Studying the effect of lubricant on laser joining of AA 6111 panels with the addition of AA 4047 filler wire

Accepted Manuscript Studying the effect of lubricant on laser joining of AA 6111 panels with the addition of AA 4047 filler wire

Guang Yang, Junjie Ma, Hui-Ping Wang, Blair Carlson, Radovan Kovacevic PII: DOI: Reference:

S0264-1275(16)31529-5 doi: 10.1016/j.matdes.2016.12.014 JMADE 2559

To appear in:

Materials & Design

Received date: Revised date: Accepted date:

19 October 2016 23 November 2016 4 December 2016

Please cite this article as: Guang Yang, Junjie Ma, Hui-Ping Wang, Blair Carlson, Radovan Kovacevic , Studying the effect of lubricant on laser joining of AA 6111 panels with the addition of AA 4047 filler wire. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Jmade(2016), doi: 10.1016/ j.matdes.2016.12.014

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Studying the effect of lubricant on laser joining of AA 6111 panels with the addition of AA 4047 filler wire Guang Yang, Junjie Ma, Hui-Ping Wang, Blair Carlson, Radovan Kovacevic Abstract: This paper explores the feasibility of laser joining of aluminum alloy 6111 panels with the

IP

T

presence of lubricant used in the preceding stamping operation. The effects of two commonly used

CR

automotive lubricants, Bonderite L-FM MP-404 and Ferrocote 61 MAL HCL lubricant, on the weld quality were investigated. Images captured by a CCD camera showed that the presence of lubricant

US

caused the poor wetting and spreading of the filler wire. The decomposition of lubricant formed hydrogen pores in the weld, promoted the plasma plume with a high intensity, and elevated the temperature of the

AN

molten pool. The simulation of rapid solidification revealed that the formation mechanism of pores was related to the temperature distributions and the solidification rates. The comparison of effects of diluted

M

lubricants in different concentrations on the weld surface quality and the mechanical properties provides a

ED

reference on an acceptable level of lubricant on the stamped panels prior to welding.

AC

CE

PT

Keyword: Lubricant; laser joining; aluminum alloy

1

ACCEPTED MANUSCRIPT 1 Introduction Lubrication is a vital part of the surface conditioning sequence during the forming of aluminum alloy sheets. Before the assembly/ joining in auto manufacturing, all sheet products are used in a prelubricated form. Inadequate lubrication can lead to galling of the workpiece in the press-shop [1].

T

Automotive manufacturers prefer to weld panels as received because post-cleaning of lubricants is costly.

IP

However, little information is provided about the feasibility of laser joining of aluminum alloy 6111

CR

panels with lubricant, as it is always suggested to remove hydrocarbons thoroughly to prevent the introduction of gas bubbles [2]. Although gas pores are often rendered as detrimental defects to

US

deteriorate the functionality of the industrial products, the application of porous welds as the lightweight non-load-bearing components could be possible if the mechanical properties are not inferior significantly.

AN

Therefore, it worth the effort to explore the effect of lubricants on weld quality. Hydrogen is the only gas known to solute noticeably in either solid or molten aluminum [3]. The

M

solubility (S) is a function of temperature (T), pressure, and alloy composition. The widely-used empirical

ED

equations for calculating the solubility of hydrogen are proposed by Ransley and Neufeld (Eq.s 1&4 in Fig. 1) [4], Opie and Grant (Eq. 2 in Fig. 1) [5], and Pebler (Eq.s 3&5 in Fig. 1) [6]. Over the past decades,

PT

the quantitative study of the dynamics of bubble nucleation, the driving force for bubble growth, and

CE

bubble entrapment and escape during solidification has already provided insights into the formation of porous materials. Analytic solution models [7], criteria function models [8, 9], and Stokes flow models

AC

using Darcy's laws [10, 11] assume that the primary mechanism of pore growth is the volumetric shrinkage that a feeding of liquid cannot compensate for. Lee [12, 13] investigated the kinetics of pore development in Al-Cu alloys using an X-ray temperature gradient stage and found that gas-diffusion models instead of Darcy's law- based models matched with the experiments. Since the shrinkage pores and keyhole-induced pores were not observed during the fusion-brazing experiments, the explanation of gas-diffusion models [14, 15] is preferred to illustrate the lubricant-induced pores.

2

ACCEPTED MANUSCRIPT This study compares the effect of two commonly used automotive lubricants in North America, Bonderite L-FM MP-404, and Ferrocote 61 MAL HCL lubricant on the quality of laser fabricated aluminum alloy joints. In order to illustrate the influence of temperature field and the solidification rates on the formation and growth of bubbles inside the molten pool, the transient heat flows were analyzed by

T

the three-dimensional finite element thermal models using ANSYS. The fundamental processing-

M

AN

US

CR

IP

microstructure-mechanical property relationship of the porous aluminum welds was explored.

ED

Figure 1 Solubility of hydrogen in molten aluminum and solid aluminum with 1 atom of hydrogen over sample according to Ref. [4-6].

PT

2 Materials and methods 2.1 Lubrication

CE

Lubrication on the degreased panels was conducted in a vent hood as shown in Fig. 2(b). Aluminum alloy 6111 panels (1.2 mm thick, 306 mm long, 129 mm wide) were washed with dish soap

AC

and then dried with a shop towel. The average surface roughness (Ra) of a clean panel was 0.275 µm (Fig. 2(a)). Low odor mineral spirits were sprayed to increase the lipophilic property of the substrate. Bonderite L-FM MP-404 (B lubricant for short, Henkel corporation) or Ferrocote 61 MAL HCL (F lubricant in abbreviation, Quaker chemical corporation) were mixed with x parts (x=1, 2, 3, 5, 7, 9) of mineral spirits (MS) to make different solvents. The toxicology reports presented that the Bonderite MP-404 has 30-60% distillates, hydrotreated heavy naphthenic and sulfonic acid; a boiling point >100

; a flash point: 137.8

; and under decomposition, this product emits carbon monoxide, carbon dioxide, and hydrocarbons. 3

ACCEPTED MANUSCRIPT Ferrocote 61 MAL HCL has 5-10% petroleum distillate, 1-5% sulfonic acids and sodium salts, <1% sulfonate, and <1% mineral oil; a flash point: 204

; and a fire point: 212.8

. The petroleum distillates

M

AN

US

CR

IP

T

contain 30-60% severely hydrotreated heavy naphthenic and less than 30% hydrotreated light naphthenic.

PT

ED

Figure 2 (a) Surface roughness profile obtained by the profilometer; (b) experimental setups for the lubrication application; (c) schematic illustration of joining AA 6111 coach peel panels with the addition of AA 4047 filler wire. Eight degreased panels in a group were submerged in the above solvents, respectively. Then they were handled by their edges to minimize the removal of the lubricant layer. After drying for 1 day on a

CE

rack, the panels were placed horizontally to minimize the lubricant loss. The film weights were measured

AC

by an analytical balance (Veritas S423) by the weight- strip- weight method and determined by the following formula:

Film weight per area= (wet weight – dry weight)/ (2× length × width of the panel)

(1)

Most organic solvents and mineral oils are Newtonian fluids [16], so any concave part on the surface was filled instantaneously. The film weight depends on the viscosity and polarity of the lubricants [17]; a high polarity is able to form an effective adsorption film while a low viscosity results in the dripping of lubricants by gravity. Fig. 3 reveals that the pure F has a stronger film formation capability than B, but due to the low viscosity the film weight decreases more quickly with the increase in dilution. 4

ACCEPTED MANUSCRIPT The heaviest film was achieved by using 1 part of lubricant+ 1 part of MS rather than the pure lubricant. This might be because mineral spirits could increase the polarity of the lubricant film, but the viscosity of the diluted film was reduced at the same time. After adding more than 3 parts of MS, the film weight per area dropped below 2 g/m2. For simplicity, panels with 1 part of B lubricant+ x part of MS (x=1, 2, 3, 5, 7,

M

AN

US

CR

IP

T

9) are called the 1BxMS panel in this paper, and the generated weld corresponds to 1BxMS weld.

Figure 3 Coating weights of Bonderite L-FM MP-404 lubricant and Ferrocote 61 MAL HCL

ED

lubricant

2.2 Joining process

PT

The cross-beam continuous wave fiber laser (IPG laser) with a wavelength of 1070 nm was

CE

applied to do the laser fusion-brazing experiments. The core diameter of the optical fiber, the focal lengths of the collimating lenses and the focal lenses were 0.4 mm, 150 mm and 200 mm, respectively. The

AC

processing parameters were adopted from our previous study of parameter optimizations to achieve the best weld surface quality on the degreased panels. Dual beam with a power density distribution of 50 /50 was arranged side-by-side with respect to the groove. Each spot diameter was 0.962 mm at the defocused plane as measured by Beam Viewer. A BINZEL precision wire feeding system was installed at the KUKA robot and the aluminum AA 4047 filler wire with a diameter of 1.6 mm was positioned at an angle of 45° from the horizontal plane (Fig. 2(c)). All experiments were conducted at a scanning speed of 60 mm/s, a wire feed rate of 75 mm/s, and a laser power of 4 kW. The molten pool was shielded by the coaxial and 5

ACCEPTED MANUSCRIPT lateral streams of argon, both of which had a flow rate of 14.163 L/min (30 SCFH). The lateral argon shielded the molten pool at an angle of 55° from the horizontal plane trailing the dual beam (Fig. 2(c)). Experiments of each surface condition were repeated 6 times to calculate the occurrence of pores and the average surface roughness.

T

2.3 Analysis techniques

IP

The surface roughness of a clean degreased surface was measured by a Nanovea Profilometer. The sectioned coupons were mounted by a hot-press mounting machine (ALLIED), and etched by the

CR

Keller’s reagent (1% HF + 1.5% HCl + 2.5% HNO3 + 95% distilled water). The microstructures were

US

observed by an optical microscope and a Leo-Zeiss 1450VPSE scanning electron microscope in both secondary and backscattered electron imaging modes (beam current: 10 mA; working distance:15 mm;

AN

acceleration voltage: 25 kV).

An Ocean Optics spectrometer (SD2000) was applied to detect the element composition in the

M

laser induced plasma: the integration time was 3 ms, and the slit width was 50 mm. The Boltzmann-plot

ED

method was introduced to calculate the electron temperature of plasma coming from aluminum and magnesium during joining of the panels with pure B and F lubricants. In order to compare the

PT

temperatures conducted to the base metal during joining, the procedure was as follows. Three surface

CE

conditions (degreased, pure B and F lubricants) were prepared on one pair of panels, and then thermocouples were placed at the distance of 1 cm from the weld centerline in each section (Fig.4(a)). In

AC

this way, the position of thermocouples with respect to the weld centerline and the beam focusing characteristics can be strictly kept constant. Finally, temperatures were measured during joining. Each joining process was carried out only when the panels were cooled down to room temperature. The tensile tests of coupons were conducted by an Instron Universal Materials Testing System at a constant cross-head speed of 1 mm/min up to the final failure. Ten bone-shaped coupons from each condition were cut by an abrasive waterjet machine, and the extension direction of each tensile sample was vertical to the weld bead. The dimensions of the coupon are shown in Fig. 4(b). 6

ACCEPTED MANUSCRIPT

T

Figure 4 (a) Positions of thermocouples; (b) dimensions of a tensile test coupon.

IP

2.4 Simulation of the lubricant-free weld

CR

In order to obtain the temperature distribution and the distribution of solidification rate along the molten pool boundary, the thermal simulation of joining of the lubricant-free panels was conducted in

US

ANSYS Mechanical APDL 16.2. After extruding the 2D model for 40 mm, an 8-node hexahedral element

AN

(SOLID70) was implemented for the thermal analysis (Fig. 6(a)). Due to the axial symmetry, only a half of geometry was modeled. The model of the cross-beam joint contained 211,820 elements and 238,164

M

nodes; the dimensions of the smallest element in the molten pool and the largest element in the flange portion of the joint were 0.09 mm×0.09 mm ×0.09 mm, and 0.3 mm (in the angular direction) ×0.18 mm

ED

(in the radial direction) ×0.09 mm (in the length direction), respectively. The temperature dependent

AC

CE

PT

thermal properties of AA 6111 and AA 4047 were cited from [18-20] as shown in Fig. 5.

Figure 5 Material properties of the AA 6111 and AA 4047 [18-20] The overlapped length of the two beams was 0.262 mm (22.66% of the beam spot diameter). The rotary Gaussian heat source model [21] was used for each laser beam: (2) 7

ACCEPTED MANUSCRIPT where η is the energy transfer efficiency coefficient; H is the height of the heat source, and can be considered as the depth of the fusion zone; y varies from 0 to H; v is the welding speed of 0.06 m/s; R0 is the radius of spot that each beam can generate on the material surface; x 0 is one-half of the inter-beam space, and x0 =(50%-22.66%)×2 R0 =0.5468 R0. Q’ is 2 kW for the cross-beam laser. The accuracy of the

T

simulation was validated by matching the predicted weld shape with the real one first (Fig. 6(b)). Then,

IP

the temperature histories at positions of 3 mm and 4.4 mm from the weld centerline were measured by the

CR

chromel -alumel (K-type) thermocouples. As Fig. 6(c) presents, the temperature profiles during cooling

CE

PT

ED

M

AN

US

match well with the data obtained by the thermocouples at the same locations.

AC

Figure 6 (a) FE model. Verification of the simulation results through the comparison of (b) the joint shape, and (c) temperature profiles.

3 Results and discussion 3.1 Feasibility of fabrication and surface quality In the automotive industry, coach peel joints prepared for the further painting require no defects on the flat weld surface. Therefore, the wetting and spreading, the occurrence of pores on the surface, surface roughness, and soot were investigated for joining panels in different surface conditions. Joining the degreased panels generated smooth weld surface without any defects. The sufficient wetting and 8

ACCEPTED MANUSCRIPT spreading indicated that the liquid filler metal was capable of dissolving or alloying with the substrate on which it flows [22]. In contrast, intermittent bonding, rough ripple edges, and holes could be formed during joining of lubricated panels (Fig. 7(a)). Besides, the occurrence of porosities on the weld surface rises with the increase in weight of the lubricant film (Fig. 7(b)). Images captured by a high-speed CCD

T

camera during joining reveal that the molten pool surface was quite agitated, the molten filler material

IP

was inclined to ball up, and the spreading to create the smooth weld edge was insufficient (Fig. 7(a)). The

CR

low surface quality might result from the poor wettability of metal on the decomposed lubricant itself or the escaping bubbles that decomposed from the lubricant on the panel surface.

US

Filler wire spread well on panels with B lubricants at different concentrations except for pure B and 1B1MS. In comparison, the F lubricant made joining more challenging than the B lubricant.

AN

Sufficient spreading could barely be achieved on the pure F, 1F1MS, and 1F2MS panels. The molten filler metal could detach to one side of panel randomly due to the combination of effects of poor wetting and

M

the existence of lift force exerted by the shielding gas. Lower concentrations of F lubricants made joining

ED

easier, but still harder than B lubricant with the same dilution. The formation of bubbles not only agitated the molten pool flow but also disturbed the arrays of surface ripples. Fig.s 7(b-c) shows that B lubricants

PT

have fewer tendencies to form pores on the weld surface, and the surface roughness of 1B5MS, 1B7MS,

AC

CE

and 1B9MS welds was close to that of the lubricant-free welds.

9

AN

US

CR

IP

T

ACCEPTED MANUSCRIPT

M

Figure 7 (a) Molten pools captured by a CCD camera. (b) The probability of occurrence of pores on the weld surface; (c) surface roughness of welds.

ED

Soot occurred at the weld surfaces when the panels were lubricated. Hee-keun Lee et al.[23] found that the soot is a deposit of metal oxides that predominately consists of aluminum and magnesium

PT

oxides, with trace quantities of carbon by using the EPMA mapping. The vapors of metallic oxides

CE

condensed on the panel surface because they were heavier than the air [1, 24]. Considering that the identical gas parameters were applied to joining the degreased panels but no soot was found, we can

AC

conclude that the discrepancy was attributed to two factors. First, the shielding effect changed. The combustion of hydrogen bubbles escaping from the molten pool changed the local temperature and gas pressure dramatically, so there was a significant entrainment of air into the argon jet and finally influenced the condensation of oxides. Second, a larger amount of aluminum and magnesium in the molten pool were evaporated and ionized because of the exothermal reaction of lubricants. Although it is hard to confirm which factor is dominant, the assumption regarding a significant temperature change

10

ACCEPTED MANUSCRIPT could be validated by calculating the electron temperatures of plasma and measuring the conducted temperatures through thermocouples. 3.2 Plasma inspection and temperature measurement The emission spectrum of the plasma was recorded in real-time by a spectrometer during joining

T

panels. The atomic hydrogen emission line at 486.13 nm was recognized from the monitoring signal

IP

according to Ref.s [25, 26]. Plenty of molecular hydrogen emission lines were found by the SpecLine

CR

Spectroscopy Software (Fig. 8(a)). As reported, the dual-beam laser could decrease the fluctuation frequency of plasma dramatically [27]. The spectrum during joining degreased panels were gradually

US

stabilized to a low intensity (Fig. 8(b)). In comparison, joining panels with lubricant exhibits the

AC

CE

PT

ED

M

AN

following characteristics:

11

ACCEPTED MANUSCRIPT Figure 8 (a) The spectrums of the hydrogen emission line during joining panels with pure B (weak signal) and pure F (strong signal). (b) The spectroscopic monitor of laser-induced plasma plumes. The maximum scales of intensity are 5000 and 600 counts in F and B lubricant groups, respectively. First, the spectral intensities under different surface conditions follow the sequence: F lubricants to B lubricants at the same dilution to degreased. The plasma intensities in the F lubricant group (except

T

for 1F9MS panels) exceeded the maximum detecting level of 4000 counts; whereas, the peak spectral

IP

intensities in the B lubricant group were below 500 counts. Second, stronger fluctuations of spectra were

CR

detected during joining panels with lubricants than for the degreased case (Fig. 8(b)). The plasma fluctuation indicates that these hydrocarbons made the joining processes unstable. Third, the average

US

plasma intensity affected by the lubricants decreases with the increase in dilution. The difference in

AN

spectral intensities reveals that more plasma was excited in the F lubricant group than B lubricant at the same dilution. As radiation, convection, and conduction are the main approaches to lose energy during

M

welding, the detection of plasma electron temperature and the temperature measurement by thermocouples can illustrate the difference in the released energy.

ED

The plasma electron temperature Te can be calculated by choosing two emission lines using Eq. 3

the wavelength

and

PT

[28]. The selection of emission lines satisfies a condition on the upper energy levels , expressed by

and

for

. Based on this relation, the

AC

CE

chosen emission lines of Al and Mg are presented in Table 1.

where k is the Boltzmann constant; E is the emitted energy;

(3) is the statistical weight of the upper level

(g= 2J+1); and J value represents the total electronic angular momentum of the upper energy level that can be found in NIST Atomic Spectra Database [29]. Joining panels with pure B lubricant produced higher electron temperatures of Mg and Al than the corresponding case of pure F lubricant (Fig. 9(a)). As Ma et al. [28] illustrated, the fluctuations of the calculated electron temperature are related to the generation of defects that disturb the stability of plasma 12

ACCEPTED MANUSCRIPT plume. Fig. 9(b) shows the temperature profiles measured at the same distance to the centerline of the molten pool. B lubricant created a higher peak temperature than F lubricant, indicating a stronger exothermal reaction and higher released energy of B lubricant. Consequently, the decomposition of B lubricant should excite a larger amount of metal vapor as plasma than the reaction of F lubricant. This

T

conclusion is not contradictory to the higher plasma intensities in the F lubricant cases shown in Fig. 8(b),

IP

because plasma includes ionized atoms of metal vapor, shielding gas (argon), and the ambient gases. The

CR

excited hydrogen plasma relies on the decomposed amount of hydrogen that strongly depends on the molecular structures, the cations, the types and length of alkyl chains, and the anions inside the lubricant

US

[17]. Fig. 8(a) shows that the emission lines of hydrogen were intense with the presence of pure F lubricant. The plasma intensities were dominant with ionized hydrogen when the panels with F lubricants

AC

CE

PT

ED

M

AN

were joined.

Figure 9 (a) Temperature profiles obtained by thermocouples; and (b) calculated electron temperature profiles of Mg and Al during joining of the panels with pure lubricants. The released energies of lubricants have two effects on the weld surface quality. First, the increase in temperature can improve wetting and spreading [30]. This is one reason why the occurrence of insufficient bonding in the B lubricant group was leaner than in the F lubricant group. Second, plasma plume has a higher intensity surrounding the laser beam; therefore, a pressure gradient exists. The 13

ACCEPTED MANUSCRIPT pressure differences result in an unbalanced surface and cause the perturbation on the molten pool surface [31]. Compared to the F panels, B panels generated weaker plasma. The induced oscillations on the molten pool surface were weaker, and the surface roughness of the B welds was lower.

CR

IP

T

Table 1 Spectroscopic information of Al and Mg to calculate Fig. 9(a) [29] Selected Statistical Transition probability, Energy of the upper Wavelength, element weight, (s-1) level, (eV) (nm) Al I 308.21 4 5.87e7 4.0214836 Al I 396.15 2 9.85e7 3.1427212 Mg I 383.83 5 4.03e7 5.9459135 Mg I 517.26 3 3.37e7 5.1078269 3.3 Formation of hydrogen bubbles

US

Both lubricants are distillates (petroleum) containing hydrotreated heavy naphthenic. Under hightemperature conditions, there is always a high possibility of thermal cleavage of a hydrocarbon chain,

AN

especially when the availability of oxygen is limited [32]. R(CH2)6R → 2[CH2CH2−CH2·] → 2RCH2CH=CH2 + H2

(4)

M

where R refers to a long-chain alkyl substituent [32]. The above formula will continue to break down into

oxygen above 120

ED

unsaturated molecules with lower molecular weight and higher volatility. If a lubricant contacts with , the cleavage of hydroperoxide will happen, and the hydrocarbon will generate CO/

PT

CO2 and a lot of volatile low molecular-weight oxidation products [32, 33]. For example, ethers

CE

(RCOOR) are formed when the rate of oxidation becomes limited by diffusion [32]. A transition metal such as Fe, Cr, Cu, or Mn in AA 6111 panels will become catalysts to accelerate the hydrogen abstraction

AC

[32]. According to Denisov [34], the degradation of mineral oils at high temperature went through four distinct stages described by the well-established free radical mechanism: (1) initiation of the radical chain reaction; (2) propagation of the radical chain reaction; (3) chain branching; (4) termination of the radical chain reaction. The effective oxygen pressures in the Al-O system are around 10-2 Pa at 1330 K and 10-9 Pa at 930 K [35]. Hence, it is hard to avoid the oxidation during laser joining. Based on the analysis above, the generated bubbles comprise a majority of hydrogen and a small portion of insoluble gases that might consist of low-molecular-weight oxidation products. The pressure in bubble, Pb, is given by [36] 14

ACCEPTED MANUSCRIPT (5) where

is the partial pressure of hydrogen gas above the molten material;

insoluble gases in the gas mixture;

is the partial pressure of

is the hydrostatic pressure at the solid/liquid interface; and

is

induced by the curvature of the gas /liquid interface.

T

The calculation of the mass of hydrogen is given by the supersaturation of hydrogen in liquid and

where

is the mass fraction of hydrogen, and and

are the concentrations of hydrogen in wt.% and

shown in Fig. 1. When all the liquid transforms into solid,

and

are the solubilities

drops from 26 cm3/100 g Al at

(obtained from Eq. 1 in Fig. 1) dramatically to 0.012 cm3/100 g Al at 400

AN

2000

(6)

are the volume fractions of liquid and solid,

US

respectively.

and

CR

IP

solid [37]:

(obtained from Eq. 4

M

in Fig. 1). Solubility of hydrogen is a function of temperature. As for laser welding, temperature distribution inside the molten pool evolves with time. Therefore, the time-dependent temperature field is a

ED

crucial factor to influence the bubble formation, and further decides the bubble distribution and

PT

morphology.

Fig. 7(a) reveals that no keyhole could be observed, and joining of lubricated panels was still in

aluminum 2467

CE

the conduction mode; therefore, the peak temperature was lower than the boiling temperature of [38] (2457

for aluminum alloys 6XXX [39]). The thermal simulation of cross-beam

AC

laser joining of degreased panels in Fig. 10 shows that the peak temperature can reach 2385 ; cooling takes 0.6615 s from the peak temperature of 2385

to 360

at the centerline (position A in Fig. 10(a)).

If a finite time is considered, the generation of pores is as follows.

15

US

CR

IP

T

ACCEPTED MANUSCRIPT

AN

Figure 10 (a) 3D view of the simulated lubricant-free weld and temperature histories at the centerline (position A) and the fusion line (position B) of the molten pool; (b) distribution of the

M

solidification rate along the molten pool boundary in the longitudinal section.

ED

Because a part of the dual-beam laser was located ahead of the filler wire along the welding direction, the thermal cleavage of hydrocarbon chains and the melting of filler wire started simultaneously. , a majority of hydrogen generated at the molten pool boundary

PT

At a peak temperature higher than 2000

is soluble into the aluminum molten metal while the insoluble gas becomes bubbles floating with the

CE

Marangoni flow. As laser moves away, the temperature quickly drops, and a majority of hydrogen atoms

AC

are partitioned between the liquid and solid phase of the molten pool. The solute atoms, stable and metastable precipitates, high-angle grain boundaries, and lattice interstices act as trap sites for hydrogen supersaturation [40]. Besides, the unavoidable oxide layer on the molten pool surface occludes the permeability of hydrogen from the aluminum to the air [41]. Therefore, the concentration of hydrogen level in the liquid continues to increase until the bubbles precipitate. Besides the temperature-dependent solubility, solidification rates inside the molten pool also influence the nucleation of pores. Whether hydrogen pores will form depends on the local hydrogen concentration levels and the rate of diffusion [42]. The increase in solidification rates gives hydrogen a 16

ACCEPTED MANUSCRIPT shorter time to diffuse, and the supersaturation will contribute to the new precipitation on entrapped nonsoluble gas or inclusions [42, 43]. In other words, the number of pores flourishes with the increase in the solidification rate, similar to the nucleation of grains. The distribution of the solidification rates (Vn) along the molten pool boundary was obtained according to Eq. 7 [44, 45]:

IP

where Vb is the welding speed along the welding direction (z-axis);

T

(7) are the three

CR

components of the temperture gradient. Fig. 10(b) shows that the solidification rates at S/L interface in the

US

longitudinal section increase as the location is closer to the molten pool surface. Therefore, more bubbles are distributed in the upper part of the fusion zone. Besides, the solidification rates near the molten pool

AN

boundary are higher than in the center. Due to the limited diffusion, a multitude of tiny pores congregate near the weld boundary, while larger pores are present in the center of molten pool (Fig. 11(a)). In

M

summary, increasing the solidification rate on the solid/liquid interface resulted in a decrease in the pore

ED

size and an increase in the number of pores per unit area. If just temperature distribution inside the molten pool is considered, the result would be the opposite. Thus, the balance between temperature-

PT

dependent solubility and the distribution of the solidification rates affects the formation of bubbles inside

AC

CE

the molten pool.

17

ACCEPTED MANUSCRIPT Figure 11 (a) Weld formed on 1B5MS panels; (b) microstructure of 1B5MS weld at the fusion /brazing interface shows dendrites in the fusion zone and Si particles in the brazing zone; (c) a sketch of the inward flow pattern from the boundary to center of the molten pool [46]; (d) pore size

AN

US

CR

IP

T

distribution.

Figure 12 (a) Weld generated on panels with different surface conditions (left: degreased; right:

M

brushed with the blend of pure B lubricant and alumina); (b) alumina (light color) congregated in the upper section of the molten pool ; (c) and (d) SEM image in the backscattered electron mode

ED

showing the micro-segregation along the isotherms.

PT

3.4 Growth of bubble in the Marangoni flow The pore morphology can be defined with the following parameters: pore size, pore aspect ratio

CE

between the length and the diameter, pore orientation, and porosity [47]. Ostwald ripening and migration/coalescence of the gas surface are the two fundamental growth mechanisms that determine the

AC

morphology of finely divided two-phase systems. Toda et al. [40] applied the X-ray tomography and found that Ostwald ripening governs the growth behavior of hydrogen micro-pores in Al-5.5 mol. % Mg alloys; the porosity is related to but not linearly proportional to the total hydrogen content. Moreover, the ratio between the maximum radius of pores to the average radius was reported to be around 2.23 in this mechanism [40]. Digital image processing of 124 pores in two cross-sections of 1B5MS weld by using Matlab shows that these closed pores have an average aspect ratio of 88.45%, the radii are in a range of 12.6 μm to 42.1 μm (Fig. 11(d)), and the weld has a porosity of 4.41%. The cross-sections of pores are 18

ACCEPTED MANUSCRIPT spherical, and no connected and engulfed pores could be observed. Except for the exceptionally high occurrence of radii between 12.6 μm and 14 μm, which might be affected by the insoluble gases, the distribution of pores follows a mean field approximation with an average value of 19 μm. A calculated ratio of 2.16 between the maximum radius of pores to the average radius further suggests that the pore

T

growth behaviors are dominated by Ostwald ripening.

IP

The distribution of pores in the molten pool was inhomogeneous but not disordered. In order to explore the movement of bubbles inside the molten pool, one panel was degreased thoroughly while

CR

another was brushed with the blend of Al2O3 powders (φ= 3~ 5 μm) and pure B lubricant before joining.

US

The advantage of adding Al2O3 particles is that they are chemically stable inside the AA 4047 molten pool, and can act as a tracer of the Marangoni flow. Extra heat was generated by the lubricant, so the fusion

AN

zone was larger at the lubricated side and an asymmetrical molten pool is presented in Fig.s 12(a-b). The solute conservation ensures that changes in the solidification rates result in the corresponding

M

shifts in the volume of solute at the L/S interface (boundary layer), and a solute-rich or solute-lean layer

ED

could be formed. Bubbles and particles in Fig.s 12(c-d) were microsegregated, tracing solute boundary layers along isotherms, the thickness of which is inversely proportional to solidification rate [48]. The [49], lower than the simulated maximum temperature of

in the degreased molten pool; therefore, not too many particles were found in the center of the

CE

2385

PT

melting temperature of Al2O3 is only 2072

fusion zone (Fig. 12(b)). One interesting observation is that alumina particles delivered by the Marangoni

AC

convection only featured the boundary layers in the fusion zone rather than in the brazing zone. Hence, it naturally comes to the idea that the Marangoni flow might be inward instead of outward. This assumption is also certificated by the formation of the brazing microstructure at the bottom of 1B5MS weld shown in Fig. 11(a). When the flow of the molten pool is agitated, the particulate reinforced brazing microstructure shown in Fig. 11(b) can be easily disrupted and transformed into dendrites; at the same time, the fusion/brazing interface disappears. Gas bubbles escaping the molten pool is a physical disturbance to the 19

ACCEPTED MANUSCRIPT stability of the molten pool. Pure B lubricant generated more hydrogen bubbles than the 1B5MS case; as a result, the Marangoni flow became more turbulent, and the fusion/brazing interface was totally disappeared in Fig. 12(a). If the flow is outward, the bubbles at the boundary of the molten pool will be delivered to the bottom first and then to the center of the weld later. In such case, bubbles should be

T

captured in the brazing zone, because the liquid/solid interface solidifies from the bottom of the molten

IP

pool first as demonstrated in Fig. 10(b). However, no bubbles were frozen in the brazing zone. Zhao et al.

CR

[46] and Heiple [50] investigated the unsteady interfacial phenomena during inward weld pool flow, and found that the surface active oxides induced an inward flow motion with the greatest surface tension

US

located at the center of the molten pool surface (Fig. 11(c)). Welding panels with lubricant generated a plenty of soot and the main ingredient MgO is the surface active oxides for aluminum [30]. Drenchev et

AN

al. [36] found that decreasing the value of surface tension could increase the pore number per unit area and reduce the average pore diameter. Drenchev’s finding is in agreement with the observation that tiny

M

bubbles were congregated at the boundary of the molten pool (Fig. 12(c)). Based on the analysis above, it

ED

can be concluded that the surface tension gradient drove an inward flow pattern from the boundary to the

3.5 Mechanical properties

PT

center of the molten pool with the presence of a lubricant.

CE

Tensile tests can provide a rigorous prediction of the effect of the bubbles on the mechanical properties of welds [51]. For the comparison purpose, true stress and true strain of the joints were still

AC

used under the assumptions that the load-bearing area in the joint did not change too much. Among all the porous welds, an enlargement in the load-bearing area and consequently a reduction in the localization of strain can contribute to the increase in Young’s modulus (E) and ultimate strength (UTS). The Young’s modulus, E, is given by [52]: (8) where p is a given fraction of porosity;

and

are Poison’s ratio and Young’s modulus of the

compacted material. 20

ACCEPTED MANUSCRIPT Due to the inhomogeneous distribution of pores, some coupons with the same lubrication had different values in Young’s modulus and UTS. It is well known that the failure mechanism of a porous metal is the growth and coalescence of cavities such as voids and pores; stress concentration at the pores could form incipient necks, accelerate the damage, cause a premature fracture, and develop plastic

T

instability. Fig. 13(b) compares the ultimate strengths of porous welds. The large scattering of errors

IP

illustrates that the damage from pores makes the plastic deformation unpredictable. It is hard to correlate

CR

the weight of the same lubricant that refers to the amount of hydrogen to the ultimate strength, because the area of the pores in the fractured surface, which reduces the load bearing area, influences the tensile

US

strength rather than the volume fraction of pores [53]. Therefore, controlling the size of pores through adjusting the welding parameters can change the strength of the porous weld with the same surface

AN

conditions. Caceres et al. [51] proved that the porosity content, including the defect shape, number, and distribution, has little correlation with the ductility and tensile strength. However, the mechanical

M

performances decrease monotonically with an increase in the area fraction of pores in the cross section.

ED

Among 1B5MS, 1B7MS and 1B9MS joints, whose surface qualities are similar to those of the degreased

AC

CE

PT

joint, 1B7MS has the highest strength.

21

ACCEPTED MANUSCRIPT Figure 13 (a) Strain-stress curves of porous and lubricant-free welds; (b) ultimate strengths and (c) the maximum absorbed energy of different welds with the lubricant-free weld as a reference. Fig. 13(a) shows that the disadvantage of the lubricant-free weld is the low ductility. In contrast, reducing density of the weld through introducing pores unexpectedly increases the extension to failure. Bubbles on a small-scale can be rendered as large voids. The effect of plastic dissipation on void growth

T

stabilizes and delays the coalescence of the voids when voids expand [54]. Weinberg, et al. [54] describe

IP

the porous plasticity in metals by a void growth model, in which the material is rendered as a two-phase

CR

composite composed of an elastic matrix and a distribution of voids. The total energy stored by the voids is the summation of the energy stored by each void. The stored energy for each spherical void equals the

US

plastic work of deformation with the expansion of this void. Therefore, all porous welds had higher

AN

energy absorption capacity, when the absorbed energy of the lubricant-free weld is taken as a base as Fig. 13(c) shows.

M

4 Conclusions

ED

Lubricant can generate black soot on the joint surface, induce the strong fluctuations of plasma spectra. Higher concentrations of lubricant make wetting and spreading of the filler wire difficult, and

PT

even form pores on the surface. Joining panels with 1B5MS, 1B7MS, and 1B9MS lubricant formed welds with a surface quality similar to the case without lubricant. The plasma generated by F lubricant was

CE

mainly composed of ionized hydrogen. The stronger plasma intensity made the joining process unstable.

AC

The temperature and solidification rate decreased from the center to the boundary of the molten pool, although these factors had the opposite effects on pore formation. Decreasing the temperature lead to a reduction in solubility, prompting the pore formation. In contrast, slowing down the solidification rate resulted in an increase in the pore size and a decrease in pore number per unit area. The Ostwald ripening governed the growth mechanism of lubricant-induced pores. The pore distribution was influenced by the inward Marangoni flow inside the molten pool. Introducing pores into welds decreased Young’s modulus and ultimate strength, but improved the strain at failure and the energy absorption capability of porous

22

ACCEPTED MANUSCRIPT welds. Considering the mechanical properties of the joint, the lubrication of 1B7MS is suitable for laser joining of aluminum panels as received. Acknowledgements

IP

T

The financial supports by NSF Grant No. IIP-1539853 and General Motors are acknowledged. The authors would like to thank Mr. Andrew Socha at Research Center for Advanced Manufacturing for his help during experiments. No conflicts of interest exist.

REFERENCES

AC

CE

PT

ED

M

AN

US

CR

[1] Association EA. The aluminium automotive manual. EAA, available at. 2013. [2] Barckhoff JR. Total Welding Management. American Welding Society. 2010:1-6. [3] Lee P, Chirazi A, See D. Modeling microporosity in aluminum–silicon alloys: a review. Journal of Light Metals. 2001;1:15-30. [4] Ransley C, Neufeld H. The solubility of hydrogen in liquid and solid aluminium. Journal of the Institute of Metals. 1948;74:599-620. [5] Opie W, Grant N. Hydrogen solubility in aluminum and some aluminum alloys. Trans AIME. 1950;188:1237-47. [6] Eichenauer W, Hattenbach K, Pebler A. The solubility of hydrogen in solid and liquid aluminum. Z Metallk. 1961;52. [7] Piwonka T, Flemings M. Pore formation in solidification. AIME MET SOC TRANS. 1966;236:1157-65. [8] Tynelius K, Major J. A parametric study of microporosity in the A356 casting alloy system. TRANSACTIONSAMERICAN FOUNDRYMENS SOCIETY. 1993:401-. [9] Roy N, Louchez P, Samuel F. Statistical analysis of porosity in Al-9 wt% Si-3 wt% Cu-X alloy systems. Journal of materials science. 1996;31:4725-40. [10] Pequet C, Rappaz M. Modeling of porosity formation during the solidification of aluminium alloys using a mushy zone refinement method. Modeling of Casting, Welding and Advanced Solidification Processes IX. 2000:719. [11] Rousset P, Rappaz M, Hannart B. Modeling of inverse segregation and porosity formation in directionally solidified aluminum alloys. Metallurgical and Materials Transactions A. 1995;26:2349-58. [12] Atwood R, Sridhar S, Zhang W, Lee P. Diffusion-controlled growth of hydrogen pores in aluminium–silicon castings: in situ observation and modelling. Acta materialia. 2000;48:405-17. [13] Lee P, See D, Atwood R. Porosity formation during solidification±a comparison of micro modelling approaches. Cutting Edge of Computer Simulation of Solidification and Casting. 1999:97-111. [14] Conley JG, Huang J, Asada J, Akiba K. Modeling the effects of cooling rate, hydrogen content, grain refiner and modifier on microporosity formation in Al A356 alloys. Materials science and Engineering: A. 2000;285:49-55. [15] Huang J, Conley JG, Mori T. Simulation of microporosity formation in modified and unmodified A356 alloy castings. Metallurgical and Materials Transactions B. 1998;29:1249-60. [16] Rebollo TC, Lewandowski R. Mathematical and numerical foundations of turbulence models and applications: Springer; 2014. [17] Zhou F, Liang Y, Liu W. Ionic liquid lubricants: designed chemistry for engineering applications. Chemical Society Reviews. 2009;38:2590-9. [18] Long X. Finite element analysis of residual stress generation during spot welding and its affect on fatigue behavior of spot welded joints: University of Missouri--Columbia; 2005. [19] Bayraktar FS. Analysis of residual stress and fatigue crack propagation behaviour in laser welded aerospace Aluminium T-joints 2011. [20] Touloukian Y, Powell R, Ho C, Klemens P. Thermophysical Properties of Matter-The TPRC Data Series. Volume 1. Thermal Conductivity-Metallic Elements and Alloys. DTIC Document; 1970.

23

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

AN

US

CR

IP

T

[21] Ma J, Kong F, Kovacevic R. Finite-element thermal analysis of laser welding of galvanized high-strength steel in a zero-gap lap joint configuration and its experimental verification. Materials & Design. 2012;36:348-58. [22] Schwartz MM. Brazing: ASM international; 2003. [23] Lee H-k, Park S-h, Kang C-Y. Effect of plasma current on surface defects of plasma-MIG welding in cryogenic aluminum alloys. Journal of Materials Processing Technology. 2015;223:203-15. [24] AlShaer AW, Li L, Mistry A. Effect of filler wire properties on porosity formation in laser welding of AC-170PX aluminium alloy for lightweight automotive component manufacture. Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture. 2015:0954405415578584. [25] Ralchenko Y, Kramida A, Reader J. NIST atomic spectra database (version 4.0). National Institute of Standards and Technology, Gaithersburg, MD. 2010. [26] White DR. In process measurement of hydrogen in welding. DTIC Document; 1986. [27] Xie J. Dual beam laser welding. Welding journal. 2002;81:223s-30s. [28] Kong F, Ma J, Carlson B, Kovacevic R. Real-time monitoring of laser welding of galvanized high strength steel in lap joint configuration. Optics & Laser Technology. 2012;44:2186-96. [29] Kramida A, Ralchenko Y, Reader J. NIST atomic spectra database (ver. 5.1). National Institute of Standards and Technology, Gaithersburg, MD. 2013. [30] Deyev GF. Surface phenomena in fusion welding processes: CRC Press; 2005. [31] Kotecki D, Cheever D, Howden D. Mechanism of ripple formation during weld solidification. WELD J. 1972;51:368. [32] Rasberger M. Oxidative degradation and stabilisation of mineral oil based lubricants. Chemistry and technology of lubricants: Springer; 1997. p. 98-143. [33] Naidu SK, Klaus EE, Duda JL. Evaluation of liquid phase oxidation products of ester and mineral oil lubricants. Industrial & engineering chemistry product research and development. 1984;23:613-9. [34] Denisov ET, Khudyakov I. Mechanisms of action and reactivities of the free radicals of inhibitors. Chemical Reviews. 1987;87:1313-57. [35] Giuranno D, Ricci E, Arato E, Costa P. Dynamic surface tension measurements of an aluminium–oxygen system. Acta materialia. 2006;54:2625-30. [36] Drenchev L, Sobczak J, Sobczak N, Sha W, Malinov S. A comprehensive model of ordered porosity formation. Acta Materialia. 2007;55:6459-71. [37] de Obaldia EE, Felicelli S. Quantitative prediction of microporosity in aluminum alloys. Journal of materials processing technology. 2007;191:265-9. [38] Peterlongo A, Miotello A, Kelly R. Laser-pulse sputtering of aluminum: Vaporization, boiling, superheating, and gas-dynamic effects. Physical Review E. 1994;50:4716. [39] Wang X, Wang H-P, Lu F, Carlson BE, Wu Y. Analysis of solidification cracking susceptibility in side-by-side dualbeam laser welding of aluminum alloys. The International Journal of Advanced Manufacturing Technology. 2014;73:73-85. [40] Toda H, Hidaka T, Kobayashi M, Uesugi K, Takeuchi A, Horikawa K. Growth behavior of hydrogen micropores in aluminum alloys during high-temperature exposure. Acta Materialia. 2009;57:2277-90. [41] Talbot D. Effects of hydrogen in aluminium, magnesium, copper, and their alloys. International Metallurgical Reviews. 2013. [42] Lee P, Hunt J. Hydrogen porosity in directional solidified aluminium-copper alloys: in situ observation. Acta Materialia. 1997;45:4155-69. [43] Lee P, Hunt J. Measuring the nucleation of hydrogen porosity during the solidification of aluminium-copper alloys. Scripta materialia. 1997;36:399-404. [44] Gäumann M, Bezencon C, Canalis P, Kurz W. Single-crystal laser deposition of superalloys: processing– microstructure maps. Acta Materialia. 2001;49:1051-62. [45] Zhang Z, Farahmand P, Kovacevic R. Laser cladding of 420 stainless steel with molybdenum on mild steel A36 by a high power direct diode laser. Materials & Design. 2016;109:686-99. [46] Zhao C, van Steijn V, Richardson IM, Kleijn CR, Kenjeres S, Saldi Z. Unsteady interfacial phenomena during inward weld pool flow with an active surface oxide. Science and Technology of Welding & Joining. 2009;14:132-40. [47] Nakajima H. Fabrication, properties and application of porous metals with directional pores. Progress in Materials Science. 2007;52:1091-173. 24

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

AN

US

CR

IP

T

[48] Baeslack W. Effect of solute banding on transformations in dissimilar titanium alloy weldments. Journal of Materials Science Letters. 1982;1:229-31. [49] Pradyot P. Handbook of inorganic chemicals. 2002. [50] Heiple C, Roper J, Stagner R, Aden R. Surface active element effects on the shape of GTA, laser and electron beam welds. Weld J. 1983;62:72. [51] Caceres C, Selling B. Casting defects and the tensile properties of an Al Si Mg alloy. Materials Science and Engineering: A. 1996;220:109-16. [52] Ramakrishnan N, Arunachalam V. Effective elastic moduli of porous ceramic materials. Journal of the American Ceramic Society. 1993;76:2745-52. [53] Cáceres C. On the effect of macroporosity on the tensile properties of the Al-7% Si-0.4% Mg casting alloy. Scripta metallurgica et materialia. 1995;32:1851-6. [54] Weinberg K, Mota A, Ortiz M. A variational constitutive model for porous metal plasticity. Computational Mechanics. 2006;37:142-52.

25

IP

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

AN

US

CR

Graphical abstract

26

ACCEPTED MANUSCRIPT Highlights: • Feasibility of joining AA 6111 panels with the presence of lubricant was explored. • The relationship between the weld quality and different concentrations of lubricant was studied. • Pores formation and growth mechanisms were investigated.

AC

CE

PT

ED

M

AN

US

CR

IP

T

• The inward Marangoni flow during joining was verified.

27