Effect of morphology and volume fraction of δ-ferrite on hydrogen embrittlement of stainless steel produced by electron beam additive manufacturing

Introduction

Additive manufacturing (AM) represents a rapidly advancing domain encompassing both science and in­dustry. It introduces a transformative approach to crafting intricate components and parts for mechanisms characterized by intricate configurations, a feat previously unattainable using conventional manufacturing techniques. The application of additive technologies holds significant potential within the burgeoning sphere of hydrogen energy. This encompasses endeavors related to hydrogen storage, transportation, and the fabrication of essential components like hydrogen fuel cells and vehicles. Stable austenitic stainless steels exhibit the highest resistance to hydrogen’s adverse effects when compared to other steel categories [1; 2]. Their favorable weldability and moderate cost make them an appealing choice for utilization as foundational materials within additive technologies [3; 4]. Nevertheless, when devising components intended for operation in aggressive hydrogen-rich environments, it becomes imperative to consider the unique microstructural attributes intrinsic to this steel category that manifest during the AM process. These attributes take into account the anisotropic nature of grain structure [5; 6] and formation of the secondary phases [7 – 9].

Currently, the subject of hydrogen embrittlement in steel samples produced through AM techniques remains inadequately explored in published literature, with occasional inconsistencies in the available data. The study [10] revealed that austenitic steel AISI 304L, fabricated using laser powder AM technology, is more resistant to hydrogen embrittlement compared to conventionally obtained rolled steel. This enhanced resistance is attributed to the development of a stable austenitic phase that is resistant to phase transitions, as well as the distinctive texture of the steel brought about during powder sintering. Research presented in [11] demonstrated the heightened vulnerability of martensitic aging steel produced through selective laser melting to hydrogen embrittlement. Stainless steel 17-4 PH, featuring a coarse crystalline ferritic structure and manufactured via laser additive technologies, is more prone to the adverse effects of hydrogen than its cast counterpart [12]. Conversely, for ferritic steel 09G2S, additively produced steel using the electron beam AM method exhibits a weaker susceptibility to hydrogen embrittlement than its cast equivalent [13].

In comprehending the reasons behind the degradation of mechanical properties induced by hydrogen in additively manufactured austenitic steels – known for their resistance to hydrogen embrittlement – grasping the role of key microstructural aspects in these processes becomes crucial. Particularly, samples of austenitic steels generated through electron beam AM methods exhibit distinctive characteristics: an anisotropic grain structure and a notable proportion of the δ-phase (approximately 20 %) [6 – 8]. This study aims to strategically manipulate the morphology and volume fraction of δ-ferrite within 08Kh19N9T steel samples produced via electron beam additive manufacturing (EBAM), while keeping other structural parameters (primarily the size of austenite grains) unchanged [6].

The objective of this research is to establish patterns concerning the hydrogen embrittlement of austenitic steel 08Kh19N9T manufactured using the EBAM technique, contingent on the morphology and content of δ-ferrite in its structure.

Experimental methods

For the EBAM process, we utilized an industrial wire composed of austenitic stainless steel (ASS) 08Kh19N9T, with the following chemical composition (wt. %): Cr 17.7; Ni 9.7; Mn 1.1; Ti 0.8; Si 0.6; C 0.08; with iron constituting the remainder (wire diameter 1.2 mm). The electron beam additive manufacturing was executed within a vacuum chamber, employing the subsequent parameters: accelerating voltage of 30 kV, wire feed speed set at 180 mm/min, beam sweep spanning 45×45 mm, scanning frequency of 1 kHz, and electron beam current of 45 mA. The accumulation of material layers transpired atop an austenitic steel substrate, resulting in a final geometric configuration of the steel billet measuring 110×6×30 mm.

For the purpose of uniaxial static tension tests, flat samples resembling double blades were fashioned, with the working part dimensions at 12×3×1.5 mm. These samples were extracted from an additively manufactured steel wall. A portion of the samples were assessed in their initial state – directly following additive growth (AM-ASS) – while another subset underwent annealing at 1100°C for 1 hour, succeeded by water quenching (AM-ASS + HT). Surface treatment encompassed mechanical grinding and electrolytic polishing using a solution of 25 g CrO₃ and 200 ml H₃PO₄.

Microstructural analysis of the obtained samples was conducted using optical (OM) and transmission electron microscopy (TEM), utilizing Altami MET 1C and Jeol JEM 2100 microscopes, respectively. The magnetic phase analysis (MPA) technique was employed, specifically with an MVP-3 ferritometer (Kropus, Russia) to determine the volume fraction of the δ phase (V_δ).

Electrolytic hydrogenation was executed in a 3 % NaCl aqueous solution containing 3 g/L NH4SCN. This procedure spanned 50 h at a current density of 50 mA/cm². Thermal desorption spectroscopy (TDS) investigations – centered on evaluating the rate of hydrogen liberation from specimens during gradual heating – were conducted using an automated Gas Reaction Controller LBP system (Advanced Materials Research, USA). This was executed with a heating rate of 360 °C/h, within a temperature span ranging from 25 to 400 °C. Tensile tests were carried out at room temperature, maintaining an initial strain rate of 5·10\(^-\)⁴ s\(^-\)¹. The testing was performed employing an LFM-125 universal electromechanical testing machine (Walter+Bai AG, Switzerland). Subsequently, the fracture surfaces of the samples were meticulously analyzed using a LEO EVO 50 scanning electron microscope (SEM) (Zeiss, Germany). To estimate the thickness of the brittle hydrogen-induced layer, SEM images acquired from the fracture surfaces of hydrogenated samples were employed. The secant method was utilized for this purpose, wherein the secants were oriented perpendicular to the hydrogenated layer.

Results and discussion

Fig. 1 presents optical microscopy (OM) and transmission electron microscopy (TEM) images of samples produced via the EBAM process. Both categories of samples exhibit a dual-phase (γ + δ) configuration. Notably, the δ-ferrite lamellae emergence of the δ phase within the structure of additively manufactured samples is attributed to distinct characteristics of the EBAM process, encompassing the intricate, multistage thermal history of layers and the overall billet, as well as the diminution of nickel content within the melt, among other factors [7; 9; 14]. Subsequent heat treatment (1100 °C, 1 h) induces a reduction in the δ-ferrite content while simultaneously altering its morphology – a phenomenon exhaustively elucidated in [6]. In specific, the heat treatment results in the formation of a dendritic structure of δ-ferrite, characterized by an average lamella thickness of 0.8 ± 0.4 µm within the AM-ASS samples. Following heat treatment, prolonged continuous branches (lamellae) of δ-ferrite experience partial dissolution, giving rise to non-equiaxed δ-phase grains (particles). These new grains exhibit an average thickness and length of 1.3 ± 0.5 and 6.2 ± 3.1 μm, respectively (Fig. 1, b). Consequently, the volume fraction of ferrite diminishes from 20 % to 5 % after the heat treatment, while the grain structure of the principal phase (austenite) remains unaffected [6]. Thus, the heat treatment exclusively influences the morphology and volume fraction of the ferrite phase, maintaining all other structural parameters unaltered.

For hydrogenated AM-ASS and AM-ASS + HT samples, their thermal desorption (TDS) spectra depict a sole peak at T_max = 160 – 165 °C (Fig. 2, a). This low-temperature peak corresponds to the release of hydrogen atoms from the crystal lattice and various weak, reversible traps possessing low activation energies. These encompass grain boundaries, interfacial (austenite – δ-ferrite) boundaries, dislocations, and similar features [15; 16]. The peak intensity for AM-ASS samples surpasses that of AM-ASS + HT samples, wherein the latter category bears a reduced proportion of the δ-phase. Hence, under analogous saturation conditions, a lesser amount of hydrogen dissolves within AM-ASS + HT samples. This disparity might arise due to the diminished presence of interphase boundaries in the annealed samples, which are capable of adsorbing hydrogen atoms. Given that the diffusion coefficient of hydrogen in δ-ferrite is significantly higher and its solubility lower than in the austenite phase [17; 18], the elevated volume fraction of the δ-phase, along with its morphology characterized by extended continuous lamellae in AM-ASS samples, expedites the profound penetration of hydrogen atoms into the samples during saturation. Put differently, the δ-ferrite branches serve as preferential conduits for the inward transport of hydrogen atoms, contributing to a more effective accumulation of hydrogen within the material (primarily within austenite). This proposition also correlates with the observation of the TDS peak shifting towards higher temperatures for AM-ASS samples in comparison to the peak for AM-ASS + HT samples. This shift could be attributed to a deeper saturation of the samples with hydrogen as the ferrite proportion increases.

Fig. 2, b delineates stress-strain diagrams in engineering coordinates for AM-ASS and AM-ASS + HT samples. Mechanical properties (σ_0.2 – yield strength; σ_u – ultimate strength; δ – elongation to fracture; \({I_{\rm{H}}} = \left( {\frac{{{\delta _0} - {\delta _{\rm{H}}}}}{{{\delta _0}}}} \right) \cdot 100{\rm{ }}\% ;\) δ₀ and δ_H – elongation to fracture of unhydrogenated and hydrogenated samples) are summarized in Table.

Examination of the provided experimental data demonstrates that post-production heat treatment yields an increase in elongation to failure and a decrease in yield strength for the examined steel. This phenomenon can be attributed to a decrease in the volume fraction of δ-ferrite, as δ-ferrite possesses higher strength compared to austenite [19]. Additionally, there is a reduction in the density of interphase boundaries (austenite – δ-ferrite), which function as barriers to the movement of dislocations during deformation [6].

The process of hydrogen charging induces alterations in the mechanical properties of EBAM steel (outlined in Table). Irrespective of the ferrite content, the yield strength (σ_0.2) of hydrogen-charged samples surpasses that of the original samples (deformed without hydrogen charging). This experimental observation signifies solid-solution hardening of the austenitic phase due to the presence of hydrogen atoms [20]. Notably, it’s noteworthy that in AM-ASS + HT samples, the hydrogen-induced enhancement in yield strength (\(\Delta \sigma _{0.2}^{\rm{H}}\)  = 73 MPa) exceeds that in AM-ASS samples (\(\Delta \sigma _{0.2}^{\rm{H}}\)  = 55 MPa), even though, as per the results of TDS analysis, the hydrogen concentration in the latter is higher (Fig. 2, a). The presence of a limited volume fraction of ferrite and interfacial boundaries, which function as traps for hydrogen atoms [21], can lead to a heightened accumulation of hydrogen within the austenite grain bodies located near the saturable surface of AM-ASS + HT samples. This accumulation contributes to their solid solution hardening. On the contrary, hydrogen transport is more restricted in AM-ASS + HT samples due to the altered morphology and reduced volume fraction of ferrite, which leads to the suppression of hydrogen transport over long distances along ferrite dendrites and a decrease in the prevalence of interfacial boundaries. Consequently, AM-ASS samples amass a greater total hydrogen concentration. In these samples, the concentration gradient across depth is evidently smaller, subsequently yielding a lower level of solid solution strengthening in the austenite phase compared to AM-ASS + HT samples where hydrogen transport is curtailed. In addition to solid solution hardening, the hydrogen concentration gradient throughout the depth is intrinsically linked to the stress gradient within the tested samples – stemming from their heterogeneous hydrogen charging. This aspect becomes more pronounced in AM-ASS + HT samples where the transfer of hydrogen atoms during charging and subsequent deformation is hindered due to the alterations in morphology and the reduction in the volume fraction of ferrite [22; 23]. Therefore, even with a lower hydrogen adsorption concentration during hydrogen charging, the yield strength of AM-ASS + HT samples, wherein hydrogen transport is limited, displays increased susceptibility to hydrogen charging: the hydrogen-induced rise in yield strength is more significant in AM-ASS + HT samples compared to AM-ASS samples. However, the hydrogen embrittlement factor I_H, which quantifies the reduction in elongation to failure due to hydrogenation, is greater for AM-ASS samples (Table).

Fig. 3 portrays SEM images of fractured surfaces from both hydrogen-charged AM-ASS and AM-ASS + HT samples. Across all samples, hydrogen charging results in the formation of a brittle surface layer, while the remaining portions of the samples exhibit a ductile transcrystalline fracture mode, comparable to uncharged samples [14].

The fracture surfaces of the brittle hydrogenated layer exhibit features characteristic of both transcrystalline and intercrystalline fracture, including ridges and flat facets. The presence of intercrystalline cleavages aligns with the aforementioned mechanisms of hydrogen adsorption, specifically pointing towards the accumulation of hydrogen atoms at interphase boundaries. The transcrystalline nature of the fracture signifies the brittle fracture of hydrogen-saturated austenite grains, along with the formation of deformation-induced martensite within them, as documented in [2; 14; 23; 24].

The thickness of the brittle hydrogenated layer is notably greater for AM-ASS samples (D_H = 55 ± 12 μm) characterized by a higher initial proportion of ferrite – than for AM-ASS + HT samples (D_H = 29 ± 7 μm). This experimental observation corroborates the findings from TDS studies, mechanical tests, and the preceding discussions.

Conclusions

The investigation delved into the hydrogen embrittlement characteristics of austenitic chromium-nickel steel samples produced via electron beam additive manufacturing in two distinct states: immediately following additive growth and subsequent post-production heat treatment. Annealing the additively manufactured samples at 1100 °C for 1 h yielded a notable reduction in the volume fraction of δ-ferrite – from 20 % to 5 % – accompanied by an alteration in its morphology. Specifically, the state after EBAM exhibited thin extended lamellae of dendrites, while the heat-treated samples showcased isolated inclusions (particles) of ferrite. This transformation in phase composition and microstructure exerted an impact on the steel’s susceptibility to hydrogen embrittlement, the solubility and distribution of hydrogen during the process of electrolytic saturation, and the dimensions of the brittle hydrogenated layer within the samples.

Even though the annealed samples with a lower proportion of ferrite contained a lower overall concentration of dissolved hydrogen, their hydrogen-induced increase in yield strength (\(\Delta \sigma _{0.2}^{\rm{H}}\)  = 73 MPa), exceeded that of samples following EBAM, characterized by a higher proportion of dendritic ferrite (\(\Delta \sigma _{0.2}^{\rm{H}}\)  = 55 MPa). This discrepancy can be attributed to the impedance of hydrogen transport deep into the samples along δ-ferrite dendrites due to modifications in their morphology and the reduction of interphase boundaries. These boundaries function as traps for hydrogen atoms within both the austenite crystal lattice and intergranular regions. Consequently, post heat treatment, hydrogen transport deep into the samples is curtailed, leading to its accumulation in the surface layers. This accumulation contributes to robust solid-solution hardening of the austenitic phase, thereby engendering a pronounced stress gradient due to the hydrogen concentration gradient in these sample.

The thickness of the brittle hydrogenated surface layer and the hydrogen embrittlement factor were found to be more significant for the initial additively manufactured steel samples (D_H = 55 ± 12 μm, I_H = 32 % for AM-ASS and D_H = 29 ± 7 μm, I_H = 24 % for AM-ASS + HT samples). The reduction in the volume fraction and the alteration in the morphology of δ-ferrite, brought about by post-production heat treatment, enhance the resistance of EBAM stainless steel to hydrogen embrittlement.