3.1. Microstructure and phase composition
The LSQ modified surface of steel is found to possess three distinct zones in quenching layers: quenched zone, intermediate zone and thermally affected zone [7, 19]. A detailed analysis of strengthening phases, element and phase distribution in quenched zone shall be important for the process optimization and performance improvement.
The microstructure of the untreated samples and all the three LSQ processed samples is shown in Fig. 2. The microstructure of the untreated steel was found to be comprised of tempered sorbite (Fig. 2 (a)). The microstructure also revealed the presence of small number of pores distributed on the ferrite, which would seriously affect the service life of the AISI P20 steel mold. It is clear from the Fig. 2 (b, c, d) that the pores were eliminated after LSQ treatment and the microstructure of treated samples was significantly improved. This could be attributed to the formation of new austenite grains because of the lattice recombination and elements diffusion during the heating process. The ferrite phase with pores possessed higher phase changing energy, which promoted nucleation and growth of new austenite. This illustrated that the LSQ process was beneficial in eliminating pores. Further, it could be shown that the microstructure of quenched zone was mainly consisted of lath martensite (LM), plate martensite (PM), retained austenite (RA) and a small number of carbide particles. Fig. 2 (b, c) reveals that the more PM with high carbon content distributed in the primary austenite grains in sample #1 than in sample #2, and the blocky RA were disappeared. The primary austenite grains formed during austenitizing in sample #1 were faintly visible and the grains were larger as well. The cooling process promoted rapid nucleation of PM in the primary austenite. Under the influence of cooling rate and alloying elements on Ms line, some untransformed RA still existed between the PM in the primary austenite. For sample #1, the laser power and laser energy density were higher, which resulted in the coarsening of primary austenite grains [18]. Furthermore, the high austenitizing temperature and heating rate caused more alloying elements (such as C) to diffuse into austenite, which helped in the high-carbon PM formation during cooling. The high-carbon PM formation has also been confirmed from the literature [19]. The temperature and heat input were relatively lower in sample #2. In sample #2, carbides were not fully dissolved during the austenitizing and suppressed the growth of austenitic grains, resulting in relatively finer grains. Moreover, the diffusion of alloying elements retarded dissolution of carbon in austenite, which led to the more formation of low-carbon martensite.
A comparison of sample #1, #2 and #3 showed that the sample #3 was comprised of coarse microstructure with poor uniformity, which contained more LM and some blocky RA distributed in a concentrated and local way (as shown in Fig. 2 (d)). The high scanning speed in case of sample #3 also resulted in a lower laser heating temperature [20], which led to lower carbon content in primary austenite and the formation of more LM. The laser power and scanning speed of sample #3 were higher (in comparison to sample #2), but the laser energy density was relatively lower, which reduced the laser-to-matrix interaction time and increased the cooling rate. Such conditions caused the non-uniform of temperature distribution and element diffusion in the localized regions of the quenched zone. Additionally, low carbon in the austenite and high cooling rate are conducive to the formation of low-carbon martensite during the phase change, which was also confirmed by XRD analysis as shown in Fig. 3.
In addition to detail microstructural investigation, XRD analyses were also performed on the surface of quenched zone, as shown in Fig. 3 and Table 3. On comparing the XRD peaks and relevant data between the untreated and LSQ samples, it was evident that the microstructure of the untreated samples contained ferrite, Cr-rich carbides M7C3 and M3C2. The XRD results of the LSQ samples were consistent with the phase composition observed by SEM, including martensite, retained austenite and a small number of carbides. However, no carbides could be detected in sample #1 and sample #3, which was probably because almost all carbides were dissolved during heating due to the high laser energy density. In comparison to the untreated samples, the XRD peaks of LSQ samples were significantly broadening. The half-height width of the martensite (α-Fe) diffraction peak of sample #1 (0.586) was larger than the ferrite (α-Fe) diffraction peak of untreated sample (0.167), and larger than the martensite (α-Fe) diffraction peak of sample #2 (0.540) and sample #3 (0.531). This indicated that the LSQ process refined microstructure to a larger extent (as shown in Fig. 3). Additionally, broadening of peak was also related to the lattice micro-strain variables generated from the solid-state phase transformation during the heating/cooling process. The peak broadening of the conventional heat treatment is similar to that of laser heat treatment [5]. Therefore, it can be concluded that the broadening of the diffraction peaks of martensite (α-Fe) in this process was mainly caused by the microstructure refinement during the phase changing process. It should be noted that after LSQ, the diffraction peaks of martensite (α-Fe) shifted to the left, in which sample #1 possessed the largest offset (2θ = 44.50°), and the crystal plane spacing d was the largest (as shown in Table 3, d = 2.0343). This may attribute to the fact that the sample #1 was irradiated under high laser energy density (150 J/mm2), more solute atoms (C, Cr, Mo, etc.) dissolved in austenite during austenitization, but these atoms could not precipitate during the rapid cooling, which would enhance the formation of high-carbon martensite. This was also confirmed in the discussion given in section 3.1 (Fig. 2(c)). Furthermore, as shown in Table 3, the diffraction peak area (19993) and peak value of martensite (α-Fe) in sample #1 were the largest, indicating that sample #1 possessed higher crystallinity and contained more high-carbon martensite. The higher the crystallinity, the more regular the arrangement of atoms in the crystal, the greater deformation resistance was obtained [21-22]. For the sample #3, the diffraction peak area (9289) and peak value of martensite (α-Fe) were the smallest, illustrating that the crystallinity of the obtained microstructure was lower, and form less martensite, so that the microstructure uniformity of sample #3 was poor as described in 3.1 (Fig. 2(d)).
Table 3
Analysis data of α-Fe peaks.
|
2θ (°)
|
d (nm)
|
Height
|
Area
|
FWHM (2θ)
|
Untreated steel
|
44.73
|
2.0243
|
1180
|
11927
|
0.167
|
#1
|
44.50
|
2.0343
|
565
|
19993
|
0.586
|
#2
|
44.58
|
2.0310
|
434
|
14159
|
0.540
|
#3
|
44.67
|
2.0268
|
290
|
9289
|
0.531
|
3.2. Residual stress
The LSQ process is capable of changing the state of internal stresses and strengthening surface. The stress distribution also significantly affects the corrosion and wear resistance. The distribution of residual stress in the longitudinal and transverse directions in the strengthened layer of the LSQ treated surface is shown in Fig. 4. The initial stress on the untreated surface was compressive and the average value of the measured stress was -48.7 MPa. The compressive stress in the as received material would have developed during surface machining. The residual stress was measured on the strengthened layer of the LSQ treated surface and found to compressive in nature and significantly increased (average value: -225.1 MPa). It may be noted that the flat-topped laser employed in this work has a uniform energy distribution across the beam cross section. The surface treated by such a beam generated a residual stress profile which was more or less uniformly distributed in the longitudinal and transverse directions. A large number of carbon atoms diffused into austenite during heating and caused distortion in the face-centered cubic (FCC) structure; on subsequent quenching the martensitic transformation changed the lattice from FCC to body-centered cubic (BCC), which developed high volumetric stress in the already distorted lattice and produced significant compressive stress. An increase in the laser energy density increased the diffusion of C atoms (associated lattice distortion too) and in turn increased the compressive residual stress. Furthermore, in addition to lattice distortion, the increase of laser energy density also promoted the formation of martensite, which also added to the residual stresses. This was also confirmed by the XRD analysis. The compressive residual stress on the surface effectively reduced the fatigue cracks and improved the fatigue and corrosion resistance of the treated surface [23-25].
3.3. Element distribution
The LSQ process usually completes in a very short time, consequently, the alloying elements usually do not get enough time to diffuse evenly, leading to fluctuations in concentration in different areas. This raises concern on the homogeneity of microstructure and surface performances. Fig. 5 shows the results of EDS of the untreated and LSQ treatment AISI P20 steel under different laser energy density. The two key elements of C and Cr were analyzed emphatically here, and the EDS results are listed in Table 4. The distribution of C is indicative of the microstructural homogeneity and distinguish the type of martensite present on sight. Similarly, the distribution of Cr is indicative of microstructural regions of poor or rich chromium which accordingly defines the evaluation of corrosion resistance property (especially to the corrosion of Cl-).
It is evident from the Fig. 5(a) that the concentration of Cr and C largely varied and were present on the ferrite boundary in the untreated steel. The microstructure of untreated steel was consisted of tempered sorbite and the C and Cr mainly existed as carbides. The laser heating during LSQ caused diffusion/dissolution/redistribution of elements (C and Cr) and the fluctuations in the concentration was evened out (Fig. 5(b, c, d)). A comparison of sample #2 with sample #1, indicated that the distribution of C in sample #1 was more uniform (Fig. 5(b)). Almost all the carbides were dissolved under the effect of higher laser energy density, the higher temperature and the lower scanning speed. These factors also increased the diffusion rate of C, prolonged the effective diffusion time [26] and resulted in uniform distribution of C. The uniform distribution of C enhanced the uniform and dense microstructure, as shown in Fig. 2(b). Combining with the point analysis revealed that the carbon content of RA in sample #1 was the highest in all the samples, which was 1.7% (point A). This indicated that more C atoms diffused into austenite and retained during the austenitizing. Meanwhile, it is also evident that more PM with high carbon content were formed in sample #1. The lower temperature and lower heat input, resulted in some carbides to remain undissolved in sample #2, which caused very high nonuniformity of C distribution than sample #1. Further, the number of C atoms in solid solution was lower which was found to be only 1.1% (point B). The distribution of C in sample #3 was found to be similar to that of untreated steel, with large fluctuation of carbon concentration and poor distribution (refer to Fig. 5(d)). Most of C atoms were found on the body and boundary of RA surrounded by LM. This was attributed to the fact that high scanning speed caused insufficient diffusion of C and significantly affected the uniformity in the microstructure. The distribution of Cr was found to be more uniform after LSQ which was desirable in reducing the local corrosion caused by the presence of a large number of poor chromium regions [12, 17]. Therefore, the corrosion resistance of mold steel can be significantly improved by LSQ treatment.
Table 4
EDS analysis of P1-P3 (wt. %).
|
Cr
|
C
|
Mn
|
Si
|
Fe
|
Point A
|
1.5%
|
1.7%
|
0.5%
|
0.2%
|
96.1%
|
Point B
|
1.2%
|
1.1%
|
0.5%
|
0.4%
|
96.8%
|
Point C
|
1.3%
|
1.3%
|
0.4%
|
0.3%
|
96.7%
|
3.4. Carbides ablation and microstructure evolution
3.4.1. Carbides ablation
The dissolution/ablation of carbide imparts significant effects on the diffusion and distribution of C. When the laser power is low, there may be some incompletely dissolved carbides remaining in the quenched zone. Fig. 6 shows the high magnification SEM micrographs of two carbides with different shapes which were distributed in the quenched zone in sample #2. The EDS point analysis data (listed in Table 5) and the results of XRD, could confirm that the block shaped carbides (as shown in Fig. 6(a)) were Cr7C3. The rod-shaped or elliptic carbides were identified as Cr3C2 (as shown in Fig. 6(b)). It can be evident that the irregular ablation layers and the particles which were broken and exfoliated due to ablation exist at the carbide boundary (Fig. 6(a)). The Cr3C2 has better thermal stability as compared to Cr7C3 [27-28]. Consequently, the Cr3C2 ablated relatively lower and the ablation layer (broken particles characterizing ablation) at the boundary of Cr3C2 was not obvious (as shown in Fig. 6(b)). Moreover, Cr7C3 underwent secondary reaction with C at high temperature and formed Cr3C2 (as given by Formula 1) [29-31].
Table 5
EDS analysis of C1-C2 (wt. %).
|
Cr
|
Mn
|
Si
|
Fe
|
C1
|
86.5%
|
0.4%
|
0.5%
|
3.0%
|
C2
|
83.4%
|
0.3%
|
0.3%
|
2.8%
|
High magnification SEM at the local regions could be effectively utilized in the analysis of undissolved carbides in the quenched zone of each sample (refer to Fig. 7). For sample #2 which was treated with lower laser power, undissolved carbides in the quenched zone were observed, most of which were rod-shaped Cr3C2 with high thermal stability (as shown in Fig. 7(b)). In comparison to Cr7C3 blocks, the Cr3C2 mostly appeared with relatively clear boundaries with slight dissolution/ablation phenomenon (as analyzed and discussed in the Fig. 6 of preceding section). The sample #1 was treated with the highest laser energy density (as shown in Fig. 7(a)), and it did not possess any trace of undissolved carbides in the quenched zone. It can be inferred that almost all the Cr7C3 and Cr3C2 were completely dissolved because of the high heat input and long interaction time. An observation of Fig. 7(c) reveals that more undissolved carbides with irregular shape were present in sample #3, especially the presence of Cr7C3 with large block shape. The carbides were embedded in the shallow surface of the matrix, and their interface with the matrix appeared fuzzy. In comparison to sample #2, the size of undissolved carbides in sample #3 was larger and the degree of ablation was lower. Furthermore, as the Cr7C3 generally dissolve completely at about 870 ℃ [32], it can be inferred that the temperature of sample #3 might have reached between Ac1 and Ac3 lines.
3.4.2. Microstructure evolution
Fig. 8 shows the microstructure evolution of the quenched zones treated under different laser energy density. During treatment the irradiation of high energy laser beam heated up the untreated surface rapidly, the nucleation of austenite began at the interface of ferrite and carbides (tempered sorbite) when the temperature exceeded Ac1. During transformation, the C atoms continuously dissolved by diffusion into austenite and promoted the growth of austenite. The diffusion and dissolution is a time and temperature dependent process, and different combinations of LSQ parameters produce conditions of different interaction time and temperatures. Consequently, different heat input and interaction time, results in a very widely different microstructure of the quenched zone. The variations in undissolved carbides, type of transformed martensite, number of RA and the carbon content (as shown in Fig. 8) were produced as a consequence of LSQ treatment under different process parameters.
The sample #2 was treated under the lowest laser power and median laser energy density. The SEM and results of EDS reveal the presence of some undissolved carbides, which restricted the growth of austenite grains and promoted the formation of fine LM with low carbon content. Moreover, if the cooling rate during quenching is lower than the critical cooling rate, some untransformed austenite will be retained. For the sample #1, the laser energy density was the highest, the carbides were fully dissolved, and complete austenitic transformation occurred. This resulted in the formation of high amount of high-carbon PM after cooling. However, the sample #3 was treated at the lowest laser energy density (the scanning speed was highest). Under such a condition, a small number of undissolved carbides were observed in the microstructure. The lower temperature and higher laser scanning speed result in non-uniform element diffusion. The heavily nonuniform diffusion causes the evolution of heterogeneously distributed blocky RA (or ferrite), produces the highly nonuniform microstructure, and the formation of low-carbon LM.
The mechanisms of laser surface quenching involves martensitic strengthening, retained austenite solution strengthening, grain refinement strengthening and residual stress strengthening, etc. The dominant mechanism depends on the phase composition, phase distribution and phase size. The results of studies show that the undissolved carbides in the transformed phase are beneficial in improving the wear resistance property, but the phase interface in multiphase microstructure increases the susceptibility of corrosion and reduces the corrosion resistance [8, 33, 34]. Having demonstrated a widely varied microstructural feature produced under different LSQ conditions, it is importance to select the appropriate processing parameters according to the in-service performance requirements.