ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Regular Article
Repulsive Nature for Hydrogen Incorporation to Fe3C up to 14 GPa
Hidenori TerasakiYuki ShibazakiKeisuke NishidaRyuji TateyamaSuguru TakahashiMiho IshiiYuta ShimoyamaEiji OhtaniKen-ichi FunakoshiYuji Higo
Author information

2014 Volume 54 Issue 11 Pages 2637-2642


We have performed in situ X-ray diffraction measurements under high pressure and high temperature to study hydrogen solubility in Fe3C carbide. Hydrogen solubility can be estimated from a volume expansion associated with hydrogen incorporation into metal. The lattice volumes and phase relations of Fe3C–H system and Fe3C were measured up to 14 GPa and 1973 K. The lattice volumes of Fe3C measured in this study are well fitted using the 3rd order Birch–Murnaghan equation of state with the reported elastic parameters of Fe3C. Obtained lattice volumes of Fe3C–H are quite consistent with those of Fe3C. No difference between the melting temperatures of Fe3C–H and Fe3C was observed. These results demonstrate that hydrogen incorporation into Fe3C does not occur and hydrogen is unlikely to coexist with carbon in iron-alloy up to 14 GPa.

1. Introduction

If hydrogen dissolves into metal or alloy, its melting temperature and physical properties are significantly changed. This hydrogen dissolution often causes hydrogen embrittlement. Although iron-carbide, Fe3C, is widely used and closely related to steel-making process, hydrogenation of Fe3C has never been studied systematically under pressure. Thus, it is important and required to clarify hydrogenation condition and hydrogen solubility in Fe3C.

Hydrogenation reaction can be expressed as,   

M+ x 2 H 2 =M H x (1)
where M is metal and x represents hydrogen content in metal hydride (MHx). If a system reach in chemical equilibrium of reaction (1) and chemical potential of hydrogen in MHx (μs) is written as μs = 1/2μf (μf is chemical potential of hydrogen gas), hydrogen content (x) can be expressed as follows,1)   
x= X exp[ h s -1/2 μ f k B T ]+1 (2)
where hs, kB, T, and X denote enthalpy of hydrogen in MHx, Boltzmann constant, temperature, and maximum solubility of hydrogen in MHx, respectively. Although μf is approximated by ideal gas equation (μf= μ0+kBTlnP/P0) up to 0.1 GPa, μf starts to increase over the ideal gas behaviour above 0.1 GPa and to increase drastically above 1 GPa.2,3) This leads effective hydrogen incorporation into metal or alloy under pressure.3,4) For example, hydrogen incorporation into Fe3,5) and Fe-10wt%Ni6) are significantly promoted above 3 GPa. In case of iron-light element alloy, hydrogen solubilities into FeSi and FeS have been studied up to 19 GPa in previous studies.7,8) Hydrogen tends to dissolve into FeSi above 10 GPa and into FeS above 3 GPa. The hydrogenation pressure of FeSi is much higher than those of Fe and FeS (~3 GPa). The hydrogen contents in FeSiHx and FeSHx are estimated to x~0.07–0.22 and 0.2–0.4, respectively. Those contents are clearly lower than hydrogen content in fcc and dhcp-FeHx (x~1). This suggests that hydrogenation pressure and hydrogen content strongly depend on the composition of iron-alloy. In this study, we have measured lattice volumes and phase relations of Fe3C–H system and Fe3C and estimated hydrogenation condition up to 14 GPa based on in situ X-ray diffraction (XRD) experiments under high pressure and high temperature.

2. Experimental

Starting material was Fe3C powder (99.9% purity, Rare Metallic Co. Ltd.). Starting powder was identified as Fe3C-cementite from X-ray diffraction pattern at ambient pressure and room temperature. The sample was enclosed in a cylindrical NaCl capsule, which was reported to seal hydrogen at high pressure.9) LiAlH4 was used as a hydrogen source and it was enclosed at the bottom of NaCl capsule. The Fe3C sample and LiAlH4 layers were separated by a thin MgO disc. After a thermal decomposition of LiAlH4 at high temperature (423–493 K at ambient pressure),9) H2 was released and then supplied to the Fe3C sample. Schematic picture of the cell assembly used are illustrated in Fig. 1.

Fig. 1.

Schematic picture of the cell assembly used for the pressure above 7 GPa.

High-pressure experiments were performed using 1500 ton Kawai-type multi-anvil device (SPEED Mk–II) installed at BL04B1 beamline, SPring-8 synchrotron radiation facility.10) High temperature was generated using cylindrical resistive heater made of graphite for the experiment below 7 GPa and of TiB2–BN composite for the experiment above 7 GPa. Experimental temperature was monitored using a W3%Re–W25%Re thermocouple inserted in the heater. Experimental pressure was determined from lattice parameters of pressure markers (BN, MgO and NaCl) combined with their equations of state.11,12,13)

In situ XRD measurement at high-pressure and high temperature was carried out using white X-ray (vertical size: 0.2 mm × horizontal size: 0.1 mm) with an energy-dispersive method. XRD of the sample and the pressure markers were collected using Ge-solid state detector with a diffraction angle of 5.5°. X-ray diffraction of the sample at high temperature was collected with 100 K step from 300 K. To minimize the effect of sample grain growth on the diffraction patterns at high temperature, the press was oscillated from 0° to 4° during the XRD measurement.14)

To obtain hydrogen content in the sample using the lattice volume of the Fe3C–H sample, we also need to know lattice volumes of Fe3C at the present experimental conditions (see details in section 3.2). Thus, we also carried out separate experiments using Fe3C sample and same cell assembly without hydrogen source (LiAlH4) at the similar pressure and temperature conditions of Fe3C–H sample. The experimental conditions are listed in Tables 1 and 2. The pressure and temperature conditions of this study ranges 5.3–15.3 GPa and 300–1973 K, respectively.

Table 1. Experimental conditions and obtained lattice volume of Fe3C–H system.
RunP [GPa]P error [GPa]T [K]PhasesV [A3]V error [A3]EOS V [A3]*
*EOS V  represents the lattice volume of Fe3C calculated using 3rd order Birch-Murnaghan EOS with reported elastic parameters.18)

Table 2. Experimental conditions and obtained lattice volume of Fe3C.
RunP [GPa]P error [GPa]T [K]PhasesV [A3]V error [A3]

3. Results and Discussion

3.1. Phase Relations and Melting Temperatures of Fe3C–H and Fe3C

Typical XRD patterns of the Fe3C–H sample are shown in Fig. 2. Based on XRD measurements, obtained phase relations of Fe3C–H and Fe3C are shown in Fig. 3. Difference in stable solid phase was not observed between Fe3C–H and Fe3C samples. In both cases, Fe3C was confirmed to have an orthorhombic cementite structure (space group: Pnma) and there was no phase transition of Fe3C in all the present pressure range before melting.

Fig. 2.

XRD patterns of Fe3C–H sample at 13 GPa and at 1473 K (bottom), 1773 K (top). At 1473 K, most of the peaks were identified as Fe3C cementite. Other peaks corresponded to NaCl and FeO as shown by “N” and asterisk, respectively. At 1773 K, Fe7C3 coexisted with liquid.

Fig. 3.

Phase diagram and melting curves of Fe3C–H and Fe3C samples. Circles and triangles represent Fe3C–H and Fe3C samples, respectively. White, gray and black symbols denote stability field of Fe3C, Fe7C3 with liquid, and liquid, respectively. Black dashed-dotted curves show solidus and liquidus temperatures of Fe3C–H sample. Solidus curve of the Fe3C–H system corresponds to a tie line of temperatures between experimentally determined upper limit point where solid Fe3C was observed and lower limit where Fe7C3+liquid were observed. In the same manner, liquidus curve corresponds to the tie line of the temperatures between Fe7C3+liquid and liquid. For reference, solidus and liquidus temperatures of Fe3C16) are also plotted as thin gray dotted curves.

The sample melting was determined on the basis of disappearance of the sample diffraction peaks and appearance of diffuse scattering signal. It was confirmed that peak position of the diffuse scattering corresponded to the position where the most intensive sample peak was observed before melting. The Fe3C melted incongruently to the liquid coexisted with Fe7C3 at 1673 K, 5.3 GPa and at 1673 K, 13 GPa as shown in Fig. 2 (see also Table 1). Fe7C3 was a hexagonal structure as reported by Herbstein and Snyman.15) This melting sequence is consistent with the result reported by Nakajima et al.16) Eutectic melting temperature (solidus) and liquidus curves of Fe3C–H and Fe3C are also drawn in Fig. 3. The solidus and liquidus temperatures of Fe3C determined in this study are mostly consistent with those reported by Nakajima et al.16) although the solidus of this study is slightly (~120 K) lower than that of Nakajima et al.16) at 12.4 GPa. Our solidus and liquidus temperatures of Fe3C–H sample agree well with those of Fe3C in the pressure range studied. Hydrogen incorporation into metal often causes a depression of the melting temperature of the metal. For instance, in case of iron, the depression of melting temperature from Fe to FeHx is reported to 600–900 K at 3–21 GPa.17) Therefore, no difference in melting temperature between Fe3C–H and Fe3C is likely to reflect that hydrogen incorporation into Fe3C does not occur at these pressures.

3.2. Lattice Volumes of Fe3C–H and Fe3C

In general, the lattice volumes of transition metals and alloys expand uniformly as a result of hydrogen incorporation into interstitial sites of metals and alloys.3) The volume increase associated with hydrogen incorporation is, therefore, closely related to hydrogen content in the alloy. The hydrogen content (x) in metal hydride can be denoted by a following Eq. (3),   

x= V M H x - V M V H (3)
where VMHx, VM, and VH represent atomic volumes of metal hydride and metal, and volume increase per hydrogen atom, respectively.3) Thus, (VMHxVM) indicates the volume increase caused by hydrogen incorporation. VMHx and VM are directly obtained from the present diffraction measurements of Fe3C–H and Fe3C samples, respectively. Thus, the hydrogen content (x) can be calculated using Eq. (3) if the VH is given.

Based on the XRD patterns, obtained lattice volumes of both Fe3C–H and Fe3C samples in the stability field of Fe3C cementite phase are summarized in Tables 1 and 2, respectively. Change in the lattice volumes of Fe3C–H and Fe3C samples with temperature are shown in Figs. 4(a)–4(d). The lattice volumes of Fe3C calculated using recently reported elastic properties of Fe3C18) with 3rd order Birch–Murnaghan equation of state were also shown in Fig. 4 by dotted curves. The lattice volume of Fe3C–H sample increases with increasing temperature due to thermal expansion as observed in Fe3C sample. The measured lattice volumes of Fe3C–H and Fe3C samples have relatively large errors below 873 K and slightly deviate from calculated volume of Fe3C (dotted curves). However, measured volumes become to be consistent with calculated value above 873 K as shown in Fig. 4. This is because deviatric stress is likely to remain still in the sample at temperatures below 873 K, which is supported by occurrence of XRD peak sharpening above 873 K. After release of deviatric stress above 873 K, it is found that there is no difference between the measured and calculated volumes. On the contrary, the Fe3C sample for the measurement at 7.6–9.4 GPa was once heated up to 1073 K before the XRD measurement at high temperatures. In this case, since the deviatric stress was released by the primary heating, the measured lattice volume completely overlaps with calculated volumes even from 300 K (Fig. 4(c)).

Fig. 4.

The effect of temperature on the lattice volumes of Fe3C–H (a, b) and Fe3C (c, d) samples at around 8 and 14 GPa. Solid circles and triangles represent the volumes of Fe3C–H and Fe3C samples, respectively. Dotted curve indicate the calculated volume of Fe3C using Birch–Murnaghan EOS and elastic parameters reported by Litasov et al.18)

The effect of pressure on the lattice volumes of Fe3C–H and Fe3C samples at 1073–1473 K are shown in Figs. 5(a), 5(b). Obtained lattice volumes of Fe3C match well with calculated compression curve of Fe3C18) (Fig. 5(b)). The elastic properties used to calculate the compression curve are a lattice volume at ambient pressure V0=154.56 Å3, isothermal bulk modulus at ambient pressure K0T=190 GPa and its pressure derivative dKT/dP=4.8. Thermal contributions used are a temperature derivative of KT (dKT/dT)=–0.029 GPa/K and thermal expansion α=(3.90×10–5) +(1.22×10–8)T K–1. Thus, the lattice volumes of Fe3C in this study are confirmed to agree with those of previous study.18)

Fig. 5.

The effect of pressure on the lattice volumes of Fe3C–H (a) and Fe3C (b) samples at 1073, 1273, and 1473 K. Dotted curve indicates the calculated volume of Fe3C using Birch–Murnaghan EOS and elastic parameters reported by Litasov et al.18) at respective temperatures.

The lattice volumes of Fe3C–H are in good agreement with those of Fe3C in the range of present measurement up to 14–15 GPa as shown in Fig. 5(a). The lattice volumes of Fe3C were calculated using the elastic properties of Fe3C18) at the same P–T conditions of the experiments of Fe3C–H sample. Therefore, there is no difference in the lattice volumes between Fe3C–H and Fe3C samples, i.e., (VMHxVM)~0. This result provides an important aspect that hydrogen incorporation associated with volume expansion does not occur to Fe3C up to 14 GPa. This result is supported by no difference in solidus and liquidus temperatures between Fe3C–H and Fe3C in the pressure range studied here.

In this study, hydrogen is supplied to Fe3C from hydrogen source, LiAlH4, as a result of thermal decomposition of LiAlH4. The decomposition temperature of LiAlH4 has been estimated to 423–493 K at ambient pressure, 573 K at 4.7 GPa, 873 K at 12 GPa, and 973 K at 18.5 GPa based on the hydrogenation temperature of iron-alloys in previous studies.9,7,17) According to these results, most of the temperature conditions in this study are well above the decomposition temperature, suggesting that hydrogen was supplied to Fe3C sample at the present conditions (only except for the conditions below 873 K at 14 GPa).

3.3. Stability of Carbide vs Hydride at High Pressure

The present results show that hydrogen does not react with Fe3C up to 14 GPa. In terms of possible coexistence of hydrogen and carbon in iron, a reaction between Fe and CnH2n+2 (paraffin) has been studied recently based on a XRD measurement using a diamond anvil cell.19) They reported that some chemical reactions were promoted at elevated temperatures and at 54 GPa. First, iron reacted with CnH2n+2 and Fe3C carbide was formed with release of H2 at 1400 K (reaction [i]: 3nFe + CnH2n+2 = nFe3C + (n+1)H2), suggesting that H2 coexists with Fe3C as different isolated phases rather than forming hydride or H2 in Fe3C crystal structure. Then, at higher temperature (1650 K), H2 became to react with Fe3C and FeH iron-hydride was formed together with diamond (reaction [ii]: 2Fe3C + 3H2 = 6FeH + 2C). This result suggests that hydrogen does not coexist with carbon in iron-alloy. In other words, both hydrogen and carbon does not incorporate together into Fe. If carbon reacts with iron and Fe3C is formed, H2 exists as a different phase (reaction [i]). Alternatively, if hydrogen incorporates into Fe and iron-hydride is formed, then carbon exists as a different phase, such as diamond (reaction [ii]). In this study, stable coexistence of Fe3C with H2 (right hand side of reaction [i]) is confirmed in all the present temperature conditions up to 14 GPa. Therefore, hydrogen is not incorporated into carbide and exists as a different phase (such as H2) at any temperatures at least up to 14 GPa and below 1650 K at 54 GPa. At higher pressure and temperature (e.g., 54 GPa and above 1650 K), hydrogen becomes to dissolve in iron-carbide and, then, carbide decomposes to diamond and iron-hydride. It is note that all these results suggest that disaffinity nature of hydrogen with carbon, i.e., hydrogen is not incorporated into iron with carbon.

There are several possible reasons for this repulsive effect of hydrogen to Fe3C. One is due to a size of available interstitial site for hydrogen in Fe3C cementite. Interstitial tetrahedral site of Fe3C structure is likely to be much smaller than that of Fe and FeSi. Small size of interstitial site and short distance between the interstitial sites tend to block hydrogen incorporation.20) The other possible reason is due to a small value of μf for the Fe3C–H2 reaction (see Eq. (2)). For future study, in situ neutron diffraction measurement under high pressure is needed to investigate possible hydrogen position in iron-alloy structures and to clarify mechanism of hydrogen incorporation in iron-alloys. In situ neutron diffraction measurement under high pressure is now possible using the multi-anvil high pressure device, which was recently installed at a beamline of the J-PARC neutron facility.21)

4. Conclusions

The lattice volumes and phase relations were studied for Fe3C–H and Fe3C based on in situ XRD measurement up to 14 GPa and 1973 K. Obtained lattice volumes of Fe3C–H matches well with those of Fe3C. There is no difference between the melting temperatures (solidus and liquidus temperatures) of Fe3C–H and Fe3C. These results suggest that hydrogen does not incorporate into Fe3C at the conditions of this study. Combining with the previous study,19) it is most likely that hydrogen does not coexist with carbon in iron-alloy.


This work was supported by Grants-in-Aid for scientific research from the Ministry of Education, Culture, Science, and Sport and Technology of the Japanese Government to H. T. (no. 23340159, 23654181, 26247089) and E. O. (no. 2200002). This work was also supported by an ISIJ (The Iron and Steel Institute of Japan) Research Promotion Grant. The experiments have been performed under contract of the SPring-8 (Proposal number: 2009B1184, 2009B1696 and 2011A1546).

© 2014 by The Iron and Steel Institute of Japan