Introduction

H13 is a typical hot work tool steel that is alloyed with elements of Cr, V, Mo, Mn and Si to enable its superior thermal strength, good red hardness and excellent resistance to thermal fatigue and wear. It resists softening up to 540 °C and is widely used to make dies for hot forging, hot extrusion or high-pressure casting of low-melting point metals and alloys such as aluminium and magnesium alloys [1].

Additive manufacturing (AM) builds parts from three-dimensional (3D) digital models typically by a layer additive process. The availability of affordable ≥400 W fibre laser since 2007 has significantly stimulated the development and application of metal AM over the last decade [2,3,4]. Die or mould making is a costly and time-consuming process. AM, however, has the potential to transform the die-making industry. Firstly, the lead time for mould making can be greatly shortened, from months to days as quality moulds can be printed directly from 3D design models [5, 6]. Secondly, the metallurgical bonding that forms between layers of metallic materials during AM can ensure nearly full density and good mechanical properties for mould applications [7,8,9,10,11,12]. Finally, the advantage to realise conformal cooling is a unique and unrivalled attribute of AM to the mould making industry [13, 14]. Consequently, AM of hot work steels including H13 has attracted increasing attention [15,16,17,18,19,20,21,22,23].

Selective laser melting (SLM) is a powder-bed-fusion-based AM process. Aside from being able to produce intricate moulds with a nearly full density, SLM is often accompanied by a fast cooling rate, which, on the one hand, can produce a refined microstructure [8, 24, 25]. This, however, can entail large residual stresses or distortion due to steep thermal gradients and/or significant phase transformation stresses [26,27,28]. For example, Lu et al. [14] found that the residual stress in SLM-fabricated Inconel 718 alloy samples was about 200 MPa, affected by the laser scanning strategy. Griffith et al. [29] used a holographic-hole drilling technique to determine the magnitude and distribution of residual stress in H13 steel samples fabricated by a laser engineered net shaping technology and reported that the residual stress was up to 260 MPa. Li et al. [30] studied the residual stress issue during SLM using a coupled thermo-mechanical model. They found that the residual stress on the top layer of the sample was related to the laser scanning strategy adopted: longitudinal residual stress (~ 350 MPa) was greater than that was measured crosswise in the vertical sequential scanning mode; the residual stress decreased more or less linearly when moving downwards along the built direction. Apart from these studies, high compressive residual stresses (~800 MPa) were measured from spray-formed (a rapid solidification-based AM process) H13 steel samples [31]. The process is similar to SLM in some ways: both processes are solidification-based layer AM processes with a high cooling rate.

From the aforementioned studies, it can tentatively be concluded that the residual stress arising from laser processing is both material and laser scanning strategy dependent, and the stress level can be significant, e.g. close to or even greater than half the yield stress of the material. The potential consequences include (a) decreased geometrical accuracy and stability achievable by the printed material [32]; (b) distortion, cracking or even breakage of parts during SLM or post-processing [14, 33,34,35]; (c) (if the stress is in tensile) decreased fatigue strength and resistance to stress corrosion [36, 37]. In particular, distortion caused by residual stresses can be a major concern in the AM of intricate dies or moulds by SLM. As such, the steel substrate used in the SLM process often needs to be sufficiently thick and strong in order to counter the potential distortion of the part being built. Also, the as-built parts need to be annealed for stress relief. Owing to the stringent requirement for dimensional accuracy, it is important to understand the magnitude and state of residual stresses in the as-built dies or moulds.

In this study, nearly full dense H13 steel samples were made using the SLM technique. High-angle X-ray radiation (XRD) was then used to characterise the residual stress in the as-built H13 samples. Optical microscopy (OM), scanning electron microscopy (SEM) and TEM were used to understand the microstructural details. We found that the residual stress in some samples exceeded even 1000 MPa. The corresponding stress relief heat treatment therefore may have to have both the microstructure of the as-built H13 and the high residual stress discovered thoroughly considered.

Experimental

SLM of the H13 tool steel

Spherical, gas-atomised H13 steel powder in the size range 25–44 μm was used. Its chemical composition is listed in Table 1. An SLM Solutions 250 HL facility (400 W Yb:YAG laser) was used. Cubic samples (10 × 10 × 10 mm3) were produced at a laser power of 150 W, a scan speed of 300 mm/s using an alternate raster pattern and a hatch spacing of 50 μm under argon (Ar). A detailed study of the SLM process of this alloy can also be found in [18]. A 316L stainless steel substrate was used and kept at 200 °C during SLM. For comparative analysis, conventionally produced H13 supplied by the UDDEHOLM ORVAR®SUPREME was analysed, referred to as ‘as-supplied H13’ hereafter. It was austenitised at 1025 °C for 30 min, followed by air cooling, and then tempered at 610 °C for 2 h.

Table 1 Chemical composition of the H13 powder used in this study
Full size table

Characterisation of residual stress, microstructure and microhardness

Residual stress can be assessed using a hole drilling approach, neutron diffraction and X-ray diffraction (XRD). We used the high-angle XRD approach, which is reliable, rapid and non-destructive with a measurement uncertainty of ~5% [38]. The experiments were conducted on an XRD instrument (at 20 kV and 4 mA) purposely designed for the stress measurement. The target material is Cr Kα with a wavelength of 2.291 Å. Peak from the (211) plane of the α-Fe phase was measured by centring at the 2θ value of 156.4°. Slice samples (~0.5 mm thick) were prepared from as-built cubes using a high-precision low-speed diamond saw. They were analysed without polishing to ensure minimised external influences on the as-built microstructure as well as residual stress. Detailed locations of the high-angle XRD sample slices are indicated in Fig. 7a. The analysis was focused on the central region of each sample slice using a ϕ2 mm XRD beam aperture.

The phase constitutions of both the as-supplied and as-built H13 samples were investigated using a lab XRD (Rigaku SmartLab) equipped with a high-flux X-ray copper source (9 kW power). The scan rate was 1.5 °min−1 in the 2θ angular range of (20–100)o. The microstructure was characterised using scanning electron microscopy (SEM, Merlin, ZEISS, Germany; 10 kV) and TEM (JEM-2010F, 200 kV). TEM samples were prepared by twin jet electropolishing at −30 °C, in a solution of 5% HClO4 + 95% C2H5OH (vol.%). A thin layer of gold coating (~ 5 nm thick) was applied on top of each TEM sample using a Q150T sputter coater for calibrating the camera length of the TEM as an internal standard. Digital Micrograph version 3.7.4 was used to analyse the TEM results. Density of the as-built H13 was measured by the Archimedes method. Hardness was measured using an HXD-1000TMC/LCD microhardness tester, with a dwell time of 10 s at a load of 500 gf.

Results

Preliminary characterisation by XRD and optical microscopy

Figure 1 shows the XRD results for both the as-supplied and the as-built H13 with insets showing their optical microstructures. The as-supplied H13 contains dispersed secondary particles, while the microstructure of the as-built SLM H13 is featured by fine strips. An analysis of the XRD data indicates that the as-supplied H13 consists of the bcc-structured α-Fe (a = 2.87 Å) as the dominating phase, and the cementite Fe3C phase (orthorhombic structure, with a = 5.09 Å, b = 6.74 Å and c = 4.52 Å), where the strongest peak from the (031)Fe3C overlaps the diffraction from the (110)α-Fe. In contrast, the as-built H13 was found to further contain retained austenite (face-centred-cubic structure, with a = 3.62 Å), denoted as the ‘Aˈ’ phase in Fig. 1, and martensite phase (Fe1.86C0.14, tetragonal structure with a = 2.85 Å and c = 3.05 Å), marked out as the ‘m’ phase in Fig. 1.

Figure 1
figure 1

XRD spectra of the as-supplied (i.e. as-purchased) and as-built H13 steel samples. In this figure, A′ stands for the retained austenite phase and m denotes the martensite phase. Figure insets are OM micrographs of the as-supplied and as-printed H13 steel samples

Full size image

The relative density of the as-built H13 reached 99.7 ± 0.1%, which is essentially pore-free and is consistent with microscopic observations. Aside from the microstructure, there is another clear difference in terms of hardness between the as-built H13 and the as-supplied. It was measured to be 57 ± 1 HRC for the as-built H13 versus 45 ± 1 HRC for the as-supplied H13.

Detailed characterisation of the as-supplied H13 by SEM and TEM

Figure 2(a) shows an SEM image of the as-supplied H13. Together with the TEM analysis shown in Fig. 2b, c and the XRD spectrum shown in Fig. 1, it can be concluded that matrix of the as-supplied H13 is composed of overwhelmingly α-Fe phase. As pointed out earlier, the secondary phases shown in Fig. 2a are predominantly cementite Fe3C particles according to the XRD spectrum. TEM characterisation, however, revealed that there is also the (Cr,Fe)7C3 carbide phase existing in the matrix. The selected area electron diffraction (SAED) patterns of the phases shown in Fig 2b are presented in Fig. 2c. The SAED patterns recorded are complicated. A detailed analysis confirmed that they can be decomposed into diffraction patterns of two phases, i.e. the matrix α-Fe phase, and the (Cr,Fe)7C3 carbide phase, as shown in Fig. 2c. In addition, it can also be seen from Fig. 2b that both the Fe3C phase and the (Cr,Fe)7C3 carbide phase are spherical precipitates of a few hundred nanometers in size.

Figure 2
figure 2

a SEM image for the as-supplied H13 showing α-Fe and cementite (Fe3C), and b TEM image reveals that there is an additional carbide phase, i.e. the (Cr,Fe)7C3 phase, except the Fe3C phase

Full size image

Detailed characterisation of the as-built H13 by SEM and TEM

Figure 3(a) shows an SEM image of the top layer of the as-built H13. Distinctly different microstructural features from those of the as-supplied H13 were observed. TEM was employed to detail the microstructures in slice samples prepared from the bottom (i.e. close to the substrate, Fig. 3b), and middle (Fig. 3c) of an as-built H13 cube, as well as a sample sliced along the build direction of the cube (Fig. 3d). Theses TEM bright field (BF) images suggest that there is a low level of preferred orientation of grain growth in the x–y plane, while a strong orientation exists along the build direction (z direction).

Figure 3
figure 3

SEM and TEM images for the as-built H13: a SEM image for the topmost surface, b TEM-BF image for the sample position located at the bottom of the build, c TEM-BF image for the sample position located at the middle of the build, and d TEM-BF image for the sample that was taken along the build direction and from centre of the build

Full size image

Further TEM microstructural observations of the as-built H13 are shown in Fig. 4a–d, along with insets to show corresponding SAED results of each phase included. These SAED results are recognised to be diffracted from the α-Fe phase (Fig. 4a), the Fe3C phase (Fig. 4b), the martensite phase (Fig. 4c) and the retained austenite phase (γ-Fe, fcc with a = 3.66 Å; Fig. 4d). The results are consistent with the XRD analyses shown in Fig. 1.

Figure 4
figure 4

TEM BF and/or DF images and corresponding SAED results for the major phases observed in the as-built H13: a the α-Fe phase, b the cementite Fe3C phase (DF image here), c the martensite phase and d the retained austenite phase

Full size image

TEM characterisation has also revealed that the lattice of the α-Fe phase in the as-built H13 is slightly distorted. Figure 5 provides the corresponding SAED results to compare the lattice parameters of the matrix phase in the as-supplied (Fig. 5a) and the as-built (Fig. 5b) H13 samples, where a thin layer of Au was used as the internal standard (plane distance of the (111) plane of the Au is known to be 2.35 Å). We found that, in the SAED patterns recorded, the ratio between the (211)α-Fe diffraction spot and the radius of the Au diffraction ring (111)Au is slightly lower for the as-built H13 than that obtained for the as-supplied, which is 1.958 versus 1.960 (error bar estimated to be ± 0.0005). This indicates that the lattice parameter of the α-Fe phase in the as-built H13 is higher than the as-supplied according to the established relationship of R·d = L·λ. We further found that Cr, Mo and V elements are enriched in the α-Fe phase in the as-built H13, see Table 2. This can be attributed to the high cooling rates encountered during SLM. In other words, the α-Fe phase in the as-built H13 is not in an equilibrium state.

Figure 5
figure 5

TEM-SAED patterns for the α-Fe phase in the as-supplied H13 a and as-built b H13 samples. A thin film of Au (~5 nm thick), whose diffraction was used as an internal standard, was coated on each sample

Full size image
Table 2 Chemical composition of the α-Fe phase in the as-built and the as-supplied H13 measured by TEM–EDX (data in at.%; error bar ±0.2 at.%)
Full size table

Residual stress measurements of the as-built H13

The normal residual stress (σx or σy where x and y are orthogonal; see Fig. 7) can be calculated from σ = m·(E/1 + ν), where E is Young’s modulus, v is Poisson’s ratio, and m is the slope of the d versus sin2 ψ curve [39, 40]. A plot of d versus sin2 ψ is shown in Fig. 6 based on the experimental data obtained. The slope is -0.00309. H13 has typical value of E = 210 GPa and ν = 0.3 [18, 41]. A high, compressive residual stress of ~ 1000 MPa was obtained from this slice sample. Six similar slice samples were analysed, with their locations in the as-built H13 cube being specified in Fig. 7a. The detailed results are listed in Table 3. Figure 7b summarises the residual stress versus build distance from the substrate, suggesting that residual stress builds up almost immediately after the first two layers during SLM (the stress-free thickness is limited to ~100 μm). It is also noted that the residual stress is distributed almost across the entire as-built material, ranging from ~940 to ~1420 MPa, compared to the yield strength of about 1650 MPa for the as-supplied H13 [18, 41] (Fig. 8).

Figure 6
figure 6

Plot of d versus sin2 ψ measured for the as-built H13 using high-angle XRD

Full size image
Figure 7
figure 7

a Schematic illustration of the locations of sample slices analysed using high-angle XRD and b residual stress distribution along the build direction (z axis) of the as-built H13

Full size image
Table 3 Residual stress of the as-built H13 measured using high-angle XRD (error bar ±5%)
Full size table
Figure 8
figure 8

Schematic illustration of the cooling conditions proposed for the as-built H13; the actual cooling rate at different locations can vary over a wide range due to the cyclic thermal conditions that occur during SLM. The M s and M f temperatures of the H13 steel are indicated in the figure

Full size image

Discussion

The experimental results and analyses presented above have indicated that: (1) AM by SLM enabled the formation of martensite and also its partial decomposition into α-Fe and Fe3C in the as-built H13; and (2) the SLM-fabricated H13 has a high-level residual stress in the as-built condition (Table 3).

Research has shown that the cooling rate during laser powder deposition of the H13 steel can reach ∼1.83 × 104 K/s (which is similarly (1.2–4.0) × 104 K/s for laser powder deposition of Ti-6Al-4 V) [42]. The critical cooling rate for martensitic transformation in the H13 steel is ~20 K/s, while its M s temperature is about 613 K (=350 °C) and its M f temperature is about 323 K (=50 °C) [18, 41]. The observation of the martensite phase in the as-built H13 is therefore understandable. In fact, the significantly higher hardness of the as-built H13 (= 57 ± 1 HRC) versus that of the as-supplied H13 (=45 ± 1 HRC) also supports the martensitic structure observed in the in the as-built H13 (a marginal increase in hardness can be attributed to the refined microstructure). Owing to different cooling rates in different regions of the cube and the cyclic thermal effects from successively build layers, some of the martensite structures may have decomposed into α-Fe and carbides (i.e. Fe3C).

With regard to the residual stress, the total strain (∆ε) developed during the cooling stage can be expressed as a combination of thermal (T), elastic (e), plastic (p) and phase transformation (Tr) factors: Δε = Δε T + Δε e + Δε p + Δε Tr [43]. The H13 steel has yield strength (σ 0.2 ) of ~1650 MPa [18, 41]. The high-level residual stress detected in the as-built H13 is thus still in the elastic deformation range. If we assume that the elastic strain is fully recovered (i.e. zero) in the as-built H13 and no plastic deformation has occurred, then the overall residual stress level can be regarded as a balanced result between the strain (Δε Tr) which is induced by phase transformation and the strain (Δε T) which is induced by the thermal gradient. The former normally leads to compressive residual stress via the relationship Δσ = E·Δε Tr, but the latter normally corresponds to tensile stress. Murakawa et al. [44] and Francis et al. [45] have shown that if the martensitic phase transformation temperature (e.g. M s ) is lower than 400 °C, the overall residual stress can be a large, compressive one. Since the M s temperature of the H13 steel is merely 350 °C, it can be concluded that the martensitic phase transformation is the key reason for the high-level compressive residual stress detected, which is in good agreement with Refs. [44, 45].

Normally the high residual stress detected in the as-built metallic materials should be relieved. Stress relieving of the as-supplied H13 is typically carried out in the range of 600–650 °C for 2 h. SLM-fabricated H13 may need a different annealing treatment to accommodate both its microstructural features and the high residual stress detected. These require another detailed, specific study to make a good investigation.

Conclusions

Detailed microstructural characterisation and residual stress measurements have been conducted for the SLM-fabricated H13 in the as-built condition, using the as-supplied H13 as a reference. The following conclusions can be drawn from this research.

  1. 1.

    High-angle XRD has been used to measure the residual stress in the as-built H13, which shows rather high compressive residual stresses that vary in the range of 940–1420 MPa. The residual stress exists almost throughout the as-built H13. The martensitic transformation that occurs during the SLM of the H13 steel is proposed to be the main contributing factor to the high compressive residual stress detected.

  2. 2.

    Detailed microstructural analysis has shown that the major crystalline phases in the as-built H13 are α-Fe, Fe3C, retained austenite (γ-Fe) and martensitic phase. In comparison, α-Fe, Fe3C and (Cr,Fe)7C3 are the three detectable phases in the as-supplied H13. The α-Fe phase in the as-built H13 has a slightly higher lattice parameter than that in the as-supplied H13 due to higher contents of Cr, Mo and V.