Overview of Hydrogen Effects on γ−ε Martensitic Transformation in Steels

Abstract

This paper presents an overview of our recent works on the effects of hydrogen on γ-ε martensitic transformations in steels. The study first discusses how hydrogen impacts these transformations. While hydrogen suppresses thermally-induced γ–ε martensitic transformation, it increases the fraction and number density of deformation-induced ε-martensite and decreases its thickness. Secondly, we discuss the effects of γ–ε martensitic transformations on hydrogen kinetics. The study also highlights the significance of low hydrogen diffusivity in the hexagonal-close-packed (HCP) lattice of pure iron, demonstrating the effectiveness of ε-martensite in resisting hydrogen. Moreover, the characteristic behavior of the HCP phase-related diffusionless transformation from a hydride is discussed. We believe that this overview will assist in developing hydrogen-resistant steels and in exploring new microstructural control concepts using hydrogen.

1. Introduction

In metastable austenitic steels with low stacking fault energies, a martensitic transformation from face-centered cubic (FCC, γ) to hexagonal close-packed (HCP, ε) structures occurs. An example of the ε-martensite plate is shown in Fig. 1,^1,2) which has a specific crystallographic orientation relationship of (111)_γ//(0002)_ε, i.e., Shojo-Nishiyama relation.³⁾ The γ–ε martensitic transformation is a key phenomenon that triggers several properties, including high work hardening capability,^4,5,6) high damping capacity,⁷⁾ excellent fatigue resistance,^8,9) and significant shape memory properties.^10,11) However, as a disadvantageous effect, ε-martensite can also act as a crack initiation site and propagation path^12,13,14,15) (e.g., Fig. 2). Therefore, understanding the behavior of γ–ε martensitic transformations is essential to design high-performance steels used in hydrogen-related atmospheres.

Fig. 1. An example of ε-martensite: (a) Bright field TEM micrograph, (b) diffraction pattern, and (c) the corresponding key diagram.^1,2) Reproduced with permission from reference 1. Copyright 2007, JIM.

Fig. 2. (a) Electron backscatter diffraction (EBSD)-phase and (b) inverse pole figure (IPF) maps showing a crack at an ε-martensite region.^14,15) Reproduced with permission from reference 14. Copyright 2016, ISIJ. (Online version in color.)

The thermodynamics and kinetics of the γ–ε martensitic transformation depend on the steel’s chemical composition. In particular, the presence of interstitial elements such as carbon and nitrogen drastically alters the transformation behavior. Therefore, the effects of carbon^16,17,18) and nitrogen^19,20,21) on this transformation have been investigated in detail. Hydrogen is another important interstitial element in steels. Although hydrogen is not typically used as an alloying element because of its significant diffusivity and desorption even at room temperature, it can easily enter metallic components when exposed to a corrosive or hydrogen gas environment. This affects the γ–ε martensitic transformation behavior and interfacial cracking.^22,23,24) As a result, the hydrogen-transformation interaction is the primary factor controlling hydrogen-related mechanical degradation in austenitic steels.

The interaction between hydrogen and γ–ε martensitic transformations has been reported as a complex phenomenon, with previous reports showing both promotion^{22,23,25,26,27,28)} and suppression²⁹⁾ effects of hydrogen on this transformation. The mismatch between previous studies arises from two factors: (1) the evolution of hydrogen-related stress during hydrogen charging,^27,28) which is beyond the scope of this paper, and (2) the presence of different criteria for evaluating hydrogen effects. In terms of the criteria, it must be noted that martensitic transformations are generally classified into three types: (1) thermally induced, (2) stress-assisted, and (3) strain-induced transformations.³⁰⁾ The effects of hydrogen on these different transformation types should not be combined, and the transformation behaviors with and without hydrogen should be analyzed separately. In particular, when hydrogen affects thermally induced martensitic transformations, the transformation temperatures and martensite fraction after cooling to a specific temperature were evaluated.^25,30) When stress-assisted and strain-induced martensitic transformations are targeted, the critical stress/strain for inducing the transformation and the martensite fraction at an identical strain were evaluated.^22,26,31)

In this paper, we present an overview of our recent studies on the effects of hydrogen on γ–ε martensitic transformations in steels. We discuss the different types of transformations in sections 2 to 4 and how the presence of ε martensite affects hydrogen kinetics, such as diffusion, in section 5. These sections describe the synergistic effects of hydrogen and γ–ε transformations. Furthermore, we introduce a martensitic-like shear transformation from hydride to γ and ε phase in section 6, which is significantly different from a typical martensitic transformation.

2. Thermally Induced Transformations

We first observed the effect of hydrogen on a thermally induced γ–ε martensitic transformation in an Fe–15Mn–10Cr–8Ni alloy. Initially, the as-quenched microstructure of the alloy only consists of austenite at ambient temperature and undergoes a γ–ε martensitic transformation upon cooling. Figure 3 shows the ε-martensite fraction plotted against temperature in the steel with and without hydrogen charging. Hydrogen charging at a higher temperature introduces a larger amount of hydrogen than that by hydrogen charging at room temperature (RT). The higher ε-martensite fraction at RT in the hydrogen-charged specimen at RT than that without hydrogen charging was attributed to the stress evolution associated with the presence of a hydrogen concentration gradient during hydrogen charging, which was an exceptional case in this result. It is evident from Fig. 3 that hydrogen charging decreases the starting temperature for γ-ε martensitic transformations and the ε-martensite fraction at each temperature. To support this finding, Table 1 lists the ε-martensite fractions after cooling to 77 K under different hydrogen charging conditions. The results show that the ε-martensite fraction decreases as the diffusible hydrogen content increases¹. The suppression of γ-ε martensitic transformations can be explained by a decrease in the chemical driving force, which is supported by ab initio calculations.^32,33) Figure 4 shows an example of the ab initio calculation results that demonstrate how the energy profile at 0 K depends on the hydrogen content along the minimum energy path from the FCC to HCP lattices in pure iron. In the calculations shown in Fig. 4(a), the hydrogen atoms in the HCP lattice were set to metastable tetrahedral sites. This is because the motion of Heidenreich-Shockley partials, which is the transformation dislocation of the γ–ε martensitic transformation, crystallographically changes the interstitial atom positions from octahedral sites in the FCC lattice to tetrahedral sites in the intrinsic stacking fault region (HCP lattice) when only substitutional atoms are assumed to move.^34,35,36) In contrast, when the hydrogen position in the HCP lattice was assumed to be the octahedron site, the effect of hydrogen on the energy change became relatively small, as shown in Fig. 4(b). Both calculations clearly indicate that the stability of the HCP phase decreases with increasing hydrogen content.

Fig. 3. Fractions of thermally induced ε-martensite with different hydrogen charging conditions measured by in situ XRD measurements.²⁹⁾ The hydrogen charging was conducted in an aqueous solution of 3% NaCl with a 3 g/L NH₄SCN at 50 A/m² at ambient temperature for 264 h or 353 K for 168 h. (Online version in color.)

Table 1. ε-martensite fraction at RT after immersion in liquid nitrogen.²⁹⁾ The result of post-hydrogen gas charging has also been provided for a more general understanding of the effect of hydrogen.

	No charge	Electrochemical H charging at RT	100 MPa H gas charging at 473 K	Electrochemical H charging at 353 K
Diffusible H content (mass ppm)	0.1	72	82	341
ε-martensite fraction	43%	32%	25%	9%
Néel temperature (K)	229	–		223
As temperature (K)	336		349	360

Fig. 4. Ab initio calculation results of minimum energy paths between FCC (γ) and HCP (ε) in pure iron with and without hydrogen. The hydrogen positions in the HCP lattice were set to be (a) tetrahedral and (b) octahedral sites. The peak of the energy can be regarded as activation energy.³³⁾ Reproduced with permission from reference 33. Copyright 2019, The Minerals, Metals & Materials Society and ASM International. (Online version in color.)

Next, we note a markedly lower transformation rate during the cooling of the hydrogen-charged specimen at 353 K compared to the uncharged specimen. A possible explanation is the effect of hydrogen on the activation energy for the γ-ε martensitic transformation. The highest peak value in Fig. 4 can be regarded as the activation energy, and hydrogen slightly increased it. An increase in activation energy suppresses the motion of transformation dislocations at a finite cooling rate, particularly at low temperatures. The combined effect of the lowered martensite-start temperature (M_s) and increased activation energy critically decreases the transformation rate.

However, it should be noted that the experimental conditions in Fig. 3 correspond to the range of hydrogen contents between Fe and Fe₁₂H (1489 mass ppm); in fact, the measured hydrogen contents were in the range of 72–341 mass ppm. The difference in the activation energies of Fe and Fe₁₂H is very slight; thus, another factor must contribute to the significant reduction in the transformation rate by hydrogen. In this context, it is noteworthy that the magnetic transition from paramagnetism to antiferromagnetism can prevent the increase in ΔG^FCC→HCP as the temperature decreases,^37,38) and the Néel temperature of the Fe–15Mn–10Cr–8Ni alloy was just below M_s, as listed in Table 1. The combined effect of the magnetic transition and the decrease in M_s by hydrogen can reduce the rate of increase in ΔG^FCC→HCP with decreasing temperature.

In summary, hydrogen-lowered M_s increased the activation energy for the transformation, and the magnetic transition can reduce the rate of γ–ε martensitic transformation at different temperatures.

It is noteworthy that, as listed in Table 1, hydrogen charging increased the starting temperature for the reverse transformation from ε to γ (A_s). This implies that hydrogen suppresses both forward and reverse transformations, which cannot be interpreted as a simple effect of hydrogen on the chemical driving force. This indicates that hydrogen acts as a barrier for the motion of transformation dislocations in both the forward and reverse transformations, as schematically shown in Fig. 5. The critical condition for the γ–ε martensitic transformation is as follows:³⁹⁾

  

$$2n\rho (\mathrm{\Delta }{\mathrm{G}}^{\mathrm{F}\mathrm{C}\mathrm{C}\to \mathrm{H}\mathrm{C}\mathrm{P}}+E^{str})+2\sigma (n)=-n{\tau }_0\mathrm{b}$$(1).

The terms on the left and right sides represent the thermal driving force and friction stress acting on the transformation dislocation motion at M_s, respectively. N is the atomic plane in thickness, ρ is the molar surface density along {111}, ΔG^FCC→HCP is the free energy change due to the transformation from γ to ε, E^str is the coherency strain energy, and σ(n) is the interfacial energy of γ/ε. Thus, the γ-ε martensitic transformation occurs when the thermal driving force is greater than the friction stress. In this regard, the hydrogen-induced increase in the gap between M_s and A_s can be explained as an increase in friction stress. Many previous studies have reported that interstitial hydrogen increases the yield strength in austenitic steels with low stacking fault energies, implying that it increases the friction stress for extended dislocation slip. Because the transformation dislocation for the γ-ε martensitic transformation is a simple Heidenreich-Shockley partial, a similar hydrogen effect must be present for the transformation dislocation motion, which acts as a factor suppressing both the forward and reverse γ-ε transformations. In addition, because hydrogen atoms increasing the friction stress are regarded as short-range obstacles, the motion of the transformation dislocations involves a thermal activation process. This may be an additional reason for the low transformation rate during cooling, but further investigation is required to confirm this. To summarize, the combined effects of hydrogen on friction stress, activation energy, and chemical driving force all contribute to the suppression of the thermally induced γ-ε martensitic transformation.

Fig. 5. Schematic for explaining the hydrogen effect on friction stress.²⁹⁾ (Online version in color.)

¹  Diffusible hydrogen is defined as hydrogen that diffuses at RT, which is present at interstitial sites and weak hydrogen trap sites. The diffusible hydrogen content is measured by thermal desorption spectroscopy.

3. Deformation-induced Transformation

In this section, we introduce the effect of hydrogen on deformation-induced γ-ε martensitic transformations. Before discussing the details of the transformation behavior, the yield behavior of the Fe–15Mn–10Cr–8Ni alloy with and without hydrogen is discussed. Figure 6(a) shows the 0.2% proof stresses at various deformation temperatures without hydrogen charging.

Fig. 6. (a) Deformation temperature dependence of 0.2% proof stress of the Fe-15Mn-10Cr-8Ni alloy without hydrogen charging, tested with an initial strain rate of 10⁻⁴ s⁻¹. (b) Yield strength-temperature relationship when deformation-induced martensitic transformation occurs.³⁰⁾ (c) Stress-strain curves at ambient temperature of the same alloy with and without hydrogen charging. The hydrogen charging was conducted at 50 A/m² and 80°C in a 0.1 N NaOH aqueous solution containing 3 g/L NH₄SCN. No martensitic transformation during hydrogen charging was confirmed. Reproduced with permission from reference 26. Copyright 2019, Elsevier. (Online version in color.)

As mentioned in the introduction, deformation-induced martensitic transformations can occur through two mechanisms: stress-assisted and strain-induced transformations, as schematically shown in Fig. 6(b). In the stress-assisted regime, if a sufficient mechanical driving force is provided within the elastic deformation stage, the critical stress for γ-ε martensitic transformation controls the yield strength. The critical stress for the transformation increases with the stability of γ-austenite, which also increases with temperature. The highest temperature at which the stress-assisted transformation occurs is $M_s^{\sigma }$.

Above $M_s^{\sigma }$, the yield strength decreases with increasing temperature because the critical stress for the slip in austenite decreases owing to the promotion of the thermal activation process of the dislocation motion. Therefore, the 0.2% proof stress shown in Fig. 6(a) increases with increasing temperature up to 373 K, which is the $M_s^{\sigma }$ of the alloy. In other words, the 0.2% proof stress at RT is regarded as the critical stress for γ-ε martensitic transformation. Hydrogen charging increased the 0.2% proof stress at RT from 120 to 131 MPa, indicating that hydrogen suppressed the stress-assisted γ-ε martensitic transformations in terms of “stress”, consistent with the effect of hydrogen on thermally induced γ-ε martensitic transformations. However, in general, the deformation-induced transformation behavior is evaluated at identical strain after significant plastic deformation. Therefore, we will next discuss the effect of hydrogen on deformation-induced γ-ε martensitic transformations post-yielding.

Figure 7 shows the deformation-induced ε-martensite plates of the specimens with and without hydrogen charging at a 4% strain. Image analysis revealed that the ε-martensite fraction increased from 7% to 10% with hydrogen charging. That is, hydrogen promotes the deformation-induced γ-ε martensitic transformation at identical strains. It is important to note again that hydrogen was observed to suppress the deformation-induced γ-ε martensitic transformation when evaluated at an identical stress but to promote it when evaluated at an identical strain after significant plastic deformation. This trend is illustrated schematically in Fig. 8. When evaluated at identical stress, only the absolute value of the critical stress for γ-ε martensitic transformation is important. However, when evaluated at an identical strain after significant plastic deformation, the ε-martensite fraction depends on the difference between the critical stresses for γ-ε martensitic transformation and slip deformation. In a previous study,¹⁶⁾ an increment of interstitial carbon content was reported to increase the critical stress for slip deformation more than that for the γ-ε martensitic transformation, leading to an increase in the dominance of γ-ε martensitic transformation in the plastic deformation. Because hydrogen is also an interstitial element in austenitic steels, a similar effect was expected. Therefore, the ε-martensite fraction after plastic deformation with hydrogen was higher than that without hydrogen at an identical strain. To fully understand the effect of hydrogen on deformation-induced γ-ε martensitic transformation, the effect of hydrogen on “strain-induced” ε-martensitic transformations, which requires the evolution of specific dislocation patterns as a precursor, has not been examined. Because hydrogen has been reported to alter dislocation substructures after plastic deformation, the effect of hydrogen-altered dislocation patterns on strain-induced nucleation of martensite will be a key factor in controlling the deformation-induced γ-ε martensitic transformation in relatively stable austenitic steels.

Fig. 7. Backscatter electron images showing ε-martensite plates of the 4%-strained Fe-15Mn-10Cr-8Ni alloy specimen (a) without and (b) with hydrogen charging.²⁶⁾ Reproduced with permission from reference 26. Copyright 2019, Elsevier. (Online version in color.)

Fig. 8. A schematic illustration for the promotion effect of hydrogen on γ→ε martensitic transformation at an identical strain. (Online version in color.)

4. Martensite Morphology Variation

Hydrogen affects not only the ε-martensite fraction but also the number density, and thickness of ε-martensite plates. The effects of hydrogen on martensite morphologies are summarized in Table 2. On average, hydrogen significantly decreases the thickness of ε-martensite plates and increases the number of ε-martensite plates per grain. To visualize the distribution of ε-martensite morphology, Fig. 9(a) shows the thickness of ε-martensite plates for each grain, plotted against the number of ε-martensite plates per grain, which are also plotted against the tensile orientation (Figs. 9(b)–9(c)). In the hydrogen-charged specimen, the thickness of the ε-martensite plates was below 0.2 μm in the major portion of grains, irrespective of the tensile orientation, whereas the uncharged specimen showed a wide variety of ε-martensite thickness (ranging from below 0.2 μm to over 0.4 μm). A significant number of grains in the hydrogen-charged specimen showed a high number density of ε-martensite plates (over 30 per grain) with thin ε-martensite plates, whereas only two grains of the uncharged specimen showed such a high number density. Hence, it is concluded that hydrogen plays a significant role in reducing the thickness and increasing the number density of ε-martensite plates, irrespective of the tensile orientation. It should be noted that solute carbon also has a similar effect on thinning ε-martensite plates,¹⁸⁾ which, in addition to the strengthening effect, is another similarity between hydrogen and carbon.

Table 2. Average thickness, number per grain, the area fraction of ε-martensite in the Fe–15Mn–10Cr–8Ni alloy with respective standard deviation.²⁶⁾

4%-strained specimen	Uncharged	H-charged
Area fraction of ε [%]	7±4	10±7
Thickness of ε [μm]	0.29±0.11	0.18±0.07
Number of ε [/grain]	18±11	30±17

Fig. 9. (a) ε-martensite thickness distribution and tensile orientation of the observed grains without (b) and with (c) hydrogen charging in the Fe-15Mn-10Cr-8Ni alloy.²⁶⁾ Reproduced with permission from reference 26. Copyright 2019, Elsevier. (Online version in color.)

As stated in the introduction, deformation-induced ε-martensite is the origin of the shape memory effect in austenitic steels. In fact, the thin and high number density plates of ε-martensite have been identified as the optimal morphology for gaining shape memory property associated with ε-γ reverse transformation. Therefore, the morphological changes and increase in the fraction of ε-martensite due to hydrogen charging are expected to improve the shape memory effect.

Figure 10 shows the effect of hydrogen on the shape recovery strain of an Fe–29Mn–7Cr–6Si alloy, known as a ferrous shape memory alloy.¹¹⁾ The addition of Si increases the dominance of the γ-ε martensitic transformation in plastic strain,^40,41) leading to a larger ε-martensite fraction of the Fe–29Mn–7Cr–6Si alloy compared to the Fe–15Mn–10Cr–8Ni alloy. Similar to the Fe–15Mn–10Cr–8Ni alloy, hydrogen (27.1 mass ppm) increased the ε-martensite fraction, decreased the average thickness of the plates, and increased the number of plates per grain.⁴²⁾ The shape memory property of the Fe–29Mn–7Cr–6Si alloy was evaluated by measuring the shape recovery strain ε_r defined as follows:

  

$${\epsilon }_r=\frac{l_r-l_1}{l_0}\times 100,$$(2)

where l₀, l₁, and l_r are the distances between the gauge lines before and after pre-straining and after heating, respectively. It is clearly shown that hydrogen improves the shape recovery property associated with the deformation-induced γ-ε martensitic transformation.

Fig. 10. Hydrogen effect on shape recovery strain with different pre-strains.⁴²⁾ The specimens were pre-charged with hydrogen at 100 MPa and 543 K for 238 h. Reproduced with permission from reference 42. Copyright 2020, The Minerals, Metals & Materials Society and ASM International. (Online version in color.)

5. Hydrogen Diffusion and Desorption in the Transformations

After understanding the effects of hydrogen on the transformations, it is important to understand how hydrogen behaves during the γ-ε martensitic transformation or in the presence of ε-martensite. We first note the diffusivity of hydrogen in the ε-phase because hydrogen diffusivity is one of the primary factors determining hydrogen embrittlement susceptibility, particularly in fatigue. However, obtaining experimental data on hydrogen diffusivity in a single ε-phase in Fe-based alloys without lattice defects is impossible at ambient pressure, thus necessitating the use of ab initio calculations. Figure 11(a) shows the hydrogen diffusivities of the BCC, FCC, and HCP structures in pure iron. Interestingly, the hydrogen diffusivity of the HCP structure was even lower than that of the FCC structure, indicating that the formation and presence of the ε-phase plays an advantageous role in hydrogen embrittlement from a viewpoint of hydrogen diffusivity. This result is consistent with a previous report showing no acceleration of fatigue crack growth by hydrogen under a 0.7 MPa hydrogen gas atmosphere in metastable austenitic alloys that undergo deformation-induced γ-ε martensitic transformations.⁴³⁾ Furthermore, hydrogen diffusivity depends on the c/a ratio of the lattice parameters of the HCP structure. The equilibrium value of c/a in pure iron is 1.58, and the c/a-dependence of the hydrogen diffusivity is not monotonic. As shown in Fig. 11(b), hydrogen diffusion occurs along the c-axis below c/a ~ 1.6 and on the c-plane above c/a ~ 1.6. Microscopic hydrogen mapping confirmed preferential hydrogen diffusion along the c-plane in the Fe–15Mn–10Cr–8Ni alloy with c/a = 1.613 (Fig. 12).³⁶⁾ Therefore, hydrogen diffusivity decreases as c/a increases until 1.6 and then increases as c/a increases beyond 1.6, resulting in minimum hydrogen diffusivity at c/a ~ 1.6. As the c/a ratio of the HCP structure in steels can be controlled by tuning their chemical composition, a c/a-based alloy design is promising for the development of hydrogen-resistant steels. However, when ε-martensite is used for designing hydrogen-resistant steels, the plasticity anisotropy of the HCP structure is problematic, limiting the slip deformation only to the basal plane (<a> slip) i.e., cracking problem. The ease of non-basal slip (<a+c> slip) in HCP alloys has been reported to depend on the c/a ratio,⁴⁴⁾ which has been confirmed as the major deformation mode in an Fe-based metastable high-entropy alloy with c/a = 1.616. Therefore, optimizing the c/a at approximately 1.60 to 1.62 may provide a new type of hydrogen-resistant metastable steel.

Fig. 11. Ab initio calculation results: (a) hydrogen diffusivity and (b) its dependence on the lattice parameter of hcp structures.⁴⁵⁾ c/a = 1.58 is the equilibrium value. AF: Antiferromagnetic. Reproduced with permission from reference 45. Copyright 2018, The Minerals, Metals & Materials Society and ASM International. (Online version in color.)

Fig. 12. Hydrogen is replaced by Ag when a hydrogen-charged specimen is immersed into an aqueous solution containing Ag ions. (a) SEM image showing Ag particles deposited along ε-martensite plates in the Fe-15Mn-10Cr-8Ni alloy. The corresponding EBSD-phase and (c) IPF maps.³⁶⁾ The specimen was cooled down to 77 K. Then, the pre-cooled specimen was hydrogen-charged and subsequently immersed in a 4.3 mM K[Ag(CN)₂] aqueous solution for 24 h. To promote hydrogen diffusion, the temperature during silver decoration was set at 323 K, which was lower than the A_s. Reproduced with permission from reference 36. Copyright 2018, ISIJ. (Online version in color.)

Fig. 13. Hydrogen desorption profile during cooling and subsequent heating in the Fe-15Mn-10Cr-8Ni alloy.⁴⁸⁾ Reproduced with permission from reference 48. Copyright 2016, Elsevier.

From the perspective of hydrogen kinetics, it is important to study hydrogen re-distribution during or after the γ-ε martensitic transformation, as this behavior can trigger hydrogen embrittlement in the case of γ-α′ martensitic transformations.^46,47) Specifically, when hydrogen supersaturation occurs due to the γ-α′ martensitic transformation, hydrogen-related cracking can easily occur due to the difference in hydrogen solubility. Two approaches can be used to analyze hydrogen distribution kinetics: in situ microscopic hydrogen mapping and in situ detection of hydrogen desorption. Although the former is a powerful method, there have been no successful reports of such results for ε-martensite thus far. However, we have succeeded in the in situ detection of hydrogen desorption related to the γ-ε martensitic transformation. By continuously measuring the hydrogen desorption rate of a hydrogen-charged specimen, we found that it decreased with decreasing temperature owing to the suppression of hydrogen diffusion. However, extra hydrogen desorption was detected during cooling at the M_s temperature of 244 K in the hydrogen-charged Fe–15Mn–10Cr–8Ni alloy. This indicates that the thermally induced γ-ε martensitic transformation results in hydrogen transport or assists in hydrogen diffusion. As mentioned above, the hydrogen diffusivity in the HCP structure is lower than that in the FCC structure. Furthermore, as shown in Fig. 14, the ε→γ reverse transformation during heating also results in extra hydrogen desorption. The extra hydrogen desorptions associated with both the γ-ε forward and reverse transformations cannot be explained by local hydrogen diffusivity changes stemming from differences in diffusion coefficient and solubility of hydrogen between the FCC and HCP phases. Therefore, we believe that the motion of the transformation dislocations (Heidenreich-Shockley partials) plays a significant role in hydrogen transport during the transformation.

Fig. 14. (a) XRD profiles of the Fe-20Mn, Fe-24Mn, and Fe-40Mn alloys after cooling to 77 K. (b) TDS profiles during heating in the three alloys pre-cooled to 77 K.⁴⁹⁾ The Fe-40Mn alloy without ε-martensite does not show an extra hydrogen desorption while the Fe-20Mn and Fe-24Mn alloys with ε-martensite do at the A_s temperatures as indicated by the arrows. Reproduced with permission from reference 49. Copyright 2015, ISIJ.

Deformation-induced γ-ε martensitic transformation also induces hydrogen desorption, as shown in Fig. 15(b). However, because there is no contribution of hydrogen diffusivity changes by the γ-ε martensitic transformation, the hydrogen desorption is an order of magnitude smaller than that of γ-α′ martensitic transformation (comparison of Figs. 15(b) and 15(c)). A comparison of Figs. 15(b) and 15(d) reveals unexpected hydrogen desorption behavior; that is, the hydrogen desorption rate was higher when γ-ε martensitic transformation occurred than when only slip deformation of austenite occurred.

Fig. 15. (a) Martensite fraction plotted against strain. Stress-time curves and hydrogen desorption profiles during deformation in the (b) Fe-15Mn-10Cr-8Ni, (b) Fe-19Cr-8Ni-0.14C, and (c) Fe-19Cr-14Ni-0.15C alloys.⁵⁰⁾ Reproduced with permission from reference 50. Copyright 2019, Elsevier. (Online version in color.)

Three factors are generally considered to represent the effects of martensitic transformation on hydrogen desorption: (1) hydrogen transport by transformation dislocations, (2) changes in hydrogen diffusivity, and (3) changes in hydrogen solubility. The transformation dislocations of ε-martensite are simple Heidenreich-Shockley partials, and thus, the nature of hydrogen transport by the dislocation motions involved in the γ-ε martensitic transformation and γ-slip deformation is considered to be similar. In addition, the hydrogen diffusivity of ε-martensite is rather lower than that of γ-austenite, as mentioned above. When the hydrogen solubility of ε-martensite is lower than that of austenite, extra hydrogen desorption at the moment of γ-ε martensitic transformation can occur due to the supersaturation of the hydrogen within the ε-martensite. Therefore, the relatively high hydrogen desorption rate involved in the γ-ε martensitic transformation implies that the solubility of hydrogen in the ε-martensite is lower than that in austenite.

In summary, the hydrogen desorption associated with both thermal and deformation-induced γ-ε martensitic transformations indicates that (1) the transformation dislocation motions significantly contribute to hydrogen transport, and (2) the hydrogen solubility in ε-martensite is higher than that in α′-martensite but lower than that in γ-austenite.

6. Hydride-related Diffusionless Phase Transformation²

As a unique hydrogen effect on ε-martensite-related transformations, we investigated the hydride effect on the microstructure evolution. When a near-stoichiometric amount of hydrogen is introduced into pure Fe or Fe-based alloys under high pressure, hydrides form as the major constituent phase. Interestingly, the crystallographic structure of the hydride is either HCP or double hexagonal close-packed (DHCP).^51,52) To investigate the hydride effects on diffusionless transformations, heat treatments were performed on pure iron and some ferrous alloys under several gigapascals, both with and without a hydrogen source (AlH₃).^53,54) For instance, in the case of the Fe–29Mn–7Cr–6Si alloy specimen, which underwent a γ-ε martensitic transformation, hydrogenation occurred during heating to 973 K under pressurization with the hydrogen source. This phenomenon is discussed in this section.⁵⁴⁾

As shown in Fig. 16(a), the Fe–29Mn–7Cr–6Si alloy without hydrogen exhibited an HCP structure as the primary phase at 9 GPa and 973 K, which remained intact even after cooling to 293 K. Some of the HCP phase transformed into the FCC phase resulting in the microstructure shown in Fig. 16(b). The plate-like morphology⁵⁴⁾ of the HCP phase indicated that it was pressure-induced ε-martensite³, whereas the numerous stacking faults in the austenite after depressurization resulted from a depressurization-induced reverse transformation from ε-martensite.

Fig. 16. (a) XRD profiles showing pressure-induced martensitic transformation and subsequent reverse transformation to austenite during cooling and depressurization in the Fe-29Mn-7Cr-6Si alloy without hydrogen. (b) EBSD phase map showing ε-martensite plates after the depressurization.⁵⁴⁾ Reproduced with permission from reference 54. Copyright 2022, Elsevier. (Online version in color.)

However, with the presence of a hydrogen source, the transformation sequence drastically changed, as shown in Fig. 17(a). The specimens hydrogenated at 9 GPa and 973–293 K comprised HCP and DHCP phases. The DHCP phase underwent a diffusionless transformation to the HCP phase with decreasing pressure, and subsequently aging at ambient temperature and pressure resulted in an ultrafine microstructure (with grain size ranging from ~100–200 nm) consisting of HCP, FCC, and DHCP phases (Fig. 17(b)). Additionally, ultrafine lamellar microstructure (width of ~11 nm) was also observed.⁵⁴⁾ This ultrafine-grained microstructure with numerous lattice defects is expected to have high strength. However, the process required high pressure and rapid desorption of near-stoichiometric amount of hydrogen, resulting in numerous microscopic and nanoscopic cracks,⁵⁴⁾ making it unsuitable for practical application. Nevertheless, we believe that this new type of diffusionless transformation from hydrides could open new paths for designing high-strength alloys.

Fig. 17. (a) XRD profiles showing phase transformation involving the formation of DHCP phase in the Fe-29Mn-7Cr-6Si alloy with hydrogen. (b) Phase map showing the distribution of HCP, FCC, and DHCP phases after the depressurization and subsequent aging at ambient temperature and pressure.⁵⁴⁾ Reproduced with permission from reference 54. Copyright 2022, Elsevier. (Online version in color.)

²  The transformation shown in this section occurs without diffusion of substitutional atoms, but with diffusion of hydrogen, which is referred to as diffusionless transformation in this paper.

³  Because pressure stabilizes the HCP phase in ferrous alloys, increasing pressure, instead of cooling, can also induce γ-ε martensitic transformation.

7. Summary

In this paper, we presented an overview of the effects of hydrogen on the γ-ε martensitic transformation in steels. Solute hydrogen can suppress the thermally induced γ-ε martensitic transformation while promoting deformation-induced γ-ε martensitic transformation at a certain strain. Specifically, when the deformation-induced γ-ε martensitic transformation is promoted, hydrogen increases the ε-martensite fraction, decreases its thickness, and increases the number density of the ε-martensite plates, thus enhancing the shape memory effect of ferrous alloys. In addition, ε-martensite has a lower diffusivity that depends on the c/a ratio of the HCP lattice. Therefore, coupled with the c/a ratio dependence of non-basal slip activity, the use of ε-martensite with the optimal c/a ratio (approximately 1.60–1.62) can be a promising alloy design concept for hydrogen-resistant steels. Moreover, using hydrides to induce a new type of diffusionless transformation can create ultrafine microstructures with different constituent phases, which is expected to be a new frontier for microstructure design research. Some behaviors mentioned above are similar to those of other interstitial elements, such as carbon. In addition, hydrogen affects not only γ-ε martensitic transformations but also γ-α′ (γ-ε-α′) martensitic transformation^22,55) and deformation twinning.^56,57,58) Therefore, we anticipate that hydrogen, as one of the interstitial elements in steels, will also be used as an effective alloying element in both structural and functional ferrous alloys.

References

• 1)   M.  Koyama,  M.  Murakami,  K.  Ogawa,  T.  Kikuchi and  T.  Sawaguchi: J. Japan Inst. Met. Mater., 71 (2007), 502. https://doi.org/10.2320/jinstmet.71.502
• 2)   M.  Koyama,  M.  Murakami,  K.  Ogawa,  T.  Kikuchi and  T.  Sawaguchi: Materials Transactions, 48 (2007), 2729. https://doi.org/10.2320/matertrans.MRA2007124
• 3)   Z.  Nishiyama: Kinzoku no Kenkyu, 11 (2013), 561.
• 4)   L.  Remy and  A.  Pineau: Materials Science and Engineering, 28 (1977), 99. https://doi.org/10.1016/0025-5416(77)90093-3
• 5)   M.  Koyama,  T.  Sawaguchi,  T.  Lee,  C. S.  Lee and  K.  Tsuzaki: Materials Science and Engineering A, 528 (2011), 7310. https://doi.org/10.1016/j.msea.2011.06.011
• 6)   Z.  Li,  K. G.  Pradeep,  Y.  Deng,  D.  Raabe and  C. C.  Tasan: Nature, 534 (2016), 227. https://doi.org/10.1038/nature17981
• 7)   Y.-K.  Lee,  J.-H.  Jun and  C.-S.  Choi: ISIJ Int., 37 (1997), 1023. https://doi.org/10.2355/isijinternational.37.1023
• 8)   T.  Sawaguchi,  T.  Maruyama,  H.  Otsuka,  A.  Kushibe,  Y.  Inoue and  K.  Tsuzaki: Materials Transactions, 57 (2016), 283. https://doi.org/10.2320/matertrans.MB201510
• 9)   Y.-B.  Ju,  M.  Koyama,  T.  Sawaguchi,  K.  Tsuzaki and  H.  Noguchi: Acta Materialia, 112 (2016), 326. https://doi.org/10.1016/j.actamat.2016.04.042
• 10)   A.  Sato,  E.  Chishima,  K.  Soma and  T.  Mori: Acta Metallurgica, 30 (1982), 1177. https://doi.org/10.1016/0001-6160(82)90011-6
• 11)   H.  Otsuka,  H.  Yamada,  T.  Maruyama,  H.  Tanahashi,  S.  Matsuda and  M.  Murakami: ISIJ Int., 30 (1990), 674. https://doi.org/10.2355/isijinternational.30.674
• 12)   S.  Takaki,  T.  Furuya and  Y.  Tokunaga: ISIJ Int., 30 (1990), 632. https://doi.org/10.2355/isijinternational.30.632
• 13)   M.  Koyama,  T.  Sawaguchi and  K.  Tsuzaki: ISIJ Int., 52 (2012), 161. https://doi.org/10.2355/isijinternational.52.161
• 14)   T.  Kaneko,  M.  Koyama,  T.  Fujisawa and  K.  Tsuzaki: Tetsu-to-Hagané, 102 (2016), 227 (in Japanese). https://doi.org/10.2355/tetsutohagane.TETSU-2015-081
• 15)   T.  Kaneko,  M.  Koyama,  T.  Fujisawa and  K.  Tsuzaki: ISIJ Int., 56 (2016), 2037. https://doi.org/10.2355/isijinternational.ISIJINT-2016-272
• 16)   K.  Tsuzaki,  Y.  Natsume,  Y.  Kurokawa and  T.  Maki: Scripta Metallurgica et Materialia, 27 (1992), 471. https://doi.org/10.1016/0956-716X(92)90213-X
• 17)   J.  Nakano and  P. J.  Jacques: Calphad, 34 (2010), 167. https://doi.org/10.1016/j.calphad.2010.02.001
• 18)   J.-B.  Seol,  J. E.  Jung,  Y. W.  Jang and  C. G.  Park: Acta Materialia, 61 (2013), 558. https://doi.org/10.1016/j.actamat.2012.09.078
• 19)   A.  Ariapour,  I.  Yakubtsov and  D. D.  Perovic: Materials Science and Engineering A, 262 (1999), 39. https://doi.org/10.1016/S0921-5093(98)01035-1
• 20)   P.  Behjati,  A.  Kermanpur and  A.  Najafizadeh: Materials Science and Engineering A, 588 (2013), 43. https://doi.org/10.1016/j.msea.2013.08.053
• 21)   L.  Mosecker and  A.  Saeed-Akbari: Science and Technology of Advanced Materials, 14 (2013), 033001. https://doi.org/10.1088/1468-6996/14/3/033001
• 22)   Y.  Hosoya,  A.  Inoue and  T.  Masumoto: Tetsu-to-Hagané, 64 (1978), 769 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.64.6_769
• 23)   S. M.  Teus,  V. N.  Shyvanyuk and  V. G.  Gavriljuk: Materials Science and Engineering A, 497 (2008), 290. https://doi.org/10.1016/j.msea.2008.07.003
• 24)   M.  Koyama,  C. C.  Tasan and  K.  Tsuzaki: Engineering Fracture Mechanics, 214 (2019), 123. https://doi.org/10.1016/j.engfracmech.2019.03.049
• 25)   N.  Narita,  C. J.  Altstetter and  H. K.  Birnbaum: Metallurgical Transactions A, 13 (1982), 1355. https://doi.org/10.1007/BF02642872
• 26)   M.  Koyama,  N.  Terao and  K.  Tsuzaki: Materials Letters, 249 (2019), 197. https://doi.org/10.1016/j.matlet.2019.04.093
• 27)   P.  Maulik and  J.  Burke: Scripta Metallurgica, 9 (1975), 17. https://doi.org/10.1016/0036-9748(75)90139-8
• 28)   P.  Rozenak: Metallurgical and Materials Transactions A, 45 (2014), 162. https://doi.org/10.1007/s11661-013-1734-7
• 29)   M.  Koyama,  K.  Hirata,  Y.  Abe,  A.  Mitsuda,  S.  Iikubo and  K.  Tsuzaki: Scientific Reports, 8 (2018), 16136. https://doi.org/10.1038/s41598-018-34542-0
• 30)   G. B.  Olson and  M.  Cohen: Journal of the Less Common Metals, 28 (1972), 107. https://doi.org/10.1016/0022-5088(72)90173-7
• 31)   C.  Hao,  M.  Koyama and  E.  Akiyama: Metallurgical and Materials Transactions A, 51 (2020), 6184. https://doi.org/10.1007/s11661-020-06021-7
• 32)   D. N.  Movchan,  B. D.  Shanina and  V. G.  Gavriljuk: International Journal of Hydrogen Energy, 38 (2013), 8471. https://doi.org/10.1016/j.ijhydene.2013.04.145
• 33)   Y.  Kuroki,  S.  Kawano,  S.  Iikubo,  H.  Ohtani,  M.  Koyama and  K.  Tsuzaki: Metallurgical and Materials Transactions A, 50 (2019), 3019. https://doi.org/10.1007/s11661-019-05237-6
• 34)   P. H.  Adler,  G. B.  Olson and  W. S.  Owen: Metallurgical and Materials Transactions A, 17 (1986), 1725. https://doi.org/10.1007/BF02817271
• 35)   S.-J.  Lee,  J.  Kim,  S. N.  Kane and  B. C. D.  Cooman: Acta Materialia, 59 (2011), 6809. https://doi.org/10.1016/j.actamat.2011.07.040
• 36)   M.  Koyama,  Y.  Abe and  K.  Tsuzaki: ISIJ Int., 58 (2018), 1745. https://doi.org/10.2355/isijinternational.ISIJINT-2018-260
• 37)   A.  Sato,  E.  Chishima,  Y.  Yamaji and  T.  Mori: Acta Metallurgica, 32 (1984), 539. https://doi.org/10.1016/0001-6160(84)90065-8
• 38)   S.  Curtze,  V. T.  Kuokkala,  A.  Oikari,  J.  Talonen and  H.  Hänninen: Acta Materialia, 59 (2011), 1068. https://doi.org/10.1016/j.actamat.2010.10.037
• 39)   G. B.  Olson and  M.  Cohen: Metallurgical Transactions A, 7 (1976), 1915. https://doi.org/10.1007/BF02659824
• 40)   K.  Tsuzaki,  Y.  Natsume,  Y.  Tomota and  T.  Maki: Scripta Metallurgica et Materialia, 33 (1995), 1087. https://doi.org/10.1016/0956-716X(95)00337-U
• 41)   M.  Koyama,  T.  Sawaguchi and  K.  Tsuzaki: Materials Science and Engineering: A, 528 (2011), 2882. https://doi.org/10.1016/j.msea.2010.12.065
• 42)   M.  Koyama,  C.  Hao,  E.  Akiyama and  K.  Tsuzaki: Metallurgical and Materials Transactions A, 51 (2020), 4439. https://doi.org/10.1007/s11661-020-05886-y
• 43)   K.  Tsuzaki,  K.  Fukuda,  M.  Koyama and  H.  Matsunaga: Scripta Materialia, 113 (2016), 6. https://doi.org/10.1016/j.scriptamat.2015.10.016
• 44)   Y.  Bu,  Z.  Li,  J.  Liu,  H.  Wang,  D.  Raabe and  W.  Yang: Physical Review Letters, 122 (2019), 075502. https://doi.org/10.1103/PhysRevLett.122.075502
• 45)   K.  Hirata,  S.  Iikubo,  M.  Koyama,  K.  Tsuzaki and  H.  Ohtani: Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science, 49a (2018), 5015. https://doi.org/10.1007/s11661-018-4815-9
• 46)   M.  Koyama,  T.  Ogawa,  D.  Yan,  Y.  Matsumoto,  C. C.  Tasan,  K.  Takai and  K.  Tsuzaki: International Journal of Hydrogen Energy, 42 (2017), 26423. https://doi.org/10.1016/j.ijhydene.2017.08.209
• 47)   B.  Kumai,  T.  Hojo,  M.  Koyama,  E.  Akiyama,  H.  Waki and  A.  Nagasaka: International Journal of Hydrogen Energy, 45 (2020), 27920. https://doi.org/10.1016/j.ijhydene.2020.07.036
• 48)   M.  Koyama,  Y.  Abe,  K.  Saito,  E.  Akiyama,  K.  Takai and  K.  Tsuzaki: Scripta Materialia, 122 (2016), 50. https://doi.org/10.1016/j.scriptamat.2016.05.012
• 49)   M.  Koyama and  K.  Tsuzaki: ISIJ Int., 55 (2015), 2269. https://doi.org/10.2355/isijinternational.ISIJINT-2015-215
• 50)   T.  Hojo,  M.  Koyama,  N.  Terao,  K.  Tsuzaki and  E.  Akiyama: International Journal of Hydrogen Energy, 44 (2019), 30472. https://doi.org/10.1016/j.ijhydene.2019.09.171
• 51)   J. V.  Badding,  R. J.  Hemley and  H. K.  Mao: Science, 253 (1991), 421. https://doi.org/10.1126/science.253.5018.421
• 52)   V. E.  Antonov,  K.  Cornell,  V. K.  Fedotov,  A. I.  Kolesnikov,  E. G.  Ponyatovsky,  V. I.  Shiryaev and  H.  Wipf: Journal of Alloys and Compounds, 264 (1998), 214. https://doi.org/10.1016/S0925-8388(97)00298-3
• 53)   M.  Koyama,  H.  Saitoh,  T.  Sato,  S.-i.  Orimo and  E.  Akiyama: Scientific Reports, 11 (2021), 19384. https://doi.org/10.1038/s41598-021-98938-1
• 54)   R. S.  Varanasi,  M.  Koyama,  H.  Saitoh,  R.  Utsumi,  T.  Sato,  S.-i.  Orimo and  E.  Akiyama: International Journal of Hydrogen Energy, 48 (2023), 10081. https://doi.org/10.1016/j.ijhydene.2022.11.274
• 55)   A.  Shibata,  M.  Enoki,  N.  Saji,  H.  Tai,  M.  Koyama,  H.  Ohtani,  N.  Tsuji and  K.  Tsuzaki: Metallurgical and Materials Transactions A, 50 (2019), 4027. https://doi.org/10.1007/s11661-019-05320-y
• 56)   K.  Yamada,  M.  Koyama,  T.  Kaneko and  K.  Tsuzaki: Scripta Materialia, 105 (2015), 54. https://doi.org/10.1016/j.scriptamat.2015.05.007
• 57)   Y.  Ogawa,  H.  Hosoi,  K.  Tsuzaki,  T.  Redarce,  O.  Takakuwa and  H.  Matsunaga: Acta Materialia, 199 (2020), 181. https://doi.org/10.1016/j.actamat.2020.08.024
• 58)   H.  Nishida,  Y.  Ogawa and  K.  Tsuzaki: Materials Science and Engineering: A, 836 (2022), 142681. https://doi.org/10.1016/j.msea.2022.142681