Alloy 718 Subsea Bolt in Relation to Surface Cracking: A Microstructural Perspective

The brittle fracture mode close to the surface of Inconel 718 subsea bolts is closely related to their nanostructure. When the bolts are used subsea under cathodic protection, hydrogen evolves. Intergranular precipitates and points of intersection of slip lines always held responsible for hydrogen enhanced decohesion. However, thus far, little attention has been paid to the bolts manufacturing method and to the characterization of the role of subsurface oversized nanoprecipitates on transgranular surface cracking. As-received subsea bolts were analyzed using multi-scale observation techniques such as focused ion beam milling, scanning electron microscopy, transmission electron microscopy, and in-air strain rate tensile tests. The results further demonstrated that under identical API aging, there was a difference in the morphological precipitation and strength of 718 as a bolt and rectangular billet. For the 718 rectangular billets, discrete intergranular stable δ/MC carbides precipitate and only γ' of 10 - 20 nm was observed for the bulk and subsurface. Whereas, for the CRA subsea bolt, γ' was approximately 30 nm, γ" was 30 - 50 nm for the bulk (370 HV); γ' was approximately 50 nm and γ" was 50 - 100 nm at subsurface (400 HV). As-received subsea bolts were investigated at subsurface (10 μm from the surface) at the thread and shank region, which revealed transgranular sheared oversized ~50 nm γ' (with dislocation networks) and metastable γ" > 100 nm particles, respectively. Thus, an effort was made to develop the hydrogen-assisted surface cracking theory for bolts.


Background Problem
Nickel alloy Inconel 718 is used in deep subsea wells in high pressure, high temperature, and corrosive environments as bolts, drill tools, hangers, packers, and subsurface safety valves for extended long-term exposure to stress (>70% yield strength) [1] [2]. Historically, bolts for subsea production systems met manufacturer quality standard in accordance with API Spec 6A applicable for surface wellhead and tree equipment; however, sometimes the installation does not conform to standards by overtightening and over loading [2]. In addition, 718 bolts in taperlock flanges of a subsea completion system are cathodically protected by aluminum anodes, which results in the ingress of hydrogen from the surface.
Hydrogen, as an interstitial solute, is considered to have an adverse effect on bolt ductility and strength. Specifically, hydrogen lowers the fracture energy (reduces fracture toughness) of the crack and promotes cleavage brittle fracture. The interaction of hydrogen atoms with the second phase is important and requires understanding at the atomic scale. Several hydrogen embrittlement (HE) mechanisms have already been proposed; however, the most important mechanism for metallic materials (consisting mainly of Ni and Fe) that do not form hydrides are hydrogen enhanced localized plasticity (HELP) [3] [4] and hydrogen enhanced de-cohesion (HEDE) [5] [6]. Thus, for decades, HE has been a problem for alloy 718 when used subsea. To mitigate this issue, alloy 718 for oilfield applications relies on unique chemistry, heat treatment (as per API 6A718 standard spec [7]), and microstructure to resist stress corrosion cracking; however, the microstructure of Inconel 718 bolts strongly depends on the manufacturing sequence. Typically, the subsea bolts of Inconel family are cold rolled and, then, heat-treated; however, this process results in an increase in the localized hardness, lower ductility, discrete grain boundary precipitation, and higher susceptibility to HE [2]. The reason for this abrupt change in mechanical and corrosive properties is attributed to the microstructure where the detrimental role of intergranular precipitates for HE has already been well established [2] [8] [9] [10] [11], and the analysis of 718 service failures has been performed by Cassagne et al. [12]. Moreover, phase kinetic simulations [13] [14] defined the alloy 718 limitations, which led to the development of controlled microstructure for specific applications [1] [7]. It is possible that for 718 subsea bolts under CP, transgranular oversized nanoprecipitates at the outer periphery (subsurface) may be the primary source for crack initiation and intersection of localized dislocation slip lines (DSLs) [15] [16] [17], which leads to the formation of nanovoids at transgranular regions; later, transgranular nanovoids coalescence propagates to intergranular cracks.

Chemistry and Microstructure
The nominal composition of Inconel 718 in wt% is 52.  Nb-3.0 Mo-0.5 Al-1.0 Ti-0.08 C [11] [18]. The matrix is composed of face cen- and Cr 2 O 3 protective oxide films that increase corrosion resistance [19]. It is primarily strengthened by the unique intermetallic, Ll 2 ordered) FCC phase, γ' Ni 3 (Al, Ti, Nb), by interacting with dislocations [19]. This ordering places Ni at the face of each unit cell and Al, Ti, or Nb at each corner. The γ' phase is more spherical at low volume fraction and tends to be cubic at higher volume fraction [20] depending upon heat treatment [21]. Hagel and Beattie [22] have observed that γ' appears as spheres at the 0% -0.2% lattice mismatch, as cubes at the mismatch of approximately 0.5% -1.0%, and as plates at the mismatch above approximately 1.25%. γ' forms coherent precipitates in the γ matrix at approximately 620˚C. γ" phase provides additional strength and is considered to be the main strengthening agent. γ" is body centered tetragonal Ni 3 Nb and precipitate in the γ matrix at approximately 720˚C. Both γ' and γ" have the solvus temperature of approximately 900˚C -920˚C [14] [23]. However, γ" is a metastable phase, which tends to be replaced by an incoherent stable orthorhombic delta (δ) phase, Ni 3 Nb, after long-term exposure to temperature above 650˚C [24] [25] [26]. It typically precipitates at 650˚C -980˚C at grain boundaries with thin plates extending to grains [27] and with the solvus temperature of 980˚C -995˚C. Both γ" and δ phases are Nb-based, which implies that δ phase precipitation occurs with the loss of the γ" phase. δ morphology contributes to the loss of strength and hardenability owing to the depletion of γ", and its presence in the structure has led to increased susceptibility of hydrogen cracking with the loss of fracture strain and notch tensile strength [28] [29] [30]. In addition to these phases, blocky shaped carbides (i.e., MC carbides, where M is mostly Nb and Ti) are homogeneously distributed throughout the matrix [31] [32] [33]. Both δ and MC carbides inhibit grain growth and prevent grain boundary sliding. Nb is essential in making up all the phases present in alloy 718. It has been determined that the Laves phase requires 10% -12% of Nb; the delta δ phase requires 6% -8% of Nb; γ" needs 4% of Nb, and γ' can form at Nb levels below 4%. The degree of Nb segregation during ingot or casting solidification depends on the rate of cooling (i.e., the faster the cooling, the lower is the segregation). High temperature homogenization can be used to produce the Laves phase solution and uniformly distribute Nb throughout the bulk matrix [34]. Sigma phase has hexagonal geometry with the composition formula of (Cr, Mo) x (Ni, Co) y ; where × and y vary from 1 to 7; typically, sigma phase grows through γ' particles. It has a detrimental effect and the plate-like morphology. Sigma phase is the source for crack initiation and propagation, which leads to low temperature brittle fracture.
It contains high amount of refractory elements, which are extracted from the γ matrix, which results in the loss of solution strengthening [27]. Thermodynamic calculations and kinetic simulations have become important in understanding alloy 718 properties. It is logical to further improve the alloy;  [34]. It is important to understand that NiFe-based superalloys behave in a more complex way than predominantly Ni-based superalloys. This occurs because the hardening mechanism in the austenitic γ matrix involves metastable γ' Ni 3 Nb along with stable γ' Ni 3 (Al, Ti). Thus, the NiFe-Super database can handle these complex alloys to predict both stable and metastable equilibrium states. The equilibrium phases, which can exist in 718 derivatives, are predicted using the thermodynamic software packages JMatPro and ThermoCalc. The predicted phases that can be present in alloy 718 can be seen in thermodynamic phase diagrams, shown in Figure 1(a) and Figure 1(b). 718 phase diagrams (a) and (b), use the NiFe-Super thermodynamic database, which is designed by Thermotech using JMatPro specifically for Ni-based superalloys containing high Fe content.
γ, γ', MC and, M 23 C 6 are commonly predicted for both 718 conditions, as per the heat treatment. Sigma and αCr phases are normally present at extended exposure to high temperatures. The time temperature transformation (TTT) diagram for alloy 718 is shown in Figure 1(c).

Purpose
Herein, an extensive microstructural study seeks to investigate alloy 718 that is heat-treated according to API specification in two different product forms, rectangular billet (RB) and subsea bolts. The manufacturing route of bolts involves the API heat treatment followed by cold rolling of threads, whereas, samples extracted from RB undergo electrical discharge machining followed by the heat treatment. The objective was to perform subsurface cracking analysis for the subsea bolts on the basis of microstructural results and to critically discuss the 718 subsea bolt vulnerability to HE in-service with the already established HAC theories for alloy 718 (which were subjected to different thermochemical treatments and product forms) [ [14]) and slightly longer (16 and 25 h) aging conditions [13] were used to observe the effect on 718 precipitation in comparison to the API 6A718 aging temperature. Other conventional solution annealing and double aging were added to observe their effect on microstructure when compared to API 6A718. The reason for selecting non-standard heating parameters was to characterize the morphological evolution of second phase particles (i.e., γ', metastable γ", stable δ, M (C, N)s, or any Laves phase) within grains and on grain boundaries. These new heating variants provide more room for observing the PLC effect under slow strain rate testing in the future.

Experimental Materials
Forged Inconel 718 in the product form as a 4× RB (100 × 30 × 20 mm 3 ) was received from the supplier, and eight separate subsea bolts (Φ of ~28 mm and length of ~150 mm), which were heat-treated and then cold rolled to produce threads before deploying them in-service, were received in as-received (AR) condition from the subsea completion system. Optical emission spectroscopy (OES) was performed with thin off cut from the bolts and RBs. The OES results are well within the specifications for the standard UNS NO7718 alloy [7] and are presented in Table 1.

Heat Treatment and Sample Preparation
The 718 specimens were subjected to heat treatment as per conditions described in Table 2. For every sample, type K thermocouples were peened into holes that

Microstructural Studies
The specimens, which were observed by light microscopy (LM) and scanning electron microscopy (SEM), was grinded and polished using the descending order of abrasive cloths up to the finer grit size of 0.25 μm. Specimens were mounted with conductive resin prior to grinding and polishing. The Kallings reagent (100 mL of HCL, 100 mL of ethanol, and 5 g of pitates was performed by SEM (spot size 5, operated at 20 KV using the backscatter electron mode) with energy dispersive X-ray (EDX) spectroscopy. The average size of precipitates within the grain and at the grain boundary was measured by evaluating 100 randomly chosen precipitates.
For the TEM analysis, the JEOL 2100 instrument at the department of engineering of University of Leicester was used and operated at the accelerating vol- (i) This is the location of region of interest.
(ii) A platinum layer with the dimensions of 20 × 2 × 4 µm is deposited at 52˚.
(iii) Milling of trenches above and below the platinum deposit.
(iv) Milling of "L" cut at 7˚ and cleaning of trenches at 52˚.
(v) Tilting back to 7˚ to check the "L" cut.
(vi) Attaching the tungsten omniprobe needle with carbon deposition to the right side of the foil. The vertical cut (red dotted line) is made to extract the sample from the trench.
(vii) The foil is attached with the omniprobe needle.
(ciii) Aerial view of the TEM copper grid where foil is to be attached.
(ix) Attachment of the foil using C deposition to the "I"-shaped holder of the TEM grid.  (xii) SEM image of the electron-transparent foil, which is less than or ~100 nm and ready for TEM imaging.

Mechanical Studies
The schematic diagram of the tensile specimen is shown in Figure 2(b). Two tensile specimens for each heat treatment were prepared using standard laboratory methods. The tensile specimens from the bolt were extracted from the bulk. The tensile specimens from each alloy were machined to the following dimensions: gage length of 32.5 mm, diameter of 5 mm, and the width of the grip section of 8 mm. The total length of the tensile specimen was 42 mm. The specimens were tested in air using an Instron tensile machine at the strain rate of 2.22 × 10 −4 s −1 .
Vickers hardness (HV) measurements were carried out for each characterization specimen. Prior to hardness testing, samples were mounted in resin followed by grinding (240 -1200 grit) and polishing up to 0.25 µm with colloidal silica. Measurements were performed using a diamond pyramid indenter with a force of 5 -10 kgf (HV5/10). The values were recorded while moving from one end to another. The distance between each indent was kept approximately greater than that of 3 d spacing, where d is the indent diameter (either in x or y direction), to prevent overlap of strain fields.

Results
LM, SEM, TEM, and mechanical results were collected for all 718 specimens (from RB and bolt). The primary focus was on the results of SEM and TEM analyses of used CRA bolts, which were used to determine the effect of nanostructure on mechanical properties and surface cracking; then, we established the surface cracking theory for subsea bolts under CP.

Variation in Grain Size with Heat Treatment
Light micrographs for each specimen are shown in Figure 3. The grain size for each heat treatment was evaluated using the LM intercept method and later confirmed with the concentric circles method outlined in the ASTM standard E-112 [38]. To use a statistical approach toward grain size and particle size measurements, at least 15 light and 15 SEM micrographs were considered. There is variation in grain size in 718 specimens from 30 µm to 85 µm. The grain size of 30 -35 µm was observed for S2 API from RB, and 30 -40 µm for S6 and S7 from CRA bolts. For non-standard heat treatments, an increase in the grain size of 70 -85 µm was observed for S3 and S4 (16 and 25 h, respectively), whereas for S5, the microstructure grain size of 55 -65 µm was observed. However, the light images of bulk and edge sections of CRA bolts (S6 and S7) show no significant change in grain size.
There was no significant difference in the grain size of S2 API from RB and of S6 API or S7-AR from API bolts. The observation of light microstructures, shown in Figure 3(a), Figure 3(b), and Figure 3(f) for S1 API , S2 API , S6 API , and AR clearly indicated the presence of finer grains with a grain size of 30 -40 µm. This implies that RB and the bolt have the same grain size after the API heat treatment. However, the evolution of grain size is observed for non-standard heat-treated specimens S5 (55 -65 µm) to S3 and S4 (75 -85 µm). For S3 and S4, this change may be due to an increase in the solution treatment temperature from 1035˚C to 1250˚C, which results in the dissolution of carbides (as mentioned earlier) and some nitrides back into the matrix, whereas for S5, a change in the grain size may be due to the double aging after the solution treatment. This treatment may have dissociated primary carbides to form intergranular M 23 C 6 . Moreover, twinning of grains occurred in some specimens (as highlighted in Figure 3), i.e., the same crystal lattice points are shared by separate grains in a symmetrical manner. The plastic deformation mechanism has always been used for Inconel 718 [29] [30]. In addition, grain boundary triple junctions are common in all specimens, both for solution-annealed and aged conditions.

Intergranular Precipitates and M (C,N)s-Carbides and Nitrides
The typical SEM images of S1 and S2 API from the 718 RB are shown in Figure 4 and Figure 5, respectively. The solution-treated specimen S1 shows that M (C, N)s are randomly distributed throughout the structure without intergranular precipitation. SEM EDX showed that these precipitates were metal carbides (white) and nitrides (black). Metal carbides were identified as NbC, and metal nitrides as TiN. Whereas, S2 API had a similar distribution pattern of M (C, N)s but showed the discrete intergranular precipitation of δ and M 23 C 6 . The solution-treated specimen S3 at 1250˚C showed the dissolution of carbides and some nitrides back into the matrix. Grain boundary precipitation, if any, also dissolved back into the matrix because their solvus temperature was less than 1000˚C. This  1250˚C produced a Nb-rich fcc γ matrix, which facilitated the continuous precipitation of δ and Laves phases. It was observed that S5 and S6 (API bolt) did not have the delta δ phase (needle platelets) at the grain boundary but S5 contained discrete M 23 C 6 , as shown in Figure 8, whereas S2 API showed the discrete delta δ phase precipitation at intergranular positions. For specimens S1 and S3, no intergranular precipitation was observed because the solvus temperature of all second phases was below 1000˚C. The solvus temperature of all second phases (e.g., γ', γ", M 23 C 6 , and MC) was calculated using ThermoCalc simulations. Typical scanning electron images from the edge and bulk of S6 and S7 CRA bolts from shank and thread regions are shown in Figure 9. Similar to other specimens, M (C, N)s are randomly distributed at intergranular and transgranular positions from the edge (shank and thread) to bulk (shank and thread) of the bolt. The image shown in Figure 9(a) is an overview of the bolt microstructure   with M (C, N)s precipitates (which appear as bright spots). They are present in the variable size range from 1 to 20 µm. They were seen at the shank edge, shank bulk, thread edge, and thread bulk of the bolt. Grain boundaries were carefully examined at each of these locations; it was determined that there was no discrete or continuous intergranular precipitation, as shown in Figures 9(b)-(e).

Transgranular Precipitates Morphology and Distribution: γ' and γ″
The TEM images of specimens from the API bolt and RB are shown in Figures  10-17 [34]. In addition to different morphology, each nanophase had a unique selected area diffraction pattern associated with its crystal structure, which were recorded as a confirmation of their existence in the fcc γ matrix. The bright primary spots are related to fcc γ, and secondary faint spots are associated with γ'. These faint spots, where they appear to be elongated, represent the γ" phase. Furthermore, EDS line scans (length depending upon the nanophase size) were conducted from the γ matrix to the nanophase to confirm phase composition. Figure 10 shows the TEM images of solution-treated sample S1 from RB with the γ' precipitation of <2 nm. In the dark field imaging mode, Figure 10(a) shows the overview of the γ matrix with the dense homogenous distribution of γ' precipitates. The electron-diffraction pattern at ( 010 ), shown in Figure 10(b), is associated with the fcc γ matrix because the size of the γ' precipitates is small and not easily distinguishable from that of the γ matrix, as evident from Figure   10(c), which shows a mottled appearance.
Sample S2 API has a dense homogenous distribution of spherical 10 -20 nm γ' precipitates. Typical TEM images from the bulk and edge (subsurface) of S2 API are shown in Figures 11(a)-(c); Figure 11(b) is the high-magnification image of Figure 11(a). The electron-diffraction pattern at (001), shown in Figure   11(d), indicates the presence of the γ' phase. Few precipitates appeared to exceed the 20-nm barrier, which occurred because γ' precipitates coalesced to form elongated precipitates, Figure 11  h, WQ to RT then 675˚C for 8 h AC), they were 14% larger, i.e., 28.0 nm in diameter. After double aging, the physical definition of γ' precipitates was enhanced compared to that after single aging, which has been also reported in [20].
However, S5 under double aging produced an additional population of γ' precipitates of 1 -10 nm along with primary γ' and γ" populations, as shown in Figure 14(c) and Figure 14(d). Compared to the single population of γ' in S2 API , Ding [40] observed two main populations in the RB of 718 that was solution-treated at 1050˚C for 1 h AC but was double-aged at 720˚C for 8 h and 650˚C for 8 h. They have reported the presence of primary γ' of 25 nm, which was homogeneously distributed throughout the matrix, and secondary γ' of 100 nm, which had a much lower density.
The TEM image of sample S3 in Figure 12(a) is a dark field image, where the γ' phase is not easily distinguishable from the γ matrix of sample S3, which is non-standard solution-treated at 1250˚C for 2 h followed by water quenching. As expected, electron-diffraction pattern at (110 ) in Figure 12(b) is more related to the γ matrix, which is evident from the distinct primary spots. S3 was non-standard solution-treated to verify whether providing higher Nb back to the γ matrix leads to the formation of γ" metastable or intergranular stable δ phases in S4 samples (which are solution-treated at 1250˚C followed by two aging treatments for 16 and 25 h). The image in Figure 13(a) was acquired in the bright field imaging mode, where the γ' precipitates (closely spaced) grew under 16 h of aging and were homogenously distributed throughout the matrix. The high-magnification image in Figure 13(c) shows a dark faint line around the precipitates, which differentiates γ' from the γ matrix, and spherical-to-spheroid-shaped γ' precipitates in the range of 10 -50 nm with a bulge effect. Whereas, during the 25-h aging process, few γ' precipitated during the aging process because there was an insufficient amount of Ti and Nb in the γ matrix for γ' precipitation. Figure 13(d) shows that γ' precipitates are widespread with greater inter-particle spacing (>200 nm); the maximum size of γ' was   Figure 13(b) and Figure 13(e). Sample S5 after the double aging heat treatment has a homogeneous distribution of γ' and γ" (from bulk to subsurface), as shown in the overview TEM image in Figure 14(a). The size of γ' precipitates was approximately 50 nm in diameter, and γ" appeared as elongated disks of 50 -70 nm. The selected area diffraction pattern shown in Figure 14(b) strongly corresponds to both γ' and γ" phases at (110 ). In the bright field imaging mode, small secondary γ' precipitates of 1 -10 nm were also observed between the primary γ' and γ", as evident from the pointed red arrows in Figure 14(c) and Figure 14(d). These precipitates likely appeared during air cooling to RT after the final aging at 621˚C for 8 h. Furthermore, a faint dark line appeared around these precipitates, which may be associated with the lattice misfit between γ and γ'. Upon closer inspection (see Figure 14(c) and Figure 14 spheroids and is coalesced at some points, whereas small γ', which is observed in dark field mode, is much more distinct than large γ', which appears to be embedded along with γ" within the γ matrix.
The TEM image in Figure 15(a) shows the typical nanostructures of the shank bulk area of the bolt (S6 and S7), i.e., γ' precipitates and few random γ". At higher magnification [ Figure 15(b)] in the dark field mode, the γ' phase appears to consist of spheres and spheroids and is coalesced (appearing elongated) as approximately 20-30-nm black precipitates. The selected area diffraction patterns strongly correspond to the presence of the γ' phase at (100) [see Figure  15(c) and γ" at ( 10 1 ) (see Figure 15(f))]. However, for S2 API and S4, γ', which is observed in the dark field mode for S6 and S7, is considerably less clear and appears to be embedded within the γ matrix. However, γ" at the shank edge, see Figure 15(d) and Figure 15(e), precipitated with a morphology that is similar to that in sample S5 from RB. Similarly, for the thread bulk, the nanostructure shown in Figure 16(a) and Figure 16(b) shows the precipitation of γ' and γ".
However, unlike the nanostructure of the shank edge, the thread edge is made of both 50-nm γ' and 50 -100-nm γ", as seen in Figure 16(d) and Figure 16(e). To substantiate the existence and morphology of γ' and γ" phases, TEM images were randomly extracted from similar locations on other bolt samples (S6 and S7), as shown in Figure 17. The results were identical to what was observed earlier, except for some sheared 50-nm γ' precipitates and >100 nm γ", at the thread and shank edge (subsurface).    Table 3 shows the strength data obtained for each specimen from in-air tensile    tests. The tensile specimens were extracted from the bulk of RB and bolt. Each heat-treated specimen, as shown in Table 2, has two tensile specimens for mechanical characterization. Thus, two measurements for yield strength, ultimate tensile strength, and percentage reduction in area were calculated from the tested tensile specimens, and the readings were averaged. The tensile specimens were machined from the bulk of the bolt (S6 and S7); therefore, the tensile property data does not relate to the microstructure at the periphery or the edge of the bolt. Whereas, the hardness data from the outer periphery (close to the surface, approximately 10 µm) compared to that from the bulk of the bolt has the difference of ∆30 HV. Thus, FIB samples were extracted and analyzed by TEM to perform detailed structural investigation. The in-air %RA values of 718 variants will help to observe the PLC effect at subsea temperatures under CP. However, these particles are the potential source of hydrogen traps; in Inconel 718, 77 -87 kJ/mol has been suggested as the activation energy for hydrogen atoms to be trapped in carbide particles [41].

Discussion
The activation energy of 58 kJ/mol has been proposed as the criterion to determine whether hydrogen, which is already trapped in a site, can be reversibly released or not [42]. However, hydrogen-assisted intergranular cracking is more common in δ decorated grain boundaries [17] [18] [29] [30] (specimen S2 had discrete, and S4 treated for 16 h had continuous δ platelets extending to grains, see Figure 5(c) and Figure 7(b)). The hydrogen binding energy to δ particles is DOI: 10.4236/msa.2020.1112053 808 Materials Sciences and Applications 30 kJ/mol, which is smaller than 58 kJ/mol [40]. To compare δ and γ", Dai and Liu [43] have used the first principles total energy pseudo-potential method that is based on the density functional theory to determine that both γ" and δ have covalent bonding between Ni-Ni atoms and the combination of covalent and ionic bonding between Ni and Nb atoms. There is covalent bonding between Nb and Nb atoms in the γ" phase, and no interaction was observed between Nb and Nb atoms in the δ phase. In addition, their results have indicated that both phases are hard and have good ductility. The bulk moduli of γ" and δ are 211 GPa and 208 GPa, respectively, which make both phases incompressible and hard. Thus, except from providing remarkable strength to the subsea bolt, the exact effect of metastable γ" on HE surface cracking remains unclear. However, solution-annealed specimens S1 and S3 only contain MCN particles, and Rezende et al. [8] have reported that for solution-annealed specimens, the HE im- In this study, twin boundaries were also observed for all 718 variants, including bolt specimens S6 and S7, as seen in Figure 9. Some researchers have reported [45] [46] that twin boundaries are considered to have low hydrogen solubility and high separation energy, which makes them less susceptible to HE.
Whereas, Seita et al. [47] have reported that for Inconel 725, crack initiation is associated with twin boundaries but at the same time is more resistant to crack propagation. However, Zhang et al. [15] have not observed crack initiation and propagation in Inconel 718. Further analysis revealed that crack path is always along the {111} crystallographic plane, which is also the plane of DSLs. Grain boundary triple junctions assist decohesion and crack initiation at grain boundary and particle interface, which has been also reported by Tarzimoghadam et al. [16]. Triple junctions were observed in all specimens, including subsea bolt specimens (see Figure 3 for details). However, bolt grain boundaries were observed to lack any discrete intergranular precipitates, which is evident from Figure  pitation, see Figure 15(d) and Figure 17(c). It is clear from the bolt microstructure that there is a morphological gradient of the second phase from the bulk to the edge; specifically, 30-nm γ' and 30 -50-nm γ" observed in the bulk, 50-nm γ' and 50 -100-nm γ" are observed at the edge. Whereas, the nearly consistent morphology for the second phase (from bulk and edge) was observed for 718 specimens extracted from RB, apart from specimen S5, where the second γ' population of 1 -10 nm exist along with primary 50-nm γ' and 50 -70-nm γ", as seen in Figure 11, Figure 13, and Figure 14, which also contributed to its remarkable strength (910 MPa) and hardness (408 HV), as seen in Table 3.
To replicate the nanostructure of the AR subsea S7 bolts (from the bulk and   [17] in hydrogen-assisted cracking for alloy 718; however, the primary nanoprecipitates responsible for surface and subsurface cracking remain unknown and need to be studied. The nanostructure at the surface cracks has not been extensively reported (including the morphology-shape and size), and only generic nanophases representing the bulk of the tensile samples are accounted for. This is the reason why the manufacturing route of the bolt is critical and determines the nanostructure at the extreme surface/subsurface of the alloy 718. By analogy, it has been reported [44] [49] that the machining of the tensile specimens (for hydrogen charging and slow strain rate testing) has been always performed after the heat treatment, which have resulted in different nanostructure at the surface/subsurface of the tensile specimens in comparison to that of the bulk lattice; both the number density and the length density of surface/subsurface cracks depend on the hydrogen charging time (mentioned earlier). Whereas, in the current study, the detailed examination of the outer periphery (surface) of the used subsea bolt (S7) (at thread and shank regions) revealed the Vickers hardness of 400 HV at the subsurface (370 HV at bulk) of the bolt, as shown in Table 3. FIB samples were lift-out from the sub-microstructure (approximately 10 µm from the surface) of the shank and thread region. The schematic of the location of the extracted FIB samples and corresponding TEM images are shown in Figure   18(b), Figure 18(c). The dotted lines represent the area within the γ' precipitate ( Figure 18(d)) where the shearing effect is most pronounced.
For DSLs, similar findings have been reported in [49], in which an FIB sample (10 µm below the fracture surface) (sub-microstructure in the HE brittle zone)  to what was observed at higher magnification at the bolt thread region, is shown in Figure 18(d). In addition to DSLs, in between the sheared γ' precipitate, the γ matrix is clearly observed owing to the shearing effect caused by the disordering of crystallographic planes within the γ' precipitate by DSLs (moving under stress from the γ matrix to the γ' precipitate) and hydrogen atoms. These sheared γ' are the potential source of surface/subsurface cracking as hydrogen accumulates under CP at these transgranular sites, which creates stress-concentrated regions.
This may also be the reason why these subsea 718 bolts in-service (under overtightening and overloading conditions) may have stud or thread fracture.
In-service pre-tensioning and loading produces high driving force, and the available slip system operates at high coherency strain regions, where hydrogen atoms accumulate close to the surface of the bolt; this phenomenon promotes the further shearing and elongation of γ' particles at the threads and shank edge (with a 50-nm size) but not in the bulk (where the maximum size of γ' is 30 nm).
Some available DSLs are induced at the manufacturing level during the rolling of shank and threads; this process may produce sheared surface γ' precipitates prior to deploying the bolts in-service. It is reasonable to infer that for 718 bolts, sheared γ' sites (see Figure 18) along with the points of intersection of slip lines undergo nanovoid coalescence to exhibit the transgranular mode of surface cracking. Thus, these transgranular nanocracks on bolt surface and subsurface under loading result in the brittle fracture mode at the threads and shank surface, compared to the ductile fracture of the bulk.
During the manufacturing, the rolling of bolts considerably alters their structure by inducing lattice mismatch regions [50]. The concentration of these regions was higher at the surface than at the bulk, which resulted in the extensive precipitation of γ" disks at the edge of the bolt, where their size surpassed even the 100-nm limit ( Figure 17, Figure 18); the cold rolling of threads after the heat treatment resulted in sheared oversized 50-nm γ' precipitates at the surface/subsurface of the bolt. The structure difference between the two phases might also have an effect on their interaction with hydrogen atoms; however, this needs to be resolved as a future work by observing the PLC effect of 718 specimens with a desired microstructure (S1, S2, S3, S4, S5, and bolt specimens from the bulk and subsurface).  [17] where the embryos of nanocracks form. Cassagne et al. [12] have postulated the presence of HEDE intergranular cracking at the grain boundary. However, until now, no attention has been paid to characterizing the individual role of nanoprecipitates at subsurface regions for crack in-itiation. Whereas, the hydrogen-assisted surface/subsurface cracking equation for API 718 subsea bolts (cold rolling of threads after the heat treatment) can be written as: CP under service loading + nanoscale oversized γ' precipitates + metastable γ" precipitates + intergranular precipitates (if any) = hydrogen-assisted surface cracking. The morphological data from this study will act as a reference in the future for observing the PLC effect in CRA 718 bolts at subsea temperature and under cathodic hydrogen charging. Therefore, this study includes some non-standard heating parameters to include more 718 variants for extensive PLC comparison.