- Email: [email protected]

Contents lists available at ScienceDirect

Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc

Full length article

The mechanism of surface nanocrystallization during plasma nitriding J.W. Yao a b

a,1

a,1

, F.Y. Yan

a,⁎

a

a

b

, M.F. Yan , Y.X. Zhang , D.M. Huang , Y.M. Xu

T

National Key Laboratory for Precision Hot Processing of Metals, School of Materials Science and Engineering, Harbin Institute of Technology, Harbin 150001, China Beijing Research Institute of Mechanical & Electrical Technology, Beijing 100083, China

ARTICLE INFO

ABSTRACT

Keywords: Plasma nitriding Spinodal decomposition Pseudo-binary solution model Nanocrystallization

Motivated by the observation of in-situ nanocrystallization in the surface layer of plasma nitrided steels, this work is carried out to establish a thermodynamic model that describes the Gibbs free energy of BCC-Fe solution using the quasi-binary solution model. The model is specifically developed for the nitriding process of the multicomponent steels, whose composition is first converted to an Fe-Creq-N ternary system using the concept of Cr equivalent through the relative elemental electronegativities with respect to nitrogen. With the thermodynamic model, the limit of stability of the BCC pseudo binary solution can be evaluated with temperatures and used to guide the selection of nitriding conditions. It has been recognized that the nitrogen-containing martensite is unstable and likely to decompose in the form of spinodal decomposition over a specific compositional/temperature range. 40CrNi steel is adopted in this work and plasma nitrided at 800 K for 8 h. Nano-scale FeN1/12 phase and high‑nitrogen martensite phase have been experimentally observed in the nitrided surface layer, which embraces the thermodynamic predictions in terms of nanocrystallization via spinodal decomposition.

1. Introduction Nitriding is an effective thermochemical treatment process for case hardening of load-bearing parts, by diffusing nitrogen atoms to the surface to form a hard compound layer (or called “white layer”) and a diffusion zone. For ferrous alloys, after the nitriding process, the surface compound layer is typically comprised of iron nitrides, i.e., γ′-Fe4N and ε-Fe2-3N [1]. With this compound surface layer, enhanced corrosion and wear resistance can be achieved. While extremely hard, the nitride layer becomes incredibly brittle as the layer thickness increases [2,3]. Therefore, there exists a hardness-ductility trade-off when controlling the steel nitriding conditions. Grain refinement has been well acknowledged as an effective metallurgical solution to this trade-off in metallic materials, and an ultra-high performance has been discovered when grain size is refined to nano-scale [4]. The most common method for grain refinement is to apply mechanical work to the materials to introduce severe plastic deformation to trigger recrystallization as the stored energy released at elevated temperatures. In terms of surface engineering, Lu et al. [5] has successfully used the surface mechanical attrition treatment (SMAT) method to synthesize a nano-structured surface layer on various metallic materials as large amount of defects are repetitively generated. By combining the SMAT method and the nitriding treatment, they are able

to promote the nitriding kinetics over a nano-structured surface layer, as the defects introduced by the SMAT method provide more nitrogen diffusion paths [6]. At the same time, some studies [7–10] have observed in-situ nanocrystallization in the surface layer of some quenched bearing steels after low-temperature plasma nitriding without prior surface deformation. And these surfaces are associated with superior wear resistance and fatigue life to conventionally nitrided counterparts. Structural characterizations of the nanocrystallized surface layer reveal the presence of nano-scale FeN0.076 phase and high‑nitrogen martensite α′N phase, instead of coarse γ′ and ε phases as observed in conventional white layers. In addition, the occurrence of nanocrystallization phenomenon is found to be strongly dependent on nitriding temperatures, as shown in Fig. 1, where nanocrystallization is only observed when M50 steel is plasma-nitrided at a lower temperature (500 °C) [7]. However, the mechanism for in-situ nanocrystallization phenomenon during nitriding has rarely been discussed. For N-containing super-hard coatings, such as ZrN-AlN, it has been proposed that it is spinodal decomposition that refines the microstructures to nano-scale [11,12]. Based on these, this work aims to develop a thermodynamic model to describe the N-containing BCC-Fe system, so that a thorough investigation on the nature of in-situ nanocrystallization during nitriding can be made by evaluating the thermodynamics of the solution system and that the selection of nitriding conditions can be further guided.

Corresponding author. E-mail address: [email protected] (M.F. Yan). 1 J.W. Yao and F.Y. Yan contributed equally to this work. ⁎

https://doi.org/10.1016/j.apsusc.2019.05.164 Received 29 December 2018; Received in revised form 17 April 2019; Accepted 14 May 2019 Available online 15 May 2019 0169-4332/ © 2019 Harbin Institute of Technology. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Applied Surface Science 488 (2019) 462–467

J.W. Yao, et al.

Fig. 1. (a) XRD pattern of the surface layer of M50 steel in the un-nitrided and nitrided conditions. (b) Phase (bright field, left) and structure (selected area diffraction (SAD), right) analysis of M50 steel nitrided at 540 °C when nanocrystallization occurs [7].

In this work, a formula to calculate equivalent Cr content is first proposed with respect to substitutional solute element electronegativities, such that the nitrided multi-component alloy can be treated as an Fe-Cr-N ternary solution for thermodynamic calculations. A thermodynamic model is then developed by postulating the system to be a pseudo-binary solution between a N-lean BCC-Fe phase and a Nrich martensite phase. GB/T 40CrNi steel is taken as an example material system to illustrate the development of the thermodynamic model and to evaluate the stability of the solution with temperatures. To support the thermodynamic predictions for the occurrence of nanocrystallizaton, 40CrNi sample steels are plasma nitrided under the calculated conditions and characterized by X-ray diffraction and transmission electron microscopy (TEM) techniques.

composition listed in Table 2. 2.2. Theory/calculation 2.2.1. The basis of thermodynamic model For a ternary solution with two substitutional and one interstitial solutes, (A,B)aCc, the molar Gibbs free energy can be described by the ideal solution of the end-members AaCc and BaCc at reference states plus an excess quantity (excess Gibbs energy) that represents the interaction between the two components and the deviation from the ideal solution [16], as in Eq. (2),

G (A,B)aCc = yA °G Aa Cc + yB °G Ba Cc + aRT (yA ln yA + yB ln yB ) + ayA yB LA,B:C (2)

2. Experiment

where LA,B:C refers to the parameter of interaction between AaCc and BaCc compounds, and yA, yB are site fractions of A, B atoms that are related to the atomic fractions of A and B atoms (xA, xB) through Eqs. (3a) and (3b).

2.1. Chromium equivalent For the sake of engineering applications, the concept of using the amount of one element to reflect the combined effects of different alloying elements has been widely adopted to simplify the predictions of properties of multi-component solutions, such as in ferrous alloys using C equivalent to calculate the hardenability of steel [13] and using the ratio of Cr equivalent and Ni equivalent to predict δ/γ solidification mode [14]. Due to strong interactions between nitrogen and substitutional solute atoms (typically 3d period of transition metals) in ferrous alloys, nitrogen solubility in Fe is determined by the alloying elements due to the effect of electronegativity [15]. Therefore, a Cr equivalent relationship can be developed based on elemental electronegativities that represent the affinity between substitutional solute elements and nitrogen. Electronegativity difference between solute element and nitrogen is scaled to that between Cr and N, as in Eq. (1),

k=

Me

N

Cr

N

Mo

V

N

Electronegativity

1.66

1.91

2.16

1.60

3.04

yB =

xB xA + xB

(3b)

(4)

At 0 K, which is taken as the reference state in this work, the excess Gibbs energy ayAyBLA,B:C is equal to the formation enthalpy of 1 mol of the ternary phase (∆HA, B:C0), as in Eq. (5). 0 HA,B:C = ayA yB LA,B:C (0K)

(5)

The interaction parameter at 0 K is therefore equal to ∆HA, B:C0/ ayAyB. The coefficient θ can then be calculated when the interaction parameter at a finite temperature is given. 2.2.2. Thermodynamic model for the Fe-Creq-N system The Fe-Creq-N ternary solution can be treated as a quasi-binary phase, i.e., a continuous solid solution of two compound phases with the same structure. With the compound model, the BCC phase can be described as (Fe,Cr)1(N,Va)3 [17]. Since only FeN0.076 (or FeN1/12) phase and high‑nitrogen martensite phase α′N are experimentally observed when nanorecrystallization occurs (Fig. 1), it is reasonable to postulate the end-members of the solution as BCC-structured FeN1/12 and α′NMmNn (M = Fe, Cr) phases. Since the average nitrogen content in the nitrided surface is tested to be around 13 at.% [18], and there is only

Table 1 Elemental electronegativity. Ni

(3a)

LA,B:C (T ) = LA,B:C (0K ) + T

where χN, χCr and χMe are electronegativities of N, Cr and alloying elements (Me = Ni, Mo, V, etc.). The values of elemental electronegativity are listed in Table 1. Cr equivalent (Creq) can thus be expressed by Σkixi (xi is the mole fraction of the element), and the multicomponent system is then converted to an Fe-Creq-N ternary system, which is later used for thermodynamic calculations. For 40CrNi alloy adopted in this work, Creq is equal to 1/24 with the nominal

Cr

xA xA + xB

As the excess Gibbs energy is directly subject to modeling, the description of LA,B:C is of paramount importance. It is assumed that the temperature dependence of the interaction parameter is linear [12], as in Eq. (4).

(1)

Element

yA =

463

Applied Surface Science 488 (2019) 462–467

J.W. Yao, et al.

Table 2 Nominal composition of 40CrNi steel. Element

Fe

Cr

Ni

Si

Mn

C

S/P

Content (wt%)

Bal.

0.45–0.75

1.0–1.49

0.17–0.37

≤0.80

0.37–0.44

≤0.035

7.6 at.% N in the FeN1/12 phase, the nitrogen content in the α′N phase must be higher than 13 at.% to ensure mass balance. According to the FeeN phase diagram, when the nitrogen content approaches 20 at.%, γ′-Fe4N tends to form [19]. It is therefore assumed in this work that the nitrogen content in the α′N phase reaches the extreme value, i.e., m = 1, n = 1/4. To be specific, the chemistry of α′N phase is described as (Fe1–tCrt)1N1/4, with t representing the site fraction of Cr in the α′N phase. To construct the Gibbs free energy model, the interaction parameter between Fe-Cr-N in the BCC system is first calculated. For simplicity, the calculation of the interaction parameter is performed based on the BCC-structured (FeCr)N phase at a composition of FeN : CrN = 1 : 1. At 0 K, the enthalpy of formation of (Fe0.5Cr0.5)N can be calculated by Eq. (6) [11], 0 H(Fe = 0.5 Cr0.5)N

0 HFeCr

0 ( HFeN +

0 HCrN )

Table 4 Gibbs free energy Fe12N, N2, Fe2N and Cr2N at different temperatures [22]. T (K)

°GFe12N (kJ/mol)

°GN2 (kJ/mol)

°GFe2N (kJ/mol)

°GCr2N (kJ/mol)

298 400 600 800 1000

−9.583 −12.357 −21.636 −33.247 −37.625

−57.128 −77.101 −97.459 −118.412 −206.708

−33.95 −45.38 −73.02 −105.93 −143.00

−128.36 −138.894 −145.392 −161.311 −179.841

Table 4. And the Gibbs free energy of MN1/4 is calculated by Eq. (11), based on the chemical reaction in Eq. (12), which indicates that as the nitriding process prolongs, the high‑nitrogen martensite phase tends to decompose into Fe2N and Cr2N, more stable and more frequently observed nitrides in the nitrided steels. The Gibbs free energies of Fe2N, Cr2N and N2 at different temperatures are also listed in Table 4.

(6)

where the formation enthalpies of FeCr, FeN and CrN at 0 K can be calculated by the density functional theory (DFT) method, which are available in the OQMD database [20,21], as listed in Table 3. Therefore, the interaction parameter at 0 K is equal to

°G(Fe1

BCC LFe,Cr:N (0K) =

4(Fe1 t Crt )N1/4 +

0 0 H(Fe /(0.5 × 0.5) = 4 H(Fe 0.5Cr0.5)N 0.5Cr0.5)N

(7)

According to Dulong-Petit law, the heat capacity at constant volume (CV) of most solids at room temperature is approximately 3R (R is the gas constant), which is 25 J/mol. It is equal to 26 J/mol when converted to the heat capacity at constant pressure (CP). With Knopp law, CP of solids can be estimated by linearly superpositioning the contributions from each component. For the (Fe0.5Cr0.5)N compound, the estimated CP at 298 K is 52 J/mol. Since CP = (∂H/∂T)P, the enthalpy of formation of (Fe0.5Cr0.5)N at 298 K is estimated to be CP × T = 15.496 kJ/mol with 0 K as the reference, and the interaction parameter at 298 K is therefore equal to 61.984 kJ/mol, approximately four times of the enthalpy of formation at 298 K. Since the temperature dependence of the interaction parameter is linear, the calculated coefficient θ in Eq. (4) is – 0.051 kJ/(mol·K) Therefore, the temperature dependence of the interaction parameter for the BCC-structured Fe-Cr-N system is BCC LFe,Cr:N

= 77.188

t=

(10) 0

where GFeN1/12 and GMN1/4 are Gibbs free energy of FeN1/12 and MN1/4 phases, yFe is the site fraction of Fe atoms in the FeN1/12 phase, and yM is the site fraction of M atoms (M = Fe, Cr) in the αN′-(Fe1–tCrt)1N1/4 phase. The Gibbs free energies of FeN1/12 at different temperatures are available in the thermochemical data handbook [22], as listed in

FeCr FeN CrN

Fm-3m Pm-3m Pm-3m

0.879 0.622 0.057

(12)

(

1 1 Cr eq

)

+ 1 (1

x)

(13)

3. Results and discussions

Table 3 Calculated enthalpy of formation (ΔHf) of FeCr, FeN, CrN at 0 K [20,21]. ΔHf (eV/atom)

2t )Fe2 N + 2t Cr2N

(11)

To testify the thermodynamic models, plasma nitriding was carried out on 40CrNi steel. Before the thermo-chemical treatments, 40CrNi cylinders with a dimension of Φ32 × 4 mm were solutioned at 980 °C for 1.5 h followed by oil quenching. The surface of the specimens before plasma nitriding was mechanically ground by 240- and 800-grit sandpapers. Plasma nitriding was performed at 520 °C (793 K) for 8 h in a 30 kW pulse plasma multi-element furnace (LDMC-30, 30 kW), with a mixed atmosphere of N2 and H2 whose flow rates were 0.3 L/min and 0.1 L/min, respectively. The phase structure in the surface layer was characterized by XRD (D/max-2200 X-ray diffraction) using Cu-Kα radiation (λ = 0.15405 nm) at 40 kV and 30 mA, and by TEM (Tecnai G2 F30) in the bright field, central dark field and SAD modes.

BCC = °G FeN1/12 + °G MN1/4 + RT (yFe ln yFe + yM ln yM ) + yFe yM LFe,Cr:N

Space group

1 N2 = (2 2

1 °G N2 2

2.3. Materials and methods

G (Fe,Cr)1 (N,Va)3

Compound

2t ) °G Fe2N + 2t °GCr2N

With Eqs. (9)–(13), the Gibbs free energy of BCC-structured Fe-Creq-N quasi-binary solution can be calculated, and it is later used to evaluate the thermodynamic stability of the system during nitriding.

According to Eq. (2), the molar Gibbs free energy of BCC(Fe,Cr)1(N,Va)3 phase is described as

0

1 (2 4

For the molar system comprised of x mole FeN1/12 and (1–x) mole MN1/4, the site fractions of Fe atoms in FeN1/12 phase and M atoms in MN1/4 phase are x and (1–x), respectively. With the calculated Creq for each material system, the BCC solution is (FeCreq)1(N,Va)3. According to the mass balance, the fraction of Cr atoms in the αN′-(Fe1–tCrt)1N1/4 phase can be related to the phase fraction of FeN1/12 following Eq. (13)

(9)

0.051T (kJ/mol)

=

t Crt )N1/4

3.1. Phase stabilities by thermodynamic calculations With Eqs. (9)–(13), the Gibbs free energy of BCC-structured quasibinary solution with a composition of xFeN1/12 - (1–x)(Fe1–tCrt)1N1/4 can be calculated at different temperatures. For 40CrNi steel, the Gibbs free energy, the second derivative of Gibbs free energy and Gibbs free 464

Applied Surface Science 488 (2019) 462–467

J.W. Yao, et al.

Fig. 2. (a) Gibbs free energy, (b) second derivative of Gibbs free energy and (c) Gibbs free energy of mixing of xFeN1/12-(1–x)MN1/4 (M = Fe, Cr) in 40CrNi at temperatures 300, 400, 600, 800 and 1000 K. Solid dots represent inflexion points of the curve. Table 5 Composition limits (xN) and range (Δx) for the miscibility gap at different temperatures. T (K)

xN

Δx

300 400 600 800 1000

0.080–0.198 0.081–0.197 0.085–0.194 0.091–0.189 0.104–0.179

0.118 0.116 0.109 0.098 0.075

solution is likely to decompose into two phases through up-hill diffusion upon compositional fluctuations. Whereas outside the inflexion points, the formation of the new phases in response to compositional fluctuations follows the regular nucleation-growth regime (or ‘downhill’ diffusion), which explains the frequent observations of Fe4N and Fe2e3N in the surface layer of the nitrided steel. According to Fig. 2, when 14.3 at.% nitrogen is introduced to the solution, i.e., FeN1/ 12:MN1/4 = 1:1, the system is most unstable and highly likely to decompose into N-lean and N-rich phases via spinodal decomposition. The calculated temperature-composition diagram that delineates the miscibility gap is shown in Fig. 3. The composition/temperature range for spinodal decomposition can be used to guide to select the nitriding conditions. The nitrogen contents that set the limits of the miscibility gap at temperatures of 300 K, 400 K, 600 K, 800 K and 1000 K are listed in Table 5. To achieve complete nanocrystallization in the nitrided layer, it is critical to restrict the nitrogen gradient in the

Fig. 3. Calculated temperature-composition diagram showing the spinodal region for the BCC xFeN1/12-(1–x)MN1/4 (M = Fe, Cr) solution in the 40CrNi system. The dash line represents the Ac1 temperature, above which the curve is dotted.

energy of mixing with respect to composition are sketched in Fig. 2. For temperatures at 300, 400, 600, 800 and 1000 K, the inflexion points of the Gibbs free energy curve define the limit of stability. Within the points of inflexion, the second derivative of Gibbs free energy is negative (∂2G/∂x2 < 0), which is typical of chemical spinodal, i.e., the 465

Applied Surface Science 488 (2019) 462–467

J.W. Yao, et al.

Fig. 4. XRD pattern of 40CrNi steel plasma-nitrided at 520 °C for 8 h with N2 : H2 = 3 : 1.

Fig. 5. Phase and structure analysis of 40CrNi steel plasma-nitrided at 520 °C for 8 h with N2 : H2 = 3 : 1 by TEM in the (a) dark field mode, (b) bright field mode, and (c) SAD mode.

surface layer within the composition limits as in Table 5. The nitrogen gradient is determined by the surface nitrogen content, which is primarily determined by the nitriding temperature and nitrogen potential in the atmosphere (e.g., N2/H2 ratio) for plasma nitriding in particular. As indicated by Table 5, the composition range Δx for spinodal reduces as the temperature increases, which indicates that the nitrogen potential in the atmosphere need to be lowered at high nitriding temperatures to ensure the surface nitrogen content within the spinodal, while at lower nitriding temperatures it is safe to increase the nitrogen potential in the atmosphere. This also explains the phenomena observed in [7] (Fig. 1(a)), i.e., for the same nitriding conditions (nitrogen potential in the atmosphere), γ′-Fe4N nitride is observed at a higher temperature (540 °C) while nanocrystallization occurs only at a lower temperature (500 °C), as the nitrogen gradient fails to completely falls

into the spinodal range at the high nitriding temperature. It also needs to be noted that the nitriding temperature should be lower than Ac1 temperature (~863 K for 40CrNi steel [23]) to avoid the formation of austenite, as the transition of crystal structure during BCC to FCC phase transformation may hinder spinodal decomposition. 3.2. Microstructures and phase analysis According to the thermodynamic calculations, the nitriding temperature for 40CrNi is set as 520 °C (~800 K), which is an intermediate temperature below Ac1 that ensures a broad composition range for spinodal as well as the kinetics. Structural characterization of the nitrided surface layer by XRD is shown in Fig. 4(a). The phase analysis indicate that the surface layer is entirely composed of FeN1/12 phase 466

Applied Surface Science 488 (2019) 462–467

J.W. Yao, et al.

and high‑nitrogen martensite αN′ phase. The partially enlarged view in Fig. 4(b) manifests the split of the main peak (~45°), which is also a typical phenomenon of spinodal decomposition [24]. With this, one can preliminarily identify the occurrence of spinodal decomposition 40CrNi steel during plasma nitriding at 520 °C. The samples were further characterized by TEM. The bright field and central dark field images in Fig. 5(a-b) shows that grains are mostly globular in the nitrided surface layer of 40CrNi and are < 30 nm. In the selected area electron diffraction mode, concentric diffraction rings induced from a substantial amount of nano-scale grains are obvious, as in Fig. 5(c), which indicates that FeN1/12 phase and high‑nitrogen martensite α′N phase are nano-scale, evidently proving the occurrence of spinodal decomposition during nitriding at 520 °C. Meanwhile, experimental results have been obtained that support the nanocrystallization nature of the segregation of the FeN1/12 phase and high‑nitrogen martensite α′N phase during the nitriding of the Cr equivalent steels.

Development Program 2018YFB200007.

of

China

through

award

number

References [1] B.C.E.S. Kurelo, G.B. de Souza, S.L.R. da Silva, F.C. Serbena, C.E. Foerster, C. Alves, Plasma nitriding of HP13Cr supermartensitic stainless steel, Appl. Surf. Sci. 349 (2015) 403–414, https://doi.org/10.1016/J.APSUSC.2015.04.202. [2] T. Bell, Y. Sun, A. Suhadi, Environmental and technical aspects of plasma nitrocarburising, Vacuum 59 (2000) 14–23, https://doi.org/10.1016/S0042-207X (00)00250-5. [3] D. Pye, Practical nitriding and ferritic nitrocarburizing, ASM International, Materials Park, 2003. [4] L. Lu, Y. Shen, X. Chen, L. Qian, K. Lu, Ultrahigh strength and high electrical conductivity in copper, Science 304 (2004) 422–426, https://doi.org/10.1126/ science.1092905. [5] K. Lu, J. Lu, Nanostructured surface layer on metallic materials induced by surface mechanical attrition treatment, Mater. Sci. Eng. A 375–377 (2004) 38–45, https:// doi.org/10.1016/J.MSEA.2003.10.261. [6] W.P. Tong, N.R. Tao, Z.B. Wang, J. Lu, K. Lu, Nitriding iron at lower temperatures, Science 299 (2003) 696, https://doi.org/10.1126/science.1080216. [7] M.F. Yan, B.F. Chen, B. Li, Microstructure and mechanical properties from an attractive combination of plasma nitriding and secondary hardening of M50 steel, Appl. Surf. Sci. 455 (2018) 1–7, https://doi.org/10.1016/j.apsusc.2018.04.213. [8] R.L. Liu, Y.X. Wang, M.F. Yan, Method for realizing steel surface layer nanocrystallization based on thermal diffusing permeation process, CN102061441B, 2012. https://patents.google.com/patent/CN102061441B/en. [9] M.F. Yan, X.A. Wang, R.L. Liu, Y.X. Zhang, Y. Yang, Kinetics and wear behaviour of M50NiL steel plasma nitrided at low temperature, Mater. Sci. Technol. 33 (2017) 370–376, https://doi.org/10.1080/02670836.2016.1213027. [10] M.F. Yan, Y.Q. Wu, R.L. Liu, M. Yang, L.N. Tang, Microstructure and mechanical properties of the modified layer obtained by low temperature plasma nitriding of nanocrystallized 18Ni maraging steel, Mater. Des. 47 (2013) 575–580, https://doi. org/10.1016/J.MATDES.2012.11.007. [11] R.F. Zhang, S. Veprek, On the spinodal nature of the phase segregation and formation of stable nanostructure in the Ti–Si–N system, Mater. Sci. Eng. A 424 (2006) 128–137, https://doi.org/10.1016/J.MSEA.2006.03.017. [12] R.F. Zhang, S. Veprek, Metastable phases and spinodal decomposition in Ti1−xAlxN system studied by ab initio and thermodynamic modeling, a comparison with the TiN–Si3N4 system, Mater. Sci. Eng. A 448 (2007) 111–119, https://doi.org/10. 1016/J.MSEA.2006.10.012. [13] T. Kasuya, N. Yurioka, Carbon equivalent and multiplying factor for hardenability of steel, Weld. Res. Suppl. (1993) 263–268. [14] N. Suutala, T. Takalo, T. Moisio, Ferritic-austenitic solidification mode in austenitic stainless steel welds, Metall. Trans. A 11 (1980) 717–725, https://doi.org/10.1007/ BF02661201. [15] V.G. Gavriljuk, H. Berns, High nitrogen steels: structure, properties, manufacture, applications, Springer-Verlag Berlin Heidelberg, New York, 1999. [16] M. Hillert, Phase equilibria, phase diagrams and phase transformations: their thermodynamic basis, 2nd ed., Cambridge University Press, Cambridge, 2008. [17] H. Ohtani, M. Hillert, A thermodynamic assessment of the Fe-N-V system, Calphad 15 (1991) 25–39, https://doi.org/10.1016/0364-5916(91)90023-D. [18] Z. Li, G. Tang, X. Ma, M. Sun, L. Wang, XPS study on chemical state and phase structure of PBII nitriding M50 steel, IEEE Trans. Plasma Sci. 38 (2010) 3079–3082, https://doi.org/10.1109/TPS.2010.2050075. [19] L. Tang, M. Yan, Microstructure and mechanical properties of surface layers of 30CrMnSiA steel plasma nitrocarburized with rare earth addition, J. Rare Earths 30 (2012) 1281–1286, https://doi.org/10.1016/S1002-0721(12)60221-X. [20] S. Kirklin, J.E. Saal, B. Meredig, A. Thompson, J.W. Doak, M. Aykol, S. Rühl, C. Wolverton, The Open Quantum Materials Database (OQMD): assessing the accuracy of DFT formation energies, npj Comput. Mater. 1 (2015) 15010, , https:// doi.org/10.1038/npjcompumats.2015.10. [21] J.E. Saal, S. Kirklin, M. Aykol, B. Meredig, C. Wolverton, Materials design and discovery with high-throughput density functional theory: the open quantum materials database (OQMD), JOM 65 (2013) 1501–1509, https://doi.org/10.1007/ s11837-013-0755-4. [22] I. Barin, Thermochemical Data of Pure Substances, 3rd ed., VCH, Weiheim, 1995, https://doi.org/10.1002/9783527619825. [23] H. Göhring, O. Fabrichnaya, A. Leineweber, E.J. Mittemeijer, Thermodynamics of the Fe-N and Fe-N-C systems: the Fe-N and Fe-N-C phase diagrams revisited, Metall. Mater. Trans. A 47 (2016) 6173–6186, https://doi.org/10.1007/s11661-0163731-0. [24] V.H.C. de Albuquerque, E. de Macedo Silva, J. Pereira Leite, E.P. de Moura, V.L. de Araújo Freitas, J.M.R.S. Tavares, Spinodal decomposition mechanism study on the duplex stainless steel UNS S31803 using ultrasonic speed measurements, Mater. Des. 31 (2010) 2147–2150, https://doi.org/10.1016/J.MATDES.2009.11.010.

4. Conclusions In this work, a thermodynamic model that describes the Gibbs free energy of BCC-Fe quasi-binary solution in the Fe-Cr-N system is established. The model is specifically developed for the nitriding process of the multi-component steels, whose composition is first converted to an Fe-Creq-N ternary system using the concept of Cr equivalent through the relative elemental electronegativities with respect to nitrogen. With the thermodynamic model, the limit of stability of the BCC solution can be evaluated with temperatures and used to guide the selection of nitriding conditions. It can be concluded that the BCC solution is unstable and likely to decompose in the form of spinodal decomposition over a specific compositional/temperature range, which mechanistically explains the occurrence of nanocrystallization in selected nitriding conditions. To achieve complete nanocrystallization instead of the formation of micro-scale nitrides during nitriding, the nitrogen concentration gradient needs to be within the spinodal, i.e., the nitrogen potential in the atmosphere for each nitriding temperature needs to be carefully controlled. The nitrogen potential in the atmosphere needs to be lowered at a high nitriding temperature to ensure spinodal decomposition throughout the whole nitrided layer, while a high nitrogen potential can be used at a low nitriding temperature. The developed thermodynamic model has been successfully applied to the nitriding of 40CrNi steel system. The Gibbs free energy curve of N-containing BCC-Fe solution is predicted to be double-well and indicates the instability of the solution within a specific range N content for each temperature. When 14.3 at.% N is introduced, the solution is most unstable. For 40CrNi plasma nitrided at 800 K, the observed nanoscale FeN1/12 phase and high‑nitrogen martensite phase embrace the thermodynamic predictions in terms of nanocrystallization via spinodal decomposition. Despite underlying presumptions made to construct the thermodynamic model, this work is especially meaningful for engineering applications, where quick and accurate predictions are advisable in directing engineering processing conditions. Furthermore, with the thermodynamic model and the introduction of Cr equivalent, this work can be quickly extended to different steel systems, which is also highly appreciated for engineering applications. Acknowledgements The authors gratefully acknowledge the supports from the National Key Research and Development Program of China through award number 2017YFB0304601, and the National Key Research and

467