Hydrogen diffusion and trapping in X70 pipeline steel

international journal of hydrogen energy xxx (xxxx) xxx

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

Hydrogen diffusion and trapping in X70 pipeline steel Alen Thomas*, Jerzy A. Szpunar Department of Mechanical Engineering, College of Engineering, University of Saskatchewan, 57 Campus Drive, Saskatoon, SK S7N 5A9, Canada

highlights
Dual-polarized hydrogen permeation testing in different layers of pipeline steel.
The rise in hydrogen diffusion by reduced trapping for larger grains.
Improvement in hydrogen microprint technique to identify reversible hydrogen traps.
Visualization of hydrogen diffusion path in degenerate pearlite microstructure.
Comparison of preferential hydrogen diffusion in steel microstructure.

article info

abstract

Article history:

In this investigation, the hydrogen permeation experiment was used to determine the

Received 24 August 2019

parameters for diffusion, and the hydrogen microprint technique was used to visualize the

Received in revised form

diffusion path in X70 pipeline steels. The samples in mid-layer at the segregation zone and

19 October 2019

top-layer of the steel were used in this study. The diffusion parameters from the hydrogen

Accepted 14 November 2019

permeation experiment enabled us to determine that the calculated density of total,

Available online xxx

reversible, and irreversible hydrogen trapping sites in the top layer of the steel decreases for larger grains. However, the irreversible and total trapping sites of the mid-layer showed

Keywords:

an initial growth and subsequent decay with grain growth due to the inclusions in mid-

Hydrogen permeation

layer. The observations from the hydrogen microprint experiment allowed us to

Hydrogen microprint technique

conclude that the preferential hydrogen diffusion increases in the order of grains, grain

Hydrogen traps

boundaries, triple junctions and cementites with cementites being the easiest path for

Diffusion path

hydrogen diffusion.

Grain boundary

© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Triple junctions

Introduction The global oil and natural gas pipeline projects had tripled since 1996, regardless of the growth with renewable energy consumption [1]. By 2040, the world population will exceed 9 billion, and hence, the total energy demand will increase by

another 27% [2]. The large population, rising gross domestic product and improved standard of living require more energy from all sources, including oil and gas. According to the International Energy Agency, oil and natural gas will acquire 48% shares of the global energy demand by 2040 in the sustainable development scenario [3]. The 50% contribution of oil and natural gas for global energy consumption in 2017 and the

* Corresponding author. E-mail address: [email protected] (A. Thomas). https://doi.org/10.1016/j.ijhydene.2019.11.096 0360-3199/© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

2

international journal of hydrogen energy xxx (xxxx) xxx

projected global energy demand by 2040 shows that pipeline networks that transport oil and gas will grow all over the world. The failures to pipeline networks by corrosion, cracking, excavation damages, natural force damages, operation errors and manufacturing mistakes can cause serious economic loss [4e6]. The cracking in pipelines occurs as a premature sign of failure. The drop in material ductility that leads to crack propagation without substantial plastic deformation is caused by hydrogen embrittlement (HE) [7]. Hydrogen is produced by surface corrosion at pipeline steel in the presence of hydrogen sulfide from the transport of oil and gas. Hydrogen atom diffuses into pipeline steel, and its rate increases with the presence of sulfide ions that act as a recombination poison. HE in steels occurs as a result of the accumulation of hydrogen at micro-voids, vacancies, dislocations, inclusions, precipitates, triple junctions, grain boundaries, microstructural phases and segregation zones during manufacturing and in-service [8]. The HE that is caused by the residual stress and fatigue loads in the presence of hydrogen come under sulfide stress cracking, whereas the HE that occurs even without any stresses in the presence of hydrogen comes under hydrogeninduced cracking [9]. Fig. 1 shows the classification of failure for the North American pipelines for short periods between 1985 and 2018 [4e6,10e12]. The internal corrosion that mostly occurs by cracking contributes to the total corrosion failures. The failures from corrosion and cracking vary from 25 to 67% of the reported crashes. The mobile and immobile hydrogen are the two forms of hydrogen in high strength materials. The mobile hydrogen is temporarily stored in a reversible trap, whereas the immobile hydrogen is more permanently kept in an irreversible trap. Findley et al. reported that the binding energies of each trap with the hydrogen are the basis of trap classification [13]. The

Corrosion 100 90

8

8

Cracking

Excavation

19

70

7 4

28

8 23 37

27

8

7

6

8

16

35

30

47

17 5 3 6

15 25

20

25

17

20 10

12

Others

5

40 30

Manufacturing

27

60 50

Geotechnical 7

15 7

80

authors also reported that 60 kJ per mole act as critical binding energy for separating a trap as irreversible or reversible. The reported differences in HE susceptibility depend on the type of traps. The distributed irreversible traps in materials show less susceptibility to HE due to the low accumulation of free hydrogen at the potential crack initiation sites [14]. However, the cracking susceptibility increases with the increase of reversible trapping sites [15]. The previous studies considered the grain boundaries as reversible traps [13]. Ichimura et al. reported a grain boundary cross effect that influences the diffusivity of hydrogen [16]. According to the same authors, hydrogen mobility increases with the increased grain boundary surface area per unit volume and decreases with the increased number of triple junction traps for smaller grains. The previous studies also reported that the highest diffusion rate is for a grain size of 46 mm, and the lowest diffusion rate is for the smallest grain [17]. The increase in grain boundary will lead to a reduction in the HE susceptibility as the quantity of hydrogen in the unit grain boundary area is decreased [18]. A recent study by Masoumi et al. contradicted the previous studies and claimed that fine grains with higher grain boundary densities are more susceptible to the HE [19]. Therefore, the literature conflictingly reported that fine grains and coarse grains showed resistance to HE in different studies. The grain refinement in materials is considered as a strengthening mechanism [20]. Also, the market for high strength pipeline steel is increased. Hence, the average grain size of the steel is expected to be smaller. Therefore, the small grain size was considered in this study on the effect of grain size on hydrogen diffusion and trapping in pipeline steel. The electrochemical hydrogen permeation (HP) testing is used to measure the diffusion parameters in materials for the prediction and understanding of HE [21]. The diffusion

31 19

18

0

1985-1995 Canada

1991-2001 Canada

1987-2006 USA

2010-2014 Canda

2013-2017 Canda

2014-2018 USA

Fig. 1 e Classification of failure for North American pipeline (in %) [4e6,10e12]. Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

3

international journal of hydrogen energy xxx (xxxx) xxx

parameters include the hydrogen permeability, effective hydrogen diffusion coefficient and apparent hydrogen solubility. The measured diffusion parameters will allow us to calculate the density of different types of hydrogen trapping sites. Previous studies reported that hydrogen diffusion in the material is affected by microstructural properties like phases, grain boundaries, grain shape, vacancies, dislocations, voids, precipitates and inclusions [22]. A couple of studies in the past visualized the preferential hydrogen diffusion through grain boundaries of the pipeline steel using the hydrogen microprint technique (HMT) experiments [23,24]. Bonab et al. reported a uniform distribution of hydrogen discharge through the grains in different grades of pipeline steel and observed an increased hydrogen concentration at grain boundaries [23]. However, there is no comparison for the preferential hydrogen diffusion through the grains, grain boundaries, triple junctions and phases of the pipeline steel microstructure. In this study, top-layer and mid-layer of as-received steel were annealed to obtain samples with different grain sizes. The two as-received samples from either layer and the eight annealed samples were initially characterized. Later, the HP experiments were performed to determine the diffusion parameters in those ten samples with five different grain sizes. Finally, the HMT experiments were used to visualize the hydrogen diffusion path in three samples.

Materials and methods Sample preparation

Table 2 e Details of the samples for HP and HMT experiments. Sample Layer of the Annealing Number of Experiment name steel plate cycle samples name T1 T2 T3 T4 T5 M1 M2 M3 M4 M5

Top Top Top Top Top Mid Mid Mid Mid Mid

As-received A1 A2 A3 A4 As-received A1 A2 A3 A4

1 1 1 1 2 2 1 1 1 2

HP HP HP HP HP and HMT HP and HMT HP HP HP HP and HMT

Characterization techniques There were microstructural characterizations of samples before the HP and HMT experiments. Also, there were microstructural and chemical characterizations after the HMT experiment. We used an optical microscope (OM) from Nikon Eclipse, MA 100 and a scanning electron microscope (SEM), Hitachi SU6600, equipped with energy-dispersive X-ray spectroscopy (EDS) detectors for microstructural characterizations. The grain sizes and phases in samples were characterized by OM and SEM. The EDS measurements were used to analyze the elemental composition of samples and reduced emulsion coating after HMT experiments.

Coating

In this study, API 5L X70 pipeline steel, supplied by Evraz, Regina, was used. The chemical composition of the supplied steel is provided in Table 1. The steel samples in this study were taken from mid-layer at center segregation zone and top-layer. Also, samples from the rolling and normal direction planes were used. The asreceived samples from either layer were annealed, using Thermo Scientific, F48015 e 60 model, muffle furnace, to obtain different grain sizes. Samples from either layer were heated for 5 min at different temperatures of 875  C, 1000  C, 1050  C and 1150  C. The different temperatures correspond to four annealing cycles, A1 to A4. After reheating, the samples were cooled to room temperature inside the oven. Hence, we obtained samples with five different grain sizes for each layer of steel. Ten samples were used in the HP experiment, whereas duplicates of three samples were used in the HMT experiment. The sample details were provided in Table 2. Samples were mounted for safe handling during polishing and etching. Polished samples were etched with a 4% nital solution for 10 s to reveal the microstructure. The mountings were broken after characterization, and samples were repolished on either side before the HP and HMT experiments. The final polished samples had a dimension of 20  20  1 mm.

The International Organization for Standardization (ISO) had recommended the use of palladium coating, on either side of the polished samples, before HP experiments to get repeatable results and to avoid oxide formation [25]. ISO states that electrochemical methods of coating the sample with a thin layer of palladium can introduce hydrogen into the sample and affect experimental results. Hence, samples were sputtered with a thin palladium layer using a Quorum, Q150T model, turbomolecular-pumped coating system and Ted Pella, 91119 palladium target. The thicker palladium coating can also reduce the quantity of diffused hydrogen. Hence, a 20 nm thickness of palladium was sputtered in the steel. It is thinner than the 100 nm coating thickness formed during the electrochemical coating [26].

Hydrogen permeation experiment The electrochemical HP experiment was developed by Devanathan and Stachurski in 1962 for the instantaneous recording of hydrogen permeated through a palladium membrane [27]. A modified test setup was used in this study to perform dual-polarized HP experiments. The typical HP setup

Table 1 e Chemical composition of the X70 pipeline steels (wt %). Pipeline Steels X70

C

Mn

Si

Mo

Ti

Cr

Cu

Ni

V

S

P

N

0.025

1.65

0.26

0.175

0.015

0.07

0.21

0.08

0.001

0.0025

0.010

0.008

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

4

international journal of hydrogen energy xxx (xxxx) xxx

has two electrochemical cells that use the same sample as a working electrode. The role of one electrochemical cell is to produce the hydrogen on one side of the sample, and another electrochemical cell is responsible for the oxidation of the hydrogen that diffuses through the sample. This process can be expressed by equations (1) and (2) [28]. Hþ þ e ¼ H

(1)

H ¼ Hþ þ e 

(2)

The ISO 17081 standard [25] is the basis for an HP cell assembly, and Fig. 2 represents the schematic diagram of the same. The production of hydrogen happens at the charging cell glass beaker on the left side, and diffused hydrogen is oxidized at the oxidation cell glass beaker on the right side. The charging cell electrolyte is a 250 ml mixture of 0.1 M sulfuric acid (H2SO4) and 3 g per liter of ammonium thiocyanate, that act as a hydrogen recombination poison. The oxidation cell electrolyte is 250 ml of 0.1 M sodium hydroxide (NaOH). The Teflon lids of each cell contain fixed electrode positions to maintain the uniformity of cell geometry in all experiments. The graphite rod of 5 mm diameter is used as a counter electrode and the saturated calomel electrode with 3 M Potassium chloride act as the reference electrode. Inert argon gas is supplied into both cells, through the plastic connectors, to deaerate the system from oxygen during entire testing. The palladium-coated sample is placed between the flanges of each cell by assembling with a nylon ring. Also, rubber O eRings are kept between the sample and flange to avoid the electrolyte leakage and to maintain a circular area exposing of 1 square centimeter between the sample and electrolyte. The sample that acts as the working electrode and counter electrode form the first electrochemical cell where the hydrogen is produced. The three-electrode system that consists of counter electrode, reference electrode and sample produces the second electrochemical cell where diffused hydrogen is oxidized. When the large current passes through an electrode, its potential will be changed. This potential change happens to the counter electrode of an oxidation cell when it is formed out of a two-electrode system. Hence, the sample and reference

Fig. 3 e Schematic diagram of HP test setup.

electrode maintains oxidation potential, and oxidized current flows through the counter electrode. The HP cell assembly is connected to the power sources to work as the permeation system, as shown in Fig. 3. The Instek direct current power supply is used to provide a 5 mA constant cathodic current to the charging cell. The Gamry Potentiostat, G750, is used to supply a constant anodic potential of 250 mV to the oxidation cell. Also, the Gamry software records the oxidation current for every 2 s to measure the amount of diffused hydrogen. The experiment begins by activating the oxidation cell when the electrolyte is poured into it. The residual hydrogen of the system starts to oxidize and continues until the Gamry records a stabilized oxidation current close to 0 mA. It usually takes 5 h to reach a low oxidation current, and it varies with the sample properties. Later, the charging cell is activated by pouring the electrolyte, and produced hydrogen is diffused into the oxidation cell. The oxidation current increases and reaches a stable value in the Gamry. The cathodic current is then

Fig. 2 e Schematic diagram of the HP cell assembly.

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

international journal of hydrogen energy xxx (xxxx) xxx

5

Fig. 4 e Typical HP curve after dual-polarization. disconnected, and the oxidized current begins to decrease until the discharge of hydrogen finishes in the sample. Hence, the oxidation current will be recorded close to 0 mA again. At the same time, the charging electrolyte is removed to avoid the sample corrosion. The cathodic charging and discharging causes growth and decay of the measured oxidized current to form a typical hydrogen permeation curve after first polarization. The experiment is continued to measure the second permeation curve as reversible and irreversible trapping sites are measured in this study. The typical hydrogen permeation curve after dual-polarization is shown in Fig. 4. The derived parameters from the permeation curve are steady-state current (I∞) and time lag (TL). Steadystate current is the stable oxidation current observed in the permeation curve for the supplied cathodic current in the provided testing conditions. The time lag is the time passed until the oxidized current becomes 0.63 times of the steadystate current. The permeability, effective diffusivity, apparent solubility and density of total hydrogen trapping sites can be calculated using the values of steady-state current, time lag and dimensions of the sample. The second permeation curve is smaller than the first one as most of the irreversible hydrogen trapping sites will be saturated by hydrogen and are not active in the second polarization. All the calculated parameters from the first polarization are also measured in the second polarization. The difference in the density of trapping sites between the two polarizations is measured as the density of irreversible trapping sites [18]. The effective diffusion coefficient is the quantity of hydrogen that will be diffusing through a unit cross-sectional area of the steel sample in unit time due to the difference in concentration between two surfaces. The steel sample act as a porous media, and hence, the effective diffusion coefficient is considered rather than the bulk diffusion coefficient. The thickness of the sample is L in centimeters, and TL is the time lag in seconds measured from the hydrogen permeation curve. Then, equation (3) provides the effective hydrogen diffusion coefficient in square centimeters per second by the time lag method [18]. Deff ¼

L2 6TL

mole [29], A is the exposed area of the sample in square centimeters and I∞ is the steady-state current in microamperes from the hydrogen permeation curve, then the hydrogen permeability, J∞ L in moles per centimeter per second is calculated using equation (4). J∞ L ¼

I∞ L FA

(4)

Apparent solubility is the volume of hydrogen that will be dissolving in a unit volume of the steel sample to form a saturated sample in the provided experimental conditions of constant pressure and temperature. Also, the permeability is the product of the effective diffusion coefficient and apparent solubility. When the surface hydrogen is in thermodynamic equilibrium with the subsurface hydrogen, the apparent hydrogen solubility, Capp in moles per cubic centimeters is calculated using equation (5). Capp ¼

J∞ L Deff

(5)

Some amount of hydrogen remains in the trapping sites of steel during diffusion. If the lattice diffusion coefficient, DL, of the trap free bcc iron, is taken as 1.28  104 square centimeters per second. The number of electrons transferred in each mole is n, and it is taken as 6.02  1023 electrons per mole. Then, the total number of hydrogen trapping sites, Nt, in a cubic centimeter of the sample, is calculated using equation (6). Nt ¼

  n Capp DL 1 3 Deff

(6)

The uncertainty, of the steady-state current and time lag, by the supplied charging current were reported with results

(3)

Permeability is the rate of hydrogen flux passing through the steel sample due to the difference in pressure between two surfaces. If Faraday’s constant, F is 96,500 Coulombs/

Fig. 5 e Schematic diagram of the HMT cell assembly.

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

6

international journal of hydrogen energy xxx (xxxx) xxx

[30]. The method of uncertainty propagation is used for the uncertainty calculation of other parameters [31].

Hydrogen microprint technique The HMT experiment is used to visualize the hydrogen diffusion path, and it is performed in standard illumination [24,28,32]. In the HMT experiment, the thin square sample is polished on both sides and etched at one side to expose the microstructure. Then, the sample will be coated with a silver bromide emulsion on the etched side. The required emulsion is obtained by mixing 5 g of Silver bromide (AgBr) powder in 10 ml of 1.4 M sodium nitrite solution. A monogranular AgBr emulsion layer is deposited on the etched face of the sample by utilizing a 5 cm diameter wire loop that is made from a 1 mm thin metal wire. The necessary time for the deposition of the emulsion layer on the sample surface is about 15 min. Later, the emulsion coated sample will be electrochemically charged with hydrogen from the uncoated side using a dedicated test setup. The diffused hydrogen will reduce the silver ions in the silver bromide crystals to metallic silver according to equations (7) and (8). Agþ þ H ¼ Ag þ Hþ 2Hþ ¼ H2

1 and H2 þ Agþ ¼ Hþ þ Ag 2

(7)

(8)

After a predetermined time of charging and discharging of hydrogen through the sample, the remaining unreacted emulsion coating is washed away using a fixing solution. The fixing solution is prepared by mixing 0.6 M sodium thiosulfate solution in 1.4 M sodium nitrite solution. Initially, the sample will be dipped in the fixing solution for about 1 min. Later, the fixed sample is cleaned by dipping in distilled water for 30 s and dehydrated by a dryer. The metallic silver that remains on the microstructure after fixing is visible as superimposed white spherical particles on the microstructure when examined under an SEM. The silver particles represent the hydrogen escaping sites from the specimen. Fig. 5 displays the schematic diagram of the HMT cell assembly for hydrogen charging, and it is made according to the test setup by Luu et al. [24]. The reaction between diffused hydrogen and emulsion also happens in the test setup. The charging cell of the test setup is a glass beaker with a small flange at the bottom. The electrolyte for charging cell is 250 ml mixture of 0.1 M sulfuric acid (H2SO4) and 3 g per liter of ammonium thiocyanate that act as a recombination poison for hydrogen gas formation. The Teflon lid of the cell has a fixed counter electrode position to maintain the uniformity of cell geometry in all experiments. The graphite rod of 5 mm diameter acts as the counter electrode. The charging cell is deaerated with inert argon gas supply through the plastic connector of the lid. The sample will be placed at the end of the beaker flange with the emulsion coated face away from it. Two nylon rings keep them firmly, and a rubber O-Ring is held between the sample and flange to avoid any electrolyte leakage and to provide a circular area exposing of 1 square centimeter between the sample and the electrolyte. The graphite counter

Fig. 6 e Schematic diagram of the HMT test setup.

electrode, electrolyte and emulsion coated sample form the electrochemical cell that produces the hydrogen for diffusion. The HMT cell assembly is connected to the Instek direct current power supply to deliver 5 mA constant current, as shown in Fig. 6. After suitable electrical connections, the electrochemical cell is activated by pouring the electrolyte. The diffused hydrogen flux through the steel reacts with emulsion for 1 h, and the electrochemical hydrogen charging is terminated. The sample remains in the test setup for the same time used for charging, and the hydrogen immediately discharges from the reversible traps. Further, the sample is removed from the cell assembly for fixing, and later, the hydrogen diffusion path is visualized.

Results and discussion Microstructural evaluation Fig. 7 shows OM and SEM images of sample T1 from the toplayer and sample M1 from the mid-layer of as-received steel. The microstructural observation from OM and SEM showed that both layers looks identical and composed of acicular ferrite (AF) and some polygonal ferrite (PF), quasi polygonal ferrite (QF) and Bainite (B). The microstructural phases were marked in the figure and agreeable with the literature [33,34]. The lineal intercept technique was used to estimate the average grain size. The measured average grain size of the top-layer, T1, was 2.30 ± 0.07 mm, and that of the mid-layer, M1, was 2.66 ± 0.10 mm. This marginal grain size increase in the mid-layer is due to an increased amount of bainite grains in mid-layer. Previous studies claimed that there is elemental segregation of carbon around the center segregation zone [23]. Bohemen et al. reported that bainite transformation is prominent from an austenitic matrix of steel with increased carbon content [35]. Thus, the increase of bainite grains in mid-layer than top-layer can be related to favorable elemental segregation for bainite transformation in the mid-layer. The measured average grain size from either layer of the steel was

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

7

international journal of hydrogen energy xxx (xxxx) xxx

Fig. 7 e Microstructure of the as-received steels (a) OM of top-layer, T1 (b) OM of mid-layer, M1 (c) SEM of top-layer, T1 and (d) SEM of mid-layer, M1.

Table 3 e Permeation parameters for annealed X70 steel in the first polarization. Average grain Steady size, Sample current, I∞ (mm) ±2.73% (mA) T1 (2.5) T2 (3.5) T3 (8.5) T4 (10) T5 (12) M1 (2.5) M2 (3.5) M3 (8.5) M4 (10) M5 (12)

70.43 71.73 88.30 117.95 144.63 70.38 77.15 85.70 100.80 147.17

Time lag, TL ±12.29% (s) 3580 1304 290 206 150 2378 14,328 21,900 8832 484

Permeability,J∞ L x Diffusion 1011 ±2.73% (mol. coefficient,Deff x 108 cm1.s1) ±12.29% (cm2.s1) 7.30 7.43 9.15 12.22 14.99 7.29 7.99 8.88 10.45 15.25

Apparent solubility, Capp x 106 ±12.59% (mol. cm3)

Trapping sites, Nt x 1018 ±17.59% (cm3)

156.78 58.16 15.92 15.11 13.49 104.06 687.30 1167.07 553.55 44.29

8621.94 1157.53 67.99 44.95 28.49 3794.21 151,680.53 393,797.73 75,260.00 321.57

Apparent solubility, Capp x 106 ±12.59% (mol. cm3)

Trapping sites, Nt x 1018 ±17.59% (cm3)

103.25 53.09 7.37 4.84 4.88 16.60 11.79 5.57 4.96 4.36

6672.10 1046.82 16.69 6.04 3.54 100.58 44.91 8.68 6.65 2.49

46.55 127.81 574.71 809.06 1111.11 70.09 11.63 7.61 18.87 344.35

Table 4 e Permeation parameters for annealed X70 steel in the second polarization. Average grain Steady size, Sample current, I∞ (mm) ±2.73% (mA) T1 (2.5) T2 (3.5) T3 (8.5) T4 (10) T5 (12) M1 (2.5) M2 (3.5) M3 (8.5) M4 (10) M5 (12)

39.46 66.09 74.05 82.80 130.93 65.76 72.96 78.65 79.72 140.28

Time lag, TL ±12.29% (s) 4208 1292 160 94 60 406 260 114 100 50

Permeability,J∞ L x Diffusion 1011 ±2.73% (mol. coefficient,Deff x 108 cm1.s1) ±12.29% (cm2.s1) 4.09 6.85 7.67 8.58 13.57 6.81 7.56 8.15 8.26 14.54

39.61 129.00 1041.67 1773.05 2777.78 410.51 641.03 1461.99 1666.67 3333.33

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

8

international journal of hydrogen energy xxx (xxxx) xxx

Fig. 8 e Effect of grain growth on diffusion after dual-polarization (a) permeability (b) effective diffusion coefficient (c) apparent solubility and (d) trapping sites.

2.5 mm for the as-received steel, whereas the annealed steels from annealing cycles A1 to A4 were measured as 3.5, 8.5, 10 and 12 mm. The average grain size in each of the ten samples was based on four measurements from two microstructural images with different magnifications. The average grain size of the steel in as-received condition and from each annealing cycle was based on eight measurements from four microstructural images. The average grain size of 2.5 mm in asreceived steel was obtained from 592 grains, whereas the average grain size of 12 mm in the A4 annealing cycle was obtained from 118 grains.

junctions and the grain boundary surface area per unit volume for larger grains. The triple junctions and grain boundaries can act as trapping sites for hydrogen and reduce hydrogen mobility in steel for small grain sizes. Hence, the

Role of grain size in hydrogen trapping by the permeation experiment Table 3 and Table 4 provides the calculated permeation parameters of ten samples from the first and second polarizations. Fig. 8 shows the effect of grain growth on diffusion parameters after the dual-polarized HP experiment. The hydrogen permeability is increased with grain growth for both polarizations in either layer of the steel. The increased permeability can be related to the decreased number of triple

Fig. 9 e Irreversible traps in top-layer and mid-layer steels with increased grain sizes.

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

international journal of hydrogen energy xxx (xxxx) xxx

9

Fig. 10 e Diffusion around inclusions after the HMT experiment (a) SEM image of M1, (b) SEM image of M5, (c) EDS maps of M1 and (d) EDS maps of M5.

trapping sites were decreased with an increase in grain size for both layers of steel. The tabulated values showed that the measured steady-state current for second polarization is less than the first polarization for both layers. Hence, permeability also follows same trend in all samples. The lower permeability in the second polarization is due to the saturation of irreversible traps by hydrogen after the first polarization. This trap saturation reduces the porosity of the steel for the second polarization. So, the real distance traveled by the hydrogen atoms through the steel in second polarization is increased to cause a reduction in the permeability. The permeability in the mid-layer is slightly higher than the top layer for most grains in either polarization. This observation can be related to the marginal increase in the mid-layer grains that reduces the trapping by triple junctions and grain boundaries. The trend for the measured time lag in inversely reflected when the effective diffusion coefficient is plotted against the grain size, as shown in Fig. 8. The reduced trapping by triple junctions and grain boundaries for larger grains increases the hydrogen mobility. Hence, the effective diffusion coefficient in either layer increases with grain growth for both polarizations. Contradictorily, there is an initial decay and subsequent growth in the effective diffusion coefficient of the mid-layer in

first polarization. It may be due to the irreversible trapping sites of the samples from the center segregation zone. The initial decay and subsequent growth of measured irreversible trapping sites of the mid-layer verify the same. The effective diffusion coefficient in the second polarization is higher than the first polarization for either layer due to the saturation of irreversible trapping sites. Therefore, the average cross-sectional area for hydrogen diffusion in the second polarization increases to raise the effective diffusion coefficient closer to the bulk diffusion coefficient. The effective diffusion coefficient in the mid-layer is marginally higher than the top-layer for second polarization, and it relates to the marginal increase in the mid-layer grains that reduces the trapping by triple junctions and grain boundaries. The measured permeability and effective diffusion coefficient show a marginal increase in mid-layer for the second polarization. It allows us to conclude that the mid-layer got a slight dominance over the top layer for hydrogen diffusion. The plot of apparent solubility against grain size is also shown in Fig. 8. The apparent solubility is decreased in larger grains as the trapping by triple junctions and grain boundaries were reduced with the increase in grain size. The explanation agrees with the literature that trapping at nodes and triple junctions reduce the movement of hydrogen and enhance the hydrogen solubility [17,36]. Conflictingly, there is an initial

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

10

international journal of hydrogen energy xxx (xxxx) xxx

Fig. 11 e SEM images of M1 after HMT (a) partial diffusion and (b) complete diffusion.

Fig. 12 e Partial diffusion in M1 (a) SEM image, (b) EDS map and (c) EDS spectrum.

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

international journal of hydrogen energy xxx (xxxx) xxx

11

Fig. 13 e Complete diffusion in M1 (a) SEM image, (b) EDS map and (c) EDS spectrum. growth and subsequent decay of apparent solubility in the mid-layer for first polarization, and this occurs by the trapping of irreversible sites. The irreversible trapping sites were saturated after first polarization, and the trapping of hydrogen by the active reversible traps during the second polarization is the reason for lower apparent solubility for the second polarization in either layer. The apparent solubility in the midlayer is slightly lower than the top layer for second polarization and related to the marginal increase in the mid-layer grains that reduces the trapping. Fig. 8 also shows the effect of grain growth on the density of trapping sites. The measured trapping sites for the first and second polarization is the total traps and reversible traps in either layer of the steel. The total traps include reversible and irreversible traps. The density of trapping sites decreases with grain growth in either layer and can be related to the decrease in

the number of triple junctions and grain boundaries for larger grains. The density of trapping sites in the second polarization is lower than that for the first polarization in either layer, and it confirms that the total trapping sites participate in the first polarization and reversible trapping sites alone contribute to the second polarization. The density of reversible trapping sites for the top-layer is slightly more than that of the mid-layer, and it can be related to the marginal increase in mid-layer grains that reduces the trapping by the lower number of triple junctions and grain boundaries. This proposition of considering the triple junctions and grain boundaries as reversible trapping sites can be validated from the HMT experiments. Remarkably, the density of total trapping sites of the midlayer shows initial growth and subsequent decay with grain growth. However, the density of reversible trapping sites of mid-layer is decreased for larger grains. These observations

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

12

international journal of hydrogen energy xxx (xxxx) xxx

Fig. 14 e Diffusion at the triple junctions of M1 (a) SEM image and (b) EDS map and (c) EDS spectrum. allowed us to conclude the formation of irreversible trapping sites in the mid-layer. Precipitates are considered a significant source of irreversible traps [37], and Simm et al. reported precipitations after annealing [38]. Hence, annealing can also induce irreversible traps for the mid-layer samples. This proposition of additional traps in the mid-layer of an asreceived and annealed steel samples can be validated with HMT experiments. Fig. 9 presents the density of irreversible trapping sites in either layer of steel. The density of irreversible traps in top-layer decreases with grain growth, whereas mid-layer shows initial growth and subsequent decay for the density of irreversible traps.

Visualizing the hydrogen diffusion path by microprint experiment In the HMT experiments, we use the duplicates of an asreceived and annealed samples from the mid-layer and an annealed sample from the top layer of the X70 pipeline steel.

The samples undergo a simultaneous EDS analysis to validate the SEM observations. There was the presence of many non-metallic inclusions of Ti, Mn, Al, or a combination of the same in mid-layer samples, M1 and M5. However, the EDS analysis of one inclusion is only reported for each sample. Fig. 10 shows the EDS analysis of midlayer samples after different HMT experiments. The as-received steel, sample M1, contains titanium-based inclusion, whereas the annealed steel, sample M5, contains manganese-based inclusion. The circular pattern of silver particles around inclusions represents the hydrogen escaping sites. The immediate discharging of reversible hydrogen, after cathodic charging in the HMT experiment, allowed us to consider that the circular pattern is formed by the diffusion of trapped hydrogen by the inclusions. The circular pattern can also be related to the cracked matrix around the inclusion. However, matrix cracking around inclusion is not identified. In any case, the as-received and annealed samples in the mid-layer contain inclusions that can affect the permeation results.

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

international journal of hydrogen energy xxx (xxxx) xxx

13

Fig. 15 e Preferential diffusion in sample T5 after HMT (a) SEM image of partial diffusion (b) SEM image of complete diffusion (c) SEM image of partial diffusion in DP and (d) EDS map of partial diffusion in DP (e) SEM image of complete diffusion in DP (f) EDS map of complete diffusion in DP.

Fig. 11 shows the SEM images from two different locations of sample M1 from the same HMT experiment. There is a partial diffusion of hydrogen through one region of the sample, as shown in the first micrograph. Another area of the same sample shows a complete diffusion in the second micrograph. The observation of white spherical particles overlaid on the microstructure illustrated the hydrogen escape. The higher number of white particles in the second micrograph compared to the first micrograph showed partial and complete diffusion.

The etching of the sample before HMT allowed the microstructural observation from the same SEM images. The first micrograph clearly showed that the grain boundaries are contributing to the escape of the hydrogen atom from the sample. The magnified SEM imaging and EDS analysis from the rectangular area of the micrographs in Fig. 11 are presented as partial diffusion in Fig. 12 and complete diffusion in Fig. 13. The particle size of several hundred nanometers to almost one micron represents the amount of diffused hydrogen. The

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

14

international journal of hydrogen energy xxx (xxxx) xxx

black and intense green spots in the iron (Fe) and silver (Ag) maps from the EDS analysis confirmed the white particles as metallic silver. The complete diffusion micrograph visualized the white particles over the grain boundaries and white particle clusters at the triple junctions. The comparison of the EDS spectrum for the partial and complete diffusion allowed us to observe an enhanced hydrogen discharge for complete diffusion. The non-uniform diffusion can also occur by uneven emulsion coating and hydrogen bubble formation at the top of the glass beaker channel in the test setup. The preferential diffusion through the grain boundaries and triple junctions after sufficient time of immediate hydrogen discharging allowed us to consider them as reversible trapping sites. Fig. 14 shows preferential hydrogen diffusion through the triple junctions of sample M1. The SEM and EDS observations allowed us to conclude that there is a higher rate of hydrogen diffusion through the triple junctions than through the grain boundaries. The preferential diffusion through triple junctions and grain boundaries in sample M1 is also observed in sample T5. Fig. 15 visualized the preferential hydrogen diffusion through two locations of sample T5 that contain degenerate pearlite (DP). The observations were after the same HMT experiment. The literature reported the occurrence of DP in X70 steel [39]. The block of ferrite and colony of lath-shaped cementite in the DP were identified. The partial diffusion micrograph indicated that cementite diffuses hydrogen, and the complete diffusion micrograph supported it with the observation of clustered white particles over the cementite. Also, the ferrite in the DP does not have any overlaid white particles. Fig. 15 also show the SEM images and EDS analysis of partial and complete diffusion through the magnified DP. The partial diffusion micrograph showed that boundaries of the cementite lath appear to discharge more hydrogen as most of the white particles were visualized at the lath boundaries. The complete diffusion micrograph showed that the grain boundaries were a preferential hydrogen diffusion path than the bulk of the grain. Also, cementites were visualized with a cluster of white particles over it. This observation allowed us to conclude a high hydrogen diffusion through cementite.

Conclusions We obtained the following results after the study on the effect of grain size on hydrogen diffusion and trapping in different layers of X70 pipeline steel. 1. The increase in grain size had decreased the trapping by triple junctions and grain boundaries. Hence, the hydrogen permeability was increased with an increase in grain size, for the tested grains, in both layers of the steel. The effective diffusion coefficient was also increased in the top-layer of steel. However, the first polarization of the mid-layer showed an initial decay and subsequent growth in the effective diffusion coefficient due to irreversible inclusion traps. 2. The density of total, reversible and irreversible trapping sites in top-layer had decreased with an increase in the grain size. On the other hand, the density of

total and irreversible trapping sites of the mid-layer showed an initial growth and subsequent decay. However, the density of reversible trapping sites of the mid-layer behaved like the top-layer for the increase in grain size. 3. The preferential hydrogen diffusion after immediate hydrogen discharging in the HMT experiment allowed us to show that grain boundaries, triple junctions and cementite were reversible trapping sites. 4. There was partial and complete diffusion of hydrogen through different regions of steel samples. Hence, preferential hydrogen diffusion through individual microstructural features was compared from the same images. We found that the intensity of hydrogen diffusion increased from grains, grain boundaries, triple junctions and cementites in steel microstructure.

Data availability The raw data required to reproduce these findings are available to download from https://doi.org/10.17632/7xkc6txwb8.1.

Acknowledgments This work was supported by the Natural Sciences and Engineering Research Council of Canada [470033, 2015]. The authors appreciate EVRAZ North America, Saskatchewan, Canada, for supplying the pipeline steels for this study and CanmetMATERIALS, Ontario, Canada, for processing the supplied materials.

references

[1] Nace T, Plante L, Browning J. Global energy monitor. April 2019. [Online]. Available: https://globalenergymonitor.org/ pipeline-bubble/. [Accessed 12 June 2019]. [2] Canadas oil and natural gas producers. Canadian Association of Petroleum Producers," March 2019. [Online]. Available: https://www.capp.ca/publications-and-statistics/ publications/333595. [Accessed 12 June 2019]. [3] International Energy Agency. World energy outlook. June 2019. [Online]. Available: https://www.iea.org/weo/. [Accessed 15 June 2019]. [4] Canadian Energy Pipeline Association. Pipeline industry performance report. Calagary: Canadian Energy Pipeline Association; 2015. [5] National Energy Board, Canada. Stress corrosion cracking on Canadian oil and gas pipelines [Online]. Available: http:// publications.gc.ca/collections/Collection/NE23-58-1996E.pdf. [Accessed 15 June 2019]. [6] National Energy Board, Canada. A comparative analysis of pipeline safety performance [Online]. Available: http:// pstrust.org/docs/spi_focusonsafety_2000_2003_0503_e.pdf. [Accessed 15 June 2019]. [7] Popov BN, Lee J-W, Djukic MB. Hydrogen permeation and hydrogen-induced cracking. In: Handbook of environmental degradation of materials, norwich, william andrew publishing; 2018. p. 133e62.

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096

international journal of hydrogen energy xxx (xxxx) xxx

[8] Kutz M. Handbook of environmental degradation of materials. Norwich: William Andrew; 2013. [9] Kittel J, Smanio V, Fregonese M, Garnier L, Lefebvre X. Hydrogen induced cracking (HIC) testing of low alloy steel in sour environment:Impact of time of exposure on the extent of damage. Corros Sci 2010;52(4):1386e92. [10] Hopkins P. "Learning from pipeline failures," WTIA/APIA welded pipe symposium. 2008. Perth. [11] Canadian Energy Pipeline Association. Transmission pipeline industry performance report [Online]. Available: https://pr17. cepa.com/performance-data/. [Accessed 15 June 2019]. [12] American petroleum institute - association of oil pipelines. In: Annual liquids pipeline report," API-AOPL, Washington; 2019. [13] Findley KO, O’Brien MK, Nako H. Critical Assessment 17: mechanisms of hydrogen induced cracking in pipeline steels. Mater Sci Technol 2015;31(14):1673e80. [14] Pressouyre G, Bernstein I. An example of the effect of hydrogen trapping on hydrogen embrittlement. Metall Trans A 1981;12(5):835e44. [15] Mohtadi-Bonab MA, Eskandari M, Szpunar JA. Texture, local misorientation, grain boundary and recrystallization fraction in pipeline steels related to hydrogen induced cracking. Mater. Sci. Eng. A 2015;620:97e106. [16] Ichimura M, Sasajima Y, Imabayashi M. Grain boundary effect on diffusion of hydrogen in pure aluminium. Mater Trans, JIM 1991;32(12):1109e14. [17] Yazdipour N, Haq A, Muzaka K, Pereloma E. 2D modelling of the effect of grain size on hydrogen diffusion in X70 steel. Comput Mater Sci 2012;56:49e57. [18] Mohtadi-Bonab M, Szpunar J, Razavi-Tousi S. Hydrogen induced cracking susceptibility in different layers of a hot rolled X70 pipeline steel. Int J Hydrogen Energy 2013;38(31):13831e41. [19] Masoumi M, Santos LP, Bastos IN, Tavares SS, Silva MJd, Abreu HFd. Texture and grain boundary study in high strength Fee18NieCo steel related to hydrogen embrittlement. Mater Des 2016;91:90e7. [20] Calhoun RB, Dunand DC. Dislocations in metal matrix composites. In: Comprehensive composite materials. Oxford: Pergamon; 2000. p. 27e59. [21] Turnbull A. Factors affecting the reliability of hydrogen permeation measurement. Mater Sci Forum 1995;192e194:63e78. [22] Lee J-Y, Lee S. Hydrogen trapping phenomena in metals with B.C.C. and F.C.C. crystals structures by the desorption thermal analysis technique. Surf Coat Technol 1986;28(3e4):301e14. [23] Mohtadi-Bonab M, Szpunar J, Razavi-Tousi S. A comparative study of hydrogen induced cracking behavior in API 5L X60 and X70 pipeline steels. Eng Fail Anal 2013;33:163e75.

15

[24] Luu W, Wu J. Effects of sulfide inclusion on hydrogen transport in steels. Mater Lett 1995;24(1e3):175e9. [25] Organization IS. Method of measurement of hydrogen permeation and determination of hydrogen uptake and transport in metals by an electrochemical technique. Switzerland: International Standards Organization; 2014. [26] Manolatos P, Jerome M. A thin palladium coating on iron for hydrogen permeation studies. Electrochim Acta 1996;41(3):359e65. [27] Devanathan MAV, Stachurski Z. The adsorption and diffusion of electrolytic hydrogen in palladium. Proceed. Royal Soc. A 1962;270(1340):90e102. [28] Ichitani K, Kuramoto S, Kanno M. Quantitative evaluation of detection efficiency of the hydrogen microprint technique applied to steel. Corros Sci 2003;45(6):1227e41. [29] Cheng Y. Analysis of electrochemical hydrogen permeation through X-65 pipeline steel and its implications on pipeline stress corrosion cracking. Int J Hydrogen Energy 2007;32(9):1269e76. [30] Yu J, Zhang T, Qian J. Electrical motor products. Cambridge: Woodhead; 2011. [31] Taylor JR. An introduction to error analysis. Sausalito: California; 1996. rez T, Garcı´a JO. Direct observation of hydrogen [32] Pe evolution in the electron microscope scale. Scr Metall 1982;16(2):161e4. s RS, Corte z VHL. Hybrid laser e arc [33] Tavitas RJF, Garce welding applied in longitudinal joints for hydrocarbon conduction pipes. MRS Proceedings 2015. [34] Alipooramirabad H, Ghomashchi R, Paradowska A, Reid M. Residual stress- microstructure- mechanical property interrelationships in multipass HSLA steel welds. J Mater Process Technol 2016;231:456e67. [35] Bohemen Sv, Sietsma J. The kinetics of bainite and martensite formation in steels during cooling. Mater. Sci. Eng. A 2010;527(24e25):6672e6. [36] Krom AH, Bakker A. Hydrogen trapping models in steel. Metall Mater Trans B 2000;31(6):1475e82. [37] Pressouyre G, Bernstein I. A quantitative analysis of hydrogen trapping. Metall Trans A 1978;9(11):1571e80. [38] Simm TH, Sun L, Galvin DR, Hill P, Rawson M, Birosca S, Gilbert EP, Bhadeshia H, Perkins K. The effect of a two-stage heat-treatment on the microstructural and mechanical properties of a maraging steel. Materials 2017;10(12):1346. [39] Sha Q, Li D. Microstructure and properties of low manganese API X70 pipeline steel for sour service application. In: Proceedings of the 8th pacific rim international congress on advanced materials and processing, cham; 2013.

Please cite this article as: Thomas A, Szpunar JA, Hydrogen diffusion and trapping in X70 pipeline steel, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2019.11.096