Accepted Manuscript

Hydrogen trapping and fatigue crack growth property of low-carbon steel in

hydrogen-gas environment

Junichiro Yamabe, Michio Yoshikawa, Hisao Matsunaga, Saburo Matsuoka

PII: S0142-1123(17)30181-0
DOI: http://dx.doi.org/10.1016/j.ijfatigue.2017.04.010
Reference: JIJF 4320

To appear in: International Journal of Fatigue

Received Date: 15 January 2017

Revised Date: 22 April 2017
Accepted Date: 23 April 2017

Please cite this article as: Yamabe, J., Yoshikawa, M., Matsunaga, H., Matsuoka, S., Hydrogen trapping and fatigue
crack growth property of low-carbon steel in hydrogen-gas environment, International Journal of Fatigue (2017),
doi: http://dx.doi.org/10.1016/j.ijfatigue.2017.04.010

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Submitted to: ECF21 special issue on International Journal of Fatigue

Hydrogen trapping and fatigue crack growth property of low-carbon steel in hydrogen-gas
environment

Junichiro Yamabea,b,c*, Michio Yoshikawab, Hisao Matsunagab,c,d, Saburo Matsuokab

a)
International Research Center for Hydrogen Energy, Kyushu University, 744 Moto-oka, Nishi-ku,
Fukuoka-shi, 819-0395, Japan
b)
Research Center for Hydrogen Industrial Use and Storage, Kyushu University, 744 Moto-oka, Nishi-ku,
Fukuoka-shi, 819-0395, Japan
c)
International Institute for Carbon-Neutral Energy Research, Kyushu University, 744 Moto-oka,
Nishi-ku, Fukuoka-shi, 819-0395, Japan
d)
Department of Mechanical Engineering, Kyushu University, 744 Moto-oka, Nishi-ku, Fukuoka-shi,
819-0395, Japan

*Corresponding author
Junichiro Yamabe
International Research Center for Hydrogen Energy, Kyushu University, 744 Moto-oka, Nishi-ku,
Fukuoka-shi, Fukuoka 819-0395, Japan
Zip code: 819-0395
Phone: +81-92-802-3247
E-mail: yamabe@mech.kyushu-u.ac.jp

Abstract
Fatigue crack growth (FCG) tests for compact-tension specimens of low-carbon steel were
performed under various combinations of hydrogen pressures (0.1–90 MPa), test frequencies
(0.001–10 Hz), and test temperatures (room temperature, 363 K, and 423 K). For quantifying the
FCG acceleration, the hydrogen trapping and diffusivity of prestrained specimens were determined.
Depending on the test conditions, the FCG was accelerated. The hydrogen-assisted FCG acceleration
always accompanied localized plastic deformations near the crack tip. The onset of the FCG
acceleration was roughly quantified by using a simplified parameter representing the gradient of the
hydrogen concentration at the crack tip.

Keywords: trapping; hydrogen gas; fatigue crack growth; low-carbon steel; elevated temperature

1
1. Introduction
Stationary fuel cells and fuel-cell vehicles have been commercialized, and related hydrogen
stations are being constructed. For widespread commercialization, it is necessary to develop
hydrogen systems with a low cost and high performance. Currently, high-pressure hydrogen gas is
actively investigated for the transport and storage of hydrogen. Hydrogen sometimes degrades the
tensile and fatigue properties of metallic materials [1–4]; therefore, for ensuring the safety and
reliability of hydrogen systems, a design method must be established in consideration of the
detrimental effect of hydrogen on materials [5–9]. To develop a safe and reliable finite-life design, it
is important to precisely determine the fatigue crack growth (FCG) property of materials in the
presence of hydrogen. For the low-alloy steel JIS-SCM435 with an ultimate tensile strength (UTS)
lower than 900 MPa in high-pressure gaseous hydrogen, there exists an upper bound of the FCG
acceleration, although there is a significant degradation in the reduction of area (RA) during
slow-strain-rate tensile (SSRT) testing; therefore, this low-alloy steel is considered to be eligible for
hydrogen service under the finite-life design [7]. In contrast, there is no upper bound of FCG
acceleration in the presence of hydrogen for high-strength steel with a UTS of ~1,900 MPa;
therefore, this steel is not eligible for hydrogen service under the finite-life design [10]. These results
imply that the FCG property in the presence of hydrogen is strongly dependent on the material.
On the other hand, hydrogen transport and storage with liquid hydrogen and organic hydrides
are being demonstrated as future technologies. The establishment and commercialization of
worldwide hydrogen supply chains that have a capability to procure low-cost and stable hydrogen
are being investigated for large-scale hydrogen utilization related to hydrogen power generation. In
the long term, the transport and storage of CO2-free production of hydrogen using a combination of
carbon capture and storage and renewable energy are planned. Considering future hydrogen society,
a wide range of hydrogen environmental conditions, such as hydrogen environments at low and high
temperatures should be targeted.
In this study, to enable the use of various low-cost steels under a wide range of hydrogen
environmental conditions, we investigated the effects of the hydrogen pressure, test frequency, and
test temperature on the FCG properties of an annealed, low-carbon steel: JIS-SM490B. The FCG
properties of the low-carbon steel in hydrogen gas at room temperature (RT) were presented in [11].
In this paper, we present for the first time the FCG properties in hydrogen gas at elevated
temperatures; then, we discuss in detail the FCG properties under various combinations of hydrogen
pressure, test frequency, and test temperature. To estimate the hydrogen-diffusion properties at the
crack tip, the hydrogen-trapping sites and hydrogen diffusivity of prestrained specimens of the
low-carbon steel were determined. In addition, a novel parameter representing the gradient of the
hydrogen concentration near the crack tip was introduced for predicting the onset for the

2
hydrogen-enhanced FCG acceleration of the steel under a wide range of hydrogen environmental
conditions.

2. Experimental Procedures
2.1 Material
The material used in this study was an annealed, low-carbon steel, JIS-SM490B, composed
of 0.16 carbon, 0.44 silicon, 1.43 manganese, 0.017 phosphorus, and 0.004 sulfur (in mass %); the
remainder was iron as shown in Table 1. Figure 1 shows an optical microscopy of the microstructure,
which exhibited a ferrite and pearlite structure. The Vickers hardness of the matrix was HV = 153,
measured (20 points) with a load of 9.8 N. The lower-yield stress, UTS, elongation, and RA at RT in
air were σLY = 360 MPa, σB = 540 MPa, δ = 17%, and φ = 78%, respectively.
To estimate the hydrogen-diffusion properties of the steel at the crack tip where severe plastic
strain was produced, cold worked JIS-SM490B steel plates were prepared, in addition to an
as-received plate (no cold working). Figure 2 illustrates the cold-working process for the steel plates.
The ratio of cold working, CW, is defined as follows:

Bp0  Bp
CW [%]  100( ), (1)
Bp0

where Bp0 and Bp represent the steel plates before and after cold working, respectively. We prepared
steel plates with CW values of 5, 10, 15, 20, 30, and 40 %. Because the change in width produced by
the cold working can be considered as negligible, the plastic strain (prestrain), εpre, in the rolling
direction can be expressed as follows:

Bp0
 pre  ln( ). (2).
Bp

2.2 Determination of hydrogen-trapping sites and hydrogen diffusivity

Cylindrical specimens with 2r0 = 19 mm and z0 = 2 mm for determining the hydrogen-trapping
sites and with 2r0 = z0 = 19 mm for determining the hydrogen diffusivity, where 2r0 is the diameter
and z0 is the height, were sampled from the cold-worked plates. The surface of the specimens was
finished with #600 emery paper. The specimens were exposed to hydrogen gas at 100 MPa and a
temperature of 358 K for 200 h to obtain a uniform distribution of hydrogen. After the exposure, the
hydrogen contents of the specimens were measured under rising or constant temperatures by gas
chromatography–mass spectroscopy (GC–MS). According to the thermal desorption-rate spectra
obtained at heating rates of 50, 100, and 200 K/h, the activation energies of the trapping sites were
determined using the Choo–Lee method [12]. The hydrogen diffusivity was determined by fitting the

3
solution of a diffusion equation to the residual hydrogen contents, CR, measured at various constant
temperatures [13–15]:

 exp[(2n  1)2  2 DtR / z02 ]  exp[ D 2 t / r 2 ]

CR (tR )  AR {  }　 {  m R 0
}, (3)
n 0 (2n  1)2 m 1  m2

where tR is the elapsed time after hydrogen exposure, AR is the fitted constant related to the saturated
hydrogen content [15], D is the hydrogen diffusivity, r0 is the radius of the specimen, z0 is the
thickness of the specimen, and βm is the root of the zero-order Bessel function.
Figure 3 shows an example of the relationship between the residual hydrogen content and the
elapsed time after hydrogen exposure for a hydrogen-charged iron-based alloy [15]. After exposure
of a cylindrical specimen with 2r0 = 7 mm and z0 = 2 mm to hydrogen gas at a pressure of 10 MPa
and a temperature of 543 K for 200 h, the hydrogen-charged specimen was kept in the GC–MS
chamber at a constant temperature of 543 K. The experimental data were fitted by Eq. (3), as AR and
D were unknown parameters; consequently, the data were successfully reproduced by Eq. (3). The
hydrogen diffusivities of various austenitic stainless steels and tempered low-alloy steels determined
by this method are consistent with data from the literature [14, 15]. Especially, the tempered
low-alloy steels shows a similar hydrogen diffusivity to the cold worked JIS-SM490B steels used in
this study.

2.3 FCG test

For the FCG test, a compact-tension (CT) specimen with a width, W, of 50.8 mm and a
thickness, B, of 10 mm was sampled from the as-received plate. Table 2 shows the FCG-test
conditions in high-pressure hydrogen gas. The FCG test was performed at a stress ratio, Rσ, of 0.1
under various combinations of hydrogen-gas pressures, pH2, ranging from 0.1 to 90 MPa; test
frequencies, f, ranging from 0.001 to 10 Hz; and test temperatures of RT, 363 K, and 423 K, in
accordance with ASTM E647-08e1 [16]. The purity of the hydrogen gas in the cylinder was
99.999%, and the measured oxygen contents were always less than 1.0 vol. ppm; therefore, the effect
of oxygen on the FCG was ignored [17, 18].
In addition to the FCG test under a constant load range, ΔP, the FCG test was performed
under a constant stress-intensity factor range, ΔK, to clarify the effect of f on the FCG rate. These
two types of tests are called ΔP-constant and ΔK-constant tests, respectively. The crack size was
determined using the compliance method with the following crack-opening displacements (CODs):

  a / W  1.0010 4.6695u  18.46 u 2  236.82 u 3  1214.9 u 4  2143.6 u 5 , (1)

EVg B
u  ([ ]1 / 2  1) 1 , (2)
P

4
where a is the crack length, W is the specimen width, B is the specimen thickness, E is Young’s
modulus, and Vg is the COD. The ΔK was calculated as follows:

P2   
K  (0.886 4.64　 13.32 2  14.72 3  5.6 4 ) . (3).
B W 1    3/ 2

The FCG tests were performed under the following small-scale yielding condition:

4 K 4 K
W  a  ( )( max )2  ( )[ ]2 , (4)
  YS   YS (1  R )

where σYS is the yield stress. After the FCG test, the slip deformations were observed by laser
microscopy (LM). The fracture surface morphologies were observed by scanning electron
microscopy.

3. Results and Discussion

3.1 Determination of hydrogen-trapping sites produced by cold working
Figure 4 shows the thermal-desorption analysis (TDA) spectra of the hydrogen-charged
prestrained specimens with CW values of 5% and 40% under a heating rate of 100 K/h. For the
prestrained specimen with a CW of 5%, there were three peaks of the hydrogen-desorption rate, at
temperatures of approximately 373 K (P1), 473 K (P2), and 743 K (P3). In contrast, for the
prestrained specimen with a CW of 40%, a new peak of the hydrogen-desorption rate was observed
at a temperature of ~443 K (P4), in addition to the three peaks (P1, P2, P3) observed for the
prestrained specimen with a CW of 5%.
Figure 5(a) shows the TDA spectra of hydrogen-charged prestrained specimens with a CW of
40% under heating rates of 50, 100, and 200 K/h, and Fig. 5(b) shows the determination of the
activation energies for P1, P2, P3, and P4 by the Choo–Lee method [12]. According to the analysis
based on the Choo–Lee method [12], the activation energy, Ea, for each trapping site satisfies the
equation

[ln(Tp 2 /  )] Ea
 , (5)
(1 / Tp ) R

where γ is the applied heating rate, Tp is the absolute temperature for the peak of the
hydrogen-desorption rate, and R is the gas constant. The activation energies for each trapping site
were calculated as 20 kJ/mol (P1), 70 kJ/mol (P2), 98 kJ/mol (P3), and 49 kJ kJ/mol (P4),
respectively. Provided that the activation energy for lattice hydrogen diffusion, ED, is 6 kJ/mol [19],
the biding energies, EB, of the trapping sites are estimated as 13 kJ/mol (P1), 64 kJ/mol (P2), 92

5
kJ/mol (P3), and 43 kJ/mol (P4). Takai [20] reported the biding energies of hydrogen-trapping sites
in body-centered cubic (BCC) iron and steel as follows: 0–20 kJ/mol for the elastic field around the
dislocations; 59 kJ/mol for the dislocation cores; 41, 45, 61 kJ/mol for the vacancies; and >84 kJ/mol
for the strained Fe3C interfaces. Li et al. [21] reported that the hydrogen-trapping sites in
secondary-hardening martensitic steel are associated with martensitic interfaces, prior austenitic
grain boundaries, and mixed dislocation cores (61 kJ/mol) and metal carbides (89 kJ/mol). In
contrast, many researchers consider that hydrogen-trapping sites with a low biding energy (<20
kJ/mol) are associated with the elastic field of dislocation [12, 22, 27].
In light of the aforementioned research, the trapping site corresponding to P1 (13 kJ/mol) is
assigned to the elastic field of dislocations. The trapping sites for P2 (64 kJ/mol) and P3 (92 kJ/mol)
are associated with grain boundaries and dislocation cores (P2) and carbide-related sites (P3),
respectively. The trapping site for P4 (43 kJ/mol) detected at a high CW is associated with vacancies.

3.2 Hydrogen diffusivity

3.2.1 Modified method for determining hydrogen diffusivity
Figure 6(a) shows the relationship between the residual hydrogen content and the elapsed time
after hydrogen exposure for the hydrogen-charged prestrained specimen with a CW of 40% at a
constant temperature of 373 K. Although the fitted line is also shown in the figure, the experimental
data could not be reproduced by Eq. (3), in contrast to the case of the iron-based alloy shown in Fig.
3. This is considered to be because slow hydrogen release, which did not follow the
diffusion-controlled process, occurred after hydrogen was released by the diffusion-controlled
process; i.e., the slow hydrogen release occurred during the desorption-controlled process. To
reproduce this complicated hydrogen release, the residual hydrogen content, CR, was fitted by the
following modified equation:

CR (tR )  C1R (tR )  C2 R (tR )

 exp[(2n  1)2  2 DtR / z02 ]  exp[ D 2 t / r 2 ]
C1R (tR )  A1R {  }　 {  m R 0
}, (6)
n 0 ( 2 n  1) 2
m 1  2
m
C2 R (tR )  A2 R exp(  tR )

where C1R is the residual hydrogen content released by the diffusion-controlled process, C2R is the
residual hydrogen content released by the desorption-controlled process, A1R is the fitted constant
associated with the hydrogen content released by the diffusion-controlled process, A2R is the fitted
constant associated with the hydrogen content released by the desorption-controlled process, and ξ is
the desorption-rate constant.

6
 Figure 6(b) shows the fitted line obtained using Eq. (6), indicating that the experimental data
were successfully reproduced by Eq. (6). The other experimental data also agree well with the fitted
lines obtained using Eq. (6).

3.2.2 Determined hydrogen diffusivity

Figure 7 shows the Arrhenius plots of the hydrogen diffusivity of the prestrained specimens
with CW values of 5, 10, and 40%, together with data from the literature [19, 23]. Asano et al.
determined the hydrogen diffusivity of a low-carbon steel with 40% rolling by using electrochemical
permeation [23]. Despite the use of the modified equation in this study, the hydrogen diffusivity of
the prestrained specimens showed a temperature dependence similar to that reported by Asano et al.
((d) in Fig. 7) [23]. These evidences suggest that the modified equation in Eq. (6) provides a
reasonable hydrogen diffusivity. Furthermore, the hydrogen diffusivity of the prestrained specimens
was significantly lower than the lattice hydrogen diffusivity ((a), (b), and (c) in Fig. 7), indicating
that the hydrogen diffusivity obtained here represents the apparent hydrogen diffusivity, Dapp,
affected by trapping.
Figure 8(a) shows the relationship between the apparent hydrogen diffusivity, Dapp, and the
cold-working ratio, CW, at temperatures of 303 K (RT), 363 K, and 423 K. Figure 8(b) shows the
relationship between the activation energy of the apparent hydrogen diffusivity, ED,app, and the
cold-working ratio, CW. The activation energy of the apparent hydrogen diffusivity was obtained by
using the following Arrhenius-type equation:

ED,app
Dapp  D0,app exp( ), (7)
RT

where D0,app is the fitted constant for the pre-exponential factor of the apparent hydrogen diffusivity.
For the prestrained specimens with a CW of <20%, as the CW increased, the apparent hydrogen
diffusivity decreased, and the activation energy increased. In contrast, for the prestrained specimens
with a CW of ≥ 20 %, the apparent hydrogen diffusivity and activation energy were not influenced
by the cold working.
The activation energies obtained from the TDA spectra shown in Fig. 5 and the Arrhenius
plots of the apparent hydrogen diffusivity in Fig. 7 were compared. As shown in Fig. 5, the
activation energies of P1 and P4 were 20 and 49 kJ/mol, respectively. On the other hand, the
activation energies of the apparent hydrogen diffusivity were 20 kJ/mol for the prestrained specimen
with a CW of 5% and 34 kJ/mol for the prestrained specimen with a CW of ≥ 20 %. For the
prestrained specimen with a CW of 5%, the peak of the hydrogen-desorption rate for P4 was not
observed, and the activation energy of the apparent hydrogen diffusivity was consistent with that for
the trapping site associated with P1. In contrast, for the prestained specimens with a CW of ≥ 20 %,

7
both peaks of the hydrogen-desorption rate for P1 and P4 were observed, and the activation energy
of the apparent hydrogen diffusivity (34 kJ/mol) was consistent with the average of the activation
energies of P1 (20 kJ/mol) and P4 (49 kJ/mol). These results suggest that within the range of test
temperatures, the apparent hydrogen diffusion of the prestrained specimens was dominated by the
trapping sites associated with P1 and P4; i.e., the effect of the cold working on the apparent
hydrogen diffusivity and activation energy can be explained by the contribution of the two trapping
sites associated with P1 and P4.

3.3 FCG properties and morphologies of crack growth and fracture surface at RT
Figure 9(a) shows the FCG rate, da/dN, as a function of ΔK in air and at various pressures of
1/2
hydrogen gas [11]. In a low ΔK regime (ΔK < 20 MPa∙m ), the FCG acceleration increased with ΔK.
1/2
Conversely, in a higher ΔK regime (ΔK > 20 MPa∙m ), the da/dN–ΔK curves in hydrogen gas were
parallel to the curves in air.
Figure 9(b) shows the relative FCG rate (RFCGR), (da/dN)H2/(da/dN)air, as a function of pH2
1/2
obtained at ΔK = 30 MPa∙m under f = 1 Hz and Rσ = 0.1, where (da/dN)H2 and (da/dN)air are the FCG
rates in hydrogen and air, respectively [11]. In addition to the experimental data in Fig. 9(a),
experimental data obtained by the ΔK-constant tests were shown in Fig. 9(b). The RFCGR was
nearly constant at pH2 = 0.7–90 MPa.
Figure 9(c) shows the relationship between the RFCGR and f [11]. Under pH2 ≤ 10 MPa, the
RFCGR gradually increased with decreasing f and then suddenly decreased close to 1.0. Similar
behavior has been reported for Cr-Mo and pipeline steels [17, 24] and austenitic stainless steels [25].
The peak of the acceleration shifted towards a lower f with an increase in pH2. At pH2 ≥ 45 MPa, the
RFCGR did not decrease in the low-frequency regime for f as low as 0.001 Hz. Notably, at pH2 = 45
MPa, the RFCGR was saturated at ~30. On the other hand, at pH2 = 90 MPa, the upper bound of the
FCG acceleration did not exist in the f region down to 0.001 Hz.
Figure 10 shows LM images of the crack morphology on the surface of the CT specimen after
1/2
the ΔK-constant test was performed at ΔK = 30 MPa∙m [11]. The test was initiated in air, after the
test atmosphere was switched to 0.7-MPa hydrogen gas. In air, extensive slip bands were observed
along the fatigue crack (Fig. 10(b)). The same proved to be the case for the crack grown in 0.7-MPa
hydrogen gas at a frequency of 0.001 Hz (Fig. 10(d)), where the crack growth rate was nearly
equivalent to that observed in air. In contrast, under the test in 0.7-MPa hydrogen gas at a frequency
of 1 Hz, where the FCG was accelerated by a factor of ~10, very few slip bands were observed along
the crack (Fig. 10(c)). This is attributed to the localization of the plastic deformation at the crack tip
under the influence of hydrogen. Conversely, under the test in 90-MPa hydrogen gas, where the FCG
was accelerated at the test frequencies of 1 and 0.001 Hz, very few slip bands were observed along
the crack in both cases.

8
 Figure 11 shows the striations observed on the fatigue fracture surface of CT specimens at ΔK
= 30 MPa∙m1/2 [11]. The figure captions explain the average striation spacing and da/dN, which were
similar. In air, typical ductile striations were observed (Fig. 11(a)). The striations were flat and
blurred in 0.7-MPa hydrogen gas at a test frequency of 1 Hz, where the FCG was remarkably
accelerated compared with that in air (Fig. 11(b)). In contrast, in 0.7-MPa hydrogen gas at a
frequency of 0.001 Hz, where the FCG rate was nearly the same as that in air, the striation
morphology was similar to that observed in air (cf. Figs. 11(a) and (c)). Under 90-MPa hydrogen gas,
the striations were flat and blurred, regardless of the test frequency.

3.4 FCG properties and morphologies of crack growth and fracture surface at elevated
temperatures
Figure 12 shows the relationship between da/dN and ΔK for the ΔP-constant tests at Rσ = 0.1
and f = 1 Hz in inner gas (air or 0.7-MPa nitrogen gas) and 0.7-MPa hydrogen gas at temperatures of
RT, 363 K, and 423 K. Regardless of the test temperatures, the FCG was accelerated in the hydrogen
gas; however, the RFCGR was lower at higher temperatures. The ΔK for the onset of the
hydrogen-assisted FCG acceleration, ΔKonset, shifted to a higher ΔK with an increase in the test
temperature: ΔKonset < 13 MPa∙m1/2 at RT, ΔKonset = 16 MPa∙m1/2 at 363 K, and ΔKonset = 19
MPa∙m1/2 at 423 K.
Figure 13 shows LM images of the crack morphologies on the surface of the CT specimen
after the ΔP-constant test. The LM observation was performed at ΔK ≈ 30 MPa∙m1/2. In 0.7-MPa
hydrogen gases at 423 K, extensive slip bands were observed along the fatigue crack (Fig. 13(c)),
where the FCG was slightly faster than that observed in air at RT. In the FCG tests in 0.7-MPa
hydrogen gas at 363 K, where the FCG was significantly accelerated, as well as that in 0.7-MPa
hydrogen gas at RT, very few slip bands were observed along the crack (Fig. 13(b)).
Figure 14 shows the striations observed on the fatigue fracture surface of CT specimens at ΔK
≈ 30 MPa∙m1/2. The striations were flat and blurred in 0.7-MPa hydrogen gas at a test temperature of
363 K, where the FCG was remarkably accelerated compared with that in air at RT. In contrast, in
0.7-MPa hydrogen gas at a test temperature of 423 K, where the FCG was slightly accelerated, the
striation morphology was similar to that observed in air and in 0.7-MPa nitrogen gas (Fig. 14(c)).

3.5 Prediction of onset of hydrogen-assisted FCG acceleration

3.5.1 Estimation of saturated hydrogen content and hydrogen diffusivity at crack tip
As shown in Fig. 8, the hydrogen diffusivity of the prestained specimens with a CW of ≥ 20 %
was not influenced by the cold working, regardless of the test temperature. Severe plastic strain was
generated at the crack tip; therefore, the hydrogen diffusivity of the present steel at the crack tip
corresponded to that of the prestained specimens with a CW of ≥ 20 %. We used Oriani’s

9
equilibrium theory [26] for estimating the hydrogen diffusivity and saturated hydrogen content
contributing to the hydrogen diffusion at the crack tip. According to Oriani’s equilibrium theory
[26], the apparent hydrogen diffusivity of the prestrained steel with a CW of 40 % in Fig. 7 were
fitted by the following equation:

DL D0 E
Dapp   exp( D ) , (8)
NX EDB NX EDB RT
1 exp( ) 1 exp( )
NL RT NL RT

where NX/NL and EDB are unknown parameters, DL is the lattice hydrogen diffusivity in BCC iron
without any traps, NL is the number of lattice sites per unit volume, NX is the number of trap sites per
unit volume, and EDB is the biding energy for the trapping site contributing to hydrogen diffusion.
–8 2
According to Kiuchi et al. [20], D0 = 7.23 × 10 m /s, and ED = 5.69 kJ/mol. The experimental data
for the prestrained steel with a CW of 40 % were fitted by Eq. (8); as a result, the following fitted
–3
parameters were obtained: NX / NL = 1.3 × 10 and EDB = 28.9 kJ/mol.
The saturated hydrogen content, CS, of the prestrained steel with a CW of ≥ 20 % was
calculated by using NX / NL and EDB, as follows:

NX E
CS  {1  exp( DB )}CLS , (9)
NL RT

3440
CLS   F exp( ) 　, (10)
T

where CLS is the saturated hydrogen content in the BCC iron with no traps, F is the fugacity, and η =
1/2
104.47 mass ppm / MPa [27]. According to San Marchi et al. [28], the fugacity, F, is expressed in
terms of the hydrogen pressure as follows:

bp
,
F  pH2 exp( H2 ) 　 (11)
RT

–5 3
where pH2 is the hydrogen-gas pressure, and b is a constant (= 1.584 × 10 m /mol). The CS and D for
the steel with severe plastic strain corresponding to the crack tip under various environmental
conditions were calculated using Eqs. (8)–(11).

3.5.2 Simplified parameter representing gradient of hydrogen concentration at crack tip

The aforementioned results indicate that the hydrogen-assisted FCG acceleration was always
accompanied by slip localization at the crack tip. Importantly, even in hydrogen gas, FCG
acceleration did not occur when the slip deformation was not localized. Similar phenomena
regarding the slip localization at the crack tip were observed in JIS-SCM435 [29]. To understand this

10
peculiar frequency dependence of the FCG rate, Matsuo et al. [29] performed FCG testing in
0.7-MPa hydrogen gas at various test frequencies, detecting the peculiar frequency dependence of
hydrogen-induced acceleration using JIS-SCM435. They explained the acceleration mechanism
according to the hydrogen-induced successive crack growth (HISCG) model [8, 24, 30], suggesting
that the acceleration is determined not by the presence or absence of hydrogen at the crack tip but
rather by the distribution of hydrogen near the tip of the fatigue crack. They suggested that a steep
gradient of hydrogen at the crack tip causes a localization of the plasticity, which prevents crack-tip
blunting and sharpens the crack tip, increasing the crack growth per cycle. In contrast, Somerday et
al. [17] performed FCG testing on the pipeline steel X52 at various test frequencies in 21-MPa
hydrogen gas containing 10, 100, and 1,000 vol. ppm oxygen, discovering that the frequency
dependence of the FCG acceleration in hydrogen changed according to the oxygen content. As
previously mentioned, the hydrogen gas in the cylinder used in this study was always less than 1 vol.
ppm, which is considerably lower than that reported by Somerday et al. [17]. Furthermore,
hydrogen-charged specimens exhibit a peculiar frequency dependence of the hydrogen-assisted FCG
acceleration [31]. Therefore, in this study, we investigated the peculiar frequency dependence of the
FCG rate according to the hydrogen distribution near the crack tip based on the HISCG model.
Figure 15 illustrates the difference in the FCG between non- and hydrogen-charged specimens
based on the HISCG model, where SZW is the stretch-zone width and s is the striation spacing [8].
The FCG of the hydrogen-charged specimen occurs with a sharp shape without crack blunting under
the loading process because of the localization of the slip deformations caused by hydrogen, which
differs from that of the non-charged specimen.
A series of experimental results suggest that a steep gradient of the hydrogen concentration
causes slip localization at the crack tip. Our previous study has proposed a simplified parameter
representing the gradient of the hydrogen concentration at the crack tip under the following
assumptions: the hydrogen diffusion can be calculated from a solution of Fick's second law for a
semi-infinite plate, the hydrogen diffusivity and solubility are not dependent on the stress state, and
the hydrogen diffusion before plastic deformations is ignored [11]. Although more exact
investigations may be needed, the simplified parameter roughly quantified the onset of the FCG
acceleration obtained by the ΔK-constant test as shown in Fig. 17(a) later; therefore, the application
of this parameter is believed to be acceptable as the first step to quantify the hydrogen-assisted the
FCG acceleration. Considering these, we expanded the simplified parameter for quantifying the
onset of the FCG acceleration even for the ΔP-constant tests in this study.
The simplified parameter quantifies the onset of the FCG acceleration due to hydrogen, in
consideration of the following two factors: (1) the hydrogen concentration on the surface, (2) the
ratio of the penetration depth of hydrogen per cycle to the ordinary plastic zone produced in air. The
hydrogen-diffusion properties of the present steel at the crack tip, having severe plastic strain, are

11
considered to be nearly constant (cf. Fig. 8). Thus, as illustrated in Fig. 16, when the initial hydrogen
content is zero, the hydrogen distribution, CH, at the crack tip based on the normalized distance from
the crack tip, x', can be approximately given as follows:

x   p x
CH ( x)  CS{1  erf ( )}　, x  , (12)
2 Dt p

where x is the distance from the crack tip, t = 1/(2f) is the loading time per cycle, and erf is the error
function. The CS and D for the prestrained steel of CW ≥ 20 % should be used to reproduce the
hydrogen-diffusion properties at the crack tip. ωp is the ordinary plastic zone for plane strain in air:

1 K max 2 1 K
p  ( )  { }2 . (13).
3  YS 3 (1  R ) YS

For quantifying the gradient of the hydrogen concentration near the crack tip in consideration
of the ratio of the penetration depth of hydrogen per cycle to the ordinary plastic zone in air, the
following parameter, G, was defined:

CSp f
G  CSp . (14).
2 Dt 2D

When the G in Eq. (14) exceeds the critical value, GC, the hydrogen-assisted FCG acceleration
begins.

f f
G  CSp  G0p  GC , G0  CS (15).
2D 2D

Under the ΔK-constant tests at Rσ = 0.1 and RT, instead of the G in Eq. (15), we can use
1/2
(pH2∙f) for quantifying the onset of the FCG acceleration, as CS is approximately proportional to
1/2
pH2 [11].
1/2
Figure 17(a) shows the relationship between the RFCGR and (pH2∙f) for the ΔK-constant tests
at Rσ = 0.1 and RT (cf. Fig. 9). The onset of the FCG acceleration was quantified by the parameter
1/2 1/2 1/2
(pH2∙f) , revealing that the FCG acceleration occurred at (pH2∙f) ≈ 0.1. From (pH2∙f) = 0.1, the
average G0 for ΔKonset at pressures ranging from 0.1 to 90 MPa at RT was 1.86 mass ppm/mm;
therefore, the G0 and ωp for the ΔKonset in the ΔK-constant tests at RT were obtained. The G0 and ωp for
the ΔKonset in the ΔP-constant tests at 363 and 423 K can also be obtained from Fig. 12. These three
relationships between G0 and ωp were fitted by G0ωp = GC, where GC is an unknown parameter.
Figure 17(b) shows the relationship between G0 and ωp for the ΔKonset in the ΔK-constant tests at
RT and the ΔP-constant tests at 363 and 423 K. As a reference, the experimental results of the
ΔP-constant test at RT, for which ΔKonset is not presented in Fig. 12, are shown here. The relationship

12
between G0 and ωp for ΔKonset was reasonably reproduced by G0ωp = GC, demonstrating that the onset
of the hydrogen-assisted FCG acceleration was satisfactorily quantified by Eq. (15) and that the
peculiar dependence of the FCG rate on the hydrogen pressure, test frequency, and test temperature
were unified by using the novel parameter representing the gradient of the hydrogen concentration
near the crack tip.

4. Conclusions
We investigated the effects of the hydrogen pressure, test frequency, and test temperature on
the FCG properties of an annealed, low-carbon steel: JIS-SM490B. To estimate the
hydrogen-diffusion properties at the crack tip, which had severe plastic strain, the hydrogen-trapping
sites and diffusivity were determined for cold-worked JIS-SM490B. Our conclusions are
summarized as follows:
(1) For the prestrained specimen with a cold-working ratio, CW, of 5%, three peaks of the
hydrogen-desorption rate (P1, P2, and P3) were observed. The activation energies of P1, P2, and
P3 were estimated to be 20, 70, and 98 kJ/mol, respectively. In contrast, for the prestrained
specimen with a CW of 40 %, a new peak of the hydrogen-desorption rate (P4) was observed in
addition to the three peaks (P1, P2, and P3). The activation energy of P4 was estimated to be 49
kJ/mol.
(2) The TDA spectra under various heating rates suggested that each trapping site was associated
with the elastic field of dislocations (P1), grain boundaries and dislocation cores (P2),
carbide-related sites (P3), and vacancies (P4).
(3) The apparent hydrogen diffusivity of the prestrained specimens with a CW of <20% was lower
with a higher CW. In contrast, the apparent hydrogen diffusivity of the prestrained specimens
with a CW of ≥20% was not influenced by the CW. This result suggests that the
hydrogen-diffusion properties of the prestrained specimens with a CW of ≥20% correspond to
those of the steel at the crack tip, which had severe plastic strain.
(4) The activation energy of the apparent hydrogen diffusivity for the prestrained specimen with a
CW of 5 % showed good agreement with that of P1, as P4 was not observed in this specimen.
On the other hand, the activation energy of the apparent hydrogen diffusivity for the prestrained
specimen with a CW of ≥ 20 % was nearly consistent with the average of the activation energies
of P1 and P4. These results suggest that the effect of the cold working on the apparent hydrogen
diffusivity and activation energy can be explained according to the contributions of the two
trapping sites associated with P1 and P4.
(5) The hydrogen-assisted FCG acceleration always accompanied localized plastic deformations
near the crack tip, and the steep gradient of the hydrogen concentration is considered to have
caused localizations of the plastic deformations at the crack tip.

13
(6) The peculiar dependence of the FCG rate on the hydrogen pressure, test frequency, and test
temperature were unified by using a simplified parameter representing the gradient of the
hydrogen concentration near the crack tip, in consideration of the ratio of the penetration depth
of hydrogen per cycle to the ordinary plastic zone in air.

Acknowledgement
This work was supported by the New Energy and Industrial Technology Development
Organization (NEDO), Fundamental Research Project on Advanced Hydrogen Science (2006 to
2012) and Hydrogen Utilization Technology (2013 to 2018).

References
[1] Nagumo M. Fundamentals of hydrogen embrittlement. Tokyo: Uchida Rokakuho; 2008.
[2] Murakami Y, Matsuoka S, Kondo Y, Nishimura S. Mechanism of hydrogen embrittlement and
guide for fatigue design. Tokyo: Yokendo; 2012.
[3] Gangloff RP, Somerday BP. Gaseous hydrogen embrittlement of materials in energy technologies.
Cambridge: Woodhead Publishing; 2012.
[4] Matsuo T, Yamabe J, Matsuoka S. Effects of hydrogen on tensile properties and fracture surface
morphologies of Type 316L stainless steel. Int J Hydrogen Energy 2014;39:3542–3551.
[5] San Marchi C, Somerday BP, Nibur KA. Development of methods for evaluating hydrogen
compatibility and suitability. Int J Hydrogen Energy 2014;39:20434–20409.
[6] Matsunaga H, Yoshikawa M, Kondo R, Yamabe J, Matsuoka S. Slow strain rate tensile and fatigue
properties of Cr–Mo and carbon steels in a 115 MPa hydrogen gas atmosphere. Int J Hydrogen Energy
2015;40:5739–5748.
[7] Yamabe J, Itoga H, Awane T, Matsuo T, Matsunaga H, Matsuoka S. Pressure cycle testing of
Cr-Mo steel pressure vessels subjected to gaseous hydrogen. J Press Vessel Technol
2016;183–011401:1–13.
[8] Matsuoka S, Yamabe J, Matsunaga H. Criteria for determining hydrogen compatibility and the
mechanisms for hydrogen-assisted, surface crack growth in austenitic stainless steels. Eng Fract Mech
2016;153:103–127.
[9] Yamabe J, Matsunaga H, Furuya Y, Hamada S, Itoga H, Yoshikawa M, Takeuchi E, Matsuoka S.
Qualification of chromium–molybdenum steel based on the safety factor multiplier method in
CHMC1-2014. Int J Hydrogen Energy 2015;40:719–728.
[10] Yamabe J, Matsumoto T, Matsuoka S, Murakami Y. A new mechanism in hydrogen-enhanced
fatigue crack growth behavior of a 1900-MPa-class high-strength steel. Int J Fract 2012;177:141–162.

14
[11] Yoshikawa M, Matsuo T, Tsutsumi N, Matsunaga H, Matsuoka S. Effects of hydrogen gas
pressure and test frequency on fatigue crack growth properties of low carbon steel in 0.1–90 MPa
hydrogen gas. Trans Jpn Soc Mech Eng A 2014;80:1–16.
[12] Choo WY, Lee JF. Thermal analysis of trapped hydrogen in pure iron. Metall Trans
1982;13A:135–140.
[13] Demarez A, Hock AG, Meunier FA. Diffusion of hydrogen in mild steel. Acta Metall
1954;2:214–223.
[14] Yamabe J, Awane T, Matsuoka S. Investigation of hydrogen transport behavior of various
low-alloy steels with high-pressure hydrogen gas. Int J Hydrogen Energy 2015;40:11075–110861.
[15] Yamabe J, Takakuwa O, Matsunaga H, Itoga H, Matsuoka S. Hydrogen diffusivity and
tensile-ductility loss of solution-treated austenitic stainless steels with external and internal hydrogen.
Int J Hydrogen Energy, accepted.
[16] ASTM, E647-08e1, 2010. Standard test method for measurement of fatigue crack growth rates.
West Conshohocken, ASTM International.
[17] Somerday BP, Sofronis P, Nibur KA, San Marchi C, Kirchheim R. Elucidating the variables
affecting accelerated fatigue crack growth of steels in hydrogen gas with low oxygen concentrations.
Acta Mater 2013;61:6153–6170.
[18] Staykov A, Yamabe J, Somerday BP. Effect of hydrogen gas impurities on the hydrogen
dissociation on iron surface. Int J Quantum Chem 2014;114:626–635.
[19] Kiuchi K, McLellan RB. The solubility and diffusivity of hydrogen in well-annealed and
deformed iron. Acta Metall 1983;31:961–984.
[20] Takai K. Hydrogen existing states in metals. Trans Jpn Soc Mech Eng A 2004;70:1027–1035.
[21] Li D, Gangloff RP, Hirth JP. Hydrogen trap sites in ultrahigh-strength AERMET 100 steel. Metall
Mater Trans A 2004;35:849–864.
[22] Novak P, Yuan R, Somerday BP, Sofronis P, Ritchie RO. A Statistical Physical-based,
micro-mechanical model of hydrogen-induced intergranular in steel. J Mech Phys Solids
2010;58:206–226.
[23] Asano S, Hara K, Nakai Y, Ohtani N. The trapping effect of dislocations on hydrogen diffusion
in mild steel. J Jpn Inst Met 1974;38:62–632.
[24] Matsuoka S, Tanaka H, Homma N, Murakami Y. Influence of hydrogen and frequency on fatigue
crack growth behavior of Cr–Mo steel. Int J Fract 2011;168:101–112.
[25] Itoga H, Matsuo T, Orita A, Matsunaga H, Matsuoka S, Hirotani R, 2014. SSRT and fatigue crack
growth properties of high-strength austenitic stainless steels in high-pressure hydrogen gas. ASME
PVP2014-28640.
[26] Oriani RA. The diffusion and trapping of hydrogen in steel. Acta Metall 1970;18:147–157.

15
[27] Hirth JP. Effects of hydrogen on the properties of iron and steel. Metall Mater Trans A
1980;11:861–890.
[28] San Marchi C, Somerday BP, Robinson SL. Permeability, solubility and diffusivity of hydrogen
isotopes in stainless steels at high gas pressures. Int J Hydrogen Energy 2007;32:100–116.
[29] Matsuo T, Matsuoka S, Murakami Y, 2010. Fatigue crack growth properties of quenched and
tempered Cr–Mo steel in 0.7 MPa hydrogen gas. Proceedings of the 18th European conference on
fracture.
[30] Murakami Y, Kanezaki T, Mine Y, Matsuoka S. Hydrogen embrittlement mechanism in fatigue of
austenitic stainless steels. Metall Mater Trans A 2008;39:1327–1339.
[31] Matsunaga H, Nakashima T, Yamada K, Matsuo T, Yamabe J, Matsuoka S, 2016. Effect of test
frequency on hydrogen-enhanced fatigue crack growth in Type 304 stainless steel and ductile cast
iron. ASME PVP2016-63536.

16
Figure captions

Fig. 1 Microstructure of an annealed, low-carbon steel

Fig. 2 Schematic of cold woking process for steel plate

Fig. 3 Relationship between residual hydrogen content and elapsed time after hydrogen exposure
of hydrogen-charged iron-based alloy [15]

Fig. 4 TDA spectra of hydrogen-charged prestained specimens with CW of 5 % and 40 % under a

heating rate of 100 K/h

Fig. 5 TDA spectra of hydrogen-charged prestained specimens with CW of 40 % under heating

rates of 50, 100, and 200 K/h and determination of activation energy by the Choo-Lee
method: (a) TDA spectra; (b) determination of activation energies by the Choo-Lee
method

Fig. 6 Relationship between residual hydrogen content and elapsed time after hydrogen exposure
for the hydrogen-charged prestained specimen with CW of 40 % at a constant temperature
of 373 K: (a) fitting by the solution of diffusion equation (Eq. (3)); (b) fitting by the
modified equation (Eq. (6)).

Fig. 7 Arrhenius plots of hydrogen diffusivity of prestrained specimens with CW of 5, 10, and
40 %, together with the literature data [19,23]

Fig. 8 Effect of the cold working ratio, CW, on the apparent hydrogen diffusivity and activation
energy: (a) relationship between apparent hydrogen diffusivity and CW; (b) relationship
between activation energy and CW

Fig.9 (a) Relationship between da/dN and ΔK in hydrogen gas at pressures of 0.1–90 MPa at RT;
(b) RFCGR vs pH2; (c) RFCGR vs f [11]

Fig. 10 Crack growth morphologies at the specimen surface after FCG test in 0.7-MPa hydrogen
gas at RT: (a) Low magnification; (b) Magnification of A; (c) Magnification of B; (d)
Magnification of C [11]

17
Fig. 11 Striations formed in air and in 0.7-MPa hydrogen gas at by SEM [11]

Fig. 12 Relationship between da/dN and ΔK in 0.7-MPa hydrogen gas at elevated temperatures

Fig. 13 Crack growth morphologies at the specimen surface after FCG test in 0.7-MPa hydrogen
or nitrogen gas at RT, 363 K, and 423 K

Fig. 14 Striations formed in air and in 0.7-MPa hydrogen or nitrogen gas at RT, 363 K, and 423 K
by SEM

Fig. 15 Schematic illustration of FCG of non-charged and hydrogen-charged specimens in air

(HISCG model): (a) non-charged specimen; (b) hydrogen-charged specimen [8]

Fig. 16 Schematic illustration of approximate gradient of hydrogen concentration at crack tip

Fig. 17 (a) Relationship between the RFCGR and (PH2∙f) 1/2 for the ΔK-constant tests at RT; (b)
Relationship between the G0 and ωp for the ΔKonset in the ΔK-constant tests at RT and the
ΔP-constant tests at RT, 363 K and 423 K

18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
 Table 1 Chemical composition [mass%]
C Si Mn P S
0.16 0.44 1.43 0.017 0.004

36
 Table 2 Conditions of fatigue crack growth tests in high-pressure hydrogen gas
Hydrogen-gas
Control mode Test frequency, f [Hz] Test temperature
pressure, pH2 [MPa]
0.1 0.001, 0.01, 0.1, 1, 5, 10
0.7 0.001, 0.01, 0.1, 1, 5
ΔK-constant
10 Room temperature
(ΔK = 30 MPa∙m1/2)
45 0.001, 0.01, 0.1, 1
90
0.1 Room temperature
0.7 RT, 363 K, 423 K
ΔP-constant 1
10
Room temperature
90

37
Highlights
 FCG tests were performed with low-carbon steel.
 The hydrogen-trapping sites and diffusivity were determined.
 Depending on the test conditions, the FCG was accelerated.
 A novel parameter representing the local hydrogen at the crack tip was introduced.
 The novel parameter quantified the FCG acceleration under various test conditions.

38