WELDING JOURNAL / JUNE 2014, VOL. 93216-s

WELDING RESEARCH

Introduction Structural materials applicationsrequire consideration of many factors,including properties, costs, and manu-facturing processes. As a single basematerial often does not provide thepreferred designs, combined struc-tures with multiple materials are oftenpreferred over a single one to providethe needed characteristics. As themost common structural ferrous and

nonferrous metals that are widely andcrossly used, effective joining of steelto aluminum significantly affects themanufacturing industry. However,they possess very different physicalproperties, which cause difficulties injoining them together soundly. Joining aluminum to steel has at-tracted significant attention from thewelding research community and vari-ous application areas. Much researchhas been done to explore and utilizenovel joining methods to achieve the

joining of these dissimilar materials,such as brazing (Refs. 1, 2), laser beamwelding (Refs. 3–5), electron beamwelding (Ref. 6), diffusion welding (Ref.7), and friction welding (Refs. 8, 9).Some methods were applied success-fully. However, improvements are stillneeded in certain areas, including highcost, limited application environment,and low efficiency. Arc welding, espe-cially gas metal arc welding (GMAW), isa highly productive joining processwidely used in industrial applications(Ref. 10). Utilizing an arc weldingprocess to achieve sound aluminum andsteel joints thus becomes a critical capa-bility for further adoption of aluminum-steel combined structures. An important issue for effectivealuminum and steel joining is the ca-pability of lowering/controlling theheat input to the workpieces to reduceformation of detrimental intermetalliccompounds at the interface. This re-quires that the average welding cur-rent be controlled at a small level whilestill maintaining the arc stability andpreferred metal transfer (Ref. 11).However, typical pulsed GMAWprocesses have difficulty in providingsuch a capability. Recently, novel arcwelding methods as modifications totraditional processes have been devel-oped to provide such a capability, in-cluding methods that control weldingcurrent waveforms to reduce the arcenergy such as Cold Arc by EWM(Refs. 12, 13) and QUINTO CP byCLOOS (Ref. 14). In particular, ColdMetal Transfer (CMT) is a process de-

Pulsed Double­Electrode GMAW­Brazing for Joining of Aluminum to Steel

Y. SHI (shiyu73@gmail.com), G. ZHANG, L. LU, and J. HUANG are with State Key Laboratory of Advanced Processing and Recycling of Nonferrous Metals,Lanzhou University of Technology, Lanzhou, P. R. China. Y. HUANG is with RoMan Engineering Services, Livonia, Mich., and Y. SHAO is with University of Ken­tucky, Lexington, Ky.

ABSTRACT Double­electrode gas metal arc welding (DE­GMAW) is an innovative process thatcan reduce the heat input to workpieces and increase the energy needed to melt agiven amount of wire. To join steel to aluminum for this study, the current in the DE­GMAW was pulsed to further reduce the heat input. This resulted in pulsed DE­GMAW, which can join steel to aluminum by brazing the steel and welding thealuminum. The resultant technology may thus also be referred to as double­electrodeGMAW­brazing. In this paper, this novel method was demonstrated through both the­oretical analysis and experimental investigation. Dissimilar materials, aluminum andsteel, were joined and major parameters affecting the metal transfer in the pulsedDE­GMAW process, including bypass arc parameters and pulsing parameters, wereanalyzed. The strength of the resultant aluminum­steel joint samples was examined.It was found that the average shear tensile strength was about 144.85 MPa, and themaximum one was around 186.73 MPa, which was about 88.5% of the strength ofthe aluminum base material. Pulsed DE­GMAW can thus be considered as an alterna­tive method for low­cost joining of aluminum to steel.

KEYWORDS Aluminum­Galvanized Steel • Metal Transfer • Pulsed Double­Electrode Gas

Metal Arc Weld­Brazing

This pulsed double­electrode welding process proved to be a viablealternative for low­cost joining of aluminum to steel

BY Y. SHI, G. ZHANG, Y. HUANG, L. LU, J. HUANG, AND Y. SHAO

Shi Supplement June 2014 new_Layout 1 5/15/14 4:21 PM Page 216

JUNE 2014 / WELDING JOURNAL 217-s

WELDING RESEARCH

veloped and patented by Fronius thatutilizes a mechanical device to controlthe movement of the welding wirebased on control of the welding cur-rent waveform. However, all thesetechnologies bring undesirable side ef-fects such as complicated designs, highcosts, and relatively narrow weldingparameter windows. Hence, exploringa novel arc welding process with lowercosts and convenient control is still es-sential for the joining of dissimilar alu-minum and steel. Double-electrode GMAW is a newlydeveloped, highly efficient, novel arcwelding process (Refs. 15–20). In thisprocess, a bypass gas tungsten arcwelding (GTAW) torch is added to thetraditional GMAW system to bypasspart of the current through the basematerial. In this way, the heat input tothe base material is reduced, and byadjusting and controlling the bypasscurrent, the heat input to the work-piece can be well controlled. The desir-able stable metal transfer can be

obtained evenwhen the heatinput is low. In ad-dition, the bypassarc not only bene-fits increasing theefficiency of weld-ing but also controlof the weldingthermal processand metal transfer(Refs. 21–25).Compared to theCMT process, thecomplicated me-chanical equipmentto pull and pushthe welding wire toassist the metaltransfer is no longer needed. From theschematic configuration given in Ref.26, two power supplies and two weld-ing guns are used. While the complex-ity is slightly increased, therequirement on the dynamic charac-teristics of the power supply is not

high and the cost of the system is re-duced. Furthermore, a specialized all-in-one power supply can be made anda controllable rectified switch bridgecan replace the bypass loop power sup-ply; the two welding guns can also beintegrated together to be compact.

Fig. 1 — Schematic of the pulsed double­electrode GMAWprocess.

Fig. 2 — Schematic image of the designed synchronized pulsedwelding current waveforms.

Table 1 — Average Main and Bypass Loop Currents Used in Experiments

Main loop average current

24 A 30 A 35 A 40 A 45 A 50 A 55 A 60 ABypass loop average current

28 A √ √ √ √ √ √ √ √32 A √ √ √ √ √ √ √ √37 A √ √ √ √ √ √ — —40 A √ √ √ √ √ √ — —45 A √ √ √ √ √ — — —50 A √ √ √ √ — — — —55 A √ √ √ √ — — — —60 A √ √ √ √ — — — —

Note: ‘√’ = satisfactory formation of Al­steel weld obtained, and ‘—‘ = no experiment.

Fig. 3 — Schematic configuration of the control system inpulsed double­electrode GMAW.

Shi Supplement June 2014 new_Layout 1 5/15/14 4:21 PM Page 217

WELDING JOURNAL / JUNE 2014, VOL. 93218-s

WELDING RESEARCH

This novel process can thus be appliedwith reasonable equipment size, cost,and structure. Unfortunately, intro-ducing the bypass gun alone does notsatisfy the requirements for the join-ing of dissimilar aluminum and steel.

For low heat input applications, theapplicable window for bypass weldingparameters is still relatively narrowand the heat input needs to be furtherreduced. Based on the previous theoretical

and application studies, to better meetthe requirements of dissimilar alu-minum and steel joining (low heatinput and stable metal transferprocess) (Refs. 27, 28), a pulsed cur-rent waveform was introduced in thispaper to the novel DE-GMAW system.For the novel pulsed double electrodeGMAW-brazing proposed in thispaper, the main and bypass arc cur-rents were both pulsed. A digital con-trol system was developed to controlthe wire feed speed, synchronizationof the main and bypass current wave-forms, and other system parameters.The required detaching forces formetal transfer and heat input to work-piece to generate the needed weld poolon the similar base metals can thus beeffectively controlled.

Experimental Setup Figure 1 schematically illustratesthe principle of the pulsed DE-GMAWsystem. Analysis shows that the cur-rents satisfy the following relations:

Fig. 4— Experimental setup.

Fig. 6 — Typical metal transfer in double­electrode GMAW for joining dissimilar aluminum­steel. A — Without current pulsing; B — with current pulsing.

Table 2 — Effects of Peak and Base Currents on Metal Transfer

Main Loop Imp = 198 A, Imp =167 A, Imp = 136 A, Imp = 106 A, Imp = 75 A, Imp = 32 A,Bypass Loop Imb = 5 A Imb = 10 A Imb = 15 A Imb = 20 A Imb = 25 A Imb = 32 A

Globular and Ibp = 104 A, Ibb = 18 A MTF:80Hz MTF: 80 Hz MTF: 72 Hz MTF: 24 Hz MTF: 19 Hz short­arc transfer,

unstable processIbp = 67 A, Ibb = 24 A MTF: 80 Hz MTF: 62 Hz MTF: 17 Hz MTF: 10 Hz MTF: 1 Hz Unstable welding processIbp = 30 A, Ibb = 30 A MTF: 65 Hz MTF: 10 Hz MTF: 1 Hz Unstable process Unstable process Unstable welding process

* MTF: free­flight Metal Transfer Frequency average; Ibypass = 30 A, Imain = 32 A, V welding = 0.5 m/min, duty ratio δ = 14%, pulse frequency ƒ = 80 Hz.Results with MTF less than 80 Hz, synchronized with the waveform frequency of the currents indicate successful free flight transfer.

Fig. 5 — Aluminum­steel bead­on­plate weld beads made with double­electrode GMAW (Ibypass = 30 A, Imain = 32 A, Imp = 167 A, Imb = 10 A,Ibp = 104 A, Ibb = 18 A, Tp = 1.75 ms, Tb = 10.75 ms, Vwire = 4.47m/min). A — Without current pulsing; B — with current pulsing.

A

A

B

B

Shi Supplement June 2014 new_Layout 1 5/15/14 4:21 PM Page 218

JUNE 2014 / WELDING JOURNAL 219-s

WELDING RESEARCH

Equation 1 ~ 5. DE-GMAW has beenproven to be able to produce the de-sired spray transfer at low base metalcurrents (Ref. 28). By introducingpulsed welding current waveforms(shown in Fig. 2), the ability to achievethe desired metal transfer can be fur-ther enhanced without changing theaverage currents in the main and by-pass loops to satisfy the requirementsto detach droplets at the low base ma-terial heat inputs that are needed tojoin dissimilar aluminum and steel. Itcan also reduce the requirements onthe dynamic characteristics of thepower supply and on the specific sig-nal processing. The desired spraymetal transfer could be achieved whenthe average current is far below thetransition current (Refs. 29, 30) in thetraditional GMAW process. To easily

realize and set up pulsed double-electrode GMAW-brazing for the alu-minum to steel joining experiments,an experimental system was estab-lished using the fast control modeltechnology, as shown in Fig. 3. In the experimental system, themain power supply is a DALEX VIROMIG-400L pulsing digital machinemanufactured in Germany, and the by-pass power supply is a digital pulsingmachine with open connectors. Otherelements include an X-Y double-axlewelding table, and MC6212 motioncontrol board. The visual acquisitionsystem includes a Panasonic CP-230CCD camera, an NI PCI-1405 imageacquisition board, a GZL-CL-22C5M-Chigh-speed CCD camera, and a high-speed image acquisition board fromPoint Grey Research Co. Optical de-

vices include a zoom lens and a lenswith a neutral dimmer filter, a narrow-band filter, and heat reduction filter.The data acquisition for electrical sig-nals and the output system consist ofa real-time target system made of twoindustrial control computers, a PCL-812PG data acquisition board thatsupported xPC environment, a PCL-728 D/A output board with isolation, aCSM400FA closed-loop current sensor,and ADAM-3014 standard voltage iso-lation modules. The host computer was connectedto the target xPC by TCP/IP protocol.The host computer was used for the

Fig. 7 — Typical current and voltage waveforms (Ibypass = 32 A, I'p = 85 A， I'b = 18.5 A； Imain = 45 A， Ip = 145 A， Ib = 20A；δ = 20；f = 80 Hz; Vwire = 5.7 m/min). A — Current; B — voltage.

Table 3 — Effects of Pulse Duty Ratio and Base Currents on Metal Transfer

Duty Ratio 18% 16% 14% 12% 10% 8%Base Main Bypass Main Bypass Main Bypass Main Bypass Main Bypass Main Bypass

Current 9 A 17 A 12 A 18.4 A 15 A 20 A 18 A 22 A 20 A 23 A 23 A 25 A

Metal Occasional Occasional Transfer 80 Hz 80 Hz 58 Hz 34 Hz short­arc short­arc

Frequency transfer, 12 Hz transfer, 5 Hz

*Ibypass = 30 A, Imain = 32 A, V welding = 0.5 m/min, pulse frequency ƒ = 80 Hz, Imp = 136 A, Ibp = 91 A. Results with MTF less than 80 Hz, synchronized with the waveform fre­quency of the currents indicate successful free flight transfer.

Fig. 8 — Weld surface in bead­on­plate dissimilar aluminum­steel joining.

Fig. 9 — Aluminum­steel lap­joint de­sign for pulsed double­electrodeGMAW­brazing.

A B

Shi Supplement June 2014 new_Layout 1 5/15/14 4:21 PM Page 219

WELDING JOURNAL / JUNE 2014, VOL. 93220-s

WELDING RESEARCH

control algorithm compiling and de-bugging, acquisition of visual images,motion control of the XY double-axlework table, and data analysis process-ing and storage. The target machinewas used to execute the setting algo-rithm, which included signal acquisi-tion and real-time display of thewelding current and voltage, the sig-nals for arc ignition, motion of wirefeeder, and shielding gas feeding, andoutput of control signals, etc. The co-operation based on division of labor ofhost and target computer machinesmade the experimental system possesshigh real-time ability and fast re-sponse speed. The system can realize control and

adjustment of the main and bypassloop welding current waveforms, andcollect the welding electric signals andvisual signals in the aluminum-steeljoining process. The visual signals canbe collected, displayed, processed, andstored. By processing and analyzingthe collected welding signals and usingthe control algorithm, the output sig-nals can be controlled in real time. Thewelding parameters for aluminum-steel welding can be optimized, andthus the joining process of these dis-similar materials can be appropriatelycontrolled. Figure 4 shows the experi-mental system. Figure 5 shows the bead-on-platedissimilar aluminum-steel welds, made

by melting and depositing aluminumwire on steel, with and without pulsingthe welding current waveforms for lowaverage currents (32 and 30 A for themain and bypass loops, respectively).The filler material used was ER5356aluminum filler metal and the steel basemetal was galvanized steel with zincthickness at 100 g/m2. The travel speedVwelding was 0.5 m/min. Figure 6 showsthe corresponding metal transfer im-ages obtained by the high-speed cam-era. From Figs. 5A and 6A where thecurrents were not pulsed, it can be seenthat when the average currents are low,it is difficult to form aluminum-steelbead-on-plate welds. The droplet sizewas too large and the desirable free-

Fig. 10 — Aluminum­steel lap­joint surface image in pulsed double­electrodeGMAW­brazing (Itotal = 50 A, Imain = 30 A, Ibypass = 20 A, Vwire = 3.05m/min, Vwelding = 0.5 m/min).

Fig. 11 — Strength distribution of aluminum­steellap­joint samples.

Fig. 12 — Ruptured position in 5052 aluminum and galvanized steel lap joint (tensile strength was 186.73 MPa) (Itotal = 54 A, Imain = 38A, Ibypass = 16 A, Vwelding = 0.6 m/min). A — Front view; B — side view.

Table 4 — Effects of Pulse Duty Ratio and Peak Currents on Metal Transfer

Duty Ratio 30% 26% 22% 18% 10% 6%Peak Main Bypass Main Bypass Main Bypass Main Bypass Main Bypass Main Bypass

Current 83 A 58 A 95 A 64 A 110 A 73 A 132 A 85 A 230 A 138 A 377 A 218 A

Metal Occasional Occasional Limited by Transfer Short­arc Short­arc 27 Hz 52 Hz 80 Hz the power

Frequency transfer, 2 Hz transfer, 10 Hz supply

* Ibypass = 30 A, Imain =32 A, V welding = 0.5 m /min, pulse frequency ƒ = 80 Hz, Imb = 10 A, Ibb = 18 A. Results with MTF less than 80 Hz, synchronized with the waveform frequency of the currents indicate successful free flight transfer.

A B

Shi Supplement June 2014 new_Layout 1 5/15/14 4:21 PM Page 220

JUNE 2014 / WELDING JOURNAL 221-s

WELDING RESEARCH

flight metal transfer was difficult toachieve. After the currents were pulsed,the desirable, stable free-flight metaltransfer was achieved as shown in Figs.5B and 6B. The diameter of the dropletwas almost the same as the one of thefiller metals resulting in smooth forma-tion of welds on the steel and alu-minum. The effectiveness in using thepulsed currents to improve metal trans-fer in the DE-GMAW process is thus ex-perimentally verified together with thedeveloped experimental system.

Pulsed DE­GMAW Process andAnalysis

The bead-on-plate weld wasadopted in the analysis for the pulsedDE-GMAW process. The base materialwas the galvanized steel (Q235 low-carbon steel base) with Zn thickness at100 g/m2, and the coupon size was300 × 100 × 2 mm. The filler metalwas 1.2-mm-diameter ER5356 alu-minum. Pure argon was used forshielding gas and the flow rates for themain and bypass loop were 20 and 5L/min, respectively. The coupons werecleaned with acetone before welding toremove oil and stains on the surface.The position of and angle between themain and bypass guns were appropri-ately adjusted to make sure that stablycoupled main and bypass arcs weremaintained. To preliminarily determine thewelding parameter window in pulsedDE-GMAW for aluminum-steel dissim-ilar joining, the following experimentswere designed: Eight values for the av-erage main and bypass currents re-

spectively were selected, and thencross-matching experiments were con-ducted. The travel speed was kept at0.5 m/min. The wire feed speed wason-line adjusted based on the changein the currents. The frequency of thepulses was constant and was set at 80Hz. Table 1 shows the average mainand bypass welding currents adoptedin these experiments. From Table 1, itwas found that there was a wide pa-rameter window for aluminum-steelbead-on-plate welding in the pulsedDE-GMAW process. With the weldingprocess controlled, satisfactory alu-minum-steel bead-on-plate weldscould be effectively achieved evenwhen the average base and bypass cur-rents were 24 and 28 A, respectively. Figure 7 shows a group of typicalcurrent and voltage waveforms sam-pled each 100 ms. From Fig. 7A, thecurrent pulses for the main and bypassloop were almost synchronized. Be-cause of the response time betweenthe welding power supply and controlloop, there was a delay in the bypasscurrent pulse, which is approximately1–2 ms, whose effect on the alu-minum-steel welding process may benegligible. As the impulses of the mainand bypass loop currents were exertedat the same time, it was the peak cur-rent that would further reduce thetransition current for spray metaltransfer of aluminum. Figure 7B is thecorresponding voltage waveform. Themain loop voltage was measured be-tween the GMAW gun (contact tip)and the workpiece, and the bypass onewas between the main GMAW gun andthe bypass tungsten. Figure 8 showsthe corresponding surface of the re-sultant aluminum-steel bead-on-platewelds. It can be seen that with this set

of parameters, uniform weld widthand smooth weld surface were produced. The main advantage of the pulsedDE-GMAW process is that it can realizestable free-flight metal transfer withquite low heat input to the base materi-als. There are two major reasons: 1) Be-cause of the bypass loop arc, beside theeffect of reducing the heat input to thebase materials, the bypass arc alsochanges the distribution of the electro-magnetic force to enhance the metaltransfer and reduce the transition cur-rent needed for the droplet to freelytransfer; 2) because of the bypass cur-rent, by utilizing the synchronized andcontrolled main and bypass currentwaveforms, the average main and by-pass currents can be kept almost thesame, even smaller, while still effectingthe desired metal transfer. As such, theaverage main and bypass currents couldbe further reduced, and the heat inputto the base materials could also be fur-ther reduced. The references (Ref. 28) analyzedthe forces exerted on the droplet inthe double bypass electrode GMAWprocess. First, the radial direction ofthe electromagnetic force due to thebypass arc could accelerate the pincheffect of the droplet, and the axial di-rection of this force can enhancedroplet detachment. Second, becauseof the bypass arc, the root of the mainarc could easily extend to wrap thedroplet, thus further making the elec-tromagnetic force on the droplet to de-tach the molten metal andstrengthening the detaching force.Last, the plasma drag force from thebypass arc will generate an additionaldetaching force and accelerate dropletdetachment, and then the plasma drag

Fig. 13 — Surface morphology of rup­tured 5052 aluminum and galvanizedsteel lap joint.

Fig. 14 — Linear scanning of the joint interface. A — Scanning position; B — element dis­tribution at the interface.

A B

Shi Supplement June 2014 new_Layout 1 5/16/14 11:12 AM Page 221

WELDING JOURNAL / JUNE 2014, VOL. 93222-s

WELDING RESEARCH

force would be increased, which willbenefit the droplet detachment fromthe wire. In addition, by adjusting andcontrolling the bypass current, the arcforce on the droplet and arc shape couldbe changed, and thus different metaltransfer processes could be achieved.The transition current for spray metaltransfer could be reduced, and thusfree-flight metal transfer could be ob-tained with low heat input to the basematerials. To further reduce the heat input tobase materials and keep the stablemetal transfer, the pulsing current wasintroduced to this novel method. Bysynchronizing the main and bypassloop current pulses and controlling theprocess parameters, the droplet wouldbe detached at the peak current in boththe main and bypass loops, and the arcwould be maintained when the currentwas at the base ones to maintain thearc, preheat base material and fillermetal, and not generate droplet trans-fer. It was essential to keep the stablemetal transfer in order to use DE-GMAW to effectively GMAW-braze alu-minum-steel. The effects of the pulse current pa-rameters, under the same averagemain and bypass currents, on thefree-flight droplet transfer was stud-ied using high-speed images. The re-sults are summarized in Tables 2–4where results with an observed trans-fer frequency synchronized with thatof the current waveform imply suc-cessful free-flight transfer was pro-duced. As can be seen, increases in thepeak currents (main and bypass cur-

rents) as well as in the durations, allbenefit the free-flight metal transfer.Further, the peak currents have moresignificant effects on droplet detach-ment than pulse duty ratio. For rela-tively small average currents, thepulse parameters can be adjusted bysuch as increasing the peak currentlevel and/or peak time in the mainand bypass loops, to obtain free-flightmetal transfer. The main effect of thebase current is to main the arcs andpreheat the base material and fillermetal. The main and bypass loop basecurrents thus could be set at low lev-els to reduce the heat input to effec-tively detach droplets at low heatinputs. As can be seen, pulsed DE-GMAWprovides an effective method for pro-ducing the desired free-flight transferat extremely low currents and heatinput, and melt aluminum wire to de-posit on steel. It promises an effectivemethod for joining aluminum to steel.As will be seen in the next section,this process can join similar metals(aluminum and steel) by welding alu-minum and brazing steel to result indouble-electrode GMAW-brazing ofaluminum­steel.

Double­Electrode GMAW­Brazing of Aluminum­SteelLap­Joint Experiments

The base metal materials were thegalvanized steel (Q235 low-carbonsteel) with zinc thickness of 100 g/m2

and 5052 aluminum alloy. The couponsize was 200 × 80 × 1 mm for bothmaterials. The aluminum was lappedon the top of the galvanized steel, andthe lap width was 10 mm. The fillermetal was 1.2-mm-diameter 4043(AlSi5) aluminum. Pure argon wasused as the shielding gas, and the flowrates for the main and bypass loopswere 20 and 5 L/min, respectively.The coupons were cleaned with ace-tone before welding to remove oil andstains on the surface. Then the twocoupons were clamped to the weldingtable. It was also necessary to make surethe main and bypass guns were at thesame plate. The angle between the twoguns also had to be appropriately ad-justed, and the tip of the welding wirepointed to the right position. Theschematic configuration of the lap jointand lap width are shown in Fig. 9. Figure 10 shows the typical forma-tion of an aluminum-steel lap jointmade using the pulsed DE-GMAWprocess. It was found that a satisfacto-rily smooth-surfaced lap joint could beobtained. The amount of the zinc burntwas quite small, and the distortion wasalso insignificant. The total (average)welding current through the filler metalwas approximately 50 A, and the aver-age main and bypass loop currents wereapproximately 30 and 20 A, respec-tively. This indicates that by optimizingthe bypass arc and pulse current param-eters, stable metal transfer and satisfac-tory formation of an aluminum-steeljoint could be achieved at low currentsin pulsed double-electrode GMAW forGMAW-brazing of aluminum-steel.

Fig. 15 — XRD analysis of the ruptured surface of aluminum and galvanized steel side samples. A — Ruptured surface on the aluminumside; B — ruptured surface on the galvanized steel side.

A B

Shi Supplement June 2014 new_Layout 1 5/16/14 8:29 AM Page 222

JUNE 2014 / WELDING JOURNAL 223-s

WELDING RESEARCH

Strengths

Strength tests were conducted onthe 5052 aluminum and galvanizedsteel lap-joints. Figure 11 shows theshear tensile strength distribution forthe 20 lap-joint samples. From Fig. 11,the lap joint shear tensile strength isin the range from 120 to 180 MPa, andthe average and maximum strengthsare 144.85 and 186.73 MPa, respec-tively. It is approximately 88.5% ofthat of the aluminum base material(the tensile strength of 5052 alu-minum is about 211 MPa). This indi-cates that the intrinsic strength of thealuminum-steel lap joint was largerthan 144.85 MPa in pulsed double-electrode GMAW-brazing. During the shear tensile tests, thesamples were mainly ruptured at theheat-affected zone (HAZ) on the alu-minum side. This is mainly becausethe crystal grains in the HAZ are proneto grow to a large size during the weld-ing process, and further become softand reduce the mechanical properties,as shown in Fig. 12. Figure 13 showsthe corresponding surface morphologyof shear tensile ruptured samples. Itwas found that the rupture mode is aductile one.

Microstructure Analysis

Figure 14A shows the SEM inter-face images of the aluminum-steeljoint. From Fig. 14A, it can be foundthat a layer of intermetallic com-pounds about 10 um thick was formedbetween the aluminum fusion zone

and base galvanized steel. These kindsof intermetallic compounds formedtheir grain cores at the interface, andgrew into the base steel in the shape ofcolumnar grains. To further analyzethe composition of the intermetalliccompound at the interface, linearscanning of the joint was conducted asshown in Fig. 14B. From Fig. 14B, itcan be seen that these compoundsmainly consisted of Al and Fe ele-ments. From the galvanized steel basemetal to the central interface in thescanning zone, the content of Fe de-creased dramatically, but the contentof Al increased rapidly. Along the cen-tral interface, the Fe and Al contentswere both relatively constant, andthere was an obvious plain stage. Thezinc was uniformly distributed in theentire joint. To further determine the inter-metallic compound at the interface,the joint was sheared apart. An X-raydiffraction test was conducted on theruptured surfaces of the aluminumand steel samples, and the results areshown in Fig. 15. From Fig. 15, itcould be found that the intermetalliccompounds consisted of Fe2Al5Zn0.4and Fe. Combined with the aforemen-tioned linear analysis results, the uni-form and continuous phase at theinterface was Fe2Al5Zn0.4. Accordingto other studies (Refs. 31–33), theFe2Al5Zn0.4 is supposed to formwhen the intermetallic compoundFe2Al5 interacts with the dissolved Znat the interface. As the melting and boiling points ofZn are very low (419.5 and 907, re-

spectively) (Ref. 34), with the heat ef-fect from the electric arc, the Zn coat-ing is prone to evaporate. In this case,the content of Zn at the center areawas very low due to the high arc tem-perature. At the surrounding area ofthe electric arc, a portion of unevapo-rated Zn was accumulated at theboundary of the weld joint at the ef-fect of electric arc force. They inter-acted with the molten aluminumdroplet and formed a Zn-rich zone atthe toe of the weld, as shown in Fig.16A. Figure 16B shows the linear scan-ning of the Zn-rich zone for elementdistribution, which also demonstratesthat the Zn indeed accumulated at thiszone and Zn-Fe solid solution wasformed.

Conclusions 1. The novel DE-GMAW process,which provides a convenient way to ef-fectively control heat input and metaltransfer, was modified by pulsing thecurrents to form the novel pulsed DE-GMAW process. The pulsed DE-GMAWprocess further reduces heat input andenables double-electrode GMAW-braz-ing of aluminum-steel. 2. An experimental system was es-tablished to implement the novelpulsed DE-GMAW process. It controlsand synchronizes the main and bypassloop current waveforms to provide theneeded parameters to GMAW-brazealuminum-steel. 3. Effective ways to realize stablefree-flight metal transfer at the low cur-rent include increasing the bypass loop

Fig. 16 — Linear scanning of the Zn­rich zone. A — Scanning position; B — element distribution in the Zn­rich zone.

A B

Shi Supplement June 2014 new_Layout 1 5/16/14 8:32 AM Page 223

WELDING JOURNAL / JUNE 2014, VOL. 93224-s

WELDING RESEARCH

current as well as increasing the peakcurrent and duration both in the mainand bypass loops. The base currents inboth the main and bypass loops can beset at relatively low levels to further re-duce heat input. 4. The average and maximumstrengths for the aluminum-steel lapjoints are 144.85 and 186.73 MPa, re-spectively. 5. The intermetallic compounds atthe interface of the aluminum-steel lapjoint mainly consist of uniformly dis-tributed Fe2Al5 or Fe2Al5ZnX at thesteel side, and the FeAl3 in needle flakeat the aluminum side. At the toe of thejoint, there is a Zn-rich zone and themain content was Al-Fe solid solution.

This work was funded by the Na-tional Natural Science Foundation ofChina (#51165023), financial commis-sion of Gansu Province of China andthe Key Project of Chinese Ministry ofEducation (#210229); Projects of In-ternational Cooperation and Ex-changes NSFC (#51210105024);Financial Commission of GansuProvince of China, and the Hong LiuOutstanding Talent Training Plan ofLanzhou University of Technology ofChina (#J201201).

1. Saida, K., Song, W., and Nishimoto, K. 2005.Diode laser brazing of aluminium alloy to steelswith aluminium filler metal. Science and Technol­ogy of Welding & Joining 10(2): 227–235. 2. Gang, D. H., Liao, C., Chen, G. et al. 2012.Butt joining of aluminum to steel by arc brazingprocess. Materials and Manufacturing Processes27(12): 1392–1396. 3. Mathieu, A., Shabadi, R., Deschamps, A., etal. 2007. Dissimilar material joining using laser(aluminum to steel using zinc­based filler wire).Optics & Laser Technology 39: 652–61. 4. Ozakia, H., Kutsuna, M., Nakagawa, N., et al.2010. Laser roll welding of dissimilar metal joint ofzinc coated steel to aluminum alloy. Journal ofLaser Applications 22(1): 1–6. 5. Rathod, M. J., and Kutsana, M. 2004. Joiningof aluminum alloy 5052 and low carbon steel bylaser roll welding. Welding Journal 83(1): 16­s to26­s. 6. Franc, A. 2007. Low energy welding of het­erogeneous joints between steel and aluminum.Soutêžní prehlídka studentských a doktorskýchprací fst. 7. Goecke, S. F. 2005. Low energy arc joiningprocess for materials sensitive to heat. EWM High­tec Welding 11: 1–5.

8. Uzun, H., Donne, C., and Argagnotto, A.2005. Friction stir welding of dissimilar Al6013­T4to X5CrNi18­10 stainless steel. Materials and De­sign 26(1): 41–46. 9. Kimapong, K., and Watanabe, T. 2005. Lapjoint of A5083 aluminum alloy and SS400 steel byfriction stir welding. Materials Transactions 46(4):835–841. 10. Tetsu, I., Seiji, S., Tsuyoshi, M., et al. 2008.Dissimilar metal joining of aluminum alloys andsteel in the spot welding by using advanced hot­dip aluminized steel sheet. Kobelco Technology Re­view 28: 29–34. 11. Shi, Y., Shao, L., Huang, J., et al. 2013. Ef­fects of Si and Mg elements on the microstructureof aluminum­steel joints produced by pulsed DE­GMA welding­brazing. Materials Science and Tech­nology 29(9): 1118–1124. 12. Bruekner, J. 2005. Cold metal transfer has afuture joining steel to aluminum. Welding Journal84(6): 38–40. 13. Furukawa, K. 2006. New CMT arc weldingprocess — welding of steel to aluminum dissimilarmetals and welding of super­thin aluminiumsheets. Welding International 20(6): 440–445. 14. Kochan, A. 2006. Welding innovations forthick and thin sheet. Assembly Automation 26(4):273–274. 15. Li, K. H., Chen, J. S., and Zhang, Y. M. 2007.Double­electrode GMAW process and control.Welding Journal 86(8): 231–237. 16. Wei, H. L., Li, H., Yang, L. J., et al. 2013.Consumable double electrode with a single arcGMAW. International Journal of Advanced Manu­facturing Technology 68(5): 1539–1550. 17. Wu, C. S., Hu, Z. H., and Zhong, L. M. 2012.Prevention of humping bead associated with highwelding speed by double­electrode gas metal arcwelding. International Journal of Advanced Manu­facturing Technology 63(5): 573–581. 18. Ma, G., and Zhang, Y. M. 2012. A novel DE­GMAW method to weld steel tubes on simplifiedcondition. International Journal of Advanced Man­ufacturing Technology 63(5): 147–153. 19. Shi, Y., Han, R.­H., Huang, J.­K., et al. 2012.Numerical simulation of temperature field of DE­GMAW and its comparison with experimentalmeasurements. Acta Physica Sinica 61(2): 020205. 20. Li, K., and Zhang, Y. M. 2010. Intervalmodel control of consumable double­electrodegas metal arc welding process. IEEE Transaction onAutomation Science and Engineering 7(4): 826–839. 21. Li, K., and Wu, C. 2009. Mechanism ofmetal transfer in DE­GMAW. Journal of MaterialsScience and Technology 25(3): 415–418. 22. Li, K. H., Zhang, Y. M., Xu, P., et al. 2008.High­strength steel welding with consumable dou­ble­electrode gas metal arc welding. Welding Jour­nal 87(3): 57­s to 64­s. 23. Li, K. H., and Zhang, Y. M. 2008. Consum­able double­electrode GMAW Part II: Monitoring,modeling, and control. Welding Journal 87(2): 44­sto 50­s. 24. Li, K. H., and Zhang, Y. M. 2008. Consum­able double­electrode GMAW — Part 1: Theprocess. Welding Journal 87(1): 11­s to 17­s. 25. Zhou, W. Z., and Zhang, Y. M. 2007. Imageprocessing algorithm for automated monitoring ofmetal transfer in double­electrode GMAW. Meas­urement Science and Technology 18(7): 2048–2058. 26. Zhang, Y. M., Jiang, M., and Lu, W. 2004.Double electrodes improve GMAW heat input con­trol. Welding Journal 83(11): 39–41. 27. Shi, Y., Liu, X. P., and Zhang, Y. M. 2007.

Dual­bypass GMAW of aluminum — Minimizingbase metal heat input with good welding quality.Proceedings FABTECH International & AWS Weld­ing Show, pp. 223–226. 28. Shi, Y., Liu, X. P., Zhang, Y. M., and Johnson,M. 2008. Analysis of metal transfer in dual­bypassGMAW of aluminum. Welding Journal 87(9): 229–236. 29. Li, K., and Zhang, Y. M. 2007. Metal transferin double­electrode gas metal arc welding. Journalof Materials Science and Technology — Transac­tions of the ASME 129(6): 991–999. 30. Shi, Y., Chen, Z., Xue, C., et al. 2010. Re­search on metal transfer in dual bypass MIG weld­ing of aluminum [J]. Journal of MechanicalEngineering 46(20): 76–79. 31. Shi, Y., Wang, Z., Huang, J., et al. 2013.Study on microstructure of fusing­brazing joint ofaluminum to galvanized steel by pulsed ED­GMAW[J]. Transactions of the China Welding Institution34(5): 1–4. 32. Zhang, H. T., Feng, J. C., He, P., et al. 2007.Interalfacial microstructure and mechanical prop­erties of alumimum­zinc­coated steel joints madeby a modified metal inert gas welding­brazingprocess [J]. Material Characteristics 58: 588–592. 33. Song, J. L., Lin, S. B., Yang, C. L., et al. 2009.Spreading behavior and microstructure character­istics of dissimilar metals TIG welding­brazing ofaluminum alloy to stainless steel [J]. Material Sci­ence and Engineering A 509: 31–40. 34. Wu, C. S. 2007. Welding thermal processesand weld pool behaviors [M]. Beijing, China Ma­chine Press.

Itotal = Imain + Ibypass (1) Itp = Imp + Ibp (2) Itb = Imb + Ibb (3) Ttb = Tmp – Tbp (4) Ttb = Tmb – Tbb (5)

where Itotal is the total current through the weld­ing wire, Imain is the average main current, Ibypass is the average bypass current, Itp is thepeak current through the welding wire, Imp is themain loop peak current, Ibp is the bypass looppeak current, Itp is the base current through thewelding wire, Imb is the main loop base current,Ibb is the bypass loop base current, Ttp is the timeof the peak pulse current through the weldingwire, Tmp is the main loop peak current time, Tbpis the bypass loop peak current time, and Ttb isthe time of the base pulse current through thewelding wire, Tmb is the main loop base currentpulse time, and Tbb is the bypass loop base cur­rent pulse time.

Acknowledgments

Appendix

References

Shi Supplement June 2014 new_Layout 1 5/16/14 8:21 AM Page 224