Introduction

In the automotive industry, the use of advanced high-strength steels (AHSS) has increased to maintain the same level of strength and formability while reducing the overall weight that helps in improving fuel efficiency (Ref 1). Dual-phase (DP) steel is a type of AHSS with the ferrite and martensite phases in the microstructure (Ref 2). Martensite is a hard phase distributed randomly in the soft and tough ferrite matrix (Ref 2). Compared to low alloy steel, DP steel offers very good combination of strength, ductility, low yield ratio, continuous yielding along with moderately high work hardening rate facilitating good strain redistribution capacity (Ref 2,3,4,5,6). All these characteristics of DP steel make it a crucial lightweight material. Generally, cold-rolled DP steels with ferrite-martensite microstructure cover grades of 490 to 1180 MPa (minimum) tensile strength which finds attractive applications in the automotive industry due to its resistance to denting, fatigue, and impact (Ref 7). Specifically, DP 600–DP 800 steel grades are commonly used at front side member, A, B, C pillar, cross member, roof bow and roof tail portion of car body (Ref 8). In automobile manufacturing, to use any material, it should exhibit excellent mechanical properties along with good weldability. In previous auto design, car bodies were fabricated using resistance spot welding (Ref 9). But nowadays due to ease of automation, high energy density, low heat input, the flexibility of laser beam, high efficiency and minimum distortion in welded joints (WJ) (Ref 4, 5, 10), laser welding has become popular in the manufacturing of auto components.

The widespread laser welding processes use continuous and pulsed wave lasers. In continuous mode, a laser emits a steady-state beam over a period of time, whereas controlled short laser pulses are emitted in pulse wave mode (Ref 11). Welding with pulsed Nd:YAG system is characterized by periodic heating of weld pool by high peak power density of pulsed laser beam which causes melting and solidification consecutively (Ref 12). Due to short pulses, minimal amount of heat is transferred onto the part being welded and cooling occurs in between two pulses. As a result, the material is not heated to a very high temperature. This renders the application of the pulsed laser suitable for welding of intricate parts. Assuncao et al. (Ref 13) compared the WJ on 12 mm thick mild steel S355 with similar values of power density and found that pulse wave lasers produced deeper penetration along with higher penetration efficiency than continuous laser. In this study, it was observed that because of higher penetration in case of pulse wave laser, the WJ profile changed from a conduction mode to keyhole mode due to a considerable increase in depth to width ratio.

Pulse mode of laser welding has low average power compared to a continuous mode which offers low heat input and higher cooling rate (Ref 14). Pulsed laser provides an opportunity to shape the temporal profile of each pulse which offers the control on welding input, penetration depth, molten pool shape and keyhole formation (Ref 15). All these features have made pulsed wave laser welding process popular for joining of thin materials (Ref 14). Due to the presence of short welding cycle, wide adjustment range and high energy input offered by pulsed mode also find an application in joining of dissimilar alloys (Ref 16). High power density manufacturing processes, like laser, electron beam processing in both continuous and pulsed mode can be used for welding as well as for surface modification (Ref 17, 18). Compared to laser treatment, very high heating and cooling rates in case of electron beam treatment result in significant changes in the microstructure and phase composition and hence improvement of the functional properties of the welded or surface-modified metals/alloys occurs (Ref 19). In the automotive industry, electron beam processing is replaced by laser technology to accommodate the manufacturing environment as it provides benefits such as increased production rate due to the absence of vacuum and reduces the evaporation loss of alloying elements.

Recently, many studies have been reported on continuous wave mode laser welding of DP steels (Ref 3, 4, 9, 20) and showed that fusion zone (FZ) consists of martensite due to rapid cooling during welding. In the heat-affected zone (HAZ), softening behavior was observed due to tempering of existing martensite (Ref 3, 4, 20, 21). On the other hand, the pulsed laser welding finds attraction due to the application of low average power and high peak power which result in decreased width of FZ, HAZ, low residual stress and reduced degree of HAZ softening (Ref 5).

Continuous wave mode lasers involve usage of straightforward processing parameters, like average power, welding speed and focus position. In contrast, pulsed laser welding considers the combination of process parameters, such as pulse energy, pulse duration (PD), pulse frequency and welding speed in deciding the mode of energy transfer during welding, such as conduction or penetration or keyhole mode (Ref 11). PD is the laser “on” time, during which the material surface gets melted or vaporized depending on peak power density. The molten material is solidified during pulse “off” time (Ref 12). Maximum depth of penetration can be achieved using an optimal combination of increased pulse duration and reduced peak power density without splashes of metals (Ref 22). The peak power, defined as that portion of energy and pulse length which determines the interaction intensity of laser beam with material for a given spot size. Peak power (Pp) is numerically calculated as the ratio of pulse energy (E) to the pulse duration (t) whereas average power (Pavg) is the product of pulse energy (E) and the pulse frequency (f) (Ref 1) \({\varvec{P}}_{{\varvec{p}}} = \frac{{\varvec{E}}}{{\varvec{t}}}\); \({\varvec{P}}_{{{\varvec{average}}}} = {\varvec{E}} * {\varvec{f}}\).

Pulse energy is a measure of all energy emitted by a single pulse of laser over one full period. With an increase in pulse energy, WJ dimensions (width and depth) though increases, excessive increase in pulse energy leads to decrease WJ cross-sectional area due to evaporation and splashes of metal (Ref 22). Liao et al. (Ref 23) investigated the effect of laser pulse energy and incident angle on pulsed laser welding of SS 304. The study showed that WJ spot volume is insensitive to the incident angle, whereas it increases with an increase in pulse energy. Pulse frequency is related to pulse occurrence or frequency with which laser pulses are emitted which should be set according to the features of the pulse laser WJ (Ref 22). Sun et al. (Ref 5) reported the effect of pulse frequency on microstructure and properties of DP 590 steel WJ. The results showed that higher frequency increased in overlapping factor, weld penetration, average power and ultimately heat input. All these together resulted in the improvement in tensile strength, ductility and formability of DP 590 steel WJ (Ref 5). The energy required for the melting of the material depends on the overlapping factor (Of), and it indicates the transferred energy to welded material, which can be calculated using Eq 1 (Ref 12):

$${\varvec{O}}_{{\varvec{f}}} \user2{ } = \left( {1 - \frac{{\left( {\frac{{\varvec{V}}}{{\varvec{f}}}} \right)}}{{\left( {\user2{D } + \user2{ V*T}} \right)}}} \right)\user2{*}{\mathbf{100}}$$
(1)

where V is the welding speed, f is the pulse repetition rate (pulse frequency), T is the pulse duration and D refers to a spot diameter of the laser beam.

During pulsed laser welding, the energy pumped to an area comes not only from a single pulse but also from the pulse overlapping at the same spot. So only the peak power density or heat input cannot fully explain the performance of pulsed laser WJ (Ref 12). Therefore, the concept of “effective peak power density (EPPD)” is introduced to analyze the real power reaching to the material surface. The EPPD is calculated by Eq 2 (Ref 12):

$${\mathbf{EPPD}} = \left( {\frac{{{\varvec{P}}_{{\varvec{p}}} *{\varvec{F}}}}{{\varvec{A}}}} \right)$$
(2)
$${\varvec{F}} = {\mathbf{1}} + {\varvec{n}}\left[ {{\mathbf{1}} - \frac{{\left( {{\varvec{n}} + {\mathbf{1}}} \right){\varvec{V}}}}{2fd}} \right]$$
(3)
$${\text{Where}}\;\left( { {\varvec{n}} = \frac{{{\varvec{D}}*{\varvec{f}}}}{{\varvec{V}}}} \right)$$
(4)

where F is the cumulative overlapping index, Pp is the peak power, A is the laser spot area, D is the laser spot diameter, f is the pulse frequency, and V is the welding speed.

Improper selection of pulsed laser welding parameters can result in the vaporization of weld metal and forms the concave shape of the WJ cross section in the fusion zone (Ref 24).

The knowledge of corrosion behavior of AHSS after joining is a prerequisite for the application in the automotive industry (Ref 25). In case of AHSS WJ, Gracia et al. (Ref 25) observed that WJ bead has a smaller cross-sectional area compared to the base metal (BM), which may act as selective attacking sites during corrosion. Wint et al. (Ref 26) reported the corrosion behavior of laser-welded high-strength low alloy (HSLA) steel joints. The results showed that during the welding process, variation in microstructure, chemical composition and grain size occurred in the base material; and such changes influenced the severity of galvanic corrosion. Galvanized steel is used in the automotive industry for car body structure and interior surface to protect against corrosion by galvanic coupling to zinc (Ref 27). However, a successful joining of these steels with modern technologies, like laser welding, is still a big challenge since at 1173 K zinc boils and generates vapor in both FZ and nearby region (Ref 28). For zero-gap/clearance WJ, the vaporized zinc that needs to be exhausted from FZ along with iron vapor formed in a keyhole leads to the formation of weld defects such as spatter, blowholes and provide poor quality WJ surface (Ref 27). Pulsed laser involved periodic spot partially overlapping with each other which refills the preceding FZ and reduce the pores in FZ due to previous pulses and periodic power-offs in pulse cycles used to produce WJ (Ref 28). But, Tzeng et al. (Ref 29) showed that completely pore-free WJ are not possible using high-power pulsed lasers. Generally, pulsed laser uses a rectangular shaped pulse with constant peak power during pulse on time. It is reported that the presence of shorter pulse duration and higher cooling rate result in unsatisfactory quality of WJ (Ref 30). The cooling rate of a weld pool can be controlled to improve the weld quality by regulating laser peak power and pulse duration with the aid of pulse shaping (Ref 30) which can provide an opportunity to join the galvanized steel and produce the defect-free WJ.

It emerges from the above survey that available information on the effect of pulse duration as a laser welding parameter of DP steels is very limited. The detailed characterization of microstructure, tensile properties and corrosion behavior of WJ with different PD is also lacking. The present work is a step toward addressing the above issues. In the present study, butt welding of DP 780 steel has been carried out using a pulsed laser welding process with different pulse durations. The effect of PD on the evolution of microstructures and microhardness in FZ and HAZ has been correlated with the thermal modeling of pulsed laser welding. Specifically, the present study aims to assess the role of PD as a pulse laser welding parameter for improving the mechanical and electrochemical performances of DP steel WJ.

Experimental Procedure

Material

Dual-phase steel of DP 780 grade was received from Tata Steel, Jamshedpur, India, in sheet form with a nominal thickness of 1.00 mm. The composition (wt%) of DP 780 steel is 0.1C–1.78Mn-0.33Si-0.007S-0.012P. The yield strength (MPa), tensile strength (MPa), % elongation and Vickers microhardness (HV0.1) of DP 780 steel are 518 MPa, 824 MPa, 14% and 250 respectively.

Laser Welding Procedure

Pulse mode laser welding was carried out using JK 600 HP (Fig. 1a), with Nd:YAG laser generator (GSI, UK) integrated with ABB IRB 1410 robotic control. The sheet was cut into dimensions of 55 mm × 25 mm × 1 mm using wire cut CNC EDM machine keeping the long dimension parallel to rolling direction (RD) of the steel sheet. To prevent the distortion of the workpiece, the sheets were carefully clamped in the fixture to perform the welding operation. The fixture was made of copper backing plates and aluminum top plates. Very high thermal conductivity of copper and aluminum ensures intensive heat transfer resulting in low angle distortions of the WJ (Ref 31). Welding was done perpendicular to the RD of the sheet in butt joint configuration with the parameters listed in Table 1.

Fig. 1
figure 1

(a) Robotic controlled JK 600HP Nd:YAG laser welding setup, (b) optical microstructure of DP 780 BM showing ferrite (F) and martensite (M) and (c) geometry of the laser-welded tensile test specimen

Full size image
Table 1 Parameters used for laser welding
Full size table

Metallography and Microstructures

The transverse cross section of WJ was machined and then polished using an automatic polishing machine, Struers TegraPol 21. The WJ specimens were etched using 2% nital solution to reveal the microstructures of FZ, HAZ and BM. Another etchant is modified Winsteard’s reagent consisting of 2 g picric acid dissolved in 10 ml of ethyl alcohol (Part A) and 5 ml of a 40% solution of sodium dodecylbenzene sulphonate (surfactant) in 200 ml of water and 5-6 drops of concentrated hydrochloric acid (Part B). The WJ cross section sample was etched with solution containing equal proportion of Part A and Part B to demarcate the boundaries of FZ, HAZ and BM.

The microstructure of the steel was observed using an optical microscope, Leica DM 2500M coupled with image analyzing software LAS V4.4 and in a field emission scanning electron microscope (FEGSEM), FEI Quanta 450.

Mechanical Tests (Microhardness and Tensile)

Vickers microhardness measurements were taken on a transverse cross section of WJ using 100 g load and 15 s dwell time in a MATSUZAWA microhardness tester.

Cross-weld tensile specimens (Fig. 1c) were prepared from WJ, keeping the loading direction of samples parallel to RD following the ASTM E8M standard. Strain-control tensile tests were carried out up to fracture in a computer controlled servo-hydraulic universal testing machine, INSTRON 8500R of ± 100 kN capacity at a strain rate of 1×10−3 s−1. The data acquisition was done using Instron Bluehill tensile testing software. The fracture surfaces were examined using the same FEGSEM as that was used for microstructural characterization.

Electrochemical Behavior

The transverse cross section of the WJ was considered for studying electrochemical corrosion behavior in 3.5% NaCl solution following potentiodynamic polarization method. Before potentiodynamic polarization study, the samples were polished by conventional metallographic procedure and a clear fusion zone was identified which was exposed to corrosive solution by covering the rest part of the sample with teflon tape. The potentiodynamic polarization curves were obtained by Gamry Echem Analyst V 5.30 potentiostat controlled by a computer with DC 105 corrosion analysis software. Before potentiodynamic tests, an open-circuit potential test was carried out in the test solution for 60 minutes to allow for the steady state to attain and then potentiodynamic scans were performed. Experiments were carried out by using a conventional three-electrode cell with a graphite rod and a saturated calomel electrode (SCE). In these experiments, the polished sample was considered as working electrode, graphite as counter electrode and SCE as a reference electrode. Potentiodynamic polarization curves were carried out at a constant scan rate of 1 mV/s within the potential range varying from -1000 to +1000 mV. The corrosion potential (Ecorr) represents the thermodynamic characteristics of material surface in a corrosive environment, and corrosion current density (icorr) signifies the rate of corrosion in a given medium. Corrosion potential was calculated by Tafel’s linear extrapolation method by using Gamry DC 105 corrosion analysis software.

Numerical Model for Predicting Welded Joint Temperature Profile

Three-dimensional finite element (FE) model was developed to analyze the effect of pulse duration on the temperature profile during the welding process using ABAQUS CAE 6.14-5 software with 5 mm × 5 mm × 1 mm dimensions of a sheet (Fig. 2). Since laser welding process results in a narrow FZ, sufficiently fine mesh is required to record the steep temperature gradient. It involves a large number of nodes and elements leading to long computation time as well as high computer storage and processor requirements. So it is problematic to maintain the real size of a sample in the model. Sbestova et al. (Ref 32) suggested that only part of the WJ is sufficient to model by applying proper boundary conditions. Laser welding is high power density and low heat input process which results in narrow sized WJ. So in the present work, the sheet width is kept as 5 mm along Y-direction, whereas the weld travel length along X-direction was restricted to 5 mm (Fig. 2). In 5 mm weld travel length, about 16 pulses were in contact with the material surface and it is expected adequate to demonstrate the WJ temperature profile. The total number of 1,20,000 linear hexahedral elements of type DC3D8 and a total number of nodes of 1,29,381 were considered for the meshing part. The moving heat load by a laser pulse was incorporated through DFLUX subroutine written in Fortran 95 to describe the spatial and temporal variation of a heat source. Kik et al. (Ref 33) compared the heat source models used in numerical simulations of laser welding. The study showed that the correct calibration of a moving heat source ensures the convergence of models and their reliability in further applications. It is important since this heat distribution and cooling rate experienced during welding have an impact on metallurgical changes in materials, resulting in different stress and strain distributions (Ref 33). Phase transformation of the investigated steel due to welding was analyzed by considering the temperature-dependent properties listed in Table 2 (Ref 34). The initial temperature of a sheet was 298 K. The model is based on the following assumptions:

Fig. 2
figure 2

Schematic diagram of the finite element model for pulsed laser welding including its mesh and heat source details

Full size image
Table 2 Thermal properties of steel used for FE analysis [34]
Full size table

(1) The top surface of the WJ was considered as flat and the concavity in weld bead was neglected; (2) welded sheets were considered as three-dimensional deformable solid bodies in FE simulation; (3) the symmetry effect was considered along the WJ centerline (Fig. 2) to reduce the computational time.

In the present set of laser welding, the laser beam crossed at a constant speed of 8 mm/s and heat source was assumed to follow Gaussian heat distribution as it maps the deep penetration while maintaining the small WJ width (Ref 35). Transient temperature field due to heat conduction during welding is described by the moving heat source and is given by Eq 5 (Ref 35):

$$\rho C_{p} \left( {\frac{\partial T}{{\partial t}} - U \frac{\partial T}{{\partial x}}} \right) = K\left( {\frac{{\partial^{2} T}}{{\partial x^{2} }} + \frac{{\partial^{2} T}}{{\partial y^{2} }} + \frac{{\partial^{2} T}}{{\partial z^{2} }}} \right) + \dot{Q}$$
(5)

where ρ is the density, T is the temperature, C is the specific heat, \(\dot{Q}\) is the rate of internal heat generation, K is the thermal conductivity, and U is the welding velocity.

The top and side of the sheet are subjected to radiation and convection heat losses. So, the boundary condition is expressed (Eq 6) as (Ref 35)

$$K_{n} \left( {\frac{\partial T}{{\partial n}}} \right) + h \left( {T - T_{0} } \right) + \sigma \varepsilon \left( { T^{4} - T_{o}^{4} } \right) = 0$$
(6)

where Kn is the thermal conductivity normal to the surface, h, To, σ and ε are the heat transfer coefficient, ambient temperature, Stefan–Boltzmann constant and emissivity, respectively. The Gaussian heat source is considered over the volume such that the flux density decreases with the depth of penetration. The effect of pulse duration is considered in the heat source model. The WJ with higher heat input cannot be accurately characterized by conical heat source especially in the bottom of WJ. To represent the shape of full penetration WJ, Baruah et al. (Ref 35) developed an hourglass-like heat source with Gaussian power density distribution and the same has been utilized in the present study for evaluation of temperature profile during welding. The volumetric heat source can be mathematically expressed by using Eq 7 and 8 as (Ref 35):

$$\dot{Q}\left( {Z,r} \right) = \frac{{3{\text{d}}\eta P_{pk} }}{{\pi \cdot \left[ {1 - \exp \left( { - d} \right)} \right] \cdot \left( {Z_{t} - Z_{m} } \right)\left[ {r_{m}^{2} + r_{m} \cdot r_{t} + r_{t}^{2} } \right]}} h\left( t \right) \exp \left[ { - d\frac{{r^{2} }}{{r_{{{\text{eff}}}}^{2} }}} \right] \;{\text{for}}\;{\text{upper}}\;{\text{cone}}\;\left( {Z_{o} \ge \, Z_{m} } \right)$$
(7)
$$\dot{Q}\left( {Z,r} \right) = \frac{{3{\text{d}}\eta P_{pk} }}{{\pi \cdot \left[ {1 - \exp \left( { - d} \right)} \right] \cdot \left( {Z_{m} - Z_{b} } \right)\left[ {r_{m}^{2} + r_{m} \cdot r_{t} + r_{t}^{2} } \right]}} h\left( t \right)\exp \left[ { - d\frac{{r^{2} }}{{r_{{{\text{eff}}}}^{2} }}} \right]\;{\text{for}}\;{\text{lower}}\;{\text{cone}}\;\left( {Z_{o} < \, Z_{m} } \right)$$
(8)

where Z and r represent the Z-axis coordinate and radial distance from the axis. The subscripts ‘t’, ‘m’ and ‘b’ represent the top, middle and bottom surface, respectively, of a heat source. d, η and Ppk represent Gaussian power density distribution factor, absorptivity and peak power, respectively. The average value of absorptivity of dual-phase steel is considered as 0.5 (Ref 34). The h(t) signifies the temporal variation of the intensity. In the case of continuous welding, h(t) is 0 for the whole cycle. For pulse laser, the value of h(t) is 1 at pulse on time and 0 at pulse off time (Ref 35). The reff is the effective radius of the cone and defined by Eq 9 and 10 (Ref 35) as

$$r_{{{\text{eff}}}} = \frac{{Z_{o} - Z_{m} }}{{Z_{t} - Z_{m} }}\left( {r_{t} - r_{m} } \right) + r_{m} :{\text{For}}\;{\text{upper}}\;{\text{cone}}$$
(9)
$$r_{{{\text{eff}}}} = \frac{{Z_{o} - Z_{b} }}{{Z_{m} - Z_{b} }}(r_{b} - r_{m} ) + r_{m} :{\text{For}}\;{\text{lower}}\;{\text{cone}}$$
(10)

Results and Discussion

Effect of Pulse Duration on WJ Surface Morphology and Bead Size

The WJ surface morphologies with different PDs are shown in Fig. 3. Shorter PD results in spattering of material and hence leads to poor-quality welding surface as shown in Fig. 3(a) to (d). Welding surface quality improved at higher PD. From the WJ surface morphology, it is qualitatively observed that with an increase in PD keyhole root decreases (Ref 5). Using Eq 1, it is estimated that with an increase in PD, overlapping factor increases marginally from 58 to 59%. Figure 3(e) indicates that with an increase in pulse duration both bead width and pulse overlap (Fig. 3c) increase up to 2.1 ms PD and then, it starts to decrease marginally.

Fig. 3
figure 3

WJ surface morphology with different PD (a) 1.5 ms, (b) 2.1 ms, (c) 2.7 ms, (d) 3.6 ms and (e) effect of PD on bead width, pulse overlap and overlapping factor

Full size image

Typical WJ bead profiles at 1.5 ms and 3.6 ms PD, shown in Fig. 4(a) and (b), indicate complete penetration. At highest PD (3.6 ms), WJ cross section macroview had an inverted conical shape with average FZ width of 0.42 mm, while for lowest PD (1.5 ms), WJ had a slightly inverted trapezoidal cross section with average FZ width of 0.63 mm. The average width of FZ as well as softened HAZ continuously decreases with an increase in PD, Fig. 4(c). From the WJ profiles (Fig 4a), two different terms SD (surface depression) and RD (root depression) at the top and bottom of the WJ are evaluated. These are related to underfill defects which usually happen due to metal ejection or displacement from the weld pool because of the high energy density of laser beam welding. Figure 4(d) shows that with an increase in PD, a decrease in peak power results in lowering of both SD and RD. With an increase in power, the increase in temperature gradient in weld pool leads to the development of Marangoni stress due to surface tension and buoyancy force from density difference in the molten metal (Ref 36). This causes turbulence in the weld pool and hence results in the formation of undercut and spattering of the molten metal. This could be the possible reason for producing the higher value of SD and RD at lower PD WJ. At higher PD (3.6 ms), lower peak power (Fig. 4d) causes a decrease in temperature gradient and hence less turbulence in the weld pool. This, as a result, decreases the undercut and crater in the WJ. Higher PD during welding decreases effective peak power density (EPPD) and raises the value of aspect ratio and thus results in keyhole mode of welding, Fig 4(e). This observation is in conformity to that of Chelladurai et al. (Ref 11) where it has been reported that an increase in aspect ratio (depth/width) of FZ greater than 1.25 results in keyhole mode welding.

Fig. 4
figure 4

Macroviews of the WJ and WJ bead size analysis with PD (a) 1.5 ms, (b) 3.6 ms, (c) comparison of FZ and softened HAZ width with PD, (d) calculated SD and RD of the WJ due to its respective peak power with varying pulse duration and (e) effect of pulse duration on the EPPD and aspect ratio of the WJ

Full size image

Thermal Analysis

Based on the chemistry of the BM, phase transformation temperatures were computed using Thermo-Calc software. The solidification start and finish temperatures were obtained as 1790 K and 1748 K, and the critical transformation temperatures Ac3 and Ac1 were found as 1096 K and 964 K, respectively.

The FE simulation was performed to observe the temperature profile and heat transfer phenomenon using the pulsed laser welding with the average power of 325 W. The model was validated by comparing the results of a computed WJ on YZ plane with the experimental results using 3.6 ms PD WJ. Figures 5(a) and (b) show a comparison between predicted and experimental results of WJ profile. The red color band describes the FZ temperature in the range of 2110 K to 2270 K which is much above than solidification start temperature obtained by Thermo-Calc software, ensuring complete melting. The softened heat-affected zone signifies the solid-state phase transformation and present in a dark and light green color band in Fig. 5(a) where the temperature is in the range of 956 K to 1120 K.

Fig. 5
figure 5

Comparison between FE simulation and experimental results as (a) numerically calculated WJ bead with thermal profile and (b) macrograph

Full size image

The WJ measurement was carried out at the mid thickness to validate the FE model. It was observed that values for both the locations and the half-widths of FZ and SHAZ shown by yellow color double arrow are found to be very close (Fig. 5a and b) with a difference of 10.2% between measured and predicted value. Baruah et al. (Ref 35) observed the maximum error of 14% between experimentally measured and computed half width of WJ using the same numerical model for pulsed laser welding of 0.5 mm thick Ti6Al4V alloy in butt joint configuration.

Microstructure of the WJ

Figure 6(a) shows the BM microstructure containing ferrite (F) and martensite (M) phase (with a magnified view in the inset) in it. The area fraction of martensite in DP 780 BM was estimated to be 32% (Fig. 1b) using Leica image analysis software. WJ microstructure depends on the distance from the weld centerline, local peak temperature, holding time at high temperature causing difference in cooling rate which significantly affects the phase transformation during welding and thus properties of the WJ (Ref 3). The HAZ can be divided into sub-critical HAZ considered as softened HAZ (S-HAZ), while intercritical HAZ (IC-HAZ), fine-grained HAZ (FG-HAZ) and coarse grain HAZ (CG-HAZ) are considered as hardened HAZ in the present study. The Rosenthal Eq 11 was used to calculate the cooling rate in FZ of laser WJ (Ref 37)

$$\left( {\frac{\partial \theta }{{\partial t}}} \right) = \frac{{2\pi K_{s}^{2} }}{\alpha } \left( {\frac{V\Delta x}{Q}} \right)^{2} \left( { \theta - \theta_{o} } \right)^{3}$$
(11)

where \(\left( {\frac{\partial \theta }{{\partial t}}} \right)\) is a cooling rate (K/s), Ks is the thermal conductivity of steel (30 W/m·K), α is the thermal diffusivity (5.613 × 10−6 m2/s), V is the welding speed (m/s), Δx is the sheet thickness (m), Q is the laser average power (J/s), θ and θo are the temperatures of FZ (2270 K) and the ambient temperature (298 K), respectively. The cooling rate of FZ was estimated to be about 4680 K/s for 3.6 ms PD WJ, and it was sufficient to form a fully martensitic structure in FZ.

Fig. 6
figure 6

Microstructures of the WJ with 1.5 ms PD of (a) BM (b) FZ, (c) CGHAZ, (d) FGHAZ, (e) ICHAZ and (f) softened HAZ with a magnified view of tempered martensite shown in the inset

Full size image

For both the WJ with PD 1.5 ms and 3.6 ms, (Fig. 6b and 7a), the FZs show predominately lath martensite (LM) microstructure. The original ferrite-martensite structure disappeared due to complete austenitization and formation of martensite (Ref 3, 4, 9, 20). The features of the FZ microstructures are summarized in Table 3.

Fig. 7
figure 7

Microstructures of the WJ (3.6 ms PD) of (a) FZ, (b) CGHAZ, (c) FGHAZ, (d) ICHAZ, (e) softened HAZ with the enlarged view of tempered martensite shown in the inset

Full size image
Table 3 Microstructural features in the FZ
Full size table

The prior austenite grain size and martensitic lath spacings were measured with the help of image analyzing software from multiple high-magnification FEGSEM images. Larger heat input can cause increased spacings between the lath martensitic bundles (Ref 4). Lower lath spacings for higher PD WJ indicate the rapid heat dissipation and limited austenite growth during solidification of weld pool. In CG-HAZ, (Fig. 6c and 7b) microstructures consist of lath martensite. The peak temperature attained in this zone observed by numerical FE model, is around 1940 K-2110 K which was much above Ac3 temperature. It leads to significant growth of prior austenite grains which subsequently transformed into lath martensite during solidification of the weld pool. The FG-HAZ microstructures (Fig. 6d and 7c) contain fine grain ferrite (FGF) and lath martensite formed within the smaller sized prior austenite grains compared to CG-HAZ. Microstructure of IC-HAZ shown in Fig. 6(e) and 7(d) contains martensite and ferrite with a higher amount of martensite (42% for 1.5 ms and 48% for 3.6 ms of PD) compared with the as-received condition (32%). The peak temperature in this zone is in between 956 and 1120 K causing partial austenitization which transforms into martensite and fine grain ferrite, depending on the cooling rate experienced during solidification. The sub-critical (softened) HAZ shown in Fig. 6(f) reveals tempered martensite (TM) and ferrite (magnified view of tempered martensite is shown in the inset). The detailed investigation of the softened HAZ is discussed in the later section.

Microhardness Profile

Figure 8(a) shows the microhardness profile obtained from the cross section of DP 780 steel WJ and BM, taken at a load of 100 g and 15 s dwell time. The average hardness value of the DP 780 BM is 250 HV. The hardness of FZ was 423.8 HV which is approximately 1.7 times that of BM. It confirms the formation of martensitic structure in the FZ. The hardened HAZ shows higher hardness than that of BM due to the formation of lath martensite and ferrite. The location of hardened HAZ is approximately 0.3 to 0.45 mm from the weld centerline which has hardness value in the range of 380 to 310 HV. This decrease in hardness at hardened HAZ is due to formation of lower amount of martensite from austenite. The region near to the base material at outer HAZ has a lower hardness value than the BM. This is known as sub-critical or softened HAZ (Ref 3). The hardness value of this zone lies within the range of 220 to 246 HV. The soft zone appears due to tempering of the pre-existing martensite (Ref 3, 4, 9, 20).

Fig. 8
figure 8

(a) Microhardness profile of WJ with varied PD, (b) average microhardness variation in FZ and HAZ of the WJ with the PD

Full size image

The hardness variation in different regions of the WJ (as FZ, hardened HAZ, and softened HAZ) with PD is presented in Fig. 8(b). The average hardness of FZ decreases from 408 to 385 HV with an increase in PD from 1.5 to 3.6 ms. The average hardness of hardened HAZ increases from 355 to 374 HV, and for softened HAZ it shows marginal increases from 235 to 242 HV. The softened HAZ was detected around at 0.6 to 0.8 mm of distance away from the weld centerline at the mid thickness cross section of the WJ. The temperature in this zone is in the range of 792 to 956 K during welding. This is the temperature range of martensite tempering where hardness becomes very low due to softening. Kundu et al. (Ref 38) experienced the peak temperature in SHAZ around 879 K for DP 780 steel laser spot WJ and also observed 12-25% lower hardness value than BM due to HAZ softening.

Effect of Pulse Duration on HAZ Softening

Critical size of soft zone with varying PD was investigated by microhardness indentation within ferrite grains (Fig. 9a and b). The reference line (Fig 9a) was considered at the start of the sub-critical HAZ. Figure 9(c) shows the hardness variation. With lowest PD, the lowest ferrite microhardness is 120.7 HV at a distance of 108 µm, while for the highest PD, the hardness is 131.2 HV at 159 µm from the reference line. The average BM ferrite microhardness is 150 HV. Therefore, the HAZ softening is not only affected by the tempering of martensite but also by a variation of hardness of the ferrite matrix (Ref 39). This minimum ferrite microhardness in outer HAZ (Fig 9c) is expected to have a direct influence on the tensile failure.

Fig. 9
figure 9

Effect of PD of WJ on extent of HAZ softening (a) position of microhardness indentation on ferrite grains for a PD of 3.6 ms cross section with respect to reference line shown by a red double arrow, (b) microstructure and indentation in coarse ferrite grain of softened HAZ and (c) ferrite microhardness profile indicating ferrite softening and position of lowest ferrite hardness (shown by blue color arrow). Martensite decomposition into cementite (red mark) and equiaxed ferrite grains (blue mark) at softened HAZ of (d) 1.5 ms and (e) 3.6 ms PD WJ (Color figure online)

Full size image

Tempering of martensite in WJ of DP steel depends on maximum peak temperature, holding time at peak temperature during welding, parent microstructure and composition of the steel. At higher temperature, it involves the formation of stable cementite, spherodization of cementite, recovery and recrystallization of martensite lath structure (Ref 40). Spherodization of cementite occurs with a decrease in surface energy. It preferentially grows and spherodizes at interlath boundaries and prior austenite grain boundaries due to greater ease of diffusion of vacancies required to accommodate the growth of cementite. During tempering of martensite in the temperature range of 623-873 K, recovery occurs with considerable rearrangement of dislocations within the martensite laths and at the lath boundaries. In the temperature range of 873-973 K, this recovery process is replaced by recrystallization which results in the formation of equiaxed ferrite grains (Fig. 9d), with spheroidal cementite particles at boundaries and within grains, marked by a yellow arrow. The coarser and needle-like cementite indicated by red arrow (Fig. 9d) also observed along the grain boundaries. The precipitation of cementite due to rejection of carbon atoms from supersaturated ferrite matrix and generation of equiaxed ferrite grains are the results of recovery phenomenon which is detected by the disappearance of pre-existing martensite laths. All these together possibly resulted in the hardness variation in ferrite matrix in the softened HAZ. Hernahdez et.al. (Ref 41) showed that a reduction in ferrite nanohardness occurred in the tempered region due to a possible reduction in dislocation density. Figure 9(d) and (e) demonstrate that the extent of decomposition of martensite into cementite is more in the softened HAZ of WJ with 1.5 ms PD due to larger holding time at high temperatures and high EPPD. As a consequence, the higher the EPPD more is the HAZ softening.

Effect of Pulse Duration of the WJ on Tensile Properties

Figure 10(a) shows the load–extension curve of both BM and WJ. It shows a continuous yielding for both BM and WJ. Figure 10(b) and (c) shows the tensile failure location of WJ with 1.5 ms and 3.6 ms PD. Almost all the welded tensile test specimens failed in the softened HAZ [except 1.5 ms PD WJ shown in Fig 10b].

Fig. 10
figure 10

(a) Load–extension curve DP 780 BM and WJ, (b) and (c) failure locations indicated by a red color arrow of DP 780 laser-welded steel joints for the PD 1.5 and 3.6 ms, respectively, under tensile loading

Full size image

The WJ in the present study contains concavity (Fig. 4a and b). The amount of concavity has been measured from the WJ profiles by taking the ratio of reduction in area from the sheet thickness in FZ to the initial sheet thickness (Ref 42). Concavity was generally created by metal ejection (or displacement) from the melt pool during welding (Ref 43). In keyhole and penetration mode of welding, evaporation of molten metal was observed, and it produces higher vapor pressure. This vapor pressure offers higher melt velocity, which promotes the ejection of weld metal from the front of the weld pool and results in the concavity. Figure 11(a) shows that with an increase in PD from 1.5 to 3.6 ms, the peak power decreases, because of which there happens a decrease in concavity from 35 to 6%. So for a more accurate comparison of the load bearing ability of WJ with varying PD, the yield and peak load obtained from the tensile tests were considered instead of representing as yield strength and tensile strength, and so also for extension of specimen gauge length. Westerbaan et al. (Ref 42) also reported the peak loads obtained from tensile tests for effective comparison of WJ. It helps to minimize the errors obtained during measurement of a cross section along the gauge length due to different amounts of concavity (Ref 42).

Fig. 11
figure 11

Effect of PD of the WJ on a variation of (a) concavity, (b) yield, peak load (N) and extension of gauge length (mm) at maximum load

Full size image

Table 4 lists the tensile test results and its failure location. The yield load of WJ with different PD is about 75-91% of BM. It was inconsistent with studies reported for DP 590 WJ where higher yield load was obtained than BM due to the presence of yield point phenomenon (Ref 8). Also the presence of tempered martensite in HAZ is responsible for a reduction in yield load of WJ. Figure 11(b) shows with an increase in PD from 1.5 to 3.6 ms, due to reduced concavity, both yield and peak load of WJ increase. A 1500 N drop in peak load was observed in DP 780 steel WJ when concavity increased from 14 to 35% (Fig. 11a and b and Table 4). Weld concavity acts as a notch and results in stress concentration and hence decreases the load bearing capacity during tensile deformation leading to premature failure of the WJ. A 740 N rise in peak load was observed for higher PD (2.7-3.6 ms) due to reduced weld concavity and HAZ softening. So, the WJ with PD greater than 2.7 ms, having the concavity in the range of 6-14%, is considered to be acceptable concerning industry standards (Ref 42). In summary, the yield load, peak load and extension of gauge length at peak load increase with an increase in PD of the WJ (Fig. 11b) due to reduced amount of concavity which avoids the premature failure of the WJ in FZ or HAZ.

Table 4 Tensile test results of DP 780 BM and laser WJ
Full size table

Effect of Pulse Duration on Tensile Fracture Surfaces

The fracture surfaces show that under tensile loading the welded specimens failed in a ductile manner involving stages of microvoid nucleation followed by their growth and coalescence. The fracture surface of the tensile specimens was characterized to measure the size and distribution of dimples in the BM, 1.5 ms and 3.6 ms PD WJ conditions. Figure 12(a) and (c) shows the fracture surface of 1.5 and 3.6 ms specimens with the fibrous area located at the center surrounded by a strip of shear lip all around. Width of shear lip is 0.095 mm and 0.112 mm for 1.5 and 3.6 ms PD WJ, respectively. The larger shear lip is related to the small fibrous zone. The dimple size distribution for fracture surface of BM, 1.5 and 3.6 ms PD WJ conditions is shown in Fig. 13. The smaller dimple size is in 1.5 ms PD WJ in comparison with BM, and 3.6 ms PD WJ specimen is believed to be related to higher degree of concavity. As concavity increased, failure occurs early with smaller gauge length extension. Table 5 gives a summary regarding the comparative assessment of pulsed mode of laser welding used in the present study with the continuous laser WJ.

Fig. 12
figure 12

Typical SEM fracture surface images of tensile tested DP 780 steel WJ (a), (b) weld specimen with PD 1.5 ms showing an overall view of ductile fracture surface, (c), (d) specimen with 3.6 ms pulse duration weld ductile fracture with dimples and voids by red color arrow and (e), (f) DP 780 BM fractograph showing coalescence of voids shown by red color arrow leading to crack propagation indicated by white color arrow resulted into ductile failure

Full size image
Fig. 13
figure 13

Dimple size distributions of the fracture surface

Full size image
Table 5 Comparative summary of published literature on the overall performance of laser-welded DP steel WJ along with the present study
Full size table

Effect of Pulse Duration on the Electrochemical Behavior

Figure 14 shows the potentiodynamic polarization curves of the BM and WJ with different PD. The immersion potential (Eim), steady-state potential (Ess) after 1 h, corrosion potential (Ecorr), corrosion current density (icorr), pitting potential, cathodic and anodic Tafel slopes were calculated from potentiodynamic polarization curves and corresponding microhardness values of WJ summarized in Table 6. The open-circuit potential (OCP) measurements with time in 3.5% NaCl solution revealed that the steady-state potential for higher PD WJ tends to shift at the more positive direction, whereas 1.5 ms PD WJ specimen was stable at a low negative value. At a first glance this behavior of OCP indicates that an increase in PD during welding improves the corrosion resistance. The stability of metals in a corrosive environment is known to depend on its thermodynamic properties and kinetics of electrochemical reactions that occur on its surface. High PD specimens exhibit nobler characteristics due to spontaneous oxide layer formation. From the potentiodynamic polarization curve, it was observed that like active-passive metals, the passivation current increases considerably in WJ formed with lower PD, though the potential of passivation peak shifted toward a positive value. The passive region was found to be larger for high PD specimens compared to 1.5 ms PD specimen. An increase in PD during welding causes a decrease in corrosion current density. Both cathodic and anodic reaction kinetics decreased with the increase in PD. The decrease in cathodic kinetics (hydrogen evolution reaction, HER) was observed with higher PD WJ. This behavior is explained in terms of the catalytic properties of the different FZ microstructure toward the HER.

Fig. 14
figure 14

Potentiodynamic polarization curve for BM and WJ with varying PD

Full size image
Table 6 Summary of the results obtained from the potentiodynamic polarization test in a 3.5% NaCl solution and comparison with FZ microhardness values
Full size table

The corrosion of DP steel depends on the area fraction of martensite, amount of epitaxial ferrite grains and the presence of internal stresses (Ref 49). In DP steel, the microgalvanic cell is formed between the ferrite and martensite (Ref 50). Galvanic corrosion is proportional to the ratio of the cathodic area (martensite) to the anodic area (ferrite) (Ref 49, 51). As shown in Table 6, the decrease in FZ hardness with an increase in PD is possible because of the formation of the second phase like upper bainite/bainitic ferrite, and the ratio of cathodic to anodic area is reduced leading to lower corrosion current density and subsequent improvement in the corrosion resistance for higher PD WJ. Hardened microstructures in low PD WJ suffer from increasing corrosion due to high density of lattice defects in weld metal (Ref 52). The higher amount of martensite not only increases the galvanic corrosion rate but also increases pitting corrosion (Ref 53). Table 6 shows the pitting potential for BM and welded samples with different PD. For BM and high PD (3.6 ms) specimens, the occurrence of metastable pitting is indicated in Fig. 14. From the polarization curve, it shows that for BM sample metastable pits formed at potential 0.18 V. Similarly, for 3.6 ms PD WJ metastable pitting was observed in between potential 0.4 V and 0.6 V. The pitting potential was found to be the lowest for a WJ with 1.5 ms PD. This is due to the presence of more carbide precipitate (indicated by yellow arrow) in the FZ of martensitic lath structure (Fig. 15). It results in greater susceptibility to pitting corrosion (Ref 54). Weng et al. (Ref 55) also observed that increased precipitation of carbides on the larger width of lath martensite promotes higher hardness value and decreases corrosion resistance for submerged arc WJ of disk rotors. As discussed earlier (Section 4.1), lower PD WJ offers a higher value of peak power leading to a higher amount of energy interaction with the BM, resulting in larger WJ bead (FZ) area (Table 3). The transformation of austenite to a martensite is accompanied by volume expansion developing internal stresses. Besides internal stresses are also formed due to the rapid cooling process. These internal stresses also increase the corrosion rate (Ref 49). In case of welding with 1.5 ms PD, the higher peak power and longer holding time at peak temperature lead to an increase in packet size of martensite as well as its lath spacing during rapid solidification of the melt pool. Table 3 shows finer lath spacing and higher martensite needle density in 3.6 ms PD WJ. The formation of martensite with finer lath spacings results in an improvement of the corrosion resistance behavior in 3.6 ms PD WJ (Ref 51).

Fig. 15
figure 15

Carbide precipitates marked by a yellow colored arrow in lath martensitic structure of FZ for PD (a) 1.5 ms and (b) 3.6 ms (Color figure online)

Full size image

The SEM images of the fusion zone of WJ with different PDs and BM after potentiodynamic polarization test are shown in Fig. 16. More severe corrosion was observed in 1.5 ms PD samples with large number of corrosion pits (marked by red color arrow). Individual pit formation was observed in 3.6 ms PD sample in which small pit formed initially and stopped in later stages. In the BM, only metastable pits were observed. Such metastable pits were repassivated immediately and did not cause any damage to the metal surface. So in the case of WJ, it was observed that the nucleation of pitting corrosion was influenced by the passivity of the FZ and amount of surface defects present in the WJ.

Fig. 16
figure 16

The surface of the specimens after potentiodynamic polarization test in a 3.5% NaCl solution with pits marked by the red arrow for the WJ with PD (a) 1.5 ms, (b) 2.1 ms, (c) 3.6 ms and (d) BM

Full size image

Conclusions

The following conclusions are drawn from the present study:

  1. 1.

    The WJ bead size decreased with an increase in PD of laser welding due to a decrease in EPPD simulating the low heat input effect on welding. Good agreement was obtained between the numerically calculated WJ bead cross section and weld macrostructure, validating the numerical model for pulsed mode laser welding.

  2. 2.

    The microstructure of the WJ with different PDs in FZ and hardened HAZ was almost similar, but variation was observed in sub-critical HAZ with martensite tempering and precipitation of carbides along prior austenite grain boundaries or within grains. The softened HAZ dimension decreased with an increase in the WJ PD. The reduced martensite tempering in the softened HAZ was responsible for higher hardness in this zone with higher pulse duration.

  3. 3.

    With an increase in PD, the yield and peak load increased due to a reduction in the percentage of concavity. The failure locations after tensile deformation changed from FZ to softened sub-critical HAZ with an increase in the PD of the WJ due to tempering of the martensite.

  4. 4.

    Formation of oxide layer in the WJ with larger PD, smaller ratio of cathodic area to anodic area and finer martensite lath spacings in the FZ offered improved resistance to pitting and galvanic corrosions. Nucleation of pitting corrosion in low PD welded specimen was influenced by the passivity of the FZ and the amount of surface defects in it.