Vigorous development of plastic mold market has significantly promoted the researches of pre-hardened mold steels with AISI P20 (America), DIN 1.2738 (Germany) and 718H (Swedish) [1,2]. The 718H series are known as pre-hardened mold steels because they are supplied after quenching and tempering (QT) processes from steelmaking plants to mold manufacturers, without additional heat treatment after processing [3].

During the electric furnace smelting process of 718H steel, aluminum block or aluminum powder is mainly used for pre-deoxidation, which results in the formation of Al₂O₃ inclusions inevitably. The hardness of Al₂O₃ inclusion is higher than that of steel matrix [4], which not only causes nozzle clogging during ingot casting, but also adversely affects the subsequent hardness uniformity and mirror polishing of the module. Researchers indicated that oxide inclusions such as large-sized Al₂O₃ were the main cause of fatigue failure of special steels [[5], [6], [7], [8]]. In addition, due to the segregation of chemical composition of ingots, large-sized MnS inclusions are easily formed during solidification and are difficult to eliminate. Liu et al. [9] demonstrated that the existence of MnS inclusions in steel promoted the nucleation of voids during deformation, resulting in the occurrence of quasi-cleavage fracture and failure of materials.

It is well known that RE elements are effective additives to improve the properties of alloys and composites. Many scholars have studied the influence of RE elements on microstructure evolution, mechanical properties, and deformation behavior of aluminium alloy [10], magnesium alloy [[11], [12], [13], [14], [15], [16]], and composite material [17], and achieved good results. In addition, the advantages of adding RE in steel are also highly attractive. After years of efforts, our research group has proposed that the oxide inclusions are the key factors leading to segregation, and purification of oxygen can effectively inhibit segregation [18]. This discovery also greatly enriches the theoretical background of the application of RE-steel. The sizes and morphologies of inclusions changed significantly after adding Ce, La, Y and Ga in steel, and highly stable oxy-sulphides, oxides, sulphides were easy to form and precipitate as solid particles [[19], [20], [21], [22], [23]]. According to the theory of heterogeneous nucleation in solidification, the mismatch of Ce₂O₂S and δ-Fe is only 3.5%. Ce₂O₂S can become an effective nucleation core because of its good wettability and fine dispersibility in molten steel [24].

With the continuous improvement of the cleanliness of steel, the microalloying effect of RE in steel is becoming increasingly significant. Hamidzadeh et al. [25] demonstrated that the M₇C₃ carbides were refined and evenly distributed with RE addition in AISI D2 steel, resulting in increasing the impact toughness by 75%. Liu et al. [26] showed that the coarse eutectic structure and eutectic carbides of 4Cr5MoSiV1 mold steel were remarkably refined after adding RE alloys (Ce and La), thereby improving the yield strength. In addition, it is found that trace RE atoms are easily enriched at grain boundaries or phase boundaries, resulting in improving the cohesion of grain boundaries to a certain extent [[27], [28], [29], [30], [31]]. The comprehensive effects of RE elements in steel can improve the impact toughness, hot ductility and wear resistance properties, etc [24,27,[31], [32], [33], [34]].

In our another paper, we displayed attractive purification and micro-alloying effects of RE in a bearing steel and a plain steel, and discussed the mechanisms behind [35]. Based on these new findings, we try to improve the mechanical properties of 718H mold steel by using the purification and micro-alloying effects of RE. However, the effects of RE elements on phase transformation behavior and inclusions evolution has not been clearly discussed of 718H mold steel. Simultaneously, the works concerning the effect of RE elements on impact toughness, strength and fatigue crack growth rate of 718H steel was also controversial and in confusion.

It will be beneficial to 718H mold steel if the mechanical properties can be improved on the basis of optimizing the microstructures by appropriate addition of RE elements. The purpose of this paper is to investigate the influences of RE contents on the microstructure evolution and mechanical properties of 718H mold steel systematically, and to further analyze the mechanisms.

Four cast ingots with RE content ranging from 0 to 0.07 wt% were manufactured by vacuum induction furnace with a capacity of 25 kg, and the chemical compositions is listed in Table 1. RE alloys containing 70 wt% Ce and 30 wt% La elements were added to molten steel in vacuum atmosphere. RE alloys were measured by inductively coupled plasma method (ICP), and Si, Mn, Ni, Cr, P, and Mo were measured by direct-reading spectrometer. The contents of C and S elements were measured by infrared carbon and sulfur analyzer. In addition, the content of total oxygen was detected by TCH600 analyzer according to GB/T11261-2006.

Four cast ingots were forged in the temperature range of 950 °C-1150 °C, with the forging size of 65 mm × 65 mm × 720 mm. The heat treatment process includes normalizing at 870 °C for 2 h, and oil quenching after holding at 860 °C for 1 h. Followed by tempering at 600 °C for 2 h.

The local chemical composition of 0.07RE steel was analyzed by secondary ion mass spectrometry (ToF-SIMS 5), which has high quality resolution and sensitivity. SIMS was performed with Bi+ as the ion source at 30 keV and 1.0 pA, and the area of 300 μm × 300 μm was investigated. SIMS samples were prepared by ion etching to reduce the contamination rate and the residual stress.

The morphology and composition of inclusions were analyzed by scanning electron microscope (SEM) and energy dispersive spectroscope (EDS). The size of inclusions and queched grains were calculated by using Image-Pro Plus software (IPP 6.0). The austenite grains were etched by immersion in saturated picric-acid solution. In this study, 20 metallographic images (500×) were selected randomly to analyze the size and the distribution of inclusions and quenched grains with different additions of RE elements. In addition, the spatial distribution of inclusions can be accurately observed by three-dimensional (3D) X-ray microtomography technique, which is based on the absorption contrast between matrix and inclusions. Fracture analysis for impact and tensile samples were performed by SEM and EDS.

Continuous cooling transformation curves (CCT) of 0RE and 0.07RE steels were measured in inert atmosphere (high purity argon) by induction heating thermal expansion instrument (Linseis L78RITA). Samples with size of ϕ3 mm × 10 mm were continuously heated to 900 °C at a rate of 2 °C/s and maintained for 900 s, then cooled to room temperature at a rate of 0.01, 0.02, 0.05, 0.1, 0.2, 0.5, 1, 2, 5, and 10 °C/s, respectively.

Samples with size of 55 mm × 10 mm × 10 mm were tested to investigate the evolution of the impact energy of test steels with different additions of RE elements. The U-notches of the impact samples are parallel and perpendicular to the forging direction, respectively. Impact test was measured using a pendulum-type impact testing machine (RKP450). At least five impact tests for each testing condition are adopted for the average value. Simultaneously, tensile samples were prepared along the forging direction with the diameters and gage lengths of 5 mm and 25 mm, respectively. Tensile test was measured using AG-100KNG tensile machine with a strain rate of 0.5 mm/min. Tensile tests are determined by means of the average values of four measurements at each parameter at room temperature. In addition, the fatigue crack growth (FCG) tests were carried out on MTS-810 fatigue testing machine at 300 °C. Specific parameters of FCG tests as described in Ref. [36]. Finally, tempered samples containing 0.022 wt% RE were tested by in-situ tensile table loaded by SEM at room temperature. The strain rate is 0.05 mm/min, which is controlled by displacement. Strain rate is continuously loaded until the samples breaks.

The field of view of 300 μm × 300 μm area was analyzed by SIMS to examine the distribution of Ce and La in 0.07RE steel, and the result is shown in Fig. 1. It indicates that the element distributions of Cr+, Mn+, Ni+, S-, CeO+, Ce+, La+, LaO+, and C- are clearly displayed. According to SIMS data, the aggregation behavior of Ce+, La+ and S- is observed in test steel. Further observation indicates that the stacking sequence of CeO+ and LaO+ signals is also visible in the same region. Meanwhile, the alloying elements (Cr+, Mn+, Ni+) are in a state of disappearance in this region. The result confirms the main form of RE elements in test steel probably is the RE-inclusions (RE-O-S).

In addition, Pan et al. [37] clarified that the solubility of RE alloys in steel is around 10^-6-10^-5 ppm based on the electrolysis and plasma mass spectrometry of RE-inclusions. In Fig. 1, there is no signal aggregation of Ce+ and La+ was separated except for RE-inclusions, which confirms that the existence of solid solution RE is extremely low in test steel.

More in-depth and detailed characterization is needed to further explore the types of RE-inclusions. Fig. 2 shows the variation of total oxygen ([O]) in test steels with different additions of RE elements. The resTult indicates that the T.O content basically shows a continuous decline trend from 15 ppm (0RE) to 6 ppm (0.022RE) with the increase of RE contents, which is mainly attributed to the active chemical properties of RE elements and innovative method of adding RE elements in molten steel.

In addition, the relationship between RE contents and average diameter as well as size distribution of inclusions is quantitatively calculated, as shown in Fig. 3. It indicates that the average equivalent diameter of inclusions decreases initially and then increases with different additions of RE elements. The average diameters are 5.42 μm, 2.97 μm, 3.10 μm and 7.18 μm, respectively. Fig. 3(b) indicates that compared with that of 0RE steel, the proportion of large-sized inclusions with diameter exceeding 10 μm obviously decreases by 11.5%, while the proportion of tiny inclusions with diameter less than 2 μm significantly increases by 34% and 41.8% in 0.012RE and 0.022RE steel, respectively. However, when the RE content reaches 0.07 wt%, the large-sized inclusions increase by 11.4%.

Fig. 4 shows the distribution of inclusions using 3D X-ray microtomography, in which equivalent volume diameters of inclusions were represented by different colors. It clearly indicates that the uneven distribution of inclusions in steel without RE addition, and the size is generally deviation (Fig. 4(a)), which is consistent with the statistical result of Fig. 3(b). For test steels, it could be clearly observed that the strip inclusions have a certain extent of reduction, and the RE-inclusions are generally tiny in size and evenly distributed by adding 0.012 wt% and 0.022 wt% RE. In addition, the maximum diameter of inclusions decreases from 38 μm to 10-11 μm within the same statistical range compared with that of 0RE steel, as shown in Fig. 4(a-c). However, when the RE content reaches 0.07 wt%, the number and size of inclusions increase significantly compared with that of 0.012RE and 0.022RE steels (Fig. 4(b-d)).

In summary, tiny inclusions can be obtained by adding a certain content of RE in steel, but when RE content exceeds the optimum value, inclusions will coarsen. For 718H steel, the optimal addition of RE alloys is considered to be 0.012-0.022 wt% in terms of size and distribution of inclusions.

Characterizations of inclusions were performed in polished samples by SEM with EDS, and the results are shown in Fig. 5, Fig. 6. Fig. 5 shows the typical SEM micrographs of inclusions in steel without RE addition. It indicates that the maximum length of irregular inclusions could reach 30 μm. EDS analysis shows that the inclusions are composite Al₂O₃+MnS (Fig. 5(a)) and MnS inclusions (Fig. 5(b)), respectively. Fig. 6(a) and (b) represents the SEM micrographs of inclusions in 0.012RE steel, and two kinds of inclusions were mainly observed by EDS analysis: ellipsoid RE-O-S-Al (Fig. 6(a)) and irregular RE-O-MnS (Fig. 6(b)). In addition, the SEM micrographs of inclusions in 0.022RE steel are shown in Fig. 6(c) and (d). Two kinds of inclusions were mainly observed: composite inclusion consisting of ellipsoid RE-S, RE-O (Fig. 6(c)) and ellipsoid RE-O-S (Fig. 6(d)). Fig. 6(e) and (f) represents the SEM micrographs of inclusions with 0.07 wt% RE addition. The chemical composition of inclusions includes RE, S and O elements. These kinds of inclusions are considered to be irregular RE-O-S (Fig. 6(f)) and RE-S (Fig. 6(e)). The probability of collision between inclusions increases, and the growth form of inclusions gradually occurs from precipitation to polymerization with the increase of RE content.

Inclusion evolution by RE treated can be inferred as the RE-O, RE-O-S, and RE-S appearing in sequence by calculating the Gibbs free energy of inclusions [38,39]. RE-O-S not only inhibits the nucleation of strip MnS inclusions, but also contributes to modify Al₂O₃ inclusions. Simultaneously, RE-O and RE-O-S can be used as heterogeneous nucleation cores, which greatly improve the nucleation rate and the degree of undercooling during solidification [24,40]. In addition, the S atom of MnS would be replaced by RE-S because of the more negative standard free energy than MnS inclusion [41]. RE-S would precipitate in the form of RE₂S₃ to RE₃S₄ and further precipitate into RES with the decrease of S content in steel.

In our study, the evolution of inclusions during solidification with different RE addition was calculated using FactSage 6.4, and the results are shown in Fig. 7. Ce element was selected to instead of RE (Ce and La) alloys for calculation according to previous research by our group [2]. It indicates that the MnS and Al₂O₃ are the main inclusions in 0RE steel during solidification, which is consistent with the EDS data in Fig. 5. Typical morphologies of composite inclusions consist of MnS around the center of Al₂O₃. According to the calculations by first-principles, α-Al₂O₃ is beneficial to capture the S^2- and combine with Mn²⁺ to initially nucleate MnS because of the strong electron hybridizations between S and its nearest neighbor O and Al on the surface [18]. Ce-inclusions begin to appear in molten steel with 0.012 wt% Ce addition. There are four kinds of inclusions mainly appears: Ce₂O₂S, Ce₂O₃, CeAlO₃ and MnS, and the precipitation temperature of the inclusions is 1550 °C, over 1600 °C, 1500 °C and 1300 °C, respectively. In 0.022Ce steel, MnS inclusions disappear because of the strong desulfurization ability of Ce element. Ce₂O₂S, CeS, and Ce₂O₃ were found in test steel during the solidification. When the Ce content reaches 0.07 wt%, Ce₂O₃ disappears and the inclusions mainly exist in the form of Ce₂O₂S and CeS.

In our study, solidification is relatively rapid during the cooling process. It is believed that the inclusions formed in molten steel should be mostly maintained without significant transformation during solidification. Fig. 7(e) is equilibrium phase diagram of the test steel without RE addition, which is calculated by Thermo-Calc software. The results showed that the solidus temperature (T_S) is about 1438 °C. Because of the high melting point of RE-inclusions, the inclusions precipitated above the solidus temperature (1438-1600 °C) are retained. Combined with the SEM observation and thermodynamic calculations of complex inclusions, a summary of the relationship between amount of RE contents and types of formed inclusions is shown in Table 2. The result indicates that the thermodynamic calculation of inclusions coincides well with the actual precipitation sequences observed by SEM and EDS in all test steels.

CCT diagrams were measured of 0RE and 0.07RE steels to study the effect of RE addition on non-isothermal transformation, and the result is shown in Fig. 8. In this study, the effect of inclusions on the nucleation of the phase transformation was not considered. It indicates that the CCT diagram of 0RE steel shows the continuous characteristics of complex microstructures and phase transformation with the increase of cooling rate from 10 to 0.01 K/s. Pearlite, bainite and martensite are involved in Fig. 8(a-d). In addition, the shape of the CCT diagram of 0.07RE steel is similar to that of 0RE steel. The difference is that the addition of RE elements narrows the range of pearlite transformation and decreases the P_s (680 °C-650 °C at the cooling rate of 0.02°C/s) and P_f (609 °C-595 °C at the cooling rate of 0.02°C/s). Simultaneously, B_s (391 °C-380 °C at the cooling rate of 0.05°C/s) and B_f (275 °C-270 °C at the cooling rate of 0.05°C/s) also decreased after adding RE elements. In addition, it indicates that the transformation range of bainite of 0.07RE steel shifts to the bottom right and prolongs incubation period compared with that of 0RE steel.

Researchers have demonstrated that the radiuses of RE atoms are much larger than that of Fe atom. RE atoms are more likely distribute at the grain/phase boundaries, vacancies and other defects than to replace the normal lattices [42]. The occupation trend of La atoms in α-Fe was calculated by using plane wave pseudopotential method and density functional theory, and the result showed that La atoms were more easily enriched at grain/phase boundaries and decreased the interfacial energy [43]. Other experimental results also revealed that the fact of the enrichment of solid solution RE elements at grain boundaries [27,29,30], which not only decreases the diffusion coefficient of carbon atoms, but also improves the stability of undercooled austenite.

For test steels, the transformations of bainite and pearlite are affected by the diffusion of carbon atoms. Liang et al. [44] also showed that the segregation of RE atoms hindered the diffusion of carbon atoms and formed carbon-depleted regions, resulting in prolonging the incubation period of bainite transformation. In addition, Hsu et al. [45] and Yan et al. [42] also revealed that RE elements can prolong the pearlite transformation and reduce the lamellar spacing of pearlite.

Various investigations such as the FCG rate, tensile property, and impact energy were carried out to evaluate the effect of RE elements on the mechanical properties of test steels. For 0RE, 0.012RE, 0.022RE and 0.07RE steels, the relationship between FCG rate and applied stress intensity factor range (da/dN-ΔK curves) was plotted in Fig. 9 when the stress ratio is 0.1. The curves are mainly the Second Stage of FCG, which can be expressed by the Paris equation [36,46]:

where da is the incremental length of crack, dN is the increment of cyclic pressure, $\frac{\text{d}a}{\text{d}N}$ represents the fatigue crack growth rate, ΔK is the increment of the each cycle crack length, c and m are empirical constants depending on test steel. The parameters of Paris equation of test steel can be obtained by fitting the data, as shown in Table 3.

The results indicate that the fitting data of FCG behavior has a strong linear relationship and a small data dispersion, with an exponent m ≈2.70-3.66. As a general trend, it can be noted that the higher the RE content (0.012 wt% and 0.022 wt%), the lower the crack propagation rate. However, when the RE content reaches 0.07 wt%, the inhibition ability of fatigue crack growth is decreased compared with that of other RE-steels, especially at high ΔK values.

There are a variety of microstructural factors that affect the FCG rate, consisiting of grain size, inclusion and precipitates. The relationship between RE contents and average diameter of grain size of quenched samples was quantitatively calculated. The results indicate that the grain size of test steels is stable at 20 ± 2.5 μm. Grain size has little effect on FCG rate in test steels. In addition, researchers show that the formation of second phase and the precipitation kinetics are always influenced by the addition of alloying elements [47]. In our study, the effect of precipitates can be ignored due to the extremely low solid solution content of RE (Fig. 1). Considering that there is no obvious difference between grain size and precipitates of test steels, the change of the FCG rate with RE addition is obviously related to the evolution of inclusions. Due to the inclusions frequently act as the preferred crack initiation sites, it has been confirmed that inclusions have the noticeable effects on the fatigue life of materials. In this paper, the significant influence of inclusions on FCG rate is that addition of RE (0.012 wt% and 0.022 wt%) can effectively refine inclusions by changing the morphology of inclusions from large-sized strip to tiny ellipsoid. This is consistent with the downward trend of index m (3.48-2.70). However, in 0.07RE steel, the formation of large number of large-sized inclusions increases the crack source, resulting in increasing the FCG rate (index m from 2.70 to 3.65).

Evolution of yield strength (YS), ultimate tensile strength (UTS), elongation and reduction of area of test steels in response to different contents of RE is shown in Fig. 10(a). The results display that the average values of YS and UTS are basically stable at 917 MPa and 1057 MPa with 0.012 wt% and 0.022 wt% RE addition. However, in 0.07RE steel, YS remained unchanged, but UTS decreased slightly to 1033 MPa. In addition, it is found that the elongation of RE-steels changed slightly compared with that of 0RE steel.

SEM micrographs of tensile fracture at low and high magnifications with different contents of RE addition are shown in Fig. 11. It is evident that the overall fracture morphologies and the area fractions of zone A and B of all samples are similar. The crack source is located in the center of the tensile specimen. The dimples of various sizes are evenly distributed in matrix by high magnification observation, which is typical ductile fracture. Inclusions are the nucleation core of dimples in the crack origin zone. EDS analysis shows that there are mainly composite Al₂O₃+MnS inclusions in 0RE steel, while RE-O-S and RE-S inclusions in 0.07RE steel. These types of inclusions are consistent with the EDS analysis in Fig. 5, Fig. 6.

It is considered that the reason why the UTS decrease in 0.07RE steel is related to the formation of large-sized inclusions. The crack initiation, propagation and shedding of inclusions in 0.022RE steel during tensile process were systematically studied by in-situ tensile test, and the relevant mechanism is shown in Fig. 12. The results display that the plastic deformation occurs in matrix and the cracks begins to initiate and aggregate at both ends of the inclusions when the load exceeds the YS. As a result of the continuing action of tensile stress, the internal and boundary cracks of RE inclusions occur. Finally, the inclusions are separated from the matrix, as shown in Fig. 12(b-d). At this time, the whole RE-inclusion can be regarded as the crack source, and the crack size is consistent with the maximum diameter of the inclusion. In summary, the larger the average diameter of inclusions, the faster the rate of crack propagation after plastic deformation. This is also one of the reasons why the UTS of 0.07RE decreased to a certain extent.

The impact energy of tempered samples was measured to investigate the influence of RE on impact toughness, as described in Fig. 10(b). The result demonstrates that the impact energy increases initially and then decreases with the increase of RE addition. In 0.012RE steel, the average value of transverse impact energy increases from 83.6 J to 95.8 J compared with that of 0RE steel. However, in 0.07RE steel, it decreases to 77.6 J. Similar varied phenomenon also occurs in the samples of the average value of longitudinal impact energy. However, instead of 0.012 wt% RE, it could achieve the maximum value of longitudinal impact energy in 0.022RE steel. Likely, 0.07 wt% RE addition achieves the minimum value of longitudinal impact energy.

In addition, isotropic property of 718H mold steel is the main property requirement and closely related to the service life. It can be expressed by the ratio of average value of transverse and longitudinal impact energy. The result demonstrates that the optimum isotropic property can reach approximate 0.89-0.92 with the 0.012 wt% and 0.022 wt% RE addition (Fig. 10(c)), which exceeds that of 0RE steel (0.83).

Impact fracture analysis of longitudinal samples were performed to display the morphology of impact fracture, as shown in Fig. 13. It demonstrates that all impact fractures are typical ductile fracture characterized by large amounts of dimples. The strip inclusions are evenly distributed in dimples of 0RE steel (Fig. 13(a)). Strip inclusions, whether MnS or MnS + Al₂O₃ inclusions, in which central axis is perpendicular or parallel to the fracture surface of longitudinal impact samples, resulting in the low isotropy property. However, in 0.012RE and 0.022RE steels, tiny ellipsoidal inclusions are distributed uniformly instead of strip inclusions (Fig. 13(b) and (c)). However, a large number of inclusions accumulated in the fracture area of matrix in 0.07RE steel by SEM observation (Fig. 13(d)).

The main reason for increasing the impact energy of 0.012RE and 0.022RE steels can be attributed to the purification of molten steel and the modification of inclusions [48]. Researchers indicated that the T.O content in steel has a negative effect on impact toughness [2]. With the decrease of T.O content, the number of crack sources caused by inclusions decreased obviously. In addition, the strip MnS and composite Al₂O₃+MnS inclusions are transformed into submicron ellipsoidal RE-inclusions by RE treatment, which can effectively reduce the stress concentration around inclusions during impact test and improve the isotropic property of test steel. Yang et al. [49] also revealed that the path of crack growth can be changed by RE-inclusions within a certain size range, resulting in improving the impact energy. Simultaneously, it was reported that the activity of tiny RE-inclusions is lower than that of many other large-sized inclusions in crack initiation and propagation [22]. Larger plastic deformation can be produced by RE-inclusions during impact fracture, which improves the toughness of materials [22]. However, there are lots of large-sized RE-inclusions with large aspect ratio in 0.07RE steel. The number of crack sources and the ability to promote crack growth increase significantly during impact test, resulting in the decrease of impact toughness.

(1) The T.O content of test steels decreases from 15 ppm to 6 ppm with 0.022 wt% RE elements addition, which is mainly attributed to the active chemical property of RE elements.

(2) Combined with the observation of complex inclusions and thermodynamic calculations, it can be inferred that the addition of RE elements contributes to modify irregular MnS and Al₂O₃ inclusions to ellipsoidal RE-inclusions. With the increase of RE content (0.012 wt% to 0.07 wt%), the formation sequence of RE-inclusions should be (RE₂O₃, RE₂O₂S and REAlO₃)→(RE₂O₂S, RES and RE₂O₃)→(RE₂O₂S and RES).

(3) MnS or MnS + Al₂O₃ inclusions have a central axis perpendicular or parallel to the fracture surface of impact samples. RE additions of 0.012 wt% and 0.022 wt% were significantly effective for refining inclusions by eliminating 11.5% large-sized inclusions with diameter exceeding 10 μm. The purification of molten steel and the modification of inclusions by RE treatment have a significant effect on the improvement of FCG inhibition ability and impact energy as well as the isotropy of the test steels.

(4) The transformation region of bainite and perlite of 0.07RE steel shifts to the bottom right and prolongs incubation period compared with that of 0RE steel. It is mainly attributed to the fact that trace solid solution RE atoms more easily enriched at grain/phase boundaries and decreased the interfacial energies. It not only decreases the diffusion coefficient of carbon atoms, but also improves the stability of undercooled austenite.

(5) Excessive RE addition (0.07 wt%) seriously deteriorates the impact energy, UTS and FCG inhibition ability due to rapidly increase of the volume fraction of large-sized inclusions. Considering the evolution of inclusions and mechanical properties, the optimum RE content of 718H steel should be controlled from 0.012 wt% to 0.022 wt%.

Hanghang Liu carried out the experiments and analysis of the results. Paixian Fu, Hongwei Liu, Chen Sun, and Ningyu Du prepared and revised the original manuscript. Yanfei Cao gave great help in revising the marked up manuscript. Dianzhong Li contributed in the interpretation and discussion of the results.