The influence of solute content and heat input on microstructure was investigated and a prediction model for solidification growth rate and thermal gradient was developed BY P. SCHEMPP, C. E. CROSS, A. PITTNER, AND M. RETHMEIER Solidification of GTA Aluminum Weld Metal: Part 2 — Thermal Conditions and Model for Columnar-to-Equiaxed Transition ABSTRACT In this study, the influence of solute content and heat input on microstructure was investigated for gas tungsten arc (GTA) bead-on-plate welding of the aluminum al- Joys 10SDA (199.5) and 6082 (Al SiMgMa). Temperature measurements in Uae so- lidtying weld pool showed that parameters such as Soidication growth rte, cooling rate, local thermal gradient, and solidification time vary lidifcation front along the s0- ‘weld centerline and weld interface), As a result, the ob- {ained thermal data were sed to explain the corresponding grain morphology from the frst pat of this study. On the basis of this comparison, an analytical approach was used to model the transition {rom columnar-to-equiaxed grain growth (CET), This model allows the prediction of critical values for both solidification growth rate and thermal gradient, at which the CET occurs. Introduction and Background ‘The weld metal microstructure has a ‘great influence on the properties of a fu sion weld. Many studies have shown that small, equiaxed grains instead of large, columnar grains can improve the mechan- ical properties ofthe weld (Refs. 1,2) and the weldability ofthe base metal (Refs. 3 4), Small, equiaxed weld metal grains are usually achieved through the addition of ‘commercial grain refine to the filler metal (Ref. 5) where A TISBL is one of the most important grain refiners (Ref. 6). It was suggested that particles like THB and ‘ALT act during solidification as very effi cient solidification auclei for Al grains (Refs. 7-9). Experimental results on grain FSCHEMPP (PSchempp@ gmc) ond A PIE earch and Testing Berlin, Germany. CE. CROSS with LorcAlamos Nabonel Laboratory, Materials Setence & Technology, Los Alamos, [N Mex M, RETHMBTER is wth Fraunhofer Inst lute for Production Systems and Design Teh ‘og and Federal Institute fr Materials Research ‘and Testing, Belin. Germans Pat 1 appeared n the Febrary 2014 issue ofthe Welding Journal refinement in aluminum GTA welds are shown in Part 1 ofthis study, where a fine ‘weld metal microstructure was achieved by fusing cast inserts with « GTA welding process, These inserts were produced with ‘casting process, in which Al base metal was enriched with varying amounts of commercial Al TISE1 grain refiner (Ref. 10). Beside the presence of effective nucle, the chemical composition of the weld metal influences the weld microstructure strongly. The alloying elements partition inthe melt during the solidification of the ‘weld pool and provide constitutional un- deteooling (Ref. 11), which is needed to activate particles such as TiB, and AlsTi for nucleation (Ref. 12). The highest de- KEYWORDS ‘Ajominum ‘Gas Tungsten Are Welding (GTAW) ‘Colunnas-to-Equiaxed ‘Transition (CET) ‘Thermal Analysis 706 of constitutional undercooling i pro ded by itanium (Rel. 13), which makesit particularly favorable for Al grain relinest (Ref. 14). Furthermore, the coasttutional _undereooling is influenced by a third con ‘rol variable regarding weld metal grain refinement, the solidification conditions that are controlled by the welding param- ters In are welds, these conditions are controled, in adaition othe weld geome tu, particulary bythe welding parameters * Welding speed v_(inmms") sArccurrent I (inA) + Arevoltage U (ia V) that define the heat input 1 vr y a ‘These welding parameters result in thermal conditions that can be expressed by the parameters + Cooling rate dTidt (inK ©") + Local thermal gradient G (in K mn") + Solidification growth rate R (in mm =) + Constitutional undercooling AT; (in K) where cooling rate, thermal gradient, and growth rate have the following relationship Tler a @ It should be noted that the above so- lidification parameters and the correspon: ding microstructure vary strongly along the solidification front. In fusion welds ‘wo different grain morphologies are ire. quently observed. At the weld interface, grains often start growing with a columnar shape. While these columnar grains grow toward the weld center, the soi conditions at their leading edge change. Both dTid and R increase and G de creases (Ref. 15). As a consequence, theFoture Copper backing Type K thermgcouple (wire 2.0.13) Ceramic insulator (@ 1.2) ig. 1— GTA welding and temperatare measurement seep (dimensions in mom. Centerline Weld Interface Thermocouple Fig, 2— Weld bead (rose section) amd location of thermocouple within the weld metal along ai) degree of constitutional undercooling changes. Ttwas suggested that the extent ‘of constitutional undercooling is inversely proportional to G/R" (Ref. 16). Thus, high GIR values can be related to very low constitutional undercooling ahead of the solid-liguid interface (Ref. 17) that favors planar or cellular growth (Ref. 18). Low ‘GIR values, however, result in a arge zone ‘of constitutional undercooling (Ref. 17) which allows columnar — dendritic, dendritic, or (at very low GIR values) equiaxed dendritic structures to form (et. 18). Accordingly, small and equiaxed _rains can be observed along the weld cen- ter, dependent upon alloy content and welding conditions, This. columnar-to- equiaxed transition (CET) is often ob- served in aluminum weld metal (Refs. 19-21) and is illustrated with micrographs from aluminum GTA welds in Part 1 of this study (Ret. 10), tis important to note that large, colum- nar grains provoke anisotropic mechanical properties of the weld and facilitate the propagation of solidification eracks (Ret. 22). Consequently, ts of interest to know critical CET conditions in order to prevent columnar grain growth, A model, which was developed for directional solidi cast structure, predict the cri gradient, under which the grain morphology ‘becomes predominantly equiaxed (Ret. 23). ‘This approach was originally developed for directional solidification in castings and is applied inthis study to weld metal solciti- cation. The eritical gradient (Geyey) is de- fined as the gradient at which the CET occurs (G < Gey: equiaxed: G > Gog colunmnat) reenT 8 Fully equiaxed growth is considered to ‘occur ifthe volume fraction of equiaxed grains is higher than 49% (Ref, 23), Iti of interest that the model behind Equation 3 predicis a second critical condition for fully columnar grain growth at a volume Inaction of equiaxed grains < 1% (Ref, 23), In between these two eritical condi- tions, both equiaxed and columnar grain {growth are expected to occur. For pur poses of simplicity and fora better under: Standing, this paper concentrates on the critical conditions for fully equiaxed grain growth (Equation 3), No s the total num- ber of heterogencous substrate particles that are available per unit volume and can bbe approximated as (Ref. 24) 1 Noe e @ where dis the weld metal grain size. ATiy is the undercooling that is needed to acti vate these paticles and AT. cpr the ext ical constitutional undercooling caused by the partitioning of solute elements and the solidification conditions. AT, cen gen- cerally be related to G and R 48 follows (Ret. 25) AROS 6 where D is the liquid diffusion coefficient. A is a materials constant that depends ‘upon chemical composition of the liquid pphase and that can be defined according to Equation 6 (Refs 23,25) where @ is the curvature undercooling (Ref. 25), my, is the slope of the liquidus line, k is the par tition coetficient between sold and liquid and Cy is the chemical composition of the alloy. 4 (- @. mult=B) cy" o In this work involving GTA welding, “Table l— Chemical Composition of Bate Metals as Measured by Optical Emission Spectrometer (ICP-OES) Alloy Sook ow 1050 oo 02 ot (A195) ose 086 042 0.09 (ASiiMeMa) (Chemical composition in w1-% Ma Mg Gr NZ 0004 0.001 0001 000.1 43075098 om 007 nT oR vow A oot 000030010001 Bal 032 ono 0.010.003 Bal PE earcr 2014, vor 93Messe eis isa i Gy Fig 3 Approximation of ran fof weld mata y= 0: centering, = 3 mom: weld interface) GTA bead-om plate weld (no grain refiner ations welding tothe lf), Alloy 6082 plate thickness 3 mom, welding speed 8 mim 5% heat input 258 mm the first term of Equation $ (GDIR) was neglected because the second term (AR?) was much higher (about 24 times at the lowest welding speed 2.0 mm/s and 70 times at 4.2 mmys) due to high R values and low G values (Res. 23,25) ATe = AR o With respect to R, one as to consider that ‘grain growth occurs parallel to the maxi ‘mum temperature gradient, which is per pendicular to the advancing weld pool boundary (Ref. 22). Ione assumes that the dendrite solidification velocity corre- sponds to the grain growth velocity due to ‘competitive growth (Ref. 22) R can be ap- pproximated for the weld pool susface as cos(a) © where «is the angle between the direc: tions of welding speed v and R at a partic ‘ular point st the solid-liquid interface, as illustrated in Ref. 10 Part 1 ofthis study presents results from GTA welding of three aluminum alloys where welding speed and grain refiner ad- ditions were varied in order to investigate their influence on microstructure and nu- leant particles (Ret. 10). In this second part, an intensive thermal analysis reveals the solidification conditions in the weld pool, dependent upor base metal and weld- ing speed, and their influence on the weld ricrostructure. On the bass ofthe obtained thermal data, the above CET approach (Equation 3) is applied to solidification in aluminum GTA bead-on-plate welds Experimental “Wo different wrought aluminum alloys were used in this study: Alloy 1050A (Al 99.5, temper H14) and Alloy 6082 (Al SilMgMan, temper T6), whose chemical compositions ate shown in Table 1. The base metals were delivered in 3-mmthick plates and ext into 140 x 60x Smmetbick ‘weld coupons that were cleaned hy de= zreasing and eiching for 15 min with an Elchant consisting of 69 mf. H0, 125 ma. 659% TINO,, and 625 mil. 4892 HE Bach ‘weld coupon was placed in a fiture and GTA bead-on-plate weld (single pass, complete penetration) was produced as lustrated in Fig lft "A backing made of copper was used to avoid unwanted melt though, The weld- ing setup was, for purposes of comparison, ‘very similar to the fist part ofthis study Rin ms |All 60ea ‘Fig. 4 Sobdifiction growth rate B dependent upon horizontal poston in weld metal (y= 0. Centelin: y= 3mm: weld interface). GTA weld ing plate sickness 3mm. (Ref, 10) and the welding (Tabie 2) were equal to those used in the first part. Table 3 contains thermal prop- erties for both alloys since they represent the ally’s solidification behavior ‘Metallographic samples were prepared from the middle of some welds (mid Tength) to obtain top-sectional views ofthe weld metal. Each of these samples was ‘round, polished mechanically, and etched anodically with a solution containing 27% HBE, and 98% HO to reveal the grain structure. Micrographs were made with « ‘microscope sing polarized light, which helped to differentiate grains ‘Temperature meanurements were ac complished in the middle of each weld ‘Table? ~GTA Welding Parameters Parameter Alloy 1050, (Al995) Welding speed in mss 20 42 go 80 tons 20 Current in A im 180 18519019295, m0 Voltage in V (= 02.V) Mi ae 167 ne 107 Polasity [AC (80% electrode negative, 20% electrode positive) Frequency soe Electrode W + 2% CeOp, diameter 82.mm, point angle 30 deg Shielding gas 50% Ar, 50% He Flow rate 26L min Distance electrode Sam coupon Alloy 6082 (alsitMeMo) 42 60 $0 100 us 1s Is i 190195, 2 u2 m2 uz us WELDING JOURNAL ESya _ Shaye : ie xo of 4 6 8 “Torch spec vin mais to 12 20 mle ~60mmis — 80 mms 05 Temperature Ti 88 8 00 10 Temperature Tin 20am <-tOmne — aoe ‘80mms Fig. 5 — Heat input Hat plate mid-ticknese (calelaed with Equation I ond daa rom Table 2), dependent upon aloy an welding speed CT ‘eld plate thickest 3 mine Fig 6— Cooling curses (onc values a weld centerline ( 0) GTA welding plate thickeners 3 rem (eid-length and depth) with a drill hole method, Both sites of a Type K thermo- ‘couple were insulated with a two-hole ee. rama insulator and fused at their end ‘This thermocouple was placed from below {nto a hole that was drilled vertically into the weld coupon — Fig. Iright, A constant dill hole depth of 1.3 mm ensured tem- perature measurements in the middle of ‘each weld (mid-depth). The horizontal po- sition ofthe drill hole, ané hence thermo- couple, was varied in order to investigate the thermal conditions during soliitiea- tion between weld centerline and interface Fig. 2. This figure shows the cross sec- tion from Fig. 1 left under higher magnili- cation after welding. Accordingly, the thermocouple position was varied from y= 0 (weld centerline) toy = 3 mm (weld interface), Results and Discussion ‘Approximation of R As mentioned previously, the soliifi- ‘Table 3- Thermal Conductivities, (Ref. 26), Equilibrium Liguidus and Equil ation Ranges for Alloys 10508 and 6082 Temperatures (Ret, 26,37), and Solidi Parameter Alloy 20504, (at995) ‘Thermal conductivity 210 W (mK) Ligquidus temperature 657°C Solidus tempecature 615°C Soliiiation range uk ium Solidus Alloy 6082 (AISUMgMn) 170. W (rk) 650°C 550°C 100K cation growth rate R varies widely along ‘the solidification front, Ta determine R de- pendent upon the position in the weld ppool, one can use micrographs that show ‘the solidified grain structure (Ref. 10). Here, the grain morphology usualy indi cates the grain growth direction, and hhonce R during soliifieation, for any point in the weld metal, R can be calculated with Equation 8, whereby the ange cc between R and welding speed v need to be deter ‘mined dependent upon the position in the weld. Therefore, micrograph of the hori zontal cy plane from the middle of each weld (mid-length and depth) were used to approximate and consequently R. Figure 3 shows an example of this approximation for one half (regarding width) of an Alloy {6082 weld, produced with a welding speed ‘of § mm sx is the welding direction (to the left) and y is the transverse direction, ‘The black curve approximates the grain growth direction, which depended on alloy and welding speed, and which was caleu lated with Equation 9. Here, a and b are ‘Table 4~ Critica Parameters doen Reer- Gear aid ATe,cr forthe Tansition from Columnar to Equlaxed Grain Growth (CET) Parameter Unit 1 Welding speedy mae! 2 Teontent wo 3 Grain sed yo 4 cer aes 5 mms! ® Kmm! 7 Kemm* 5 K 8 mm? 0 K nob ma’ ¢1 ROA Kame Bo STocer K wat °K mame? 8 ATecer K Equation Alloy 1050, . ’ 10 i oo 02 i 3 2 ao) 748 @) of 70 : 53 30 i 3300 © rs 45554 re) 30 o er : 1208 O 3024 Alloy 082 6 sw So. 02 002 st oo as 60 8 Ta S32 aT S80 Tb 10 | 12 ets 45047 4a 1.003 20 49 sa 5 rs 7 a7 45 4n 48 [2 earcr 2014, vor 93snondimensional parameters that were ad justed in each cate on the grain morphol ogy of the corresponding weld as demonstrated in Fig. 3, ‘After approximating the grain growth direction foreach alloy and welding speed ‘according to Equation 9, the angle a be- twveen v and R was determined with Equa. tion 10, which originates from Equations 8 and 9, For purposes of simplicity, the {ain growth curvature in the vertical y-2 plane (which also influences R) was neg- lected. This is a good approximation be cause here, the grain growth direction was ‘observed to be most often parallel to the (horizontal) y axis. In a last step, R was calculated with Equa- tion 8 dependent upon welding speed v and the position in the weld pool (y). In this calculation step, it was assumed that dendrites are oriented in the same direc: tion as grain growth due to competitive growth and that grains grow normal tothe solid-liquid imerface (Ret. 22). This is con- sidered to be a reasonable approximation Fig. 5. In other words, welding speed provides in Equation 8 an order of magni- tude upper lint for R at each position in the weld pool. [Note that y isthe direction transverse to the welding direction (y = 0: centerline, y = 3 mm: weld intesface) — Fig. 3. As ‘one can see in Fig. 4, the calculated R val tues are maximum a the centerline (= 0), where they correspond to welding speed, and they are minimum at the weld inter face (y 3 mm). Interestingly, for Alloy 16082, higher minima were observed at the weld interface than for Alloy 1050. This is due to the angle o that was found atthe weld interface to be generally lower for Alloy 6082 than for Alloy 1050. ‘Thermal Conditions ‘After having determined R by means of ricrographs, the temperature measure: ‘ments revealed the thermal conditions in the weld pool. The most important influ- ceacing weld parameter is the heat input (H), which was calculated with Equation 1 ‘on the basis of data from Table 2, Figure 5 shows for both alloys that H decreased strongly with increasing welding speed, al ‘though the weld current was raised with increasing welding speeds to allow similar weld bead sizes. Asa result, the weld bead width decreased slightly with increasing welding speed, being approximately 5 to 6 A B 13 200 ¢ J? 100 | =e to0 mie 9:86 hie TEeomme, 2-42 mms S088 inms ° 1 @ : 3 vine yin me ‘Fig 7 — Cooling rte Tito huss tomperatare, dependent upon hortzontel prion im weld pool OF cot = Sm: wl ary CEA dg pa km x Al a 8 oy a “ wo) (Ebadi ' gp | Eom i : a i ede Se 6 ie S20 mms =-e-4 2 minis et “lems sbnm o (sitemt, ot ya — Fon Fig. £— Thermal gradient Gy at iquidus temperature, dependent upon horizontal postion in weld metal (y= O- centri y ram in the middle of the weld (sid-epth) ‘Also, the copper backing (Fig. 1) coa- trolled the weld bead width because it caused a dominating directional heat flow from the weld coupon downward. “To determine the thermal conditions for the whole weld pool, he position of the thermocouple was varied between y 0 and y = 3 mm (Fig. 2). The messure- nent iechnique allowed an approximate adjustment of the thermocouple position (). This approach complicated temper {ure measurements paticulasly at the weld interface (where temperature reaches liquidus temperature). This ex- plains the limited experimental data for the range betweeny = 25 mm andy = 3.0 ram in the following diagrams. The exact thermocouple position was always deter- mined through measurements after weld- ing and was related to the recorded temperature profile Figure 6 reveals the cooling curves for ‘both alloys and different welding speeds at y= 0 mm (centerline); the start tempera tures (at (15) are the maximum tempere- tures foreach thermal eele measurement. Both diagrams suggest a slightly faster cooling for Alloy 1050A than for Alloy 6082 welds. This can be explained by their different thermal conductivities (Table 3) high for Alloy 10504 due to its chemical purity and comparably low for Alloy 6082. ‘mt weld interface) GTA welding plate thickest 3 mk. ‘This suggestion is further confirmed by the cozzesponding cooling rates that are shown in Fig. 7 dependent upon the posi tion in the weld pool (y axis). These valies are the cooling rates at liquidus tempera- ture (Table 3) since the solidification starts at this moment. The cooling rates were ob- served to be maximum af the centerline (y = 0) and are supposed to be minimum at the weld interface (y = 3 mm). Figure 7 confirms the observation that high cooling rales (at centerline and at high welding speeds) generally result in finer grain structure (Refs. 28-30). This isin aevor- dance with results from the first part of this study where small, equiaxed grains formed particulary inthe weld center and thigh welding speeds (Ret. 10), In Figs, 6 to 11, each eurve (Fig. 6) or ‘each data point (Figs. 710 11) is based on at least wo single temperature measure ments at constant welding parameters. ‘The data scatter in Figs. 7 o 11 was ob- served to increase with increasing welding speed and was caused on the one side by fluctuations in the welding process. While welding speed and are current were con- stant in all welds, the arc voltage varied about + 0.2 V, dependent upon the shape ‘of the electrode ip, which fad tobe shaxp- ‘ened frequently, The resulting differences in the heat input have probably caused changes in cooling rate and other solidi WELDING JOURNAL ESA 1 1 é é | =sseeeera seme B 200 loy 6082) 160 E E = 20mm e-AT me 2 100 | |-8-G0nme = 80mmie = so100mms é Fig, 9—Thermal gradient Ga solidus temperature (y= Os cenetine'y mom weld interface). GTA welding plate thickness 3m. dependent upon horizontal postion in weld metal A aa Guin Keim 250 = a0 242 ‘GuR In Keim se 8 8 acest Fig 10 — Ratio GR at Iiguu emperature Ty dependent upon horizontal porto in weld metal (75 0: coterie: ¥ cation parameters. On the other side, banding (linear solute-rich bands in the weld metal with very fine grain structure (Ref. 28)) was observed in the weld metal ‘This is further evidence for pool fluctua- tions during welding and hence thermal conditions (Ref. 30) and cannot be avoided completely ‘An additional and likely more impor: tant point is the thermocouple, Small but ‘unavoidable fluctuations of both shape fand Vertial position of the thermocouple tip have probably caused a large part of the observed data scattering. Further: ‘more, for purposes of simplicity, the liq uuidus and the solidus temperatures (Table 3) are taken as equilibrium values that ‘may change with rapid solidification. Nev certheless, Figs. 6 to 11 offer order of mag: nitude values for solidification conditions such as cooling rate or thermal gradient at the solidification front. ‘Ina further calculation step, the cooling rates were used to determine the thermal gradient G according to Equation 2. For purposes of comparison, both gradients Gi, (alculated with dTidt at liquidus temperae ture) and Gs (calculated with Tit at solidus temperature) were determined — [2bearcr 2014, vor 93 ‘mm weld interface). GTA welding plate thickness 3 mm Figs. 8 and 9, From centerline (y = 0 mm) to Weld interface (y = 3 mm), G did aot change significantly owing to the fact that ‘both R (Fig 4) and dTidt (Fig, 7) clearly de- creased, Increasing welding speeds and de- creasing heat inputs, however, reduced Gy, Significantly (by up to 40%) — Fig. 8. This observation explains the results from the first pat of this study (Ref. 10) and former studies (Refs, 31-35), where equiaxed grains formed particularly at high welding speeds, low heat inputs, and hence, low thermal gradients. Forthermore, the com- parison between Figs. 8 and 9 shows clearly {thatthe most important stage during soi ification isthe start at liquidus temperature, regarding grain growth and subsequent grain morphology. Here, the influence of ‘welding speed/heat input on G is for Alloy (6082 significant (Gy, Fig. 8B) and it be- comes almost negligible at solidus temper- ature (Gs, Fig. 9B). Due to its low tolidiication range, commercially pure Al (Alloy 1050) does not show this pro- nounced behavior ‘The above G values, however, cannot explain solely why the grain morphology often changes from columnar at the weld interface to equiaxed atthe centerline, In- stead, the parameter G,/R can be used to explain this phenomenon — Fig. 10. For oth alloys, Gy/R was, as expected, the lowest at the centerline and the highest next tothe weld interface with an increase in between, mainly for Alloy 1050A. Fig- ture 10 further discloses that the dominant factor in GiR isthe growth rate R and not the thermal gradient G, which confirms former results on thermal conditions in Al weld metal (Ref. 35. Teas suggested thatthe quotient Gi isan indirect measure of the amount of constitutional undercooling ahead of the solidification front (Ret. 16), where low GIR values favor the transition from columnar to equiaxed grain growth (Ret, 18). Also, it was argued thatthe size ofthe constitutionally undercooled zone in- creases with decreasing G/R (Reis. 31, 36), Furthermore, a recent study about Ai weld metal grain refinement reported for base metal composition (no grain refiner additions) @ grain size drop al a critical GIR value (Ref. 21), which emphasizes the significance of GiR. These observa- tions help to relate the above thermal data to the micrographs from the first part of this stady (Ref. 10), which revealed for three different Al alloys a clear transition {rom columnar to equiaxed growth with inereasing welding speed. ‘AL this point, 1 is important to note that, beside grain refiner additions, the alloy’ chemical composition is a key [ac tor with regard to the columnar-to- ‘equiaxed transition (CET), The fsst part ‘of this study (Ref. 10) showed that the ten- dency for equiaxed growth increases strongly with increasing alloy content, due to the increasing supply of constitutional ‘undercooling through solute paztitioning uring solidification (Ref. 11) that facil: tates equiaxed grain growth (Ret. 18). Be- side alloy content, one important parameter that reflects upon the chemical ‘composition isthe solidification time A fy. ‘As was determined on the basis of cool ing curves (Fig. 6) by calculating the time period between the moments when the temperature T was T, (liquidus tempera ture) and 7; (solidus temperature). As one expects, Avy was much higher in Alloy {6082 welds than for commercially pure Al (Alloy 1050) welds — Fig. 11. This ob: servation emphasizes the need for suit cient time atthe beginning of solidification to activate the particles present for nucle. ation, It was argued elsewhere that equiaxed grains have to grow to a sufficient size in order to block columnar grain growth (Ret. 23), ‘A comparison of Fig. 11 with micro: ‘graphs from the corresponding welds (Ref, 10), however, shows the following: Above a ‘minimurn solidification time, the influence ‘of Afg on grain morphology seems to get eclipsed by the strong influence of thermalparameters such as dTiat, G and GIR. Ace cordingly, the grain morphology was pre- dominantly equiaxed at high welding speeds and thus high d7idr, low GIR and low A values (Ref. 10). Nevertheless, are- cent study on Al fusion welding revealed, at constant chemical composition, for laser ‘beam (LB) welds a three times lower solid- ification time than for GTA weld metal ‘This large disference explains why the mean weld metal grain size and the tendency for columnar grain growth were found to be ‘generally higher in LB welds than in GTA welds (Ref. 37). Columnar-to-Equiaxed Transition (CET) Iisofnote thatthe above temperature measurement technique disclosed a vata tion in the thermal condition along the solidification front, This, in tora, allows determination of eritical soliification con ditions forthe columnar-to-equiated tan sition (CET; this was accomplished is th su according tothe analytical model in Equation 3 (Ref. 28). Fist, the ential angle, CET and the critical sliiication rate Repr were determined with Equa tions 10 and 8 for two (Alloy 1050) and three (Alloy 6082) diferent welding speeds where the CET was observed — “Table 4, For this purpose, micrographs from the first part of this study (Ref 10) were analyzed and the weld metal mean train size (line 3 in Table 4) was deter ‘ined. Note that the five analyzed welds hha the same T content (ine 2in Table 4) to prevent variations inthe grain refining clements, titanium and boron, from inf encing the results. ‘Then, the critical thermal gradient Gogr was determined by comparing Fig. 8 wih the corresponding micrographs (Ref, 10) and the ratio Goer/Ropr Was calew lated (lines 6 and 7m lable 4). Afterward, these (experimentally determined) Gory vlies were compared with the analytical model thet predicts Goer (Equation 3) (Ref. 23) The fact that Goer and Ate cer depend on each other (Equations 3 and5) led to the following calculation procedure: ‘The experimental Gey values Were fist used to calculate the corresponding ci cal undercooling Acer with Equation 3 (nes 8 to 10 in Table 4). The paremeter [ATiy was taken from literature (Ref. 24) and N; was calculated with Equation 4 Afterward, ATecep Was caleslated for ‘onsparison with Bgation S (lies 11 to 13 Jn Table 4). For this purpose, Equation 5 ‘sas simplified to Equation 7, which isa very good approximation for solidification Infusion welds hecause here, Ris igh and Gislow (Refs. 23,25). The parameters D and A were taken from literatuze (Rels 24, 38) based on the suggestion that they ‘are typical values for many commercial Al alloys (Ref. 38) Aw + loy 10508 oe +20 nnie oa Tae g 06 | -e-samms > 80mm z 08 7 Be ome Ears Soae. eka oa) = Somme samme 02 Gomme 2 80mm oo Pesan. oo | elem rine rine is 11 Stein te gpd pene tion eel y= etn on weld mera) GTA welng plat cess 3 me 7° = rs A ‘> Eauianed B : © Equaved E aka 3 © ‘canoer cer cer A cer ¢ er . ‘ : E 2 4 a 2 ral ‘yp 5 inoy 10808) : [oy 02a om mH ow io @cinkimen Guinkinm {Fig 12— Predominant mierosiuctre in RG space and columnar to equiaxed anstion (CET), eae lated with Equations 3 and 7 using A (CHT) and the adused A* (CET), CET and CET" are mean his foreach alloy a a constant Ti content of 02 wt. GTA welding plate thickness 3mm ‘The two different Arccer values (ines 10 and 13 in Table 4) were finally compared to each other. As a result, both caleulations produced similar Arg cy Vale ues, particularly for Alloy 6082. This sug- gests, on the one side, that the thermat data in Figs, 60 11 and the experimentally determined Gexvalues (lin 6 in Table 4) are realistic. Furthermore, the tempera- ‘ute measuzement technique (Fig. 2) 8 ap- propriate for describing. the thermal conditions of the fusion welds of ths study ‘with sufficient accuracy. On the other side, the results show that the analytical CET approach from Equation 3, which was de- veloped for slow cooling in castings (Ref. 23), ean be applied to more rapid solidi cation in Al welds, ‘One possible reason for the different resulls from both calculation procedures regarding Arc cpr may be an overestima- tion of Np that was approximated with Equation &, A more likely reason for di ferences lies, however, in parameter 4 in Equation 7. A is a materials constant i fluenced by the chemical composition but was defined here for both Alloy L050A, and Alloy 6082 to be 20s K mm, a value taken from the literature (Refs. 24, 38). It was argued that this is a typical value for many commercial aluminum loys within the 2x, 6xex, and Trax series (Ref. 24). However, A= 2.0 °K mm may not be appropriate for commercially pure Al (Alloy 1050A), as the results in ‘Table 4 show. For this reason, a fitting eal culation step was done to determine opti- ‘mum A* values that produce in Equation 7 Atccoy* Values that are equsl to those caleulated with Equation 3, Lines 14 snd 15 in Table 4 show the obtained data, which indicate that the optimum A values ared* = 11" Kmm" for Alloy 10S0A and A* = 1.7 8K mm for Alloy 6082, ‘The low A* value for Alloy 1050A ean be related diteetly to its low solute content. ‘After having determined the critical CET conditions (Table 4), the CET was ‘modeled for both alloys in the R-G space Fig. 12. The data points in these two di ‘agrams originate from both parts of this study (Ref. 10 and this paper); they repre- sent welds with predominantly columnar ‘or equiaxed grain morphology, which were produced at different welding speeds and hheat inputs and thus different G and R val ues. The two curves in each diagram are the graphs of Equation 3, one of them cal- culated with @ constant 2.0 s!2K mm? (CET) and the other one with the ad- justed A* value (-+CET*). In other words, ‘each curve represents the critical thermal WELDING JOURNAL ESconditions (R and G) for the transition {rom columnar to equiaxed grain growth ‘One can see clearly in Fig. 12 that the CET" curve separates both equiaxed and columnar regions in the R-G space better than the CET curve, particularly for Alloy 1050A. Thus, we can summarize from “Table 4 and Fig. 12 that Equation 3 (Ret 23) is appropmite to predict the CET for aluminum GTA weld metal when A is ad justeé in Equations 5 and 7 to Suit the hemical composition, as demonstrated here. Interestingly. the adjusted CET curves in Fig. 12 ate very simular for both alloys. This emphasizes how strongly the local solidification parameters G and in fluence grain morphology, reducing the in uence’ of the chemical composition. Farthermoze, the slope of the CET curves in Fig. 12 can Be used to predict sxitical GIR values. Accordingly, te recip rocal value ofthe slope in the viinity of the data points (approximately at G, = 50 to 60 K mn) corresponds t0 the Geey/Ropy vases a Table line 7, which ate 109 Ksmm"! Teis of note that several former studies related the predominant grain morphol- ‘oy in aluminsm weld metal to welding parameters such as ate current, voltage, and welding speed (Refs 31-35)."The tem perature measurement technique and the analytical approsch (Ref, 23) used in tht, study, however, permit the prediction of the critical solidification conditions for the CET. Moreover, Fig. 12 provides impor- tant information about the extical values for solidification rte R snd thermal gra dient G that are based on experimental data, Consequently, one can now predict, the location of the CET in alsinum GTA Welds from the compasison of thermal data (Bigs. 3 to 11) with microstructural data (Ref, 10). turn, the cial welding parameters can be deduced from the en {cal solidification parameters in order to minimize or to. prevent unfavorable columnar grain stfucture in aluminum Welds, Hence, the above results a a im. portant extension of a former study on {GTA welding of AL-Cu alloys, where the CET prediction was achieved by simula tion and aot, asin this study, by the com. pasion of micrographs with results from femperature measurements (Ret. 39) Conclusions A thermal analysis of aluminum GTA ‘bead-on-plate welds (Al Alloys 105DA and (6082) with Type K thermocouples has re vealed the thermal conditions along the solidification froat of the weld pool. In comparison to the weld centerline, tem- perature measurements next to the weld interface disclosed the following: + Lower solidification growth rates (R) + Lower cooling rates (d7idt) [Ei earcr 2014, vor 93 + Slightly higher thermal gradients (@) + Ahigher solidification ime (As) With similar weld bead sizes, increas- ing welding speeds from 2 mm s* to 11.5 ram $7 decreased the heat input by up to 807% and resulted in a strong increase in ATi and a strong reduction of G, GIR. and Ars “The obtained thermal data were used together with data of the corresponding grain morphology (from the frst part of this study) to model the columnar-io- equiaxed transition (CET) according to.an existing analytical approach, This model allowed the prediction of critical R and G values, at which the CET occurs in alu- ‘minum welds Acknowledgments The authors are grateful to P. Gude from KBM Afflips BV, The Netherlands, forthe very kind donation of grain refiner (KEM AMfilips). They also would like to thank H. Streblau (ICP-OES chemical analysis) for her great support at BAM. The authors are very thankful to the Ger- man Research Association on Welding and Allied Processes of the DVS for its Support and to the Program for Funding of Industrial Research and Technology (IGF) of the German Federal Ministry of Economics and ‘Technology for funding the research project 16.242N, References 1. SehomppP Cons, ., Bn, A and Rethineen M 2013 atvats of ruse on techni properties of alunintts GIA weld Stal Welg the Wort 373) 303-308 2 Jana Ham, D. Mit TK, Shank ¥, and Sundari, $2108 Nirostral Encment though nocltion of pe 720 A Zorg aly wel sd eetet on ot eacing tea properties Jounal of Material Pre aig Tecnology WO) TE 3S Schenpn, P, Crom, CB, Sebwenk, C. and Rothe, M3012 laisse of Hand B Miionson gain ave and welby of frum ally 6082 Teding nthe Word Soe). 95108 4. Mouse MG, Coss, and Grong ©, 198 fect of seid vad tatu horob tu grin efnement and ht seckng of sina lo 7108 Scece and Teele of Tiling and ning € (0): 38148 5 Schon, Ps Crom, C:,Sewenk C sod Retin M2011, Weld etl gan fnement of sminom Alloy 3083 through eon trolled addons of Ti ané BMP Marenats Teng 530) 0-50, Br Esson M.A. an St obo, Df 2001 {A model of gai tefinemcat inonpratng Sly consonant of etree ont flea aster ht Mena 000) isomers 4. Cn, A. 199,The mechani of en sine fan cuts nln sone Seumal of he nate of Mears). 32 6 Crossley, F.A., and Mondolfo, L. F, 1951. ‘Mechanism of grain refinement in aumsinum ae loys. Journal of Metal 191: 11831148, 5, Mohanty, PS, and Gruslesk, J. 1995 ‘Mechanism of grain refinement in aluminum. Acta Metallugica et Moteriaha43(3) oot-2012. 10, Schempp, P, Cross, C.B. Pittner, A, (Oder, G., Salivan Neumann, R., Rooch, arte 1, Ostere, W, and Rethmeier, M2014 Solidification of GTA aluminum weld metal: Part 1— Grain morphology depeadeat upon alloy composition and grain refiner coatent. Welding Journal 95(3) 53-10 58. 1 Rater, J. W, and Chalmers, B. 1953, prismatic substructure formed during solidi ation of metals. Canada Journal of Physic 33 15538, 12, Backerud L, Kr, , and Tamminen, J. 1985, Soluifeation Charecterticr of Al ‘minum Alloye — Val, Wrought Alloys. Oso, "Norway, Skanafuminin, 13, McCartney, D. G. 1989. Grain cefining of ajusinum and its alloys using inocula Jn femational Materials Resiens 34(8): 247-260, 14, Sebloz, .D. 2010. Fundamentals of ‘ran efning aluminum alloys. Light Metal Age 43007. 15, Daves, G.I Solidiication strut son weld, Intermtional Metallurgical Reviews 20 8-106, 1 Tiller, W.A, and Ratter,1.W.1956,The «fet of growth conditions upon the soldifiese tion of a binary alloy. Canadian Journal of Physic 3: 96-121 7 Tiller, W.A., Jackson, K.A., Rutter. \W, and Chalmers, B, 1853. The redistribution offolute atone during the slidiation of et- als. Acta Metallurgica 8) 428-437. 18. Winegard W-C., and Chalmers, B.1954, Supercooling and dendritic freezing in alloys Trancactons ofthe ASM 48: [214-1204 19. Savage, W.F, Lundin, C. D., and irubee, KJ. 1968, Segregation and hot crack- ing in Towalloy quench and tempered steels eling Journal $7: 2s to 425+ 20. Kou, 8, and Le, ¥. 1986, Nucleation ‘mechanisms and grain refining of weld meta. Welding Journal 95(12) 305-80 3158, 21, Schemapp,P, Cross, C. , Pitne, A. and Rethmeier, M. 2013, lnfvence of solute ‘content and solidification parameters on grain refinement of aluminum weld metal. etal (geal and Matenaly Transactions 6 44(T) Si9s-a2. 22. Savage, W-F, and Aronson, A. HL. 1966, Preferred orientation inthe weld fusion zon. Welding Journal 45(2) 85-510 88, 25. Hust J-D. 1984, Steady state columnar and equiaxed growth of dendnites and eutectic Material Science and Enginecring 85 (1). 75-83 24, Grong, @, and Cross, C. B, 1999, A model for predicting weld metal grain refine- ‘ent ia G- space. Proc Materials Research So ety Spot 578, 9p 431138, 25. Burden, MH, and Hunt, J.D. 1974 cellular and dendritic growth H. Joumal of Cotal Growth 2: 108-116 2b. Hee, W 2008, Aluminum Schell minim Verlag, Disseldor, Germany 27. Balitchev, E2004, Thermochemsieehe wad ineliche Modelierang ir Lepierungsauswahl mehsphasiger Systeme far des Thivoforming und zur Optimierang ihrer Formgebungsprozeste. PHD disertation. Aachen, Germany, RWTH.24, Fleming, M.. 1974 Soliton Pro cesing. New York, NY, McGraw-Hill 29, Manta, A. 1985. Mirostruetue of rap- idly soldiied laser molten AI45 We Pet Cu Susfaces. Metallurgical Transactions B16 (1) 18-161 30. Kou, $. 2003, Welding Metallurgy. New etsy, NI, Wiley. 31. Afala, Y., Matsuda, F, and Mats, A 1974, Bifect of welding condition om slides sion structure in weld mctl of aluminum alloy sheets. Transactions ofthe WRU: 8-57, $32. Ganuha,T, Pearce, B.P, and Ker, \W_ 1080, Grain structures in aluminiom alloy GTA welds, Metallurgical Tansactions A 11 1351-1358. 33, Kou, S, and Le, ¥. 1988. Welding pa- rameters andthe graia structure of weld metal —Athermodynamic consideration, Metal al Tansactions A19 (3) 1075-1082. 34, Yuajia HL, Frost, RIL, Olson, D. Ly and Edwards, G.R. 1989, Gain eefinement of Sluminum weld metal. Welding Journal 68(7) 28045 t9 2805. 35. Dvorak, M.J, Frost, H.and Olson, D. L, 1991. Influence of solidification kinetics fon aluminum weld grain refinement. Welding Jounal 7010) 271-810 276, "6. Kato, M, Matauda, F, and Senda, T 1972, Solidication mode in amin weld smelal Pansactons ofthe Japan Welding Sociey 30):2533 537, Schempp,P, Tang Z, Cross, © Pate ter, A, Seetel, T, and Rethmeies, M_ 2012. Iniluence of hate metal and roidiication po- rameter on grain refinement in alum weld etal due to noeslation, Proc. Int Conf on Tends in Welding Research, pp. 98-107 “38 Kurz W. and Fisher DJ. 1986. Funda mental of Solidification. Zesich, Switzerland, ‘Tans Tech Publiations. 138. Clatke, J. 1988, Columnarto-quioxed ‘run transition in gas cangsten are welds in al smimum-coppe alloys. PAD dissertation. Water Jo, Canada, Universiy of Waterloo, Appendix Lit tambo Spiel Dinan oo tee oo 7 sn D mm? s Geer Kea som 1 A k Ne x meee Aly v ye! Meaning Angle between ditectons of v and Material constants Grain growth direction parameters Chemical composition ofthe alloy Grain size Liquid difusioncoeticient Cooling rate Thermal gradient (local) ‘Thermal gradient at liquidus temperature ‘Thermal gradient at solidus emperature (Critical thermal gradient for CET Heat input Arc current Partition coefficient (between solid and guid) Slope oflguidus line “otal numberof heterogeneous substrate particles Solitcation growth rate Curvature undereooling Soldieation tine ‘emperatare Liquids temperatore Sodus temperature Constitutional undercooling Critical constitutional undercooling for CET Undercooling required for nucleation ‘Undercooling required for steady state grain growth Are voltage Welding speed Welding direction Horizontal direction perpendicular to welding ditection WELDING JOURNAL ES