Introduction

Optimizing the shaping of metallic materials with the goal of obtaining components manufactured closer to their final shape (near net shape) while minimizing the number of machining or assembling steps [1] is a real technological challenge. This optimization has been addressed these last years by the use of Additive Manufacturing (AM) that builds three-dimensional pieces from 3D digital model by progressively adding new layers of material [2]. It appears very interesting to use AM in order to engineer high performance components for different industry fields, like aerospace, automotive, medical or energy [2]. In consequence, a large number of authors report on the optimization of manufacturing parameters such as laser power, scanning speed, hatching space, powder bed thickness.

These works also focus on the influence of the processing parameters on the microstructure and the mechanical properties of the additively manufactured metallic materials. Differences were pointed out between conventional and additively manufactured components, the latter presenting new typical microstructures [3] related to the high thermal gradients and cooling rates involved in AM solidification processes [4]. These new microstructures revealed to highly impact the mechanical behavior [5]. For AISI 316L stainless steel, the typical microstructure of dense materials shaped by SLM is recognized to improve mechanical properties [6].

When elaborated via SLM, 316L presents melt pool traces at the macroscopic scale containing cellular substructures with different orientations without transition zones [3] and with a cell diameter ranging from 0.5 to 1 µm [3]. The literature reports that these cells are defined by the accumulation of large amounts of dislocations as cell walls, where higher amounts of Cr and Mo were also observed as compared to the nominal composition. Some spherical nano-inclusions are also dispersed in the microstructure of SLM 316L components, often at cell boundaries [3].

Despite many industrial applications of AISI 316L, relatively few works have been published on its high temperature oxidation behavior, whether issued from conventional casting [7] or shaped by powder metallurgy [8]. In all these studies, the protection against high temperature oxidation is ensured by the growth of a mixed chromia-spinel oxide layer which loses its protective character after some time, when iron oxides appear. However, different oxidation rates and temperatures above which iron oxides start to grow were reported, as detailed in a previous study [9]. Moreover, high temperature corrosion resistance is influenced by the oxidizing atmosphere composition, such as water vapor presence.

Several authors studied the impact of water vapor on the oxidation behavior of chromia-forming alloys [10,11,12,13,14,15,16]. In general, the presence of water vapor accelerates the oxidation rate. Indeed, after an incubation period during which protective chromium oxides are formed, a breakaway phenomenon appears due to the formation of iron oxides [14]. The oxidation rate before the breakaway seems to be independent of the H2O content. However, the explanation of the breakaway phenomenon diverges [11]. In some studies, it was attributed to volatile Cr-rich species such as CrO2(OH)2 [15] while in other works it was related to dissociation mechanism [16].

Buscail et al. [7] studied isothermal oxidation resistance of AISI 316L in dry air and in air containing 10 vol.% H2O for 96 h in 800–1000 °C temperature range. Results show that at 800 and 900 °C, water vapor presence does not have negative impact, but slightly decreases the parabolic oxidation rate as compared to dry air exposure. At 1000 °C, the kinetics remained parabolic all over 96 h of exposure in dry air, but breakaway occurred after 30 h of oxidation in wet air. In this latter case, growth of hematite Fe2O3 in the outer part of the oxide scale and layered FeCr2O4 in the inner part was reported. The authors estimated that the water dissociation is the main reason of the iron oxides formation. Once water is dissociated on the surface of the sample, hydrogen containing species are likely to diffuse into the oxide scale (initially composed of chromia and Cr-Mn spinel), as shown by Ardigo et al. [17]. When H+ and/or OH− reach the metallic substrate, H2 and FeO form. FeO then reacts with chromia to form FeCr2O4 oxide spinel. In addition, the accumulation of protons inside the oxide layer might be responsible for the change in growth mechanism by modifying the oxide defect chemistry.

Cheng et al. [13] reported breakaway for 316 stainless steel aging at temperatures above 950 °C in air containing 0.1 atm H2O and related their observation to the internal oxidation occurring in the accelerated stage. Alternate oxide layers are formed at the outer part of the substrate because of Cr depletion under a critical value, allowing Fe oxidation. Internal oxidation of Fe and Cr leads to Ni-rich metallic phases in the substrate below. The authors conclude that the Ni-rich phases might catalytically assist the decomposition of the ingresses water vapor, increasing the local partial pressure of hydrogen and oxygen. The internal oxidation conditions became then similar to those of the first step of external oxidation, explaining the rapid formation of the multilayered structure.

The aim of the present paper is to evaluate the impact of water vapor presence in the exposure atmosphere on the high temperature oxidation behavior of wrought and SLM AISI 316L stainless steel components. The two materials were oxidized up to 3000 h in laboratory air and up to 1000 h in wet air (10 vol.% H2O). Corrosion products were then analyzed by XRD and SEM–EDX, via surface and cross-sectional observations.

Experimental Procedures

Wrought AISI 316L, provided by Arcelor Mittal with 1 mm thickness, was used as a reference to compare the corrosion resistance of AISI 316L additively manufactured by Selective Laser Melting (SLM). A spherical metallic powder obtained by argon atomization and provided by Aubert & Duval was employed by BV PROTO to shape SLM samples by means of EOS M270 instrument. SLM plates of 1 mm thickness were cut off perpendicular to the building direction from parallelepiped pillars of dimensions 10 mm × 10 mm × 50 mm. All the samples were grounded on SiC papers up to 1000 grid and ultrasonically cleaned in ethanol for 2 min.

The high temperature oxidation of wrought and SLM materials was evaluated at 900 °C in two atmospheres:

  • in laboratory air up to 3000 h, by using Carbolite muffle furnace

  • in air enriched with 10 vol.% water vapor up to 1000 h, by using the experimental set-up described elsewhere [18].

All the samples were placed in alumina crucibles. They were then discontinuously weighted, after different oxidation times: 100, 200, 500, 1000 and 3000 h. In order to test the reproducibility, two samples were used for each exposure time. After oxidation, kinetics constant rates values were determined using the following equation:

$$t = c + \frac{1}{{k_{l} }}*\frac{{\Delta m}}{A} + \frac{1}{{k_{p} }}*\left( {\frac{{\Delta m}}{A}} \right)^{2}$$
(1)

where (Δm/A) is the normalized mass gain per surface area, t is the oxidation time, kp is the parabolic rate constant and kl is the linear rate constant.

The surface of the oxidized materials was characterized by scanning electron microscopy (SEM) using HITACHI SU8230 set-up with Thermo-Scientific UltraDry EDX probe and by X-ray diffraction (XRD) products at 2° fixed angle of incidence by using BRUCKER D8 Discover equipped with a Cu tube (λ = 0.154056 nm) and a LYNXEYE XE detector. After specific preparation described in a previous paper [9], cross-sectional characterizations were done by using JEOL JSM-7600F SEM with field emission gun (FEG) and Oxford Instruments EDS detector. The data were then analyzed with INCA software (Oxford Instruments).

Results in Laboratory Air

Figure 1 shows the mass gain per surface area as a function of time for wrought and SLM samples. The results highlight that wrought samples present parabolic behavior at least, up to 1000 h of oxidation at 900 °C and then undergo breakaway between 1000 and 3000 h. In contrast, SLM plates follow parabolic rate law throughout the 3000 h of exposure. During the first 1000 h, the oxidation kinetics is very similar for both types of samples (\(k_{p} = 1.4~.10^{{ - 13}} g^{2} .cm^{{ - 4}} .s^{{ - 1}} ~\) for wrought steel and \(k_{p} = 1.7~.10^{{ - 13}} g^{2} .cm^{{ - 4}} .s^{{ - 1}}\) for SLM plate).

Fig. 1
figure 1

Normalized mass gain as function of time for AISI 316L wrought and SLM samples aged at 900 °C in laboratory air

Full size image

SEM observations (Fig. 2) of oxidized AISI 316L samples agree with the mass gain curve. Up to 500 h aging, wrought coupons exhibit 3 µm thin homogeneous oxide scale (Fig. 2a), composed of (Cr,Mn)3O4 spinel grains in top of chromia Cr2O3 layer, as identified by EDS and XRD analyses. After 1000 h exposure, the corrosion scale becomes heterogeneous as iron oxides nodules are present in addition to 6–8 µm thin homogeneous layer (Fig. 2b). After 3000 h oxidation at 900 °C (Fig. 2c), the wrought sample shows very thick (~ 600 µm) oxide scale made of (Fe,Mn)2O3, Fe3O4, (Fe,Ni,Cr)3O4 and (Fe,Mo)3O4. The scale contains a large number of cracks and pores, is not adherent to the metallic substrate and therefore undergoes spallation during the oxidation process. Moreover, deep internal oxidation (~ 150 µm) mainly composed of Fe–Cr rich oxides can be observed under the non-protective oxide scale.

Fig. 2
figure 2

SEM cross-sectional images of AISI 316L wrought (a–c) and SLM (d–f) samples aged in laboratory air at 900 °C up to 3000 h

Full size image

In contrast, the behavior of SLM samples is very different. The oxide scales are thin, homogeneous and adherent to the metallic substrate for all the exposure times. They are composed of Cr2O3 up to 1000 h oxidation; a little amount of Cr–Mn spinel is present in the outer part of the scale after 3000 h of exposure. The thickness of the scales increases with the oxidation time, from 4 to 8 µm after 500 h up to 10–15 µm after 3000 h of exposure.

Pegs made of silicon oxide can be observed at the outer part of wrought and SLM metallic substrates. As shown in Fig. 2, bright precipitates, rich in Cr and Mo, are present up to 1000 h of oxidation at 900 °C and in larger amount for the wrought samples (Fig. 2a, b, d and e).

Figure 3 presents concentration profiles (in wt. %) measured after 1000 h exposure at 900 °C in laboratory air below the metal-oxide interface. Wrought samples exhibit Cr depletion below 12 wt.% (derived from the Wagner’s criterion for maintaining a Cr2O3 scale [19]) in the outer part of the metal (in the first 20 µm below the oxide scale). Even if SLM plates show a slight decrease in the Cr content in the same area, they do not evidence decrease of Cr content below12 wt.%.

Fig. 3
figure 3

Cr content evolution within the metallic substrates as function of the distance from the oxide-metal interface after aging for 1000 h at 900 °C in laboratory air

Full size image

Results in Wet Air

Kinetics evolution in air enriched with 10 vol.% H2O (Fig. 4) highlights a clear difference between the two types of specimens. Wrought samples exhibit very fast oxidation kinetics throughout the 1000 h of exposure with a tremendous mass gain that reaches 40.5 mg⋅cm−2 after 1000 h. In comparison, oxidation of SLM coupons follows parabolic rate law with low mass gain (of 0.5 mg⋅cm−2 after 1000 h exposure). The parabolic rate \((k_{p} = 7.0.10^{{ - 14}} g^{2} .cm^{{ - 4}} .s^{{ - 1}} )\) is twice smaller than the value determined for laboratory air oxidation.

Fig. 4
figure 4

Normalized mass gain as function of the oxidation time for AISI 316L wrought and SLM samples aged at 900 °C in air + 10 vol.% H2O

Full size image

After 100 h aging in wet air, wrought sample (Fig. 5a) exhibits an external oxide layer of about 10 µm, identified as Fe2O3. An inner oxide area can also be observed on a depth of about 85 µm at the outer part of the metallic substrate. As shown in Fig. 6a by cross-sectional EDS elementary mapping, the inner oxide region is composed of alternating layers of Fe–Cr oxides and Ni-rich metallic particles. Below the last Fe–Cr rich oxide layer, Cr depletion, below 12 wt.% is evidenced by the Cr content evolution presented in Fig. 7. Deeper into the substrate, precipitates rich in Cr and Mo are observed. After 1000 h of exposure in wet air (Fig. 5b), the corrosion scale formed at the surface of wrought 316L is extremely thick (~ 600 µm), porous and fractured. The metallic material was almost entirely consumed during the oxidation process (only 200 µm over the initial thickness of 1 mm were not affected by the oxidation). The multilayered oxide structure observed after 100 h of exposure (Fig. 5a) is not present anymore. The outer part of the thick scale is mainly composed of Fe2O3. The middle part of the scale contains Fe, Ni, Cr and Mn and probably corresponds to (Fe,Ni,Cr,Mn)3O4 with variable Fe, Ni, Cr and Mn contents. The inner part corresponds to (Fe,Cr)2O3 with an increase in Cr content toward the oxide-metal interface.

Fig. 5
figure 5

SEM cross-sectional images of AISI 316L wrought (a, b) and SLM (c, d) samples aged in air + 10 vol.% H2O at 900 °C up to 1000 h

Full size image
Fig. 6
figure 6

Cross-sectional EDS elementary mapping of 316L wrought (a) and SLM (b) samples aged for 100 h and, respectively, 1000 h at 900 °C in air + 10 vol.% H2O

Full size image
Fig. 7
figure 7

Cr content evolution within the metallic substrates as function of the distance from the oxide-metal interface after aging at 900 °C in air + 10 vol.% H2O

Full size image

Up to 1000 h, SLM plates (Fig. 5a, b) show dense and adherent oxide layers mainly composed of Cr2O3 with a small quantity of (Cr,Mn)3O4 in the outer part (Fig. 6b). Moreover, the oxide scales are rather thin, of about 3 µm after 100 h and, respectively, 6 µm after 1000 h exposures. In contrast to wrought 316L, any (Cr,Mo)-rich precipitates is not present in the metallic matrix. However, a large amount of silicon oxide pegs can be observed at the oxide-metal interface (Fig. 6b). As for air exposure and in contrast to wrought samples, the Cr depletion near the metal-oxide interface never falls below 12 wt.% for SLM plates even after 1000 h exposure at 900 °C in wet air (Fig. 7).

Discussion

Results in air agree with those obtained in a previous study performed for 100 h in dry air at temperatures between 700 and 1000 °C [9]. Indeed, as observed for a short period in dry air, the present study shows that SLM 316L steel presents a better oxidation resistance than the wrought. Excellent high temperature oxidation behavior of SLM 316L is explained by the growth of thin, compact and adherent chromia scale throughout the 3000 h time exposure acting as diffusion barrier. In contrast, wrought 316L exhibits the formation of non-protective iron oxide nodules. After 1000 h exposure, the iron oxide growth became very fast, leading after 3000 h aging to a large, cracked, porous and non-adherent iron oxide scale. In addition, as for short tests previously reported [9], Cr content measurements at the outer part of the metallic substrates aged for 1000 h at 900 °C falls below the critical value (calculated from the Wagner’s steady-state equation [19] and the assumption that the scale is pure chromia) of 12 wt. % for maintaining a Cr2O3 scale, only for the wrought 316L explaining the formation of iron oxides. As discussed in the previous paper [9], it seems that dislocations induced by SLM process play an important role in Cr diffusion from the bulk toward the subsurface ensuring pure chromia scale growth, even after long aging times. For wrought 316L, the lower number of dislocations in the raw substrate leads to fast Cr consumption and Cr depletion below the oxide scale, allowing iron oxide formation.

The oxidation behavior in air enriched with water vapor is very different for wrought and SLM 316L samples, as shown in Table 1. For wrought samples, oxidation kinetics is very rapid from the first hours of oxidation, testifying of the growth of porous and non-protective oxide scale. After 1000 h at 900 °C in wet air, normalized mass gain is very high (40.5 mg.cm−2) and even larger than after 3000 h laboratory air oxidation (32 mg.cm−2). The detrimental effect of water vapor is confirmed by the oxide scale characterizations. After only 100 h aging at 900 °C, inner oxides were formed on an important depth (up to 85 µm) and lead to a multilayered structure made of (Fe,Cr) oxides and Ni-rich metallic phases. Several authors reported similar behavior of Fe–Cr–Ni alloys after exposure at high temperatures in wet air [7, 10, 13, 20, 21]. Buscail et al. [7] and Otsuka et al. [20] observed internal oxidation with similar multilayered structure after oxidation of 316L for 65 h at 1000 °C and, respectively, of Fe–15Ni–18Cr alloys for 1000 h at 700 °C. When aging for 1000 h at 900 °C in the present study, the internal oxidation phenomenon causes almost complete consumption of the metallic material. In contrast, presence of 10 vol.% of water vapor slightly improves the oxidation resistance of SLM plates as they exhibit slower parabolic kinetics throughout the 1000 h of oxidation in wet air as compared to laboratory air. The resulting oxide scales are made of pure chromia.

Table 1 Normalized mass gain, parabolic constant rate, oxides phases and oxide scales thickness after exposure for 1000 h at 900 °C in laboratory and wet air of wrought and SLM 316L samples
Full size table

Some authors attributed the loss of protectiveness under wet atmosphere to the volatilization of Cr species by the formation of CrO2(OH)2 [15, 22]. In the present exposure conditions, Cr evaporation should occur similarly for both wrought and SLM samples and therefore is not sufficient to explain the poor corrosion resistance of the wrought alloy. Cr species volatilization can be considered in order to discuss the lower kp value determined for SLM plate in wet air as compared to laboratory air. As shown by Peng et al. [21] or Opila et al. [23], the competition between chromia growth and volatile Cr species formation might lead to a weight loss and thus slight decrease in oxidation rate, in agreement with Peng et al. [21] or Opila et al. [23]. Nevertheless, any volatilization phenomenon was not clearly evidenced to occur in the present study, either by the analysis of the kinetics curves or by the observation of the oxide surface morphologies.

Other studies [16] identified the water dissociation at the specimen surface as the main reason for oxide scale failure. Diffusion of hydrogen species in the oxide scale may change the oxide defect chemistry and then the growth mechanism [12, 24, 25]. It is well known that chromia growth in dry air takes place mainly by cationic diffusion. Galerie et al. [26] showed that H-rich species largely increase the inward diffusion for Fe–18Cr steel aged at 900 °C in wet air, leading to higher oxidation rate as compared to dry air. After diffusion within the chromia layer, hydrogen species may react with the metallic substrate and form FeO that will lead to formation of FeCr2O4. This phase was not clearly identified in the present study, but corresponds to (Fe,Cr)-rich oxides identified in the internal oxidation inner oxides area of the wrought alloy, in agreement with Buscail et al. [7] et Cheng et al. [13]. In addition, Ni-rich metallic particles observed in the present study to be present in the wrought metallic substrate after 100 h exposure may act as catalysts for the decomposition of water vapor following the mechanism suggested by Cheng et al. [13], leading to the tremendous mass gain measured after 1000 h aging in these conditions.

Several authors explained that accelerated oxidation occurs for Fe-based chromia-forming alloys because of the Cr depletion of the metallic matrix near the oxide scale [14, 26], whether the oxidation is carried out in air or in another atmosphere. Indeed, after initial growth of a protective layer generally composed of Cr2O3 and (Cr,Mn)3O4, the protective character of the oxide scale is altered by iron oxides formation. Water vapor presence has a detrimental effect on chromia-forming steels, not only by increasing the oxidation rate, but also by decreasing the time when breakaway occurs [10, 11, 20]. These conclusions perfectly describe the behavior observed in the present study for the wrought 316L steel. Indeed, parabolic kinetics occurring up to 1000 h in laboratory air is not observed in wet air. In the case of SLM 316L, breakaway oxidation is not observed in any of the two atmospheres of the present study, indicating again higher resistance to corrosion of the additively manufactured steel. The better behavior of SLM steel might again be related to its particular microstructure containing a large number of dislocations that represent Cr fast diffusion paths. Such conclusion is in agreement with the study of Peng et al. [21] on the oxidation of 304L steel in wet air showing that increased Cr supply leads to improved oxidation behavior.

Conclusions

AISI 316L samples elaborated by SLM were aged at 900 °C in air and air-10 vol.% H2O and compared with wrought coupons. SLM process has beneficial impact on the high temperature oxidation resistance of AISI 316L in both atmospheres. In laboratory air, SLM samples exhibit parabolic behavior throughout the 3000 h of exposure thanks to the growth of dense and protective chromia layer. In contrast, the wrought coupons present breakaway phenomenon after 1000 h of exposure as result of non-protective iron oxide formation.

The impact of water vapor (10 vol.%) is clearly different depending on the sample type. Indeed, for wrought 316L, water vapor has detrimental effect since iron oxides are formed from the early stages of oxidation. Hence, a very high oxidation rate was observed, in connection to a thick, brittle and non-protective oxide layer. In the case of SLM samples, water vapor has no negative impact on corrosion; it even slightly improves it.

In addition, Cr concentration profiles below the oxide scales showed Cr depletion below 12 wt. % for the wrought samples, but Cr content remained above 12 wt. % for SLM plates aged in the same conditions. In both atmospheres, the better corrosion resistance of SLM samples was related to better Cr supply from bulk to the subsurface because of much larger number of dislocations as compared to the wrought alloy.