Scripta Materialia 121 (2016) 18–22

Contents lists available at ScienceDirect

Scripta Materialia

j ourna l homepage: www.e lsev ie r .com/ locate /scr ip tamat

Fusion zone microstructure evolution of fiber laser weldedpress-hardened steels

Dulal Chandra Saha a,⁎, Elliot Biro b, Adrian P. Gerlich a, Norman Y. Zhou a

a Centre for Advanced Materials Joining, Department of Mechanical & Mechatronics Engineering, University of Waterloo, 200 University Avenue West, Waterloo, Ontario, N2L 3G1, Canadab ArcelorMittal Global Research, 1390 Burlington Street East, Hamilton, ON, L8N 3J5, Canada

⁎ Corresponding author.E-mail address: dcsaha@uwaterloo.ca (D.C. Saha).

http://dx.doi.org/10.1016/j.scriptamat.2016.04.0321359-6462/© 2016 Elsevier Ltd. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 4 March 2016Received in revised form 20 April 2016Accepted 22 April 2016Available online xxxx

Fusion zone microstructures of fiber laser welded Al-Si coated press-hardened steels at pre- and post-press-hardened conditions were analyzed using transmission electron microscopy (TEM). TEM and nanoindentationstudies suggest that dislocation-free δ-ferrite phase formed in the weld metal (by ingress of the Al-Si surfacecoating into the fusion zone) exhibits lower nanohardness compared to α-ferrite transformed after presshardening. The nanoscale properties of δ-ferrite, α-ferrite, and martensite and their crystallographic orientationrelationships are reported.

© 2016 Elsevier Ltd. All rights reserved.

Keywords:Press-hardened steelFiber laser weldingCoating mixingTransmission electron microscopyNanoindentation

Boron containing press-hardened steels (PHS) are widely usedfor automotive crashworthy structural components such as A-pillars,B-pillars, and door rings [1–4]. Boron is added to the steel to increaseits hardenability and retard the heterogeneous nucleation of ferriteat austenite grain boundaries [5,6]. After press hardening, this gradeof steel reaches an ultimate tensile strength between 1500 and2000 MPa. In order to prevent corrosion and high temperature oxida-tion, PHSs are often coated with various alloy coatings including Al-Si,Zn, and Zn-Ni [7–9]. Laser beam welding is usually employed to joinvarious grades of PHS with different thicknesses and dimensions. Astudy on laser welding of PHS [10–14] confirmed the formation ofsecondary phase in the fusion zone (FZ) due to mixing of Al-Si coatingin the weld pool. The secondary phase found in the FZ of laser beamwelded PHS is reported to be either brittle Fe-Al intermetallic com-pound [10–12] or soft ferrite phase [13–15]. However, no supportingevidence for either case has been provided in the literature. Thesephases were demonstrated to be rich in ferrite stabilizing elementssuch as Al [10–15]. The existence of the brittle intermetallic compoundor soft ferrite phasewithin the FZ of PHSwas reported to be deleteriousto the mechanical performance of the weld [10–15].

Despite the dramatic impact on mechanical performance, little isknown about the formation mechanism of the secondary phase andits crystallographic relationship with the martensite during laser beamwelding of Al-Si coated PHS. In our recent investigation [15], the forma-tion mechanism of the secondary phases and their structure-properties

relationshipwas systematically reported. The aimof the present study isto investigate the crystallographic and nanoscale properties of thephases formed in the FZ of laser welded PHS samples before and afterpress hardening.

The study was conducted on steel (thickness: 1.00 mm) with acoating composed of Al (90 wt.%) and Si (10 wt.%), having a yieldstrength and ultimate tensile strength of 424MPa and 570MPa, respec-tively. The composition of the used steel is reported in Table 1. Theedges of 200 mm × 100 mm (length × width) sheets were milled andwelded in a butt joint configuration using a IPG photonics ytterbiumfiber laser system (model: YLS-6000-S2) with a power and speed of4 kW and 4 m/min, respectively; detailed description of the laserwelding system can be found in previous studies [16]. The weldedcoupons were austenitized in a furnace at 930 °C for 5 mins and thenpress-hardened (at an approximate average cooling rate of 30 °C/s)between two flat dies, where a fullymartensitemicrostructurewas pro-duced [1,17,18] with an ultimate tensile strength of above 1500 MPa. Ascanning electronmicroscopy (SEM), transmission electronmicroscopy(TEM), and high-angle annular dark-field scanning transmission elec-tron microscopy (HAADF-STEM) imaging were used to study resultantmicrostructure. For TEM sample preparation, 3mmdiscs were punchedfrom the foil (thickness: b50 μm);with the FZ at the foil center;with thefoils lightly etched with 2% Nital solution to observe the FZ clearly priorto punch discs. Finally, the samples were thinned by electrolytic jetpolishing at 15 °C using a solution of acetic acid and 5% perchloric acidat 70 V with 60 mA. TEM study was performed in a JEOL 2010F, withan operating voltage of 200 keV. Nanoindentation was conductedusing a Hysitron Triboindenter TI-900 equipped with a scanning probe

Table 1Chemical compositions (wt.%) of the press-hardened steel (PHS) used in this investigation

C Mn P S Si Cu Ni Mo Cr

0.23 1.22 0.013 0.001 0.27 0.02 0.037 0.02 0.20

Co V Al Sn Ti N B Fe

0.008 0.008 0.039 0.02 0.037 0.0054 0.0032 Bal.

19D.C. Saha et al. / Scripta Materialia 121 (2016) 18–22

microscope in a load control condition with a constant loading rate of500 μNs−1 up to a maximum load of 5000 μN; 10–12 indents weremade on each distinct phase in the FZ and the average values arereported (the tolerance limit represents the standard deviation).

Representative SEM micrographs of as-received base metal (BM) inthe pre- and post-press-hardened conditions are shown in Figs. 1a and1b, respectively. The as-received BM microstructure delineated by twodistinct morphologies of ferritic (α)-pearlitic (P) structure (Fig. 1a);the pearlite phase exists as a banded structure along the rollingdirection in a ferritic matrix. The sheets were welded using theweldingparameters as described in the experimental section; the FZmicrostruc-ture is presented in Fig. 1c which is referred to as a pre-press-hardenedcondition. Thewelded structure was austenitized in a furnace, followedby rapid quenching between two flat dies; the BM and FZ microstruc-ture are shown in Fig. 1b and 1d, respectively. Typical lath martensitewith evenly distributed intralath carbides (which are the productof autotempering [19]) are found in the BM after press hardening(Fig. 1b). In the pre-press-hardened condition, the FZ microstructurecomprising of lathmartensite (α1′) and non-equilibrium high tempera-ture ferrite phase (known as δ-ferrite or δ [20,21]) is observed due tomixing of a large amount of Al (Fig. 1c). The δ phase was confirmed byTEM and HAADF-STEM imaging analysis, as discussed in a later section.

It is generally accepted that the addition of Al into steels acceleratesthe kinetics of ferrite and bainitic ferrite transformation [22] and retards

Fig. 1. Representative SEM micrographs illustrating (a) ferritic-pearlitic as-received BM, (b) fmaterial, and (d) a combination of α and α2′ at FZ of press-hardened material.

themartensite transformation kinetics [23] by decreasing the free ener-gy change associated with the austenite to martensite transformation.The α1′ phase embedded in a massive δ matrix with no visible tracesof grain boundaries; the area fraction of each phases was estimatedto be about α1′ (40%) and δ (60%), respectively. The absence of grainboundaries also implies that δ phase formed as a primary phase athigh temperature during solidification, afterward α1′ phase appearedwhen temperature reaches the martensite start temperature. Somesharp δ-α1′ interfaces with a cusp shape (shown with black arrows inFig. 1c) can be found. It is unlikely that a mixture of slender finger-likemartensite (α2′) nucleated at prior-austenite grain boundaries, andtriple point junctions after press hardening; the α grains are dividedwith grain boundaries (marked with red arrows in Fig. 1d) suggestingthat the α phase transformation occurred mainly by the process ofnucleation and grain growth. However, the martensite phase fractionwas estimated to decrease from 40% to about 20% after press hardening.The α2′ appears in this case is free of autotempered carbides which canbe attributed to lowermartensite start temperature. It can be noted thatas quenching was done from the temperature (930 °C) which isbetween Ac1 and Ac3 where only α-ferrite and austenite are present,which allows non-equilibrium δ phase to be equilibriumduring soakingfor 5 mins; therefore, the remaining δ will transform to more stable αphase (i.e., the mixture of δ and α1′ transformed to α and α2′). As aresult, a uniform distribution of α and α2′ structure is obtained in theentire FZ of the press-hardened sample.

Crystallographic analysis of the phases present at FZ of the pre- andpost-press-hardened conditions were performed using TEM with SADpatterns analysis as shown in Figs. 2a-2e and Figs. 2f-2i, respectively.It can be noted that the crystal structure of α, and δ is considered tobe a body centered cubic (BCC) structure whereas α1′ and α2′ is bodycentered tetragonal (BCT) structure. During solidification, when δ isformed at high temperature dislocation networks may generate in theδ phase due to transformation stresses associated with the phase

ully martensitic press-hardened BM, (c) a combination of δ and α1′ at FZ of as-received

Fig. 2. Bright-field TEM study of the FZ of pre-press-hardened sample showing: (a) a mixture of δ andα1′ phase, (b) lower bainite microconstituents. Diffraction patterns of (c) δ in ½�111�projection, (d)α1′ in [001] projection, and (e) an orientation relationship between δ and α1′ ½111�δ||[001]α1

' , and post-press-hardened sample showing: (f) a mixture of α and α2′ phase,(g) magnified view of α-α2′ interface. Diffraction patterns of (h) α in [012]α projection, and (i) α2′ in ½�11�1�α0

2projection.

20 D.C. Saha et al. / Scripta Materialia 121 (2016) 18–22

21D.C. Saha et al. / Scripta Materialia 121 (2016) 18–22

transformation. However, at elevated temperature, these dislocationscan be reorganized into sub-grain boundaries through recovery andpolygonization [24]. Therefore, these sub-grain boundaries can act assites for austenite nucleation [25] and subsequent cooling to roomtemperature promotes the austenite to martensite transformation;therefore, δ is likely to be dislocation-free [26] as observed in Fig. 2a.Sample tiltingwas conducted in order to confirm that only a lowdensityof dislocationswas present in this phase. This specific feature differenti-ates it from the transformed α phase after press hardening as observedin Fig. 2g, where a high density of dislocations can be observed on theαphase and in the vicinity of theα-α2′ interfaces. Bright-field images of δand α1′ and their corresponding SAD patterns taken from the regions(c), (d), and (e) of Fig. 2a are shown. The morphology of δ observed inFig. 2a is consistent with the SEM micrograph presented in Fig. 1c. Theα1′ structure found in the FZ consists of two different kinds of lathswhich can be classified as coarse (width 900–1100 nm) and fine(100–300 nm) laths, depending on their thickness [27,28]. A high den-sity of dislocation (in the order of 1015 m−2 [19]) is observed whichagrees well with the literature in the case of low carbon lath martensite[29]. SAD patterns that were obtained from δ and α1′ were indexedbased on a BCC crystal structure with a derived zone axis of ½111�δ and[001]α1

' , respectively. The lattice parameter of δ was calculated fromthe TEM diffraction spots to be (a) 2.934 ± 0.011 Å, and martensitea = 2.925 ± 0.018 Å and c = 2.925 ± 0.018 Å; it can be noted thatthe calculated a- and c-axis lattice parameter for martensite wasfound to be exactly the same representing a negligible tetragonality(i. e: low c/a ratio). The diffraction spots corresponding to retained aus-tenite that may exist at the interlath locations [30,31] were not detect-ed, which implies that the cooling rate [16,32,33] was sufficiently fastto prevent carbon partitioning into retained austenite phase duringlaser beam welding. In order to determine the orientation relationship(OR) between δ andα1′, SAD patternwas taken from the δ-α1′ interface(zone (e) in Fig. 2a) confirmed theOR ½111�δ||[001]α1

' . The δ phase had anangular deviation of about 14° between ð111Þδ and (001)α1

' projectionplanes. In addition to δ and α1′ phases, a small fraction of lower bainitelike microconstituents also revealed via TEM observation (Fig. 2b). Thecarbideswith a fewnanometers sizewere detected in the bainitic ferritesheaf (sheaf width: 0.5–2 μm). However, detailed characterization ofthe lower bainite was not performed owing to their low fraction.

The post-press-hardened welded coupon was characterized usingTEM and corresponding SAD patterns were obtained from each phasesas represented in Figs. 2f–2i. A slender finger-like α2′ phase which isfound to be highly dislocated is embedded in a αmatrix. The interfacesbetweenα andα2′ phase is outlinedwith dotted lines in Fig. 2g; a pile ofdislocations is observed at the α-α2′ interface as indicated by the ar-rows. Themartensite phase observed in this kindof sample is not typicallath martensite; the common features of martensite such as laths,

Fig. 3. HAADF-STEM images of the FZ showing: (a) δ and α1′ phase at pre-press-hardenedview of α-α2′ interface.

blocks, and packets [34] are rarely visible in this case. SAD patternsthat taken from both the α and α2′ phase were indexed as a BCC andBCT structure showing the projection planes of [012]α and ½�11�1�α0

2,

respectively.The FZ microstructures were further characterized using HAADF-

STEManalysis as presented in Fig. 3. HAADF-STEM imageswere obtainedby correcting z-contrast together with diffraction contrast which clearlyrevealed δ, lath kind ofα1′,α, andfinger-likeα2′phase. In addition, othermicrostructural features such as micro twins (T) in martensite and a pileof dislocations in the vicinity of interfaces are also visible. The δ phaseshown in Fig. 3a is free of dislocations due to high transformation tem-perature [26]. On the other hand, a dislocated α phase is found in thepost-press-hardened sample (Fig. 3b); Fig. 3c shows the magnifiedview of α-α2′ interface where an area with highly dislocated structureis highlighted with arrows. The dislocations introduced into α phaseandα-α2′ interfaces due to the effect of a large residual stress generatedby means of the volume change during austenite to martensite transfor-mation [35]. The compositional analysis carried out in HAADF-STEMmode shows that the δ phase is rich in Al (2.52±0.08wt.%)which is de-creased to 1.70±0.21wt.% inαphase after press hardening;while the Sicontent (about 0.71 ± 0.07 wt.%) remains similar for both of the condi-tions. From the elemental analysis, it may be reported that the diffusionof Al occurred during press hardening; Kang et al. [36] also demonstrateda similar phenomenon. The diffusion of Al increases the carbon activityand its diffusivity in austenite; therefore, a uniform distribution of α2′is observed in a α matrix after press hardening (Fig. 1d, and Figs. 2fand 2g). Unlike, α1′ phase that has typical lath kind of morphology, theα2′ phase contains a few micro twin-like features (highlighted withyellow arrows in Fig. 3b) which are formed in order to accommodatethe distortion energy during martensitic transformation [19].

In order to assess the nanomechanical properties of the phases thatpresents in the FZ of pre- and post-press-hardened samples, instrumen-tation nanoindentation study were conducted. The load-depth (P-h)curves and their corresponding indent impressions are shown inFig. 4. The nanohardness measured on α1′ phase possesses highestvalue ranges from 7.14 to 9.06 GPa, with an average of 7.88 ±0.82 GPa. Conversely, both the ferrite phases were identified to havesimilar nanohardness (δ: 4.96 ± 0.21 GPa, and α: 5.05 ± 0.42 GPa);however, δ phase deformed more severely (average indent depth:150.14 ± 5.56 nm) compared to α phase (average indent depth:142.57 ± 5.19 nm) due to lack of dislocations as observed via TEMand HAADF-STEM study (see Fig. 2a and Fig. 3a). The initial part of theindentation curves is plotted inset of Fig. 4a; the indent impression onthe δ phase shows pop-in at 9.14 ± 3 nm (indent load: 135 ± 50 μN)nm whereas this value extended to 22.72 ± 10 nm (indent load:467 ± 35 μN) for the α phase. The pop-in behavior was confirmedby plotting a Hertzian elastic contact solution according to references

condition, (b) α and α2′ phase at post-press-hardened condition, and (c) magnified

Fig. 4. Nanoscale hardness of the phases present in the FZs of laser beam welded PHS steels showing: (a) load-depth (P-h) curves for the phases of δ, α1′, α, and α2′, (b) and (c) theircorresponding indent impressions.

22 D.C. Saha et al. / Scripta Materialia 121 (2016) 18–22

[28,37,38] in Fig. 4a. The P-h curve on δ phase shows a deviation fromthe Hertzian elastic solution at very early stage of loading representingfully plastic deformation. On the other hand, the indent impression onα phase shows elastic deformation up to 22 nm depth and 467 μNload. The differences in pop-in behaviors is likely to be the influencesof dislocation density which may impede the applied load. Conversely,the P-h curves of α1′ and α2′ shows similar trends up to about3000 μN load afterward the indent on α2′ phase shows a pop-in at adepth of 85 ± 12 nm due to being closed to the α-α2′ interface(Fig. 4c). Unlikely, the indent on α1′ does not experience a pop-in as itlocated on a fine lath and shows an average nanohardness value ofabout 7.88 ± 0.82 GPa (Fig. 4b), which is higher than the one obtainedon the α2′ phase (7.14 ± 0.61 GPa). The nanohardness value of themartensite phase decreased after press hardening due to diffusion ofmartensite hardenable elements such as C andMnwhich are estimatedto be lower at α2′ phase compared to α1′ ones.

In conclusion, the fusion zone microstructure of fiber laser weldedpress-hardened steels were characterized using TEM, HAADF-STEM, andnanoindentation study at pre-press-hardened and post-press-hardenedcondition. TEM analysis reveals a predominant non-equilibrium hightemperature δ-ferrite, martensite, a very negligible amount of lowerbainite structure at pre-press-hardened condition, and a mixture of α-ferrite and martensite structure at post-press-hardened condition. Theδ-ferrite phase is found to be dislocation-free whereas α-ferrite, andα-ferrite -martensite interface contains a pile of dislocations. The nano-scale hardness testing revealed early pop-in behavior on δ-ferrite phasesuggesting it is slightly softer than the α-ferrite due to the absence ofdislocations.

References

[1] H. Karbasian, A.E. Tekkaya, J. Mater. Process. Technol. 210 (2010) 2103–2118.[2] D. Cornette, T. Hourman, O. Hudin, J.P. Laurent, A. Reynaert, SAE Trans. J. Mater.

Manuf. 110 (2001) 37–47.[3] H.Y. Kim, J.K. Park, M.-G. Lee, Int. J. Adv. Manuf. Technol. 70 (2014) 1787–1801.[4] P.O. Marklund, L. Nilsson, Struct. Multidisc. Optim. 21 (2001) 383–392.[5] H.K.D.H. Bhadeshia, Bainite in Steels, The Institute of Materials, London, 1992.

[6] S.K. Banerji, J.E. Morral, Boron in Steel: Proceedings of the International Symposiumon Boron Steels, TMS-AIME, Warrendale, PA, 1980.

[7] D.W. Fan, B.C. De Cooman, Steel Res. Int. 83 (2012) 412–433.[8] D.W. Fan, H.S. Kim, B.C. De Cooman, Steel Res. Int. 80 (2009) 241–248.[9] D.W. Fan, H.S. Kim, J.K. Oh, K.G. Chin, B.C. De Cooman, ISIJ Int. 50 (2010) 561–568.

[10] R. Vierstraete, W. Ehling, F. Pinard, L. Cretteur, A. Pic, Q. Yin, Industrial LaserSolutions 25 (2010) 6.

[11] W. Ehling, L. Cretteur, A. Pic, R. Vierstraete, Q. Yin, In, Proceedings of 5th Internation-al WLT-Conference on Lasers in Manufacturing, Munich, 2009 409–413.

[12] C. Kim, M.J. Kang, Y.D. Park, Procedia Eng. 10 (2011) 2226–2231.[13] F. Briand, O. Dubet, C. Choet, in, US Patent 20110226746 A1, 2011.[14] M.S. Kwon, Y.G. Kim, Y.J. Kim, C.Y. Kang, J.P. Kong, M.H. Oh, H.J. Shin, S.T. Oh, US

Patent 20140154521 A1, 2014.[15] D.C. Saha, E. Biro, A.P. Gerlich, Y.N. Zhou, Weld. J. 95 (2016) 147–156.[16] D.C. Saha, D. Westerbaan, S.S. Nayak, E. Biro, A.P. Gerlich, Y. Zhou, Mater. Sci. Eng. A

607 (2014) 445–453.[17] M. Merklein, J. Lechler, J. Mater. Process. Technol. 177 (2006) 452–455.[18] A. Bardelcik, C.P. Salisbury, S. Winkler, M.A. Wells, M.J. Worswick, Int. J. Impact. Eng.

37 (2010) 694–702.[19] H.K.D.H. Bhadeshia, R.W.K. Honeycombe, STEELS Microstructure and Properties,

third ed. Butterworth-Heinemann, Oxford, UK, 2006.[20] S.S. Babu, J.W. Elmer, S.A. David, M.A. Quintana, Proc. Roy. Soc. Mat. A 458 (2002)

811–821.[21] M. Amirthalingam, M. Hermans, I. Richardson, Metall. Mater. Trans. A 40 (2009)

901–909.[22] C. Garcia-Mateo, F.G. Caballero, H.K.D.H. Bhadeshia, ISIJ Int. 43 (2003) 1821–1825.[23] R. Kuziak, R. Kawalla, S. Waengler, Arch. Civ. Mech. Eng. 8 (2008) 103–117.[24] R.J. Dippenaar, D.J. Phelan, Metall. Mater. Trans. B Process Metall. Mater. Process. Sci.

34 (2003) 495–501.[25] T. Furuhara, T. Maki, Mater. Sci. Eng. A 312 (2001) 145–154.[26] D.V. Shtansky, K. Nakai, Y. Ohmori, Acta Mater. 48 (2000) 1679–1689.[27] L. Morsdorf, C.C. Tasan, D. Ponge, D. Raabe, Acta Mater. 95 (2015) 366–377.[28] B.B. He, M.X. Huang, Metall. Mater. Trans. A 46 (2015) 688–694.[29] G. Thomas, Metall. Trans. 2 (1971) 2373–2385.[30] M. Xia, Z. Tian, L. Zhao, Y.N. Zhou, ISIJ Int. 48 (2008) 483–488.[31] J. Chen, K. Sand, M.S. Xia, C. Ophus, R. Mohammadi, M.L. Kuntz, Y. Zhou, D. Mitlin,

Metall. Mater. Trans. A 39 (2008) 593–603.[32] M.S. Xia, E. Biro, Z.L. Tian, Y.N. Zhou, ISIJ Int. 48 (2008) 809–814.[33] J.E. Gould, S.P. Khurana, T. Li, Weld. J. 85 (2006) 111–116.[34] S. Morito, X. Huang, T. Furuhara, T. Maki, N. Hansen, Acta Mater. 54 (2006)

5323–5331.[35] C.-N. Li, G. Yuan, F.-Q. Ji, D.-S. Ren, G.-D.Wang,Mater. Sci. Eng. A 665 (2016) 98–107.[36] M.J. Kang, Y.-M. Kim, C. Kim, J. Mater. Process. Technol. 228 (2016) 137–144.[37] T.-H. Ahn, S.B. Lee, K.-T. Park, K.H. Oh, H.N. Han, Mater. Sci. Eng. A 598 (2014) 56–61.[38] Y. Kim, T.-H. Ahn, D.-W. Suh, H.N. Han, Scripta. Mater. 104 (2015) 13–16.