Study on failure evolution of 15CrMo steel used for aviation kerosene hydrogenation unit

In aviation kerosene hydrogenation unit, the 15CrMo steel with good resistance to hydrogen embrittlement performed obvious pitting corrosion with certain micro-cracks in H2/H2S environment. In this paper, the failure behavior of 15CrMo steel in H2/H2S environment had been studied using electrochemical method and microscopic morphology. The results showed that within 10 days, the 15CrMo steel performed good resistance with a smaller max pit depth (15.48 μm on the 10th day) in H2/H2S environment, proved by the passivation zone (−0.75 ∼ −0.62 V on 10th day). Furthermore, 15CrMo steel exhibited activated corrosion with a significantly increasing max pit depth and corrosion current density, especially, reaching 131.91 μm and 17.29 μA m−2 on 30th day. Moreover, the stress corrosion cracking caused by H2S appeared on 30th day.


Introduction
Aviation kerosene hydrogenation technology was an important method for improving oil quality [1][2][3][4]. The process of hydrogenation was to transform the impurities especially for S and O in raw oil into the corresponding hydrocarbons such as H 2 S and H 2 O under certain pressure, temperature, reacting space velocity and hydrogen-oil ratio conditions [5].
However, the hydrogen embrittlement (HE) of materials was the key to restrict the development of aviation hydrogenation [6][7][8]. At present, the production equipment was mostly made by ferritic steel in body-centered cubic structure and austenitic stainless steel in face-centered cubic structure. Moreover, considering the higher HE sensitivity of ferritic steel and the lower strength of austenitic steel, duplex stainless steel (DSS) were put forward to be fully competent in the hydrogen environment [9,10]. For example, Zakroczymski et al [11] found that HE of DSS was closely related to hydrogen concentration, and the brittleness index decreased exponentially as the average hydrogen concentration in 23Cr-5Ni-3Mo steel gradually decreased, indicating that DSS showed great potential to reduce the HE sensitivity by reducing local hydrogen concentration.
The HE resistance of metals can be improved by grain refinement [12]. Treating 31Mn-3A1-3Si twinninginduced plasticity (TWIP) steel by severe cold-rolling (SCR) and annealing treatment (AT), Bai et al [13] obtained grain samples with the size of 19, 1.5 and 0.58 μm and found that HE decreased with the decrease of grain size. Focusing on Fe-16Cr-l0Ni steel, Macadre et al [14] prepared six kinds of grain size using SCR and AT, and found that when the grain size was >6 μm, the fracture presented a mixed mode of dimply, quasi-cleavage (QC) fracture and intergranular (IG) fracture, but when the grain size was <6 μm, HE could be ignored, indicating the existence of critical grain size, lower than which could ensure the uniformity of plastic deformation of steel. In addition, Cr, Ni and other elements [15,16] played a significant role in inhibiting HE. For example, Yang et al [17] found that Ni atoms could decrease the free surface energy and the stacking fault energy simultaneously. The findings suggested that a low concentration of Ni might degrade the physical property of Fe-Ni alloy, and the increasing the Ni atomic concentration above specific critical values, e.g., 0.03 or 0.05, could enhance the fracture toughness.
Therefore, 06Cr18Ni11Ti and 15CrMo were used as main materials to construct the aviation kerosene hydrogenation unit. As well-known, the 15CrMo steel was a pearlite-structure heat-resistant steel with high thermal strength and oxidation resistance, and performed a good ability to resist HE. However, the 15CrMo steel showed obvious pitting corrosion with certain micro-cracks during regular inspection after running more than two years under the conditions proposed in this paper. It was supposed that the interaction between H 2 S and H 2 may be the main cause of serious failure of 15CrMo steel because the high concentration of H 2 S was found. Therefore, this paper adopted accelerated test to simulate the working environment of aviation kerosene hydrogenation unit to analyze the failure behavior of 15CrMo steel.

Methods and materials
Focusing on the on-site reaction conditions, the reaction temperature was 231°C-246°C, pressure was 4.0-5.0 MPa, hydrogen partial pressure was 3.0-4.0 MPa, volume space velocity was 4 h −1 , and hydrogen-oil ratio was largely more than 50 (V/V). The purity of the pumped hydrogen was 98.1%, and the remaining of 1.9% was CH 4 as the main component while the other impurity gas can be ignored. The molar ratio of H 2 to H 2 S detected at the corrosion position of 15CrMo steel was 3.2:1.
In this paper, analytically pure Na 2 SO 4 and deionized water were adopted to prepare the Na 2 SO 4 solution with a mass fraction of 3%, aiming to simulating the corrosion environment for the accelerated experiment. Corrosion tests were carried out according to ASTM G2/G2M-19. A static high-pressure reactor made by 316 stainless steel with a volume of 5 l was used for corrosion test at 240°C in Na 2 SO 4 solution with a pressure of 5.0 MPa. The experiment period was 30 days. According to the above ASTM standard, at the beginning of the test, when the temperature in reactor rose to 150°C, the valve was opened to discharge a certain amount of water vapor and the remaining air in the reactor. This de-oxygenation operation can reduce the dissolved oxygen content in the solution to be less than 40 μg l −1 . A mixture of H 2 /H 2 S gas with a molar ratio of 3.2:1 was continuously injected from the bottom of the reactor, and the pressure in the reactor was controlled not to exceed 5.0 MPa through the upper vent valve.
Before the corrosion experiment, the 15CrMo steel was made into 25 × 25 × 2 mm 3 samples (the main components were shown in the table 1) by wire cutting. After being successively cleaned by acetone, deionized water and anhydrous ethanol, the samples were dried and then polished by 800#−1200# sandpaper.
The experiment was carried out as following, as shown in figure 1. (1) Under different experiment time during 30 days, the 15CrMo steel was taken out to remove the surface corrosion products, and then the corrosion morphology with max pit depth was analyzed by Zeiss 3D microscope. Furthermore, the corrosion morphology was binarized to determine the pit size and crack at different times, which would reveal the failure characteristics of 15CrMo steel in H 2 /H 2 S environment from a microscopic perspective. (2) The electrochemical curves, including open-circuit potential, polarization curves and electrochemical impedance spectroscopy (EIS) of 15CrMo steel with corrosion products under different experiment time were measured by three electrode method, in which the scanning rate of polarization curve was 0.1 mV s −1 , and the scanning range of EIS frequency was 10 -2 -10 5 Hz. In this stage, the corrosion dynamic process of 15CrMo steel in H 2 /H 2 S environment could be described. (3) For the corrosion areas of pits, cracks and intact corrosion product layers, the distribution of elements in different areas was carried out. Therefore, the failure mechanism was defined by combining microscopic morphology and electrochemical characteristics. Figure 2 showed the variation of pitting morphology on the surface of 15CrMo steel under different experiment time, and figure 3 showed the max pit depth. As can be seen from figure 2(a), in the accelerated experiment environment, obvious pitting features appeared on the surface of 15CrMo steel on 1st day, with a smaller max pit depth, only 3.58 μm. As the experiment continued to carry out, the size of the pitting pit increased, while there were dense pits appearing on the surface of 15CrMo steel, for example, on the 5th day. However, during this period, the max pit depth of these pitting pits increased to a small extent, only 15.48 μm on the 10th day with a larger size of the pits. Furthermore, the max pit depth increased rapidly, enlarging 265.76% on the 15th day comparing with that on the 10th day, and then reached 131.91 μm on the 30th day. Therefore, during 15-30 days of this experiment, the pitting pits of 15CrMo steel were more obvious, and continued to develop, showing that the corrosion degree was seriously deepened [18][19][20][21].

Development of corrosion pits
In order to further analyze the distribution of pitting pits on the surface of 15CrMo steel within 15-30 days, binarization treatment [22] was carried out as shown in figure 2(b). The light colored area was represented which was not corroded or generally corroded, while the dark colored area was focused on the pits. As the experiment went on during 15-30 days, the number of pitting pits increased gradually, but the area of single pitting pits was small. Especially when the experiment reached the 30th day, the number and depth of pits on the surface of 15CrMo steel increased significantly, and cracks were appeared on the surface [23]. Figure 4 showed the open-circuit potential and polarization curve including the fitting parameters of 15CrMo steel under different experiment time.

Electrochemical analysis
Generally speaking, with the increase of experiment time, the potential (corrosion potential, E corr , and opencircuit potential, E OCP ) and corrosion current density (I corr ) gradually increased, indicating that the corrosion  rate of 15CrMo steel was accelerated [24]. In detail, the polarization curves of 15CrMo steel on 1st and 3rd day basically coincided, and there was an obvious passivation region in the anode segment of these polarization curves, in which the passivation potential and over-passivation potential were −0.90 V and −0.65 V, respectively [25][26][27]. Furthermore, the polarization curve still showed obvious anode passivation zone on the 10th day, as shown in figure 4(b). However, the passivation potential and over-passivation potential both performed significant positive deviation, indicating −0.75 V and −0.621 V, respectively. Under this circumstance within 10 days, 15CrMo steel still performed good characteristic of anti-corrosion, as shown by a relatively smaller max pit depth on 10th day in figure 3(a). However, compared with E OCP on 3rd day, the E OCP  showed a larger positive deviation on 10th day, reaching 140 mV. Meanwhile, the passivation current density and over-passivation potential density increased 8.46 times (1.75 × 10 −7 A m −2 on 3rd day and 1.48 × 10 −6 A m −2 on 10th day) and 5.72 times (1.75 × 10 −7 A m −2 on 3rd day and 1.48 × 10 −6 A m −2 on 10th day), respectively, indicating that the anti-corrosion effect of 15CrMo steel was sharply weakened [28][29][30][31]. Therefore, when the experiment was lasting to 20th day, the polarization curve of 15CrMo steel showed complete activated corrosion characteristics, and the passivation zone thoroughly disappeared. Furthermore, the polarization curve moved significantly to the right on 30th day, suggesting that the corrosion current density increased significantly.
In conclusion, it should be noted that during the experiment period of 1-10 days, the potential was significantly shifted positively, indicating that the general corrosion site was dominating [32], while the potential  remained basically unchanged as shown in figure 3(c), indicating that pitting was the main corrosion [33] during 10-30 days. Figure 5 Showed the EIS curves and fitting parameters of 15CrMo steel under different experiment time.
In the accelerated experiment environment, due to the passivation characteristics of 15CrMo steel, Nyquist diagram showed complete reactance characteristics on 1st day. With the development of the experiment, the shape of the Nyquist diagram on 3rd day basically remained unchanged, but the radius decreased. Considering Bode plots and passivation characteristic of 15CrMo steel comprehensively, the equivalent electric circuit was chosen as R s (Q(R p (C dl R ct )), as shown in figure 6(a), where R s was solution resistance, Q was constant phase angle element (including the capacitance Q-Y 0 and index n), R p was surface resistance, C dl was double-layer capacitor, and R ct was charge-transfer resistance. As experiment carried out, pitting pits gradually occurred on the surface of 15CrMo steel, destroying the passivation layer, leading to the diffusion of corrosion products. Therefore, Warburg diffusion impedance appeared in the Nyquist diagram on 10th day, and a larger phase angle appeared at f = 10 −2 Hz in the logf-j curve, which was represented by mass diffusion. Therefore, the equivalent circuit was chosen as R s (Q(R p W(C dl R ct ))), as shown in figure 6(b), where W was Warburg diffusion impedance.
Due to the higher content of Cr and other alloying elements in 15CrMo steel, passivation occurred according to the polarization curve, and passivation layer such as CrO 3 was formed on the surface of 15CrMo steel to inhibit the corrosion of metal matrix [34,35]. Therefore, for example, the capacitances (Q-Y 0 = 2.581 × 10 −5 F cm −2 and C dl = 2.136 × 10 −7 F cm −2 ) were smaller and the resistances (R p = 19.2 Ω·cm 2 and R ct = 6804 Ω·cm 2 ) were larger on 1st day. Furthermore, the destroyed passivation layer diffused, indicated by that The frequency ( f θ-max ) of the max phase angle (θ max ) shifted to the intermediate frequency, i.e., f θ-max = 2043.36 Hz on 1st day and f θ-max = 13.74 Hz on 10th day, while the mass diffusion resistance (|Z 0.01Hz |) decreased from |Z 0.01Hz | = 269.46 Ω to |Z 0.01Hz | = 160.27 Ω, reaching the minimum [36,37]. This suggested that this stage of 1-10 days was mainly the failure process of passivation layer was dominating, so the surface resistance (R p = 10.38 Ω·cm 2 ) and charge-transfer resistance (R ct = 140.9 Ω·cm 2 ) also reached the minimum. With the progressing of the experiment, the max phase angle decreased, and the frequency of the maximum phase angle shifted to low frequency, which was θ max = 48.6°at f θ-max = 1.48 Hz on 30th day. Moreover, from the 10th day of the experiment, mass diffusion impedance appeared at the position of f = 10 −2 Hz, indicating that the surface of 15CrMo steel at this time was mainly activated corrosion, leading to the continuous increase of capacitance (Q-Y 0 and C dl ) and a smaller charge-transfer resistance. However, due to the pitting characteristics, the accumulation of dense corrosion product layer at local locations on the surface led to the increase of surface resistance. Therefore, the main reason for the development of pitting corrosion of 15CrMo steel in H 2 /H 2 S environment was that the resistance formed by passivation film was destroyed and the electrochemical process was intensified [38].

Failure mechanism
As well-known, severe stress corrosion cracking of low alloy steel and austenitic stainless steel will occur under wet H 2 S condition, and the crack size was much larger than that of hydrogen crack appeared in pure hydrogen environment [39][40][41]. In this paper, focusing on the 15CrMo steel in H 2 /H 2 S environment, both the corrosion morphology and electrochemical experimental results showed the incubation and development process of pitting corrosion, until serious stress corrosion cracks appeared on 30th day.
In order to further analyze the failure process of 15CrMo steel in H 2 /H 2 S environment, the microscopic morphology of corrosion product after 30 days of the experiment was analyzed, as well as the number of different atoms, as shown in figure 7. It can be seen that in zone (a), there was lots of cracks on the surface of corrosion product, which greatly reduced the protective effect. Subsequently, Zone (b) was characterized by severe pitting corrosion, while zone (c) was characterized by a dense corrosion product layer that provided better protection to the matrix. Atom distribution scanning was carried out for the position shown in figure 7(1), and the results were shown in figure 7(2). It can be seen that the number of Cr, Mo and O in zone (a) and zone (c) was larger than that in zone (b), i.e., pitting pit, which indicated that corrosion product containing Cr and Mo formed on the sample surface, such as CrO 3 and MoO 3 , can protect the matrix and slow down corrosion. Meanwhile, the S content in zone (c) was much lower than that in zone (a). As shown in figure 8, the microstructure of 15CrMo steel in high-pressure hydrogen environment, was mainly dominated by intergranular cracking [42][43][44], which crack size was much smaller than that of hydrogen crack appeared in H 2 /H 2 S environment studied in this paper. Furthermore, S content in the cracking position shown in figure 7 was relatively higher, indicating that the failure of 15CrMo steel was mainly caused by the stress corrosion cracking of H 2 S.

Conclusions
In this paper, the failure behavior of 15CrMo steel in H 2 /H 2 S environment had been studied using electrochemical method and microscopic morphology. The following conclusions were drawn.
(1) In H 2 /H 2 S environment, the dense corrosion product layer formed by Cr and Mo performed a good protection on 15CrMo steel. Under this circumstance, the pitting corrosion mainly occurred, and the max pit depth gradually increased. Furthermore, the stress corrosion cracking caused by H 2 S appeared on 30th day.
(2) Focusing on the electrochemical test, passivation zone existed in the polarization curve of 15CrMo steel on 1-10 days, leading to a smaller max pit depth. After 10 days, 15CrMo steel exhibited activated corrosion with a significantly increasing max pit depth, in which the corrosion current density and capacitance increased continuously, as well as the charge transfer resistance decreased. However, because pitting corrosion was dominant at this time, the potential basically remains unchanged.
15CrMo steel as a pearlite-structure heat-resistant steel also showed good hydrogen brittleness resistance, as well as corrosion resistance. However, it was interesting that severe pitting with cracks occurred on 15CrMo steel in H 2 /H 2 S environments studied in this paper. The anti-corrosion of 15CrMo steel was mainly provided by the dense corrosion product layer on the metal surface. Under the action of S element, the compact product layer was destroyed, and the interaction of H 2 /H 2 S with the metal matrix led to serious pitting and cracking, and the crack size was much larger than that in the pure H 2 environment. Grain refinement and addition of effective elements were always been the key method to increase the hydrogen brittleness resistance of materials. More importantly, in complex miscible environment, improving the stability of passivation layer or dense corrosion product layer on metal surface should also be a concern.