18Ni-300 maraging steel manufactured using the industrial SLM device was plasma nitrided in combination with and without a prior-heat treatment in order to investigate their interacting influence on the wear and corrosion resistance. For a better understanding of the microstructure evolution and the formation of the nitride layer, the conventionally produced counterparts were thermochemically treated under the same conditions and compared with the AM samples. Due to the rapid solidification, the microstructure of the SLM as-built maraging steel differed significantly from the conventional steel and has a cellular/dendritic solidification microstructure4,17. Light micrographs (Fig. 1) present microstructures of the nitride layer on the as-built sample AM+N, differently thermochemically treated samples AM+DAT+N, AM+SAT+N and a conventional sample treated under the same conditions CM+SAT+N. The conventional maraging steel in the delivered state was solution treated; therefore, it does not make sense to study CM+DAT because it is identical to CM+SAT. The samples were etched in two different etchants (Nital left part of images, ferric chloride right part of images) in order to reveal all the details in the microstructure. The microstructure observed in the nitride layer as well as in the bulk is martensitic in all the samples, although in the CM sample the prior austenite grains that transform to martensite are significantly different. The prior austenite grains of the CM sample are polygonal, while in the AM samples the prior austenite grains have a more irregular shape due to the rapid solidification, which result in the different martensitic microstructure. Etching with Nital shows the melt pools and the morphology inside them, especially the cellular structure, which is better seen at higher magnifications and in detail with the SEM micrographs (Fig. 2). Melt pools can be observed in the as-built AM sample (Fig. 1a) as well as in the aged AM sample (Fig. 1b), with the melt-pool boundary running across the grains and having a slightly different chemical composition44. However, after the solution treatment the melt pools and boundaries disappear due to the diffusion process. The 8- to 10-µm-thick upper layer is not sensitive to the Nital etchant and most probably corresponds to the Fe4N compound layer45. The ferric chloride clearly reveals the nitride-layer thickness and also the grain structure in this layer, but only in the solution-treated AM sample (Fig. 1c) and the conventional reference sample (Fig. 1d). The depth of the nitride zone is very similar for all the samples and is between 115 µm and 130 µm. Visually, the nitride zone resembles the diffusion-layer structure with two regions of approximately equal thickness, with a different sensitivity to the ferric chloride etching (upper dark and bottom light), most probably due to the edge effect of the etching and the maximum residual stresses in the surface. The transition from the nitride layer to the bulk is smooth and visually restricted to a narrow range.
The bulk microstructure was investigated by SEM, where the melt pools are clearly visible and are 100–150-µm wide and 50–90-µm high (left inset in Fig. 2a). In Fig. 2a the cellular microstructure is visible and present in the whole volume. The dendrite cellular structure is a consequence of the rapid solidification and nano-segregation and is well described in the literature31,46. A higher-magnification image (right inset in Fig. 2a) shows very small (~100 nm) precipitates forming at the triple cell junction during the AM process. The precipitates after aging (DAT) form within the whole volume; however, the growth is faster at the prior cell boundaries with a higher density of dislocations47. These features give the impression that the cell structure is still present after aging (Fig. 2b). After the solution treatment and aging (SAT) the cellular structure completely disappears, revealing the morphology of a martensite structure with fine precipitates (Fig. 2c). As the AM+SAT and CM+SAT undergo identical heat treatments, an identical microstructure is also expected. However, from the micrographs shown in Fig. 2c and Fig. 2d, coarser martensite laths can be seen for the AM sample. Furthermore, no precipitates are formed along the grain boundaries of the prior austenite grains and therefore these boundaries are almost invisible. On the other hand, in the CM sample the prior austenite grains are clearly visible with very fine precipitates located along the boundaries. The main reason for the coarser microstructure in the AM sample is the nature of the AM process, where elongated columnar grains are created. Therefore, the size of grains defines the size of the martensite laths, which form as martensite packages inside those grains. The finer the grains, the finer are the packages.
Fig. 3 presents the SEM image of the upper part of the nitrided zone (~ 30 µm). The thickness of the top, amorphous-like layer is approximately 5 to 10 µm, although it seems to be slightly thicker in the CM+DAT+N sample and most probably corresponding to the compound layer, which is very hard and prone to cracks. In the sample AM+N (Fig. 3a) small cracks are present in the compound layer, which can be assigned to the high concentration of internal stresses in the as-built AM material48,49. After aging no cracks are visible in the AM+DAT+N sample due to the stress release. However, some small pores (100 to 200 nm) can be observed in the surface and near-surface zones (up to a depth of 5 µm). Among the pores, some connected pores along the grain boundaries are also observed (Fig. 3b). The previously explained cellular morphology in the AM+N and AM+DAT+N samples is still visible. The solution treatment (SAT) removes the internal stresses and chemically homogenises the AM sample. The result is a disappearing of the cellular structure and the absence of cracks in the compound layer (Fig 3c). The samples AM+SAT+N and CM+SAT+N have similar porosities in the top nitride layer, as well as connected pores (Fig. 3c and Fig. 3d). The transformation of the ε (Fe2-3N) phase to γ' (Fe4N) cause an excess of nitrogen gas and forms the trapped pores beneath the surface50,51. The most reliable explanation for the absence of pores in the AM sample is a high dislocation density46,44 which allows the faster diffusion of nitrogen.
The EBSD performed on the AM, AM+DAT, AM+SAT and CM+SAT samples indexed martensite with a small amount of austenite (retained/reverted) (Fig. 4), also reported by other authors29,31,52. The difference in the bulk microstructure between the AM and CM samples is the result of typical conditions during the SLM process, characterized by rapid solidification and multiple reheatings, which cause elongated grain growth in the building direction. The microstructure consists of lath martensite inside columnar grains (Fig. 4a, 4b and 4c). The heat-treated CM sample (CM+SAT) contains 1–2 % of retained austenite and has small polygonal prior austenite grains without an isotropic structure, which is typical for AM samples. The as-built AM sample, on the other hand, contains 3 % of retained austenite. Aging increases the amount of austenite in the microstructure, since a larger amount of austenite is in equilibrium at the aging temperature, which remains in the microstructure after quenching and is known as reverted austenite31. Accordingly, the amount of austenite in the AM+DAT sample increased up to 11 % (Fig. 4b). In order to avoid austenite after aging a slower cooling rate should be used. However, when the AM sample is solution treated and aged (AM+SAT), again no more than 1–2 % of austenite was detected (Fig. 4c). The significant difference in the amount of austenite between AM+DAT and AM+SAT can be explained by nano-segregations during the AM process, which cannot be eliminated only by aging. Therefore, a larger amount of Ni in certain areas, known as a -stabilizing alloying element, increases the formation of austenite. The absence of nano-segregations in the conventional sample is attributed to its chemically homogeneous structure. The selected SLM process parameters did not lead to any texture formation, as shown in Fig. 4. The previously described difference in the grain morphology, especially between AM+SAT and CM+SAT, is even better seen in the EBSD IPF maps. The packages of martensite formed inside the large columnar grains in the AM material are much larger than the packages of martensite inside small polygonal grains in the CM material (Fig. 4d).
The nitride layers were characterized using an EDS line-scan analysis to explain the elemental distribution, especially the depth of the nitrogen penetration. The EDS line analyses performed on the cross-section samples are shown in Fig. 5. In all the samples the nitrogen signal correlates well with the depth of the nitride layer, which is 115 µm to 130 µm, as seen from the LM micrographs. The content of nitrogen starts to increase in the top 20 µm, which corresponds well with the compound layer visible in the SEM images (Fig. 3) and the hardness depth profile (Fig. 6). The Ti, Mo and Co values are similar and almost constant in all the samples. The signals for nickel show a high scatter, assigned to the redistribution of the nickel due to the higher solubility of nickel in austenite. It is well known that Ni stabilises the austenite phase. The highest scatter is observed for the AM+DAT sample (Fig. 5b), which contains the largest amount of austenite.
Bulk hardness measurements of the samples before nitriding show that the as-built sample (AM) has a lower hardness of 39 HRC (387 equivalent HV), compared to the heat-treated ones (AM+DAT, AM+SAT and CM+SAT) with a hardness of 49–50 HRC (500–515 equivalent HV). The AM sample shows the lowest hardness value because the precipitates do not form (or at least not completely) during the SLM process14. The sample AM+DAT contains the largest amount of austenite; therefore, a lower hardness value would be expected. However, the dislocation cellular structure is not completely removed during aging and this contributes to the higher hardness value, which in the end gives similar hardness values to all the heat-treated samples. The Vickers microhardness depth profiles of the nitrided layer for the AM+N, AM+DAT+N, AM+SAT+N and CM+SAT+N are shown in Fig. 6. The bulk hardness for all the nitrided samples is very similar, in the range of 520 HV0.01. However, slightly lower values as well as a larger scatter within the nitride layer can be observed for the AM+DAT sample, which is attributed to the largest amount of austenite phase in the microstructure. It also shows a more pronounced drop in the surface hardness, which agrees with the higher austenite content and the slightly thinner compound layer, as visible from the SEM cross-section micrographs (Fig. 3). In terms of the surface hardness of the nitride layer, the AM+N sample shows the highest hardness of the compound layer, around 1150±35 HV0.01, while the hardness of the other three samples is about 850±40 HV0.01. The as-built sample has no precipitates, which corresponds to the lowest bulk hardness value. However, during nitriding, both processes (the formation of precipitates and nitrides) take place and increase the hardness to the highest value of all the studied samples. The higher hardness is also due to the cellular structure with its high dislocation density and the nano-oxide particles pinning the dislocations46. On the other hand, in all the samples the hardness decreases with the depth, reaching the value of the base material at a depth of approximately 120 µm, correlating well with the LM images (Fig. 1).
In terms of wear resistance (Fig. 7a), the highest wear rate (wear volume divided by normal load and total sliding distance) in the range of 3.0·10-5 mm3/Nm was obtained for the AM sample. This is due to the high internal stresses, nano-segregations and the small number of precipitates, resulting in a lower bulk hardness (40 HRC) and combined abrasive/adhesive wear, as shown in Fig. 7b. Although the sample is several times thermally exposed during the laser melting of its upper layers, it is just for very short periods of time, which cannot replace the post-aging procedure, which gives the maraging steel its final properties (i.e., high bulk hardness of 48–50 HRC) due to the precipitates’ formation (Ni3(Ti,Al). Interestingly, the aging treatment itself does not significantly improve the wear resistance of the AM sample (AM+DAT). The wear rate is reduced by less than 15 %, with the adhesive wear component still being very strong (Fig. 7c). The main reason lies in the high retained-austenite content. A further improvement in the wear resistance of the AM samples, on the other hand, is obtained by a combination of solution and aging treatment (AM+SAT), providing a high hardness and a more homogeneous microstructure. This results in a reduced abrasive-wear component and an about 40 % improvement in the wear resistance, as compared to the as-built AM sample. However, regardless of the heat-treatment conditions the AM samples with a reduced microstructure homogeneity show an about 15 % lower wear resistance than the conventional samples treated under the same conditions and showing the same bulk hardness. Although the conventional maraging steel is delivered in the soft-annealed condition, the solution treatment additionally eliminates the macro-segregations, thus providing a further improvement in the wear resistance (Fig. 7d). An even more drastic increase in the wear resistance, obtained for all the samples, was provided by plasma nitriding, as shown in Fig. 7a. The wear rates were reduced by 4–7 times. The high hardness of the nitride layer eliminated the adhesive-wear component, at the same time providing a superior abrasive wear resistance (Fig. 7e). The formation of the hard nitride layer more-or-less eliminates the negative SLM effects as well as the heat-treatment history, with all the samples showing very similar wear rates of about 0.4·10-5 mm3/Nm and the wear being concentrated within the top 10 µm of the nitride layer. However, for the AM material, the presence of microcracks in the nitride layer of the as-built sample (AM+N) and the significant number of pores in the near-surface area of the AM+SAT+N sample result in somewhat higher wear rates and more scatter (Fig. 7a). A detailed examination of friction behaviour of tested samples is explained in Supplementary text with appropriated Fig. S1 and Fig. S2.
Potentiodynamic polarization curves for all the AM and CM samples measured in 3.5 % NaCl are presented as Supplementary Fig. S3. The corrosion potentials (Ecorr), corrosion current densities (icorr) and corrosion rates (vcorr) are listed in Table 1. Based on the results of the corrosion measurements, the as-built AM sample has a similar corrosion resistance to the AM+DAT and AM+SAT samples. The as-built AM sample has a high dislocation density, internal stresses49 and chemical nano-segregations31. The internal stresses and chemical nano-segregations make a negative contribution to the corrosion properties, while the absence of precipitates has the opposite effect. The slightly increased corrosion of the AM+DAT and AM+SAT samples is probably the result of precipitate formation during the aging. The nitriding improves the corrosion resistance, due to the formation of a compound layer, which is known to be corrosion-resistant53. Our results show that the nitrided samples exhibit a passive region in the potentiodynamic curves, compared to the samples prior to nitriding, where this region was not observed (see Supplementary Fig. S3). In fact, nitriding without a prior heat treatment achieves more-or-less the same corrosion rate. However, without the prior heat treatment, due to the high internal stresses some cracking of the nitride layer is observed (Fig. 3a). In general, the corrosion rates for AM are higher than for the CM samples due to the appearance of local stresses caused by the multiple melt cavities, the dendritic cellular structure and the increased roughness, which results in increased corrosion in these areas54. A comparison between the AM+SAT and CM+SAT samples shows the improved corrosion resistance of the conventionally produced material due to its better chemical homogeneity, particularly since pores are not present in such material. In the SLM-produced materials, the number of pores is small, but they cannot be completely avoided, which results in a slightly lower corrosion resistance. The sample CM+SAT shows a similar corrosion resistance to the nitrided samples, which can be attributed to the fine-grained structure with fine martensite laths. This is also an explanation for the better corrosion properties of the CM+SAT sample compared to the AM+SAT sample, although in both cases the same heat treatment was performed.
Table 1 Electrochemical parameters determined from the potentiodynamic curves.
Material
|
Ecorr (mV)
|
icorr (uA/cm2)
|
vcorr (um/year)
|
AM
|
-319 ± 4
|
1.30 ± 0.05
|
15.0 ± 0.5
|
AM+DAT
|
-326.5 ± 4
|
1.35 ± 0.03
|
15.5 ± 0.5
|
AM+SAT
|
-231.2 ± 2
|
1.65 ± 0.04
|
18.9 ± 0.5
|
CM+SAT
|
-348.8 ± 4
|
0.87 ± 0.01
|
10.0 ± 0.2
|
AM+N
|
-314 ± 3
|
0.88 ± 0.02
|
10.0 ± 0.3
|
AM+DAT+N
|
-271 ± 2
|
1.06 ± 0.03
|
12.2 ± 0.4
|
AM+SAT+N
|
-307 ± 3
|
0.87 ± 0.02
|
9.9 ± 0.2
|
CM+SAT+N
|
-255.8 ± 2
|
0.93 ± 0.03
|
10.7 ± 0.3
|