro-duced over a limited range of trailingramp-down gradients; however, intermit-tent solidification cracking recurred whenthe gradient was further decreased. Thesolidification cracking susceptibility wasalso found to increase with increasingpeak power density of the main weldingsector. Use of a trailing ramp-down pulseshape was found to affect the solidifica-tion morphology of the welds. The widthof the initial planar grain growth layer atthe fusion boundaries and the dendriteand cell spacing increased with decreasingramp-down gradient. EDS measurementsof the overall chemical composition of thesolidification crack surfaces of weldsshowed that microsegregation of Mg, Si,Fe, and Cu to the grain boundaries in-creased with decreasing ramp-down gra-

dients. This was thought to be the reasonfor the return of intermittent crackingwhen low ramp-down gradients were used.

Introduction

Encapsulation of electronic and opto-electronic components or assemblies inhermetic enclosures or packages is per-formed to improve the reliability of thesecomponents by providing mechanical sup-port, shielding from electromagnetic inter-ference, better thermal management, andenvironmental protection of the environ-mentally sensitive components (Refs. 1, 2).Hermetically sealed electronic packagesare widely used in the electronics, commu-nications, automotive, computer, and med-ical industries. Figure 1 shows an exampleof an electronic package consisting of a boxmachined from 6061-T6 aluminum and a4047 aluminum alloy sheet lid. Once theelectronic circuits are assembled into thebox, the package is placed in a glove-boxenvironment with the desired enclosure gasand the lid is welded to the box. Wroughtaluminum alloys such as 6061-T6 are widelyused for microwave amplifier and opto-electronic packages because they are light-weight, and have high corrosion resistance,

good workability and machinability, andhigh thermal and electrical conductivity(Ref. 3).

The quality and durability of the her-metic seal in these packages must be highas failure of the seal can lead to corrosionof the package contents and prematurefailure of the device (Ref. 4). PulsedNd:YAG laser welding has become themethod of choice for hermetic sealing ofthe most critical devices, because it haslow and precise heat input that produces asmall heat-affected zone, low distortionand residual stresses, is a noncontactprocess with an ability to weld a widerange of materials, has good process flex-ibility for automation, and is capable ofmaking reproducible, high-quality welds(Refs. 47).

Many wrought aluminum alloys, espe-cially 6xxx series aluminum alloys, areknown to be susceptible to hot cracking dur-ing fusion welding due to their relativelyhigh thermal expansion coefficient, largesolidification shrinkage, and wide solidifica-tion temperature range (Refs. 816). Thereare generally two main categories of cracksgenerated during welding of such aluminumalloys; either solidification cracking withinthe weld fusion zone or liquation crackingin the partially melted zone (PMZ) adjacentto the fusion boundary (Refs. 819). Figure2 shows an example of a pulsed Nd:YAGlaser seam weld made in 6061-T6 aluminumusing a series of overlapping on/off or rec-tangularly shaped laser pulses. This weldhas severe solidification cracking (Fig. 2A,B) and also liquation cracking in the PMZ(Fig. 2B) and would, therefore, be unac-ceptable for electronic packaging or anyother applications.

For solidification cracking to occur,

Effects of Temporal Pulse Shaping onCracking Susceptibility of 6061-T6

Aluminum Nd:YAG Laser WeldsUse of temporal pulse shaping may make it possible to laser weld 6061-T6

aluminum hermetic packages without the need for filler material

BY J. ZHANG, D. C. WECKMAN, AND Y. ZHOU

J. ZHANG was with the University of Waterloo.He is now director, Welding Technology Center,Lincoln Electric, China. D. C. WECKMAN andY. ZHOU are with Department of Mechanical andMechatronics Engineering, University of Water-loo, Waterloo, Ont., Canada.

KEYWORDS

CrackingAluminum AlloysNd:YAG LaserLaser Beam WeldingTemporal Pulse Shaping

Zhang Supplement January 2008corr:Layout 1 12/10/07 4:34 PM Page 18

WELDING RESEARCH

-s19WELDING JOURNAL

two conditions must exist at the solid-liquid interface of the solidifying alloy:one is related to the thermomechanicalconditions present and the other is the so-lidification morphology. Solidificationcracking will occur when tensile thermo-mechanical strains in the material due tothermal contraction of the recently solidi-fied metal causes adjacent dendrites orcellular-dendrites to pull away from eachother near the base of the dendrites (Refs.9, 20, 21). If the distance between the tipand base of the dendrite is large and thereis significant restriction to fluid flow downthe relatively long narrow passage be-tween the dendrites, then a void will format the base of the dendrites. Continued so-lidification will cause growth of this void inthe direction of solidification and the cre-ation of a solidification crack similar tothose shown in Fig. 2.

The susceptibility of alloys to solidifi-cation cracking is influenced by the weldmetal composition, the welding processand weld process parameters used, andthe restraint to thermal contraction due toweldment and joint design or clamping.For example, solidification cracking sus-ceptibility is known to increase with alloyfreezing range, which in turn is influenceddirectly by the alloy composition, the pres-ence of impurities, and microsegregationdue to nonequilibrium solidification ef-fects (Refs. 8, 9, 1821). A longer freezingrange will promote the formation of long,thin dendritic morphology with poor fluid

flow between the den-drites. Cracking sus-ceptibility is also influ-enced by thesolidification mi-crostructure, theamount and distribu-tion of liquid at thefinal stage of solidifi-cation, the primary so-lidification phase, thesurface tension of thegrain boundary liquid,the grain structure,and the ductility of thesolidifying weld metal(Refs. 1721). For agiven alloy composi-tion, the dendritespacing, nonequilib-rium freezing range,and the distance fromstart to finish of solid-ification are all af-fected by the localgrowth velocity andtemperature gradient(Refs. 20, 21). All ofthese factors affect the resistance to fluidflow between the dendrites and, therefore,influence cracking susceptibility.

Several techniques have been reportedto be effective in reducing or eliminatingsolidification cracking in laser welded alu-minum alloys. The first approach involvesuse of a filler metal with a different com-

position and shorter freezing range. Nor-mally, 4xxx and 5xxx series welding wiresor Al-Si alloy foil can be used to move theweld metal composition and freezingrange away from the crack-sensitive range(Refs. 8, 9, 14, 15, 22, 23). When fabricat-ing hermetic packages by laser welding,this can be also be done by making the lid

Fig. 2 Cracks in 6061-T6 aluminum pulsed Nd:YAG laser seam weld. A Top weld bead showing overlapping spot welds and solidification cracks;B transverse section solidification and liquation cracks.

Fig. 1 Example of an electronic package consisting of a 6061-T6 alu-minum machined box and 4047 aluminum sheet lid.

Table 1 The Nominal Composition (wt-%) of AA6061-T6 Aluminum (Ref. 32)

Mg Si Cr Fe Cu Mn Zn Ti Al0.81.2 0.40.8 0.040.35 0.7 0.150.40 0.15 0.25 0.15 balance

Fig. 3 Schematic diagram of the welding jig and Nd:YAG laser beam de-livery head.

A

B


WELDING RESEARCH


out of the filler metal alloy as shown in Fig. 1 or by making a special insert for thejoint out of the filler metal alloy. However,filler metal alloys are not always available

in the desired shapesand the extra partsand manufacturingsteps required aretime consuming andcostly. Laser weldedaluminum alloy her-metic packages havealso been made suc-cessfully by Ni platingone or both aluminumparts prior to welding(Refs. 24, 25). Theweld metal can bechanged to a zerofreezing range Al-Nieutectic and solidifi-cation crackingavoided by plating anoptimized thicknessof Ni on the alu-minum part prior tolaser welding; how-ever, this also requiresan additional manu-

facturing step that can introduce otherproblems such as hydrogen porosity fromthe plating operation.

Temporal pulse shaping has also beenused to reduce or eliminate defects in weldsmade in various alloys using the pulsedNd:YAG laser spot welding process (Refs.2630). A conventional pulsed laser weldingbeam pulse is simply on/off, i.e., it has a sim-ple rectangular shape on a plot of laserbeam power vs. time. However, temporalpulse shaping involves varying the laserbeam power with time during the pulse in apredetermined manner. Temporal pulseshaping has been reported to be an effectivemethod of modifying weld fusion area andpenetration (Ref. 26) and reducing or elim-inating a number of commonly observed de-fects in pulsed laser welds, such as porosity(Ref. 26) and solidification cracks (Refs.

2830).Matsunawa et al. (Refs. 28, 29) devel-

oped a laser pulse shape composed of twodistinct pulses, the initial welding pulsefollowed by a second pulse a short timelater. They found that using this pulseshape with a controlled decrease in beampower at the end of each pulse was bene-ficial in reducing cracking in A5083 alu-minum and A7N01 stainless steel alloywelds. More recently, Michaud et al. (Ref.27) developed an optimized ramp-downpulse shape that significantly reduced so-lidification cracking in individual spotwelds in a pure binary Al-3.75 wt-% Cualloy and produced crack-free seam welds.A thermal analysis of the optimized pulseshape on the pulsed laser welding condi-tions showed that solidification times andinterface velocities were lower, and tem-perature gradients were higher during thesolidification process (Ref. 31). Such con-ditions are known to have beneficial ef-fects on the solidification microstructureand solidification cracking susceptibility(Refs. 9, 20, 21).

Temporal pulse shaping has the advan-tage over other methods of preventingvarious weld defects because it does notrequire the addition of a filler metal or useof extra manufacturing process steps suchas plating. This simplifies the manufactur-ing process and makes it possible to fabri-cate the hermetic package using one sin-gle material, thereby improving thecorrosion resistance of the package. How-ever, most previous studies of the use oftemporal pulse shaping during laser spotwelding of aluminum alloys have beendone using 2xxx and 5xxx series of alu-minum alloys (Ref. 27). These alloys havelower solidification cracking indexes com-pared to 6061-T6 aluminum alloy (Refs.813). The objective of the present study,therefore, was to study the effects of tem-

Fig. 4 Ramp-down pulse shapes used the following: A Schematic of waveform design; B measured waveforms.

Fig. 5 Schematic diagram showing the cross-sectioning positions used formetallographic examinations of the laser seam welds.

Table 2 Laser Welding Process ParametersUsed in This Study

Pulse frequency 5 HzPulse width of main welding 4 ms

sectorPeak power of main welding 1.262.21 kW

sector1/(ramp-down gradient) 019.7 ms/kWWelding speed 0.3 mm/sFocal position 0.0 mm1/e2 beam diameter 437 mOverlap rate ~87%Argon shielding gas flow rate 0.24 L/s

A B


WELDING RESEARCH

-s21WELDING JOURNAL

poral pulse shaping on solidificationcracking susceptibility of pulsed laserseam welds on 6061-T6 aluminum. In thiswork, the influences of both ramp-downgradients and the peak power density ofthe main welding sector on characteristicssuch as total solidification crack length,solidification morphologies, and severityof microsegregation were examined.

Experimental MaterialsApparatus and Methodology

The base metal used for this study was6061-T6 aluminum with a nominal chemi-cal composition as shown in Table 1 (Ref.32). All samples were cut from 3.175-mm

(0.125-in.) sheet into 50- 50-mmcoupons. Before welding, the as-receivedsurface of each specimen was cleaned withacetone in an ultrasonic bath followed bya methanol rinse and air drying. The spec-

imens were then clamped on the weldingjig and laser seam welded.

All laser seam welds were made usinga Lumonics JK702H pulsed Nd:YAG laserwelding machine with a fiber-optic beam

Fig. 6 Severe solidification cracking in a weld made with a rectan-gular pulse shape with a peak power density of 11.5 GW/m2: A Weldbead during steady welding; B end of weld bead with crater crack-ing; and C transverse section.

Fig. 7 Intermittent solidification cracking in a weld made with a ramp-down pulse shape with main welding sector peak power density of 11.5 GW/m2 and ramp-down gradient of 420 kW/s: A Weld bead duringsteady welding; B end of weld bead with crater cracking; and C trans-verse section.

Table 3 EDS Measured Compositions (wt-%) at Locations to in Fig. 15

Location Mg Si Cr Fe Cu Al1 1.7 11.2 0.3 5.9 3.1 77.72 2.2 12.0 0.0 4.0 2.3 79.53 0.4 4.2 0.8 4.9 2.2 87.54 1.4 10.3 0.7 6.8 2.6 78.25 1.3 9.9 0.5 4.2 2.5 81.66 1.2 7.5 0.6 3.8 4.4 82.7

Base Metal 0.5 0.6 0.5 0.8 0.3 97.4

A

B

C

A

B

C


WELDING RESEARCH


delivery system. A schematic diagram ofthe laser welding head and specimenclamping fixture is shown in Fig. 3. As in-dicated, the laser beam delivery head wastilted 15 deg from the normal to avoidbackreflection of the laser beam. TheNd:YAG pulsed laser welding machinewas capable of up to 350 W mean power,0.5 to 20 ms pulse widths, maximum peakpulse power of up to 5000 W, and energyup to 55 J/pulse. The laser beam deliverysystem consisted of a 600-m fiber-opticcable feeding into a beam delivery headwith a 200-mm focal length collimator lensand a 120-mm final process lens. The in-tensity distribution of the laser beam atthe focal plane was measured using a ro-tating-wire-type laser beam analyzer (Ref.33) and a LeCroy 9410 digital oscillo-scope. The beam profile was found to bewell described by a Gaussian distributionwith the 1/e2 beam diameter of 437 m.The spot size of the pulsed laser beam wasverified by the measurements of burn pat-terns produced in 25-m-thick Kaptonfilms using the technique developed byWang et al. (Ref. 34). The internal laserenergy meter was calibrated indepen-dently with an Ophir Optronics laserpower/energy meter. Thus, theenergy/pulse and peak power actually de-livered to the surface of the material waswell characterized. The nominal peakpower density was calculated using the relation

where PD is the peak power density(GW/m2), PP is the peak power (W), andw0 is the 1/e2 beam diameter (m).

The laser welding process parametersused in this study are listed in Table 2. Allbead-on-plate seam welds were made withthe laser beam focused on the surface ofthe specimen. The welding speed was kept

at 0.3 mm/s and a pulsefrequency of 5 Hz wasused. This combinationof welding speed andpulse frequency re-sulted in about 87%spot weld overlap rate,which is consistent withthe normal require-ment of 8590% over-lap to ensure the her-meticity of electronicpackages (Ref. 35). Acoaxial flow of argonshielding gas was sup-plied to the workpieceat a flow rate of 0.24 L/sthrough an 8-mm-diameter ceramicshielding gas nozzle set about 6 mm abovethe workpiece.

The temporal pulse shape techniqueused by Bransch et al. (Ref. 26) to changethe pulse shape on the JK702H pulsedNd:YAG laser welding machine whenwelding 304 stainless steel was used, i.e.,each laser pulse was divided into sectorswhere the peak power and width of eachsector were adjusted to obtain the desiredtemporal variation of power during apulse. In the present study, an initial seriesof welds was produced using no pulseshaping, i.e., a normal rectangular oron/off shape laser pulse was used. How-ever, following the arguments of Branschet al. (Ref. 26), Michaud (Ref. 27), andMatsunawa and Katayama et al. (Refs. 28,29), pulse shapes were then developedthat involved a controlled rate of decreaseof beam power after the initial rectangu-larly shaped welding pulse. Figure 4Ashows a schematic of the idealized wave-form design used in the present study forthe ramp-down temporal shaping experi-ments. The initial rectangular portion ofthe pulse can be considered the weldingportion of the pulse while the trailingramp-down portion was considered thesolidification control portion of the

pulse. Both the peak power of the mainwelding sector and the ramp-down rate ofthe solidification portion of the pulse wereadjusted to evaluate the effect of ramp-down gradient, Rg, and peak power density,Pd, on solidification cracking susceptibility.Weckman et al. (Ref. 36) found that pulsedlaser weld dimensions in 1100 aluminumincreased rapidly for the first 2 ms of laserpulse and changed little thereafter; there-fore, a 4-ms main welding sector pulsewidth was used in the present study.

Figure 4B is an example of a series ofmeasured ramp-down pulse shapes. Thespike in power at the leading edge of thewelding sector is characteristic of manysolid-state pulsed laser welding machinesand has been shown by Weckman et al. (Ref.36), Bransch et al. (Ref. 26), and Michaudet al. (Ref. 27) to have no measurable effecton the final weld dimension or weld quality.Note that the actual ramp-down portion ofthe pulse is not linear, rather it consists of aseries of individual pulse sectors of stepwisedecreasing peak powers. As indicated inTable 2, laser seam welds were made usingpeak powers for the main welding sectorranging from 1.26 to 2.21 kW and 1/Rg val-ues of 0 ms/kW (rectangular pulses) to 19.7ms/kW (Rg = 50.8 kW/s).

Laser seam welds produced using arange of peak power densities of the main

PP

wD

P=

41000 (1)

02

Fig. 8 A sound weld made using a ramp-down pulse shape with mainwelding sector peak power density of 11.5 GW/m2 and ramp-down gradi-ent of 100 kW/s: A Weld bead during steady welding; B end of weldbead with crater area; and C transverse section.

A B

C


WELDING RESEARCH

-s23WELDING JOURNAL

welding sector and ramp-down gradientswere transversely sectioned, as shown inFig. 5. All sectioned samples weremounted, ground, and polished to a 0.06- m colloidal silica finish. An opticalmetallographic microscope with imageanalysis system was employed to charac-terize the severity of cracking throughmeasurements such as total crack length,number of cracks, and cracking area onpolished, unetched specimens. The totalcrack length on transverse sections wasfound to have the least experimental scat-ter in data and most sensitivity to the ef-fects of the various weld process parame-ters on cracking. Therefore, the totalcrack length on transverse sections wasused to characterize the solidificationcracking susceptibility of the laser welded6061-T6 aluminum. Following cracklength measurements, the specimens wereetched using both Kellers reagent (Ref.37) (5 mL HNO3, 3 mL HCl, 2 mL HF, and190 mL H2O) and Graff and Sargentreagent (Ref. 37) (31 mL HNO3, 1 mL HF,6 g CrO3, and 168 mL H2O) in a two-stepetching process to reveal the weld mi-crostructure. Finally, surfaces of solidifi-cation cracks in welds were examinedusing a JEOL JSM-6460 scanning electronmicroscope (SEM) and composition mea-surements were made using energy dis-persive spectroscopy (EDS). All SEM im-ages were taken at 20 keV and a 13-mmworking distance. The EDS measure-ments were conducted at 20 keV and a 20-mm working distance.

Results

Effect of Ramp-Down Gradients on Solidification Cracking Susceptibility

Figures 69 show bead-on-plate pulsedlaser seam welds made on 6061-T6 alloywith a peak power density of the mainwelding sector of 11.5 GW/m2 and variousramp-down gradients. Figure 6 shows aseam weld produced with a conventional

rectangular pulsewhere welding wasperformed from left toright. In Fig. 6A, B, theoriginal oxide layerwas vaporized fromthe middle of the weldbead where the powerdensity of the Gauss-ian distributed laserbeam was greatest.However, the originalsurface oxide was notremoved but remainedfloating on the weldpool surface towardthe outside edges ofthe weld pool where the power density wasmuch lower. The actual fusion zone widthwas greater than the width of ripplesshown due to the much lower meltingpoint of the aluminum base metal (925 K)relative to that of the aluminum oxide,Al2O3 (2373 K) on the specimen surface.

In Fig. 6A, B, severe solidificationcracking initiated in one pulse and contin-ued into subsequent spot welds followingthe local direction of solidification. Cies-lak and Fuerschbach (Ref. 12) observedthe same pattern of solidification crackingin pulsed Nd:YAG laser welds made inthis same alloy. Significant crater crackingwas evident in the crater at the end of thewelds Fig. 6B. Examination of thetransverse sections such as shown in Fig. 6C revealed that severe solidificationcracking had occurred in the weld metaland liquation cracking had occurred in theHAZ. The semicircular bands evidentfrom the bottom to the top of the weldzone in Fig. 6C are the fusion boundariesof the sequential overlapping spot weldsthat form the seam weld.

As shown in Fig. 7, there was a reduc-tion in severity of cracking observed whenramp-down pulse sectors were added tothe initial rectangular pulse sector. In Fig. 7A, the solidification cracking was re-duced to a single centerline crack that oc-

curred intermittently along the seam weld.Notable crater cracking still existed in thecrater at the end of the weld Fig. 7B.The transverse section shown in Fig. 7Cexhibited both solidification cracking inthe weld metal and liquation cracking inthe HAZ, but only down the centerline ofthe weld. In addition, some evidence ofgrain boundary liquation toward the sidesaway from the centerline was observed inboth the weld metal and the HAZ.

As shown in Fig. 8, crack-free seamwelds were achieved with further decreaseof the ramp-down gradient to 100 kW/s.As may be seen in Fig. 8A, B, there was noevidence of solidification cracking on thesurface of the weld bead during steadywelding or in the crater area. The crack-like patterns in the crater in Fig. 8B werebreaks in the oxide layer only. As shown inFig. 8C, there were no cracks in the trans-verse sections, but there was some evi-dence of grain boundary liquation. Thiswas confirmed by examination of the as-polished, unetched specimens as well asexamination of the specimens using theSEM. A higher magnification photomi-crograph of the interface region of one ofthese welds is shown in Fig. 13C.

Figure 9 shows a weld made with fur-ther reduction of the ramp-down gradientto 36 kW/s. As may be seen, intermittent

Fig. 9 Intermittent solidification cracking in a weld made using a ramp-down pulse shape with main welding sector peak power density of 11.5 GW/m2 and ramp-down gradient of 36 kW/s: A weld bead duringsteady welding; B end of weld bead with crater cracking; and C trans-verse section.

A B

C


WELDING RESEARCH


solidification cracking was evident again.The intermittent solidification crackingoccurred along the centerline of the weldbead (Fig. 9A) and there was crater crack-ing in the weld crater (Fig. 9B). Mean-while, there was both solidification and li-quation cracking located close to thecenterline of the seam weld. As shown inFig. 9C, some evidence of grain boundaryliquation existed in both the weld metaland the HAZ.

Effects of Ramp-Down Gradient and PeakPower Density of Main Welding Sector onSolidification Cracking Susceptibility

Figure 10 shows a plot of total cracklength as a function of 1/Rg for welds madewith a peak power density of the mainwelding sector of 8.1 GW/m2. Note that1/Rg is used in these plots because the

value of Rg for a rectan-gular pulse approachesinfinity whereas 1/Rg is0. There are three iden-tifiable regimes for so-lidification crackingsusceptibility evident inthis plot. In Regime I,severe solidificationcracking to intermittentcracking was observedin welds producedusing rectangularpulses and relativelyhigh ramp-down gradi-ents such as thoseshown in Figs. 6 and 7.In Regime II, crack-free welds were pro-duced using intermedi-ate Rg values such as theweld shown in Fig. 8. Fi-nally, in Regime III, in-termittent cracking was

exhibited again in welds produced using lowRg values such as the weld shown in Fig. 9.Note that the first two data points in thisregime are zero only because, by chance, thethree transverse sections were made in un-cracked or sound portions of the welds withintermittent cracking evident on the weldbead surfaces.

The results in the first regime in Fig. 10,which showed decreasing cracking ten-dency with decreasing Rg (increasing 1/Rg),is in good agreement with the results of thestudy by Michaud et al. (Ref. 27) on the useof temporal pulse shaping in laser weldingof Al-Cu alloys and Matsunawa et al. (Refs.28, 29) when working with 5083-O alu-minum alloy and SUS 310S stainless steel.However, in Regime III, there was increas-ing solidification cracking susceptibilitywhen Rg was further decreased. This behav-

ior has not been previously reported. As shown in Fig. 11, when the peak

power density of the main welding sectorwas increased from 8.1 to 11.5 GW/m2, thesame trends in cracking behavior shown inFig. 10 were observed; however, there wasa noticeable decrease in the range ofRgvalues for the crack-free Regime II. Fi-nally, crack-free welds could not be pro-duced when the peak power density wasfurther increased to 14.7 GW/m2 and 1/Rgvalues from 0 to 18.0 ms/kW were used.

Figure 12 is a plot of the range of Rg val-ues that could be used to produce crack-freewelds vs. peak power density of the mainwelding sector. There was a clear trend thatthe range of Rg values that could be used toproduce crack-free welds decreases with in-creasing peak power density of the mainwelding sector. These results indicated thatthe solidification cracking susceptibility wasnot only affected by Rg, but was also influ-enced by the peak power density of the mainwelding sector.

MetallographicExamination Results

Solidification cracking is known to beaffected by alloy composition, solidifica-tion rate, and thermal gradient at thesolid-liquid interface during weld metalsolidification (Refs. 9, 20, 21). To help bet-ter understand the effect of Rg on solidifi-cation cracking susceptibility, it is useful toexamine the effects of Rg on the solidifica-tion morphologies and microsegregationof alloy elements within the weld metal.For example, Fig. 13A shows the mi-crostructure at the fusion boundary of aweld produced with the conventional rec-tangular pulse shape. There are fusionboundaries from three subsequent spotwelds also evident in this photomicro-

Fig. 10 Total crack length vs. 1/(ramp-down gradient) for welds madewith a welding sector peak power density of 8.1 GW/m2.

Fig. 11 Total crack length vs. 1/(ramp-down gradient) for welds madewith a welding sector peak power density of 11.5 GW/m2.

Fig. 12 Range of ramp-down gradient for crack-free region vs. peakpower density of the main welding sector.


WELDING RESEARCH

-s25WELDING JOURNAL

graph. The solidification was initially pla-nar over a narrow band at the fusionboundary adjacent to the base metal. Thisband was about 1.5 m wide. Beyond thisband of planar solidification, the interfacebroke down to a cellular-dendritic solidifi-cation structure that extended in a bandthat was about 58 m wide Fig. 13A.This transition of weld solidification mi-crostructure can be explained in terms ofthe G/V ratio, where G is the temperaturegradient in the liquid at the solid-liquid in-terface and V is the local growth rate(Refs. 9, 20, 21). Since V is expected to beinitially small at the fusion boundary, G/Vis large and planar growth takes place.However, as the laser spot welds begin tocool, V increases, G decreases (Ref. 31),and the growth morphology changes fromplanar to cellular-dendritic (Refs. 9, 20,21). Planar growth was evident only whenthe weld bordered unmelted base metal.At fusion boundaries of subsequent spotwelds in previously solidified weld metal,a planar growth region was less evident, al-

though a band was often present Fig.13A. The dendritic structure was not visi-ble beyond this point, suggesting a transi-tion to a very fine solidification mi-crostructure that was beyond theresolution of the optical microscope.There was severe solidification cracking inthe weld metal and intergranular liquationcracking in the HAZ adjacent to the fu-sion boundary. There was also evidence ofgrain boundary liquation in the HAZ.

Welds produced with ramp-down pulseshapes exhibited very different mi-crostructures. Figure 13B shows the mi-crostructure of a weld produced with apeak power density of the main weldingsector of 11.5 GW/m2 and Rg of 420 kW/s(1/Rg = 2.4 ms/kW). The microstructure atthe very beginning of weld metal solidifi-cation was similar to that in the weld madewith a rectangular pulse. However, theband of planar solidification adjacent tothe fusion boundary was wider at about3.5 m. Beyond this band, the solidifica-

tion morphology was cellular-dendritic

Fig. 13B. As solidification proceeded, theprimary dendrite arm spacing (DAS) be-came finer until the cellular-dendritesreached the boundary of a subsequentspot weld. The solidification microstruc-ture was coarser with larger primary DASin this weld compared with that in thewelds made with a rectangular pulse shapeFig. 13A, B. The solidification crackingexhibited was less severe and was reducedto a single centerline crack that propa-gated along the grain boundary in the di-rection of solidification. The observed re-duction of solidification cracking may bethe result of improved ability of liquidmetal to flow back to cellular-dendrites ordendrite roots at the final stage of weldmetal solidification with larger DAS. Fi-nally, as shown in Fig. 13B, there was someevidence of grain boundary liquation in theHAZ and continued solute segregationalong the grain boundaries in the weldmetal.

As shown in Fig. 13C, when the ramp-down gradient was further decreased to

Fig. 13 Effect of ramp-down gradient on solidification morphologies of welds made using a main welding sector peak power density of 11.4 GW/m2 and thefollowing: A A rectangular pulse shape and pulses with ramp-down gradients of B 420 kW/s; C 100 kW/s; and D 36 kW/s.

A B

C D


WELDING RESEARCH


100 kW/s (1/Rg = 10 ms/kW), there was anotable increase of the width of the initialplanar solidification band adjacent to thefusion boundary and the primary DAS inthe weld metal. The width of the initialplanar solidification band increased toabout 5.5 m. There was no solidificationcracking or liquation cracking evident inthis weld; however, there was noticeableevidence of grain boundary liquation inthe HAZ and solute segregation along theweld metal grain boundaries.

As expected, the width of the initial pla-nar solidification band increased when theramp-down gradient was further decreasedto 36 kW/s (1/Rg = 28 ms/kW) Fig. 13D.The width of the band of planar solidifica-tion in this weld is about 8.0 m. Both so-lidification cracking and liquation were evi-dent again. In the photomicrographs shownin Fig. 13AD, the width of the initial pla-nar growth band at the fusion boundary in-

creased and the primary dendrite arm spac-ing in the remaining weld metal increasedwith decreased ramp-down gradient. This isindicative that lower solidification rates andhigher thermal gradients exist during solid-ification of laser spot welds produced usingdecreasing ramp-down gradients in thelaser pulse (Refs. 9, 20, 21).

SEM/EDS Investigation of FractureSurface Compositions and Microsegregation

SEM/EDS analysis of the fracture sur-faces of cracked welds made using a peakpower density of the main welding sectorat 11.5 GW/m2 and a series of ramp-downgradients was performed to evaluate themorphology of the fracture surfaces, toobtain a measure of the element segrega-tion that had taken place near the cracks,and to identify the chemical composition

of second-phase particles observed onthese crack surfaces. In order to provide abasis for comparison for the EDS results,freshly polished and cleaned samples ofbase metal were also analyzed.

Typical crack surfaces in the weld metaland the HAZ are shown in Fig. 14. Figure 14A shows well-developed cellulardendrites of the primary phase. As maybe seen at , some second-phase particlesformed on the primary dendrite surfaces.The dendritic morphology of the cracksurfaces further suggests that these cracksare solidification cracks. Figure 14B showsa liquation crack surface in the partiallymelted zone (PMZ) in the HAZ adjacentto the fusion boundary. There is evidenceof solidification of the liquated thin film inthe lower half of the image. The second-phase particle at is on the intergranularsolidification crack in the weld metal ad-jacent to the fusion boundary.

Fig. 14 SEM photographs of crack surfaces in welds made using a rectangular pulse shape with peak power density of 11.5 GW/m2: A Solidification crackin the weld metal; B liquation crack in the HAZ adjacent to the fusion boundary.

Fig. 15 SEM photographs of crack surfaces in welds made using a ramp-down pulse shape with peak power density of the main welding sector of 11.5 GW/m2and ramp-down gradient of 3 kW/s: A Overall crack surface; B higher magnification for area of interest in in A.

A B

A B

Zhang Supplement January 2008corr:Layout 1 12/13/07 9:43 AM Page 26

WELDING RESEARCH

-s27WELDING JOURNAL

SEM images of a solidification cracksurface in a weld made using a ramp-downpulse shape with a peak power density ofthe main welding sector of 11.5 GW/m2 andthe slowest ramp-down gradient of 36 kW/sis shown in Fig. 15. Liquid that was drawnout into spikes or thin sheets by thermalstrains during the last stages of weld metalsolidification was also observed on cracksurfaces Fig. 15A. In addition, therewere significantly more second-phase par-ticles on the crack surface Fig. 15A, B.As shown in Table 3, EDS analysis of thesesecond-phase particles at locations inFig. 15A showed that they were Mg, Si, Fe,and/or Cu-rich phases.

EDS measurements of the overall chem-ical composition of the crack surfaces ofwelds with hot cracking were done at 500magnification. This lower magnificationwas chosen as it provided higher accuracyand less scatter of the EDS results. It mustbe recognized that these EDS measure-ments cannot be expected to be quantita-tively accurate as their resolution is poorwhen element concentrations are less than1% because the energy peak associated withthat element is very small and close to thenormal background noise of the detector.Also, the electron beam excites not only thesurface elements but a small volume ofmetal below the surface. Thus, the EDSmeasurement is an average measure of thecomposition in the excitation volume. If thesurface film on the surface of the crack isthinner than the excitation volume, thenEDS measurement of composition will notbe quantitatively accurate. These measure-ments should, however, be qualitatively accurate.

As shown in Fig. 16, the average Fe, Mg,and Si concentrations on the crack surfacesincrease slightly with increasing 1/Rg (de-creasing ramp-down gradient) below the1/Rg values that produce crack-free welds,i.e., in Regimes I and II. However, inRegime III, there is a dramatic increase inthese solute concentrations on the crack

surface when 1/Rg exceedsabout 15 ms/kW. Most no-tably, the Fe concentra-tion increased by about4.5 times the nominal basemetal composition, theMg concentration in-creased by about 3 timesthe nominal base metalconcentration, and the Siconcentration increasedby about 10 times thenominal base metal Siconcentration when apulse shape with a 1/Rgvalue of 28 kW/s was usedcompared with a rectan-gular pulse with the samepeak power density.These increased soluteconcentrations can be ex-pected to significantlylower the melting temperature and increasethe freezing range of the liquid in the inter-granular regions during solidification,thereby dramatically increasing the propen-sity for solidification cracking.

Discussion

Changes of laser welding process para-meters can significantly influence thecooling rate, solid/liquid interface veloc-ity, V, temperature gradient in the liquid,G, and microsegregation during weldmetal solidification. These changes will inturn result in different solidification times,dendrite arm spacing, mushy zone width,dendrite growth rate, strain rate, cell orcellular dendrite length, and solidificationtemperature range. All of these factors ul-timately affect the propensity for solidifi-cation cracking in 6061-T6 aluminum dur-ing solidification of laser spot welds.

Effects of Solidification Time

Solidification cracking is caused in part

by difficulties in feeding liquid from themolten weld pool to the cellular-dendriteor dendrite roots during the solidificationprocess and by tensile strains transverse tothe solidification direction due to thermalcontraction as the weld metal solidifiesand cools (Refs. 9, 20, 21). If liquid canflow back freely between the dendrites asthey are pulled away from each other dur-ing cooling, then voids created by the ac-cumulation of thermomechanical strainfrom thermal contraction can be healed,thereby preventing solidification cracking.The thermal conditions present duringrectangular and ramp-down pulse shapescan have a significant influence on thesefactors. Based on the work by Michaud etal. (Ref. 27) on Al-3.75 wt-% Cu for botha rectangular and an optimized ramp-down pulse shape, the solidification timefor the ramp-down pulse shape was muchlonger than that of the rectangular pulse.Solidification cracking was not observedin this alloy when using the ramp-downpulse shape. Similar results were obtainedby Matsunawa et al. (Refs. 28, 29) for an

Fig. 16 Weight-percent of alloy element detected on the solidificationcrack surface of a weld made with a main welding sector peak powerdensity of 11.5 GW/m2 vs. 1/(ramp-down gradient): A Fe; B Mg;C Si.

A B

C


WELDING RESEARCH


A5083-0 aluminum alloy using a pulse witha trailing pulse in comparison with a rec-tangular pulse shape. The longer solidifica-tion time experienced in welds producedwith the ramp-down pulse shapes wouldallow more time for liquid to flow back tothe cellular-dendrites or dendrite rootsthereby reducing the solidification crackingsusceptibility as was observed in Regimes Iand II in Figs. 10 and 11; however, this doesnot explain why intermittent cracking re-turned with further decrease of the ramp-down rate in Regime III.

Effects of Interface Velocities during Solidification

From solidification theory, it is knownthat the interface velocity and tempera-ture gradient at the solid-liquid interfaceduring solidification affect the resultingsolidification microstructure (Refs. 20,21). A planar interface morphology will beformed at extremely high solidificationrates where there is no time for diffusionin the liquid ahead of the interface orwhen the thermal gradient is high and thesolidification velocity is low. At intermedi-ate solidification rates and temperaturegradients, a cellular-dendritic or dentricsolidification morphology will be pro-duced where the primary dendrite armspacing, 1, can be calculated based on theequation proposed by Hunt (Ref. 38)

where C is a material constant whose valueis dependent on the alloy system, G is thetemperature gradient, and V is the solid-liquid interface speed during solidifica-tion. For the ramp-down pulse shape, the

interface velocities arelower than that for a rec-tangular pulse shape (Ref.31), resulting in increasedprimary dendrite armspacing and a coarser mi-crostructure comparedwith the welds producedwith a rectangular pulseshape. This is in qualita-tive agreement with theobserved differences inweld metal microstructureof the welds producedusing the rectangular andramp-down pulse shapesshown in Fig. 13AD. Thecoarser microstructureobserved with the ramp-down pulse shape wouldlead to a larger cell ordendrite interstices and,thus, larger passages forthe flow of liquid back tothe cellular-dendrites or

dendrite roots thereby resulting in re-duced solidification cracking susceptibil-ity. This is consistent with the observed de-crease in crack length with decreasingramp-down rates (increasing 1/Rg) inRegimes I and II in Figs. 10 and 11; how-ever, this also does not explain why inter-mittent cracking returned with further de-crease of the ramp-down rate in RegimeIII.

Effects of Temperature Gradient on Cellor Cellular Dendrite Length

Based on a numerical thermal analysisby Michaud et al. (Ref. 31), the tempera-ture gradients for ramped down pulseshapes are higher than those made by arectangular pulse shape because heat isconstantly being introduced into the sur-face of the weld pool during the ramp-down phase of the laser pulse and a higherthermal gradient is required to conductthis heat into the surrounding base metal.From solidification theory, the tempera-ture gradient, G, directly influences thecell or cellular dendrite length, lc, in themushy zone. This is given approximatelyby (Refs. 20, 21)

where T is the difference between thecell tip temperature, T*, and the root tem-perature Ts, and G is the temperature gra-dient. According to Equation 3, the in-creased thermal gradient at the interfacewith decreasing laser pulse ramp-downrate would result in shorter cells or cellu-lar dendrites and therefore shorter feed-ing distance for liquid to flow back to thedendrite root. Consequently, cracks could

be more easily healed by the liquid flow-ing back to the dendrite roots and thepropensity for solidification crackingwould be reduced. This reduced solidifi-cation cracking susceptibility with de-creasing ramp-down gradient and higherthermal gradients during solidification isconsistent with the results observed byMichaud et al. (Ref. 27) with a binary Al-3.75 wt-% Cu alloy and by Matsunawa etal. (Refs. 28, 29) with A5083-O aluminumand SUS 310S stainless steel. The datashown in Regions I and II in Figs. 10 and11 are also consistent with these effects,that is, solidification cracking susceptibil-ity in the 6061-T6 aluminum alloy de-creased with decreasing ramp-down rate(increasing 1/Rg). However, the reason forthe increase in intermittent solidificationcracking observed with further decrease ofthe ramp-down rate in Region III is not ex-plained by Equation 3.

Effects of Cooling and Strain Rate duringWeld Metal Solidification

The solidification rates and coolingrates of a spot weld produced using a rec-tangular laser pulse is greater than that ofa ramp-down pulse shape (Ref. 31). Thisleads to higher deformation rates (strainrates) for welds produced using a rectan-gular pulse shape. According to a combi-nation of Borlands generalized theory(Ref. 39) and the critical strain rate theory(Refs. 9, 11, 40, 41), an alloy at high tem-perature and most notably within the so-lidification temperature range is affectedby a low-ductility brittle temperaturerange (BTR) near the solidus. In the BTR,cracking can occur in the material in whichcontinuous liquid films separate grains orin which some solid-solid bridges exist be-tween grains when localized tensile strainsin the materials exceed its resistance tocracking or when the material is subjectedto tensile strain at a rate faster than thecritical tensile strain rate. For a normalrectangular pulse shape, the higher cool-ing rate and higher interface velocity willlead to higher strain rates as the metalcools through the BTR. Alternatively, thelower cooling and strain rates presentwhen using the ramp-down pulse shapewill decrease the tendency for cracking asthe material cools through the BTR. Theexperimental results in Regimes I and II inFigs. 10 and 11 are consistent with this the-ory for solidification cracking susceptibil-ity; however, the reason for the return ofintermittent cracking with further de-crease of the ramp-down rate in RegimeIII remains unexplained.

Microsegregation

Hot cracking of aluminum alloy welds isinfluenced metallurgically by the freezing

1

21

4 (2)= C G V

l T GT T

Gcs

= =

( ) /

*(3)

Fig. 17 Schematic illustration of the factors that affect solidificationcracking in pulsed laser welds in AA 6061 aluminum and the resultant so-lidification cracking susceptibility as a function of 1/(ramp-down gradient).


WELDING RESEARCH

-s29WELDING JOURNAL

temperature range of the dendrites duringnonequilibrium solidification and the typeand amount of liquid available at the rootof the dendrites during solidification.These are affected mainly by the weldmetal composition and microsegregationthat occurs during solidification (Refs.1719, 22, 41). Microsegregation, in turn,can be affected by the cooling rate duringsolidification. The cooling rate during so-lidification of a laser spot weld made usinga ramp-down pulse shape is lower than onemade with a rectangular pulse shape, sothere is more time available for microseg-regation to take place in the interdentriticand intergranular regions of the mi-crostructure. In most cases, this will resultin a solute-rich liquid film in the interden-dritic and intergranular regions that has amuch lower melting temperature andgreater freezing range than the base mate-rial. This low-melting-point liquid film isunable to resist the thermomechanical con-traction strains that take place during cool-ing, and voids and solidification crackingwill occur. Low ramp-down gradients mayalso provide enough time for low-melting-point eutectics to form at cell or grainboundaries. Low-melting-point eutecticscan be formed in 6xxx series A-Mg-Si alloysystems such as Al-Mg2Si, Al-Si, Al-Mg2Si-Fe-Mg3Si6Al8-Si, Al-Mg2Si(CrFe)4Si4Al13-Si, Al-CuAl2-Mg2Si, AlCu2Mg8Si6Al5-CuAl2-Si, etc. The meltingtemperatures of these eutectics range from787 to 868 K (Ref. 42), which are muchlower than the liquidus temperature of6061-T6 (925 K) (Refs. 32, 42). Hatch (Ref.43) has reported that coarse Mg2Si was pre-sent in 6061 aluminum and a Al-Mg2Si eu-tectic could be formed with a eutectic tem-perature of 868 K in this alloy. Also, Huangand Kou (Ref. 44) observed Si-, Fe-, andCr-rich particles in the base metal of 6061aluminum and segregation of Si up to 20wt-% and Mg up to 10 wt-% at the grainboundaries in the partially melted zone(PMZ) of gas tungsten arc welded 6061-T6aluminum alloy in addition to Fe-, Si-, Cr-,and Mn-rich particles in the PMZ. The for-mation of such eutectics and solute-rich liq-uid films will increase the freezing range ofthe alloy and increase the susceptibility ofthis alloy to solidification cracking.

As shown in Fig. 16, the severity of Fe,Mg, and Si microsegregation on the solid-ification crack surface increased with a de-crease of the ramp-down gradient espe-cially in Regime III where the ramp-downgradient used was lowest (highest 1/Rg). Inaddition, solidification cracking suscepti-bility increased in Regime III as the ramp-down gradient was decreased (see Figs. 10and 11). This suggests that reappearanceof solidification cracking in Regime IIImay be caused by severe interdendriticand intergranular microsegregation and

the formation of low-melting-point liquidfilms that occurred when the ramp-downgradients were reduced to low levels andsolidification times were increased.

Summary of the Effects of TemporalPulse Shaping on Solidification CrackingSusceptibility

Based on the previous results and dis-cussions, it is clear that the effects oflonger solidification time, larger dendritearm spacing, narrower mushy zone, andsmaller strain rate during welding metalsolidification would benefit the reductionand elimination of solidification crackingwhen decreasing the ramp-down gradient.However, longer cells and more severe mi-crosegregation would cause feeding prob-lems for the remaining liquid metal to flowback to the dendrite roots and widen thesolidification temperature range, result-ing in increased solidification crackingtendency, when decreasing the ramp-down rate further. Figure 17 illustrates thecombined effects of the above-mentionedfactors as a function of the ramp-downgradient. The resultant solidificationcracking susceptibility would follow thetrends of decreasing to increasing crack-ing tendency when decreasing the ramp-down gradient (increasing 1/Rg) as was ob-served in Figs. 10 and 11. As shown in Fig. 17, a crack-free or less severe crack-ing region may be reached when the cu-mulative influences of both effects areminimized.

Conclusions

In the present study, the effects of tem-poral pulse shaping, specifically the effectsof ramp-down gradient and peak powerdensity of the main welding sector, on so-lidification cracking susceptibility ofpulsed Nd:YAG laser beam welds made in6061-T6 aluminum alloy were examined.The results showed that it was possible toeliminate solidification cracking in 6061-T6 pulsed Nd:YAG laser seam welds usingtemporal pulse shaping. The solidificationcracking susceptibility was found to de-crease with decreasing ramp-down gradi-ent. A limited, intermediate range oframp-down gradients resulted in sound,crack-free welds; however, intermittentsolidification cracking occurred againwhen the gradient was further decreased.

There were three identifiable regimesof solidification cracking susceptibilitywith decreasing ramp-down gradients forwelds made with the peak power densitiesof the main welding sector less than 11.5GW/m2. Regime I showed decreasingcracking tendency with decreasing ramp-down gradients. This is in good agreementwith previous studies of the use of tempo-

ral pulse shaping during pulsed laser weld-ing of Al-Cu and Al-Mg alloys. It isthought that cracking in the laser spotwelded 6061-T6 was reduced with de-creasing ramp-down gradient in Regime Ibecause of the beneficial effects of longersolidification time, larger dendrite armspacing, narrower mushy zone, andsmaller strain rate during weld metal so-lidification. Regime II was the crack-freeregion. However, intermittent solidifica-tion cracking was exhibited again inRegime III, in which the solidificationcracking susceptibility increased with afurther decrease of the ramp-down gradi-ents. In this case, it was thought thatlonger cells and more severe microsegre-gation caused feeding problems for the re-maining liquid metal to flow back to thedendrite roots and widened the solidifica-tion temperature range, thus resulting inincreased solidification cracking tendencyin Regime III when the ramp-down rate isdecreased further beyond that used inRegime II.

The crack-free range of ramp-downgradients that produced welds decreasedwhen the peak power density of the mainwelding sector was increased. Crack-freewelds could not be produced when thepeak welding power density was greaterthan 11 GW/m2. Thus, the solidificationcracking susceptibility was not only af-fected by the ramp-down gradient, but wasalso influenced by the peak power densityof the main welding sector.

The weld metal solidification structuresfor welds made using a ramp-down pulseshape were different from those madeusing a rectangular pulse. The width of theinitial planar grain growth layer at the fu-sion boundaries and cell spacing increasedwith decreasing ramp-down gradient in thelaser pulses. Following the initial planarsolidification, there was a transition to cel-lular and then cellular-dendritic solidifica-tion. The typical solidification crack sur-faces in the weld metal revealedwell-developed cellular dendrites of theprimary phase. Some Cu-, Fe-, Mg-, and/orSi-rich second phases formed on the pri-mary dendrite surfaces. EDS measure-ments of the overall chemical compositionof the solidification crack surfaces of weldsshowed a clear trend that Cu, Fe, Mg, andSi microsegregation to the grain bound-aries increased with decreasing ramp-down gradients. This increased microseg-regation was thought to be the cause of thereturn of intermittent solidification crack-ing in Regime III where low ramp-downgradients were used.

In summary, temporal pulse shapingcan be used to affect the solidificationtime, the dendrite arm spacing, the size ofthe mushy zone, the strain rate, and theseverity of interdendritic and intergranu-


WELDING RESEARCH


lar microsegregation of alloying elementsduring weld metal solidification. These inturn influence the resultant solidificationcracking susceptibility. Crack-free pulsedlaser seam welds could be made in 6061-T6 aluminum alloy by using a limitedrange of intermediate ramp-down gradi-ents where both the positive and the detri-mental effects of laser pulse ramp-downrate were at an optimum level.

Acknowledgments

This work was supported by the On-tario Research and Development Chal-lenge Fund (ORDCF) Project Centre forAutomotive Materials and Manufacturing(CAMM)-Waterloo.

References

1. Johnson, B. C. 1991. Electronic MaterialsHandbook, Vol. 1, Packaging, ed. M. L. Minges.Materials Park, Ohio: ASM International. pp.397503.

2. Wong, C. P., and Fang, T. 2001. Funda-mentals of Microsystems Packaging, ed. R. R.Tummala. New York, N.Y.: McGraw-Hill. pp.580610.

3. Sampson, R. N., and Mattox, D. M. 1991.Materials for electronic packaging. ElectronicPackaging and Interconnection Handbook, ed.C. A. Harper. New York, N.Y.: McGraw-Hill,Inc. pp. 1.11.73.

4. Khanna, P. K., Bhatnagar, S. K., and Gust,W. 1999. Analysis of packaging and sealingtechniques for microelectronic modules and re-cent advances. Microelectronics Intl. 16(2):812.

5. Rossi, D. 1986. Choosing the proper de-vice sealing technique. Electronic Packaging &Production 26(2): 188194.

6. Harris, D. B. 1991. Electronic assemblies.Handbook of Electronic Package Design, ed. M.Pecht. New York, N.Y.: Marcel Dekker, Inc. pp.153238.

7. Steen, W. M. 1991. Laser Materials Pro-cessing. New York, N.Y.: Springer-Verlag.

8. Metals Handbook, Vol. 6, Welding, Braz-ing, and Soldering, 10th ed. 1993. MaterialsPark, Ohio: ASM International.

9. Kou, S. 2003. Welding Metallurgy, 2nd ed.Hoboken, N.J.: Wiley-Interscience, John Wiley& Sons.

10. Arata, Y., Matsuda, F., Nakata, K., andSasaki, I. 1976. Solidification crack susceptibil-ity of aluminum alloy weld metal (report I).Trans. JWRI 5(2): 5367.

11. Arata, Y., Matsuda, F., Nakata, K., andShinpzaki, K. 1977. Solidification crack suscep-tibility of aluminum alloy weld metal (reportII). Trans. JWRI 6(1): 91104.

12. Cieslak, M. J., and Fuerschbach, P. W.1988. On the weldability, composition, andhardness of pulsed and continuous Nd:YAGlaser welds in aluminum Alloys 6061, 5456, and5086. Metall. Trans. B 19B: 319329.

13. Matsuda, F., and Nakata, K. 1995. Eval-uation of ductility characteristics and crackingsusceptibility of Al alloys during welding. Trans.JWRI 24(1): 8394.

14. Zhao, H., While, D. R., and DebRoy, T.1999. Current issues and problems in laser

welding of automotive aluminum alloys. Int.Mater. Reviews 44(6): 238266.

15. Cam, G., and Kocak, M. 1998. Progressin joining of advanced materials. Int. Mater. Rev.43(1): 144.

16. Ion, J. C. 2000. Laser beam welding ofwrought aluminum alloys. Sci. Technol. Weld.Join. 5(5): 265276.

17. David, S. A., and Vitek, J. M. 1989. Cor-relation between solidification parameters andweld microstructures. Int. Mater. Reviews 34(5):213245.

18. Katayama, S. 2000. Solidification phe-nomena of weld metals: Solidification crackingmechanism and cracking susceptibility (3rd re-port). J. Light Metal Welding & Construction38(9): 1323 (also 2001, Welding Intl., 15(8):627636).

19. Kou, S. 2003. Solidification and liqua-tion cracking issues in welding. JOM 55(6):3742.

20. Flemings, M. C. 1974. Solidification Pro-cessing. New York, N.Y.: McGraw-Hill Inc.

21. Kurz, W., and Fisher, D. J. 1992. Funda-mentals of Solidification, 3rd ed., Aedermanns-dorf, Switzerland: Trans Tech Publications Ltd.

22. Dudas, J. H., and Collins, F. R. 1966.Preventing weld cracks in high-strength alu-minum alloys. Welding Journal 45(6): 241-s to249-s.

23. Iikawa, T., Sakai, T., Okamoto, S., Na-tori, K., and Nagai, T. 1986. Aluminum alloypackage for microwave amplifier. IEEE Trans.,Components, Hybrids, and Manufacturing Tech-nologies CHMT-9 (4): 513517.

24. Sakai, T., Okamoto, S., Iikawa, T., Sato,T., and Henmi, Z. 1987. A new hermetic sealingtechnique for aluminum packaging. IEEETrans., Components, Hybrids, and Manufactur-ing Technologies CHMT-10 (3): 433436.

25. Biro, E., Zhou, Y., Weckman, D. C., andEly, K. J. 2001. The effects of Ni and Au/Ni plat-ings on laser welding of thin sheets. J. LaserAppl. 13(3): 96104.

26. Bransch, H. N., Weckman, D. C., andKerr, H. W. 1994. Effects of pulse shaping onNd:YAG spot welds in austenitic stainless steel.Welding Journal 73(3): 141-s to 151-s.

27. Michaud, E. J., Kerr, H. W., and Weck-man, D. C. 1995. Temporal pulse shaping andsolidification cracking in laser welding Al-Cualloys. Proc. 4th Intl. Conf. Trends in Welding Re-search, eds. H. B. Smart, J. A. Johnson, and S.A. David, Materials Park, Ohio: ASM Interna-tional. pp. 153158.

28. Matsunawa, A., Mizutani, M., andKatayama, S. 1996. Mathematical modeling offusion and solidification in laser welding andevaluation of hot cracking susceptibility. Trans.JWRI 25(2): 161169.

29. Katayama, S., Mizutani, M., and Mat-sunawa, A. 1997. Modelling of melting and so-lidification behaviour during laser spot welding.Sci. Technol. Weld. Join. 2(1): 19.

30. Kim, H. B., and Lee, C. H. 1999. Effectof Nd:YAG laser pulse shape on welding char-acteristics of STS 310S stainless steel. Sci. Tech-nol. Weld. Join. 4(1): 5157.

31. Michaud, E. J., Weckman, D. C., andKerr, H. W. 1994. Effects of pulse shape on pre-dicted thermomechanical strains in Nd:YAGlaser welded aluminum. Proc. ICALEO94, LIAVol. 79, eds. T. D. McCay, A. Matsunawa, andH. Hgel. Orlando, Fla.: Laser Inst. America.pp. 461470.

32. Metals Handbook, Vol. 2, Properties andSelection: Nonferrous Alloys and Special-Pur-pose Materials, 10th ed. 1990. Materials Park,Ohio: ASM International.

33. Hirak, D. M., Weckman, D. C., and Kerr,H. W. 1994. Measuring the spatial intensity dis-tribution of high-power focused laser beamsusing a rotating-wire type laser beam analyzer.Meas. Sci. Technol. 5: 15131532.

34. Wang, Z. Y., Liu, J. T., Hirak, D. M.,Weckman, D. C., and Kerr, H. W. 1993. Deter-mining the spot size and Gaussian distributioncoefficient of pulsed laser beams using Kaptonfilms. J. Laser Appl. 5(1): 512.

35. Kugler, T. R. 2001. Nd:YAG pulsed-seam welding. LIA Handbook of Laser Materi-als Processing, eds. J. F. Ready and D. F. Farson.Orlando, Fla.: Laser Inst. America. pp.339342.

36. Weckman, D. C., Liu, J. T., and Kerr, H.W. 1997. The effect of process variables onpulsed Nd:YAG laser spot welds: Part II. AA1100 aluminum and comparison to AISI 409stainless steel. Metall. Trans. B 28B: 687700.

37. Vander Voort, G. F. 1999. Metallography,Principles and Practice. Materials Park, Ohio:ASM International. pp. 165199 and 610614.

38. Hunt, J. D. 1979. Cellular and primarydendrite spacings. Proc. Solidification and Cast-ing of Metals. London, UK: The Metal Soc. pp.39.

39. Borland, J. C. 1960. Generalized theoryof super-solidus cracking in welds (and casting).Br. Welding J. 7(8): 508512.

40. Matsuda, F., and Nakata, K. 1995. Eval-uation of ductility characteristics and crackingsusceptibility of Al alloys during welding. Trans.JWRI 24(1): 8394.

41. Katayama, S. 2000. Solidification phe-nomena of weld metals (2nd report): solidifica-tion theory, solute redistribution and microseg-regation behaviour. J. Light Metal WeldingConstruction 38(4): 1224 (also Welding Intl.,2001, 14(12): 952963).

42. Mondolfo, L. F. 1976. Aluminum Alloys:Structure and Properties. London, UK.: Butter-worth & Co. Ltd. pp. 759805.

43. Hatch, J. E. 1984. Aluminum: Propertiesand Physical Metallurgy. Materials Park, Ohio:ASM International.

44. Huang, C., and Kou, S. 2002. Liquationmechanism in multicomponent aluminum al-loys during welding. Welding Journal 81(10):211-s to 222-s.


Documents

Effects of Temporal Pulse Shaping On