materials

Article
Quantification of Temperature Dependence of
Hydrogen Embrittlement in Pipeline Steel
Xiao Xing 1, *, Jiayu Zhou 1 , Shouxin Zhang 1 , Hao Zhang 2 , Zili Li 1 and Zhenjun Li 3
1 College of Pipeline and Civil Engineering, China University of Petroleum (East China),
Qingdao 266500, China; S17060737@s.upc.edu.cn (J.Z.); b15060157@s.upc.edu.cn (S.Z.);
cygcx@163.com (Z.L.)
2 Department of Chemical and Materials Engineering, University of Alberta,
Edmonton, AB T6G 1H9, Canada; Hao7@ualberta.ca
3 Petrochina West Pipeline Company, Wulumuqi 830001, China; xbgdlizhj@petrochina.com.cn
* Correspondence: xiaoxingupc@outlook.com; Tel.: +86-13475593096




Received: 21 January 2019; Accepted: 12 February 2019; Published: 15 February 2019


Abstract: The effects of temperature on bulk hydrogen concentration and diffusion have been tested
with the Devanathan—Stachurski method. Thus, a model based on hydrogen potential, diffusivity,
loading frequency, and hydrostatic stress distribution around crack tips was applied in order to
quantify the temperature’s effect. The theoretical model was verified experimentally and confirmed
a temperature threshold of 320 K to maximize the crack growth. The model suggests a nanoscale
embrittlement mechanism, which is generated by hydrogen atom delivery to the crack tip under
fatigue loading, and rationalized the ∆K dependence of traditional models. Hence, this work could
be applied to optimize operations that will prolong the life of the pipeline.

Keywords: hydrogen embrittlement; hydrogen diffusion; stress rupture; fatigue

1. Introduction
Hydrogen embrittlement [1,2] is the most severe degradation mechanism in buried pipeline
steel and is an issue involving the loading, environment, and physical property of the steel [3–6].
Temperature is a determining factor in hydrogen diffusivity and bulk hydrogen concentration co and,
thus, also determines the level of hydrogen embrittlement [7–9]. Hence, crack propagation in pipeline
steel has a strong temperature dependence. However, traditional models (like Paris’ law) only quantify
the relationship of the crack growth with the stress intensity range (∆K) [10], or verify the correlation
of crack growth with loading frequency f [11–14]. Additionally, no model is available that quantifies
the effect of temperature on crack growth.
When hydrogen is not present in steel, the toughness of the steel would increase with temperature,
meanwhile, the hardness would decrease. Specifically, the crack resistance of steel would increase with
temperature [15,16]. However, when hydrogen is introduced, the temperature dependence of the crack
resistance of steel remains debatable, and further studies are required in this field. Therefore, the hydrogen
diffusion model has been applied to quantify the effect of temperature. Hydrogen atoms are driven
by hydrostatic stress to accumulate around the crack tip [17,18]. The trapped hydrogen atoms near a
crack lower the free surface energy [19], thus enhancing cleavage-like failure. This postulation is the
hydrogen-enhanced decohesion (HEDE) mechanism [20–22]. The hydrogen-enhanced local plasticity
mechanism (HELP) [23–26] can be interpreted in a way that hydrogen atoms enhance the dislocation
generation and emission, thus increasing the local plasticity and leading to the subsequent failure by
exhaustion of the material strain capability. These two mechanisms provide a cleavage mechanistic
framework for predictive models: (1) hydrogen atoms will be trapped in plastic deformations and will

Materials 2019, 12, 585; doi:10.3390/ma12040585 www.mdpi.com/journal/materials

 Materials 2019, 12, 585 2 of 11

diffuse into the plastic zone and saturate plastic defects [27–29]; (2) the hydrogen atoms will aggregate
at theMaterials
crack 2019, 12, x FOR
tip after PEER REVIEW
saturating the plastic zone, diminishing the energy for cleavage [18]; 2and of 12(3) as

the atomic hydrogen concentration (an atomic ratio of number of hydrogen atoms to iron atoms) at the
45 diminishing the energy for cleavage [18]; and (3) as the atomic hydrogen concentration (an atomic
crack tip approximates 1, every bond between adjacent iron atoms would be weakened by the hydrogen
46 ratio of number of hydrogen atoms to iron atoms) at the crack tip approximates 1, every bond
atom,between
47 and theadjacent
bindings ironare easily
atoms broken
would to form free
be weakened by thesurfaces
hydrogenfully covered
atom, and thebybindings
hydrogen are atoms
easily [17].
Typically,
48 brokenthetotemperature will not
form free surfaces affect
fully the plastic
covered zone size
by hydrogen because
atoms [17].yield strength
Typically, has little temperature
the temperature will
49
dependence when compared with hydrogen diffusivity and concentration
not affect the plastic zone size because yield strength has little temperature dependence [11]. As the temperature
when
50
rises compared
from 293 withto 373 K, the yield
hydrogen strength
diffusivity only changes[11].
and concentration oneAspercent [30]. Therefore,
the temperature the293
rises from temperature
to 373
51 K, the yield
dependence strength diffusivity
of hydrogen only changes andone percent [30].
equilibrium areTherefore, the temperature
critical factors that shoulddependence
be determined of to
52 hydrogen
quantify diffusivity and
the temperature equilibrium
dependence are critical embrittlement.
of hydrogen factors that should be determined to quantify the
53 A temperature
crack growth dependence of hydrogen
model based embrittlement.
on hydrogen diffusion [17] can be described as
54 A crack growth model based on hydrogen diffusion [17] can be described as
 
d Δa= 4 DΩHH (1√
+ v ) K I  π co  ∆a−−11/4
. d∆a 4 DΩ (1 + v)K I πco
a= , (1)
a = dt =π k B T 3 2π  ao  Δa 4 (1)
dt π k BT 3 2π  ao 
. ,
where a is the crack growth rate; ∆a is the crack length change; D is the hydrogen atom diffusivity; m2 /s;
55 where 𝑎 ̇ is the crack growth rate; Δa is the crack length change; D is the hydrogen atom diffusivity;
Ω is the partial volume of hydrogen atom; v is Poisson’s ratio; co is the bulk equilibrium hydrogen atomic
56 H m2/s; ΩH is the partial volume of hydrogen atom; v is Poisson’s ratio; co is the bulk equilibrium
concentration;
57 ao is the concentration;
hydrogen atomic
original lengthaoofisthe crack; k is the Boltzmann constant; and T is temperature. This
the originalB length of the crack; kB is the Boltzmann constant;
modeland
58 establishes a quantitative
T is temperature. relationship
This model between
establishes the temperature
a quantitative and crack
relationship growth
between the rate because both
temperature
59
D and c
and are dependent on temperature T. However, this model is only applicable
o crack growth rate because both D and co are dependent on temperature T. However, this model in static loading and in a
60
brittleiscondition because
only applicable of theloading
in static plastic and
zonein[31,32] which
a brittle is generated
condition because of in the
cyclic loading,
plastic zone as the aggregation
[31,32] which
61 is generated
of hydrogen atoms in by
cyclic loading,
plastic as the aggregation
deformations is ignoredofinhydrogen
this model. atoms by plastic
However, thisdeformations
model involves is the
62 ignored
physical in thisof
properties model. However,
the materials andthis model involves
environmental the physical
parameters, properties
as well as theof the materials
loading factors,and
whereas
63
only environmental
the loading factors parameters, as well as
are considered the loading models.
in traditional factors, whereas
To improve onlythe
thetheoretical
loading factorsmodelareso that
64 considered in traditional models. To improve the theoretical model so that it can be applied in
it can be applied in industrial operations, hydrogen-diffusion related crack growth theory was applied
65 industrial operations, hydrogen-diffusion related crack growth theory was applied to a new model
to a new model where hydrogen atoms saturate the plastic zone before they diffuse to the crack tip. As
66 where hydrogen atoms saturate the plastic zone before they diffuse to the crack tip. As Figure 1
Figure
67 1 shows,
shows, the the stress
stress andand dislocation
dislocation density
density are small
are small outside
outside the plastic
the plastic zonezone
whenwhen compared
compared with with
inside,
68 and the hydrogen concentration outside the plastic zone is approximated
inside, and the hydrogen concentration outside the plastic zone is approximated to be the bulk to be the bulk equilibrium
69
concentration
equilibrium co . An illustrated cannulus
concentration region outside
o. An illustrated annulus theregion
plasticoutside
zone will theoffer hydrogen
plastic zone will atoms during
offer
70
loading and deplete
hydrogen partialloading
atoms during hydrogen atoms during
and deplete unloading.
partial hydrogen If the
atoms annulus
during region
unloading. is large
If the enough
annulus
71 region isof
(the resource large enoughatoms
hydrogen (the resource
is enough)of hydrogen atoms istime
and the loading enough) and
is long the loading
enough, time growth
the crack is long rate
72
is maximized as hydrogen atoms have enough time to saturate the plastic zone during loading.the
enough, the crack growth rate is maximized as hydrogen atoms have enough time to saturate As the
73 plasticconcentration
hydrogen zone during loading. As the unity
approximates hydrogen concentration
at the free surfacesunity
crack tip, theapproximates at the crack tip, the
are generated.
74 free surfaces are generated.

75

76 Figure 1. Schematic
Figure 1. Schematicshowing
showingthethe
evolution
evolutionofofa pre-existing
a pre-existingmicrocrack;
microcrack; the yellowcircular
the yellow circularzone
zoneisis the
77 plastic
thezone where
plastic zonethewhere
stressthe
is concentrated and plastic
stress is concentrated anddeformations (dislocations)
plastic deformations are aggregated.
(dislocations) are
78 The hydrogen
aggregated.atoms need to distribute
The hydrogen atoms needaccording toaccording
to distribute the stresstointensity
the stressand saturate
intensity and the dislocations
saturate the
79 core (T symbols),core
dislocations then(Tdiffuse to then
symbols), the crack tip.
diffuse If the
to the atomic
crack tip. Ifhydrogen
the atomicconcentration ahead ofahead
hydrogen concentration a crack tip
80 of a crack 1,
approximates tipfree
approximates 1, free
surfaces are surfaces(red).
generated are generated
The stress (red). The stress
intensity andintensity and density
dislocation dislocation
outside
the plastic zone is low, hence, the hydrogen concentration outside the plastic zone approximates the
bulk hydrogen concentration, co .
Materials 2019, 12, 585 3 of 11

2. Temperature Dependence of Critical Loading Frequency

Critical loading frequency was identified back in the 1880s and is a frequency in the loading
spectra below which the crack growth rate reaches the maximum and remains constant [33]. This
threshold is only established as hydrogen atoms exist in the specimen. It is natural to associate this
threshold with hydrogen diffusion and aggregation. The movement of hydrogen atoms dependent on
axial stress Fr , and Fr near the crack tip, is related to the hydrostatic stress σ hyd , Fr = −Ω H ∇r σ hyd . Ω H
is the partial volume of the hydrogen atom and the average velocity of the hydrogen atom is defined
as [28]
R r p +req R Kmax DFr
rp Kmin k B T dK I dr
Vr = , (2)
(Kmax − Kmin )req
where rp is the size of the plastic zone and req is the size of an annulus region around the plastic zone
to supply sufficient hydrogen atoms into the plastic zone. The saturation number of hydrogen atoms
N(KI ) is related to the stress, the size of plastic zone, and the bulk equilibrium hydrogen concentration,
co . It is also equal to the number of hydrogen atoms that the virtual hydrogen supply area, encircled
by the dotted line in Figure 1, can offer:

2πco lz 
2 2

N (Kmax ) − N (Kmin ) = r p + r eq − r p , (3)
a3o

where ! !
σ hyd Ω H
Z rp  
N (K I ) = co exp 2πrlz / a3o /2 dr, (4)
0 kB T

where lz is the thickness of specimen, and the time that requires hydrogen atoms to saturate the plastic
r
zone as the stress intensity rises from Kmin to Kmax is tc = Veq . The threshold for the loading frequency
r
is fc = 1/(2tc ). This threshold is sensitive to temperature because some of the parameters have a strong
temperature dependence. For example, the diffusivity of hydrogen atoms is in an exponential function
with T and is expressed as D = Do × 10−8 × EXP(− k ET ). The diffusion coefficients in iron have a large
B
scatter because hydrogen atoms can be trapped by impurities. Typical Do values can range from 3.35
× 10−8 to 2.2 × 10−7 m2 /s, and E is the diffusion barrier which ranges from 0.035 to 0.142 eV. The
specimen was assumed to have a bcc structure, hence, the theoretical D expression, which is deduced
from first principles in the bcc lattice, is shown as [34]

D = 1.5 × 10−7 exp(−0.088 eV/k B T ) m 2 s −1 . (5)

The predicted threshold of loading frequency (f critical ) corresponds to a scenario where the plastic
zone is saturated with hydrogen atoms, and it is also expected that less time is required to saturate
the plastic zone if the movement of hydrogen atoms (V r ) is enhanced. The inset of Figure 2 shows
that the predicted velocity of the hydrogen atoms increases with temperature, and the prediction was
verified in Figure 4 in the next section. Basically, more hydrogen atoms can aggregate in the plastic
zone during the same time interval and under the same stress intensity as T is increased. In other
words, the time required to saturate the plastic zone is smaller and the temperature is higher under the
same loading condition. In Figure 2, the filled symbols were tested for critical loading frequency in
HY-130 steel and the open symbols are the predicted values. The results showed rough agreement
and verified the temperature enhancement to critical loading frequency. The predicted values and
empirical values did not converge as the stress intensity was large (∆K = 30.8 MPa * m0.5 ), because the
space between the adjacent iron atoms was enlarged as the stress intensity increased. Thus, hydrogen
diffusion was enhanced because of the extra volume induced by a larger stress intensity [35]. This kind
of enhancement is negligible in this calculation, however, predictions under common stress intensities
(∆K = 14 or 20.9 MPa * m0.5 ) showed good agreement with the empirical values.
 115 and verified the temperature enhancement to critical loading frequency. The predicted values and
116 empirical values did not converge as the stress intensity was large (ΔK = 30.8 MPa * m0.5), because the
117 space between the adjacent iron atoms was enlarged as the stress intensity increased. Thus, hydrogen
118 diffusion was enhanced because of the extra volume induced by a larger stress intensity [35]. This
119 kind of
Materials enhancement
2019, 12, 585 is negligible in this calculation, however, predictions under common4 of stress
11
120 intensities (ΔK = 14 or 20.9 MPa * m ) showed good agreement with the empirical values.
0.5

121
Figure 2. The predicted and tested critical loading frequency versus temperature, where the filled
122 Figureare
symbols 2. The predicted
the tested and
values in tested critical
the HY-130 loading
steel, and thefrequency versusare
open symbols temperature, where
the predicted theBoth
values. filled
123 symbols are the tested values in the HY-130 steel, and the open symbols are the predicted
sets of data showed the same tendency, which verified the predictive capability of the model. The values. Both
124 sets of data showed the same tendency, which verified the predictive capability of
inset predicted the hydrogen movement velocity at different temperatures and showed that hydrogen the model. The
125 inset predicted the hydrogen movement
movement was enhanced by temperature [33]. velocity at different temperatures and showed that hydrogen
126 movement was enhanced by temperature [33].
3. Temperature Dependence of Hydrogen Concentration
127 3. Temperature Dependence of Hydrogen Concentration
The hydrogen-assisted failure mechanism can be facilitated either by increasing the hydrogen
128 concentration or increasing the
The hydrogen-assisted hydrogen
failure movement.
mechanism can beThe temperature
facilitated dependence
either by increasingofthehydrogen
hydrogen
129 concentration
concentration and ordiffusion canthe
increasing be hydrogen
tested withmovement.
Devanathan–Stachurski
The temperature double electrolysis
dependence ofcells. An
hydrogen
130 X80 steel specimen (the hydrogen density and diffusivity are not related to the strength
concentration and diffusion can be tested with Devanathan–Stachurski double electrolysis cells. An of the steel,
131 soX80
the steel
X80 test results(the
specimen can hydrogen
be applieddensity
to otherand
steels) with a thickness
diffusivity of 1200
are not related the and
toµm diameter
strength of theofsteel,
35
132 mmso is
theshown
X80 testin Figure
results 3a.
canThe sample to
be applied was sealed
other by polytetrafluoroethylene
steels) with a thickness of 1200 with
µm and740diameter
silica gel of
on35
133 one
mm surface and was
is shown electroplated
in Figure 3a. The with
samplenickel
wason the other
sealed surface to prevent the recombination
by polytetrafluoroethylene with 740 silicaofgel
theon
134 hydrogen
one surfaceatoms.andAfter the electroplating
was electroplated with was completed,
nickel the polytetrafluoroethylene
on the other was removedof
surface to prevent the recombination
Materials 2019, 12, x FOR PEER REVIEW 5 of 12
135 from
the the specimen
hydrogen and the
atoms. specimen
After was cleaned with
the electroplating wasdeionized
completed, water.
the polytetrafluoroethylene was
136 removed from the specimen and the specimen was cleaned with deionized water.
137 Afterwards, the dried steel specimen was located between two electrolysis cells as shown in
138 Figure 3b. The corrosion test on the naked side was performed in a near neutral pH solution which
139 was employed to charge the hydrogen atoms under a polarized electrical potential of 300 mV.
140 Initially, the hydrogen atoms were generated on the naked side and moved to the other side of the
141 specimen afterwards. A PARSTAT2273 was applied to test the current density of the specimen on the
142 test side, which was covered with nickel. The tested current density, as shown in Figure 4, is related
143 to the hydrogen concentration. As the current approaches a constant value, the surface hydrogen
144 concentration approximates the bulk equilibrium hydrogen concentration.

(a) (b)
145 3. (a) The X80 specimen. (b)
Figure 3. (b) The
The equipment
equipment designed
designed for
for this
this study, which consisted of a
146 the PARSTAT2273,
constant current, the PARSTAT2273,and
anddouble
doubleelectrolysis
electrolysiscells.
cells.

147 Afterwards, the effect

The facilitating dried of
steel specimen was
temperature located between
on hydrogen movement twowas
electrolysis
confirmedcellsbyasthe
shown in
results
148 Figure
shown 3b. The corrosion
in Figure test on
4. The sharp risethe
of naked side was
the current performed
density could beindetected
a near neutral pH solution
and recorded which
in a shorter
149 time at high temperature. To be specific, the length of the initial horizontal line was negatively
150 proportional to the temperature, which indicates that less time was spent by the first hydrogen atom
151 travelling to the testing side. The positive correlation of hydrogen concentration with temperature
152 was verified using Figure 4. As osmotic current densities of hydrogen atoms were tested at different
153 temperatures, there was a terrace value for each temperature where the current density stayed
 Materials 2019, 12, 585 5 of 11

was employed to charge the hydrogen atoms under a polarized electrical potential of 300 mV. Initially,
the hydrogen atoms were generated on the naked side and moved to the other side of the specimen
afterwards. A PARSTAT2273 was applied to test the current density of the specimen on the test
side, which2019,
Materials was12,covered with
x FOR PEER nickel. The tested current density, as shown in Figure 4, is related
REVIEW 6 of 12
to the hydrogen concentration. As the current approaches a constant value, the surface hydrogen
concentration approximates the bulk equilibrium hydrogen concentration.

161
Figure 4. The current density curves of X80 steel. The stable current density increases with the
162 Figure 4.which
temperature, The current density
means more curvesatoms
hydrogen of X80aresteel. The stable
generated currentofdensity
at the surface increases
the specimen. with the
Namely,
163 temperature,
the supply whichatoms
of hydrogen means more hydrogen
increases atomsasare
in the specimen thegenerated
temperature at increases.
the surfaceAs of
thethe specimen.
hydrogen
164 density and diffusivity are not related to the strength of the steel, the X80 test results can be applied As
Namely, the supply of hydrogen atoms increases in the specimen as the temperature increases. to the
165 hydrogen
other steels. density and diffusivity are not related to the strength of the steel, the X80 test results can
166 be applied to other steels.
The facilitating effect of temperature on hydrogen movement was confirmed by the results shown
in Figure 4. The sharp rise of the current density could be detected and recorded in a shorter time at
 4 (1 + v ) K I Ω H 
high temperature. To be specific, the length ceq =ofcothe
exp  horizontal line was negatively proportional to(7)
initial
the temperature, which indicates that less time wasspent 3π kby
BT the2πfirst
r hydrogen atom travelling to the
testing side. The positive correlation of hydrogen concentration with temperature was verified using
167 Figure 4.Consequently, the experiments
As osmotic current densities ofsuggest
hydrogenthatatoms
increasing temperature
were tested will facilitate
at different the movement
temperatures, there
168 wasofa terrace
hydrogen atoms and bulk hydrogen concentration, however, Equation (7)
value for each temperature where the current density stayed constant, and this current suggests that as
169 density
temperature increases
corresponded the bulk
to the co, it still diminishes
hydrogen the exp(1/T)
concentration in set.
the Hence, the relationship
steel. The specific valueofofcurrent
ceq should
170 density
be calculated as determining the crack
i with bulk hydrogen concentration is
∞
growth rate.

171 4. Temperature Dependence of Crack Growthi∞ Rate

lz
co = , (6)
FD
172 To quantify the temperature dependence of crack growth, a model based on hydrogen-enhanced
173 where
decohesion (HEDE)constant
F is the Faraday was developed. Here,
and D is the we employed
diffusivity the criteriaatoms.
of the hydrogen that when the atomic
The current ratio
density i∞of a
174 increased
hydrogen atom ahead
remarkably withoftemperature,
a crack reached
hence1, cthe free surfaces
o increased withformed, and the
temperature andcrack propagated.
a specific co valueThe
175 could
corresponding
be calculatedfree surface
with Equationlength
(6) at different temperatures.
stress intensities can be shown
However, as
the equilibrium hydrogen
concentration near a crack has an exponential relation with T, as shown2 in Equation (7), which suggests
that increasing temperature may diminish the 4 (1 + vconcentration
 hydrogen ) Ω H  Knear a crack.
L ( KI ) =   I
2
(8)
 3πk4B(T1 +2vπ)Kln Ω( 1 c

o ) 
 ,
ceq = co exp √I H (7)
3πk B T 2πr
and the HEDE model can be indicated as  L ( K max ) − L ( K min )  ( f / f critical )
0.1
176

 da   1+ R  ΔK 2
  = A   (9)
 1 − R  ( f f critical )
0.1
 dN  HEDE
,
Materials 2019, 12, 585 6 of 11

Consequently, the experiments suggest that increasing temperature will facilitate the movement of
hydrogen atoms and bulk hydrogen concentration, however, Equation (7) suggests that as temperature
increases the co , it still diminishes the exp(1/T) set. Hence, the specific value of ceq should be calculated
as determining the crack growth rate.

4. Temperature Dependence of Crack Growth Rate

To quantify the temperature dependence of crack growth, a model based on hydrogen-enhanced
decohesion (HEDE) was developed. Here, we employed the criteria that when the atomic ratio of a
hydrogen atom ahead of a crack reached 1, the free surfaces formed, and the crack propagated. The
corresponding free surface length at different stress intensities can be shown as
" #2
4(1 + v ) Ω H
L(K I ) = √ K2I , (8)
3πk B T 2π ln(1/co )

and the HEDE model can be indicated as [ L(Kmax ) − L(Kmin )]/( f / f critical )0.1

∆K2
   
da 1+R
=A , (9)
dN HEDE 1−R ( f / f critical )0.1

where A is an environmental factor related to temperature and pH, and can be illustrated as A =
2
4(1+ v ) Ω H
  
− R ∆K is also a component of energy density in the Griffith
[19,27]. The set 11+
√ R 2
3πk B T 2π ln(1/co )
model where all strain energy is converged to free surface energy [36,37]. The energy density is
shown as
1 − 2v 1 + R
 
∆s = ∆K2 . (10)
4πG ( T ) 1 − R
Hence, this model verified that the ∆K dependence of crack growth under the near neutral pH
stress corrosion cracking (NNpHSCC) condition is a brittle mechanism and can be applied to
rationalize the traditional crack growth models. However, the hydrogen bubble theory [38–40],
which consumes hydrogen atoms, was neglected in this derivation. Since both mechanisms are
related to hydrogen diffusion, the total hydrogen-assisted cracking rate is naturally expected to have
a power law relationship with the crack growth rate based on HEDE. The total crack growth rate
is (da/dN ) T = (da/dN )nHEDE . As the loading frequency f is smaller than f critical , the plastic zone is
saturated with hydrogen atoms and the crack growth rate is not dependent on the loading frequency
and reaches the maximum. HY130 steel is a high strength steel and the n value approaches 1. In
Figure 5, the experiments were carried out in C-2 (0.0035 KCl, 0.0195 NaHCO3 , 0.0255 CaCl, 0.0274
MgSO4 ·7H2 O, 0.0606 CaCO3 g/L) solution. The specimen was gritted to 9 mm thickness and a notch
in the compact tension (CT) specimens was orientated in a direction perpendicular to the mode I
loading direction. The dimensions of the specimen should be 50 × 50 × 9 mm3 . The specimen was
polished to 600 grit finish and then pre-cracked in air by fatigue to generate a sharp crack from the
notch. Afterwards, the specimen was sealed in the test cells and pinhole-loaded with computer load
control. The crack growth rate was recorded at an R ratio of 0.25, a loading frequency of 0.1 or 4.9,
√
and the ∆K value was constant at 14 MPa * m, and the temperature ranged from 280 to 350 K [31].
The temperature threshold to maximize the crack growth under NNpHSCC is indicated in Figure 5,
where f is equal to 0.1. Elevating temperature will facilitate hydrogen movement and increase the
surface hydrogen concentration, however, the equilibrium hydrogen concentration near a crack has an
exponential relation with T. More specifically, elevating temperature will enhance hydrogen generation
and diffusion but restrict hydrogen accumulation near defects. Consequently, the threshold was
generated in the vicinity of 320 K, as shown by the black symbols in Figure 5. If there is insufficient
diffusion time for the hydrogen atoms to accumulate (loading frequency is large), the restriction effects
of hydrogen accumulation caused by the increase in temperature are not significant when compared
 Materials 2019, 12, 585 7 of 11
Materials 2019, 12, x FOR PEER REVIEW 8 of 12

with the enhancements of hydrogen diffusion and generation. Hence, as the loading frequency f was
bigger than the f critical , the predicted threshold was not detected, as shown by the red symbols.

209 Figure 5. This schematic illustrates the temperature dependence of crack growth in HY130 steel [33].
210 As f <Figure
f critical5.
, the
This crack growthillustrates
schematic rate was maximized at 320dependence
the temperature K. of crack growth in HY130 steel [33].
2115. Crack Growth
As f < fcritical , the crack growth rate was maximized
Rate Prediction in Different Steels at 320 K.

212 InCrack
5. FigureGrowth
6, the crack
Rategrowth rate in
Prediction in X65 was tested
Different
√ Steels
at different R ratios and loading frequencies,
and the ∆K value ranged from 10 to 25 MPa * m [41]. The yield strength σys of X65 steel was 448
213MPa, andIn Figure 6, the crack growth rate in X65 was tested at different R ratios and loading frequencies,
was much smaller than that of the HY130 steel. Typically, steel with a smaller yield strength
214has aand the ΔK value ranged from 10 to 25 MPa * √𝑚 [41]. The yield strength σys of X65 steel was 448
larger plastic zone and generates more plastic deformations before failure as the plastic zone size
215 MPa, and was much smaller than that of the HY130 steel. Typically, steel with
1
 a smaller
Kmax
2 yield strength
216is negatively related to yield strength under the plain strain condition, r p =
has a larger plastic zone and generates more plastic deformations before failure6π σ . The size of
ys as the plastic zone
the plastic zone is smaller for steel that possesses a higher yield strength. The plastic zone is the area 2
1  K 
217concentrated with defects caused by plastic deformations, which would trap significant amount
size is negatively related to yield strength under the plain strain condition, rp =  max of. The
hydrogen atoms. The theoretical model was based on HEDE theory; therefore, the trapping 6π  effects
σ ys of
plastic deformation-caused defects had been ignored. Hence, for a high strength steel in which the
218 size of the plastic zone is smaller for steel that possesses a higher yield strength. The plastic zone is
219plastic zone is small, the model is more accurate and the fixing parameter n approximates unity. Hence,
the area concentrated with defects caused by plastic deformations, which would trap significant
the n value of X65 steel was smaller than that of HY130. In Figure 6, all predicted crack growth data fit
220 amount of hydrogen atoms. The theoretical model was based on HEDE theory; therefore, the
221very trapping
well with the n value of 0.92.
effects of plastic deformation-caused defects had been ignored. Hence, for a high strength
222 steel in which the plastic zone is small, the model is more accurate and the fixing parameter n
223 approximates unity. Hence, the n value of X65 steel was smaller than that of HY130. In Figure 6, all
224 predicted crack growth data fit very well with the n value of 0.92.
 Materials 2019, 12, x FOR PEER REVIEW 9 of 12
Materials 2019, 12, 585
Materials 2019, 12, x FOR PEER REVIEW 98of
of12
11

225Figure 6. The schematic shows the comparison of the predicted crack growth rates (open symbols)
225and experimental
226 Figure 6. values
The schematic
[41]. Theshows the comparison
R ratio and loading of the predicted
frequency crack
f are growth rates
illustrated (open
in the symbols)
chart. At a
227
226smaller R ratio,
and experimental
Figurethe 6. The values
schematic
predictive [41]. The
shows the
accuracy R ratio
wascomparisonand
better as the loading
of the frequency
predicted
lattice parameter f are
crackof illustrated
growth in the
rates (open
the material chart. At is
(thissymbols)
effect a
228
227usually negligible)
smaller R ratio,
and experimental the predictive
and thevalues
accuracy
[41]. The
diffusivity of H
was
R atoms
ratio andbetter as the lattice
loading frequency
is increased
parameter
with stress
of
f areintensity.the material
illustratedPredicted (this
in the chart. effect
At a
values
229
228are smallerissmaller
usually R negligible)
ratio, the and the diffusivity
predictive accuracy of
wasH atoms
better is increased
as the latticewith stress intensity.
parameter of the Predicted
material (thisvalues
effect
than the experimental values at a large KI .
230
229 are smallernegligible)
is usually than the experimental valuesofatHaatoms
and the diffusivity large KisI.increased with stress intensity. Predicted values
230 arerule
smaller
canthanbe the experimental values at a large KI. which possesses a smaller yield strength
231The sameThe applied to another steel, X52,
same rule can be applied to another steel, X52, which possesses a smaller yield strength
when
232 compared with X65. InX65.
FigureFigure
7, the7,Rthe ratio was wasconstant, the loading frequency ranged from
231 whenThe compared
same rule with can be In applied to anotherR ratiosteel, X52,constant,
whichthe loading
possesses √afrequency
smaller yield ranged from
strength
233
0.001 to 0.005,
232 0.001
whento
and the
0.005, and
compared
stress intensity
theX65.
with stress was
In intensity
distributed
Figure 7, was the Rdistributed
from 15 to
from 15 to
ratio was constant,
19the
MPa
19loading* m [42].
𝑚 [42]. All
MPa * √frequency All predictive
predictive
ranged from
234
values
233 were located
values
0.001 towere in the
andrange
located
0.005, in the
the of error
range
stress of bar
errorof
intensity theof
bar
was empirical
the empirical
distributed values
from 15 and
values 19confirmed
to and * √𝑚 that
confirmed
MPa thatall
[42]. Allpredictions
all predictions
predictive
235
234 matched
matched well with
values were the
well experiments
with
located the rangewith
experiments
in the an
with
of error n an
bar value
of nthe of 0.88.
value A
of 0.88.
empirical triangulation
A triangulation
values and confirmed rationalized
rationalized that the
that
that all predictions the
236
235 crack
brittle brittle
growth
matched crack growth
model
well model
with based based
on on hydrogen
hydrogen
the experiments an nembrittlement
withembrittlementvalue of 0.88. isAapplicable
is applicable inin
triangulation high
high strength steel.
strength
rationalized steel.
thatThe
The
the
237
236 hypothesis
hypothesis of the
brittle crack of
model,the model,
growth modelwhere
where abased
sharp aonsharp
rise of
hydrogenrise of
hydrogenhydrogen
embrittlement concentration
concentration
is applicable causes
causes
in high the
the formation
formation
strength steel. of
of aa
The
238
237 hypothesis
nanoscale nanoscale
hydrogen-rich hydrogen-rich
of theregion region
model,inwhere in
whicha thewhich
sharp the hydrogen
rise of hydrogen
hydrogen concentration
concentration concentration approached
approached causesten ten thousand
thethousand times
formationtimes of a
239 of
of 238
the bulkhydrogen-rich
nanoscale equilibrium hydrogen
region inconcentration,
which thewas was verified.
hydrogen concentration approached ten thousand times
the bulk equilibrium hydrogen concentration, verified.
239 of the bulk equilibrium hydrogen concentration, was verified.

240
240
Figure 7. The schematic shows the comparison of the predicted crack growth rates (open symbols) and
experimental values in X52 steel [42]. The R ratio and loading frequency f are recorded in the chart.
The predictive values fit well with the experimental data at an n value of 0.88.
Materials 2019, 12, 585 9 of 11

6. Conclusions
In the presented study, we tested the temperature dependence of hydrogen generation and
diffusion in high strength steel. Based on a mechanism of hydrogen-assisted brittle failure, a theoretical
model was developed to predict crack growth rate in different loading spectra and at different
temperatures. The model yielded rough agreement with the empirical values in different steels
with fixed n values and suggests that the n value in the brittle model is positively correlated with the
degree of brittleness of the steel, and is approximately 1 in high strength steel. This conclusion verified
the brittle failure mechanism in pipeline steel and suggests that a nanoscale hydrogen-rich region may
exist ahead of a crack tip in the NNpHSCC condition.
The model also served to rationalize the empirical models and provide a framework of ∆K
dependence of crack growth. As the loading frequency was smaller than the f citical , the temperature
threshold to maximize crack growth was predicted in the model and verified by experiment. These
results confirmed the predictive capability of the model and suggest that this model can be applied to
optimize steel operations.

Author Contributions: X.X.: General ideas and wrote the paper. J.Z.: Experimental tests. S.Z.: Experimental tests.
H.Z.: Valuable suggestions and word editing. Z.L.(Zili Li): Valuable suggestions and word editing. Z.L.(Zhenjun
Li): Experimental tests and data gathering.
Funding: China Postdoctoral Science Foundation (2017M622316), the Qingdao Postdoctoral Applicable
Foundation (BY20170214), and the Innovative Fellowships of the China University of Petroleum (East China)
(18CX02175A).
Acknowledgments: The authors would like to acknowledge the financial support from the China Postdoctoral
Science Foundation (2017M622316), the Qingdao Postdoctoral Applicable Foundation (BY20170214), and the
Innovative Fellowships of the China University of Petroleum (East China) (18CX02175A).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Robertson, I.M.; Sofronis, P.; Nagao, A.; Martin, M.L.; Wang, S.; Gross, D.W.; Nygren, K.E. Hydrogen
Embrittlement Understood. Metall. Mater. Trans. A 2015, 46, 2323–2341. [CrossRef]
2. Wang, S.; Martin, M.L.; Sofronis, P.; Ohnuki, S.; Hashimoto, N.; Robertson, I.M. Hydrogen-induced
intergranular failure of iron. Acta Mater. 2014, 69, 275–282. [CrossRef]
3. Chen, W.; Sutherby, R.L. Crack Growth Behavior of Pipeline Steel in Near-Neutral pH Soil Environments.
Metall. Mater. Trans. A 2007, 38, 1260–1268. [CrossRef]
4. Dmytrakh, I.M.; Smiyan, O.D.; Syrotyuk, A.M.; Bilyy, O.L. Relationship between fatigue crack growth
behaviour and local hydrogen concentration near crack tip in pipeline steel. Int. J. Fatigue 2013, 50, 26–32.
[CrossRef]
5. Meisnar, M.; Vilalta-clemente, A.; Moody, M.; Arioka, K.; Lozano-perez, S. A mechanistic study of the
temperature dependence of the stress corrosion crack growth rate in SUS316 stainless steels exposed to PWR
primary water. Acta Mater. 2016, 114, 15–24. [CrossRef]
6. Zhao, J.; Chen, W.; Yu, M.; Chevil, K.; Eadie, R.; Van Boven, G.; Kania, R.; Been, J. Crack Growth Modeling
and Life Prediction of Pipeline Steels Exposed to Near-Neutral pH Environments: Dissolution Crack Growth
and Occurrence of Crack Dormancy in Stage I. Metall. Mater. Trans. A 2017, 48, 1629–1640. [CrossRef]
7. Syrotyuk, A.M.; Dmytrakh, I.M. Methods for the Evaluation of Fracture and Strength of Pipeline Steels and
Structures Under the Action of Working Media. Part 2. Influence of Hydrogen-Containing Media. Mater. Sci.
2015, 50, 475–487. [CrossRef]
8. Kanezaki, T.; Narazaki, C.; Mine, Y.; Matsuoka, S.; Murakami, Y. Effects of hydrogen on fatigue crack growth
behavior of austenitic stainless steels. Int. J. Hydrog. Energy 2008, 33, 2604–2619. [CrossRef]
9. Beck, W.; Bockris, J.O.; McBreen, J.; Nanis, L. Hydrogen Permeation in Metals as a Function of Stress,
Temperature and Dissolved Hydrogen Concentration. Proc. R. Soc. A Math. Phys. Eng. Sci. 1966, 290,
220–235. [CrossRef]
Materials 2019, 12, 585 10 of 11

10. Yu, M.; Chen, W.; Kania, R.; Van Boven, G.; Been, J. Crack propagation of pipeline steel exposed to
a near-neutral pH environment under variable pressure fluctuations. Int. J. Fatigue 2016, 82, 658–666.
[CrossRef]
11. Yu, M.; Xing, X.; Zhang, H.; Zhao, J.; Eadie, R.; Chen, W.; Been, J. Corrosion fatigue crack growth behavior
of pipeline steel under underload-type variable amplitude loading schemes. Acta Mater. 2015, 96, 159–169.
[CrossRef]
12. Skorupa, M. Load Interaction Effects During Fatigue Crack Growth Under Variable Amplitude
Loading—A Literature Review. Part I: Empirical Trends. Fatigue Fract. Eng. Mater. Struct. 1998, 21,
987–1006. [CrossRef]
13. Skorupa, M. Load interaction effects during fatigue crack growth under variable amplitude
loading—A literature review. Part II: Qualitative interpretation. Fatigue Fract. Eng. Mater. Struct. 1999, 22,
905–926. [CrossRef]
14. Egbewande, A.; Chen, W.; Eadie, R.; Kania, R.; Van Boven, G.; Worthingham, R.; Been, J. Transgranular crack
growth in the pipeline steels exposed to near-neutral pH soil aqueous solutions: Discontinuous crack growth
mechanism. Corros. Sci. 2014, 83, 343–354. [CrossRef]
15. Díaz-Fuentes, M.; Iza-Mendia, A.; Gutiérrez, I. Analysis of different acicular ferrite microstructures in
low-carbon steels by electron backscattered diffraction. Study of their toughness behavior. Metall. Mater.
Trans. A 2003, 34, 2505–2516. [CrossRef]
16. Tanguy, B.; Besson, J.; Piques, R.; Pineau, A. Ductile to brittle transition of an A508 steel characterized by
Charpy impact test Part I: Experimental results. Eng. Fract. Mech. 2005, 72, 49–72. [CrossRef]
17. Song, J.; Curtin, W.A. Atomic mechanism and prediction of hydrogen embrittlement in iron. Nat. Mater.
2013, 12, 145–151. [CrossRef]
18. Song, J.; Curtin, W.A. A nanoscale mechanism of hydrogen embrittlement in metals. Acta Mater. 2011, 221,
1557–1569. [CrossRef]
19. Xing, X.; Yu, M.; Chen, W.; Zhang, H. Atomistic simulation of hydrogen-assisted ductile-to-brittle transition
in α-iron. Comput. Mater. Sci. 2017, 127, 211–221. [CrossRef]
20. Shih, D.S.; Robertson, I.M.; Birnbaum, H.K. Hydrogen embrittlement of α titanium: In situ tem studies.
Acta Metall. 1988, 36, 111–124. [CrossRef]
21. Oriani, R.A.; Josephic, P.H. Equilibrium aspects of hydrogen-induced cracking of steels. Acta Metall. 1974, 22,
1065–1074. [CrossRef]
22. Daw, M.S.; Baskes, M.I. Semiempirical, Quantum Mechanical Calculation of Hydrogen Embrittlement in
Metals. Phys. Rev. Lett. 1983, 50, 285–288. [CrossRef]
23. Zhou, X.; Song, J. Effects of alloying elements on vacancies and vacancy-hydrogen clusters at coherent twin
boundaries in nickel alloys. Acta Mater. 2018, 148, 9–17. [CrossRef]
24. Robertson, I.M.; Birnbaum, H.K. Hydrogen Effects on Plasticity. Dislocations Solids 2009, 15. [CrossRef]
25. Song, J.; Curtin, W.A. Mechanisms of hydrogen-enhanced localized plasticity: An atomistic study using α-Fe
as a model system. Acta Mater. 2014, 68, 61–69. [CrossRef]
26. Birnbaum, H.K.; Sofronis, P. Hydrogen-enhanced localized plasticity—A mechanism for hydrogen-related
fracture. Mater. Sci. Eng. A 1994, 176, 191–202. [CrossRef]
27. Xing, X.; Chen, W.; Zhang, H. Prediction of crack propagation under cyclic loading based on hydrogen
diffusion. Mater. Lett. 2015, 152, 86–89. [CrossRef]
28. Xing, X.; Chen, W.; Zhang, H. Atomistic study of hydrogen embrittlement during cyclic loading: Quantitative
model of hydrogen accumulation effects. Int. J. Hydrog. Energy 2017, 42, 4571–4578. [CrossRef]
29. Espinosa-Medina, M.A.; Carbajal-De la Torre, G.; Sánchez Castillo, A.; Ángeles-Chávez, C.;
Zeferino-Rodríguez, T.; González-Rodríguez, J.G. Effect of chloride and sulfate ions on the SCC of API-X70
pipeline welds in diluted carbonated solutions. Int. J. Electrochem. Sci. 2017, 12, 6952–6965. [CrossRef]
30. Li, W.; Zhang, X.; Kou, H.; Wang, R.; Fang, D. Theoretical prediction of temperature dependent yield strength
for metallic materials. Int. J. Mech. Sci. 2016, 105, 273–278. [CrossRef]
31. Lynch, S. Some fractographic contributions to understanding fatigue crack growth. Int. J. Fatigue 2017, 104,
12–26. [CrossRef]
32. Moriconi, C.; Hénaff, G.; Halm, D. Cohesive zone modeling of fatigue crack propagation assisted by gaseous
hydrogen in metals. Int. J. Fatigue 2014, 68, 56–66. [CrossRef]
Materials 2019, 12, 585 11 of 11

33. Thomas, J.P.; Wei, R.P. Corrosion fatigue crack growth of steels in aqueous solutions I: Experimental results
and modeling the effects of frequency and temperature. Mater. Sci. Eng. A 1992, 159, 205–221. [CrossRef]
34. Jiang, D.E.; Carter, E.A. Diffusion of interstitial hydrogen into and through bcc Fe from first principles.
Phys. Rev. B 2004, 70, 1–9. [CrossRef]
35. Liu, X.; Xie, W.; Chen, W.; Zhang, H. Effects of grain boundary and boundary inclination on hydrogen
diffusion in α-iron. J. Mater. Res. 2011, 26, 2735–2743. [CrossRef]
36. Mousavi Nezhad, M.; Fisher, Q.J.; Gironacci, E.; Rezania, M. Experimental Study and Numerical Modeling
of Fracture Propagation in Shale Rocks During Brazilian Disk Test. Rock Mech. Rock Eng. 2018, 51, 1755–1775.
[CrossRef]
37. Pathirana, M.; Lam, N.; Perera, S.; Zhang, L.; Ruan, D.; Gad, E. Risks of failure of annealed glass panels
subject to point contact actions. Int. J. Solids Struct. 2017, 129, 177–194. [CrossRef]
38. Hornlund, E.; Fossen, J.K.T.; Hauger, S.; Haugen, C.; Havn, T.; Hemmingsen, T. Hydrogen diffusivities
and concentrations in 520 M carbon steel under cathodic protection in 0.5 M NaCl and the effect of added
sulphite, dithionite, thiosulphate, and sulphide. Int. J. Electrochem. Sci. 2007, 2, 82–92.
39. Simonetti, S.; Lanz, C.; Brizuela, G. Hydrogen embrittlement of a Fe–Cr–Ni alloy: Analysis of the physical
and chemical processes in the early stage of stress corrosion cracking initiation. Solid State Sci. 2013, 15,
137–141. [CrossRef]
40. Hayward, E.; Fu, C.-C. Interplay between hydrogen and vacancies in α-Fe. Phys. Rev. B 2013, 87, 174103.
[CrossRef]
41. Been, J.; Eadie, R.; Sutherby, R. Prediction of Environmentally Assisted Cracking on Gas and Liquid Pipelines;
IPC2006; IPC: Calgary, AB, Canada, 2006.
42. Kania, R.; Worthingham, R. Frequency Dependence of Fatigue and Corrosion Fatigue Crack Growth Rate; IPC2010;
IPC: Calgary, AB, Canada, 2010.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).