Effects of Mo and Cu Contents on Sigma Phase Precipitation in 25Cr-5Ni-Mo-Cu-1Mn-0.18N Duplex Stainless Steel

Abstract

Recently, a new duplex stainless steel UNS S82551 (25Cr-5Ni-1Mo-2.5Cu-0.18N) has been developed to overcome the drawbacks in super martensitic stainless steel, conventional 22Cr and 25Cr super duplex stainless steels in terms of the productivity and the cost. The characteristic of the alloy design of S82551 is to use Cu, instead of Mo, addition to ensure the corrosion resistance and strength. In addition, S82551 is expected to be less sensitive to sigma phase precipitation during single or multi-pass welding compared with conventional duplex stainless steels and super duplex stainless steels due to the significant decrease in Mo content. There is a trade-off relationship between achieving better properties and avoiding sigma phase precipitation when increasing alloying elements such as Mo, Cr and Cu. In order to utilize the new material S82551 in industry by welding in a similar manner to conventional and super duplex stainless steels, the prevention of sigma phase precipitation is an important subject.

This work investigated the effects of Mo and Cu contents on sigma phase precipitation in S82551 using time-temperature-precipitation (TTP) diagram generated by experimental work and calculation. The effects of Mo and Cu contents on the nose temperature and time of C-curve were clarified and compared with S31803 (22Cr-5Ni-3Mo) and S32750 (25Cr-7Ni-4Mo), which indicates the critical heating condition to be free from sigma phase precipitation in duplex stainless steels.

1. Introduction

Duplex and super martensitic stainless steels have been exploited by a wide range of industrial sectors for many years because of their availability, workability, strength, toughness and corrosion resistance.^1,2,3) Girth welded joints using super martensitic stainless steel (13Cr SMSS) are susceptible to stress corrosion cracking (SCC).^4,5) Although post weld heat treatment (PWHT) is effective in preventing SCC in oil and gas transportation systems,⁶⁾ PWHT could have a negative impact on the efficiency of pipe laying operations. On the other hand, conventional 22Cr duplex stainless steels (e.g. UNS S31803) and 25Cr super duplex stainless steels (e.g. UNS S32750) have also been utilized, in the as-welded condition, by the oil & gas industries for many years. However, these higher grades incur greatly increased cost due to their rich alloying elements necessary for these excellent properties. Recently, a new duplex stainless steel UNS S82551 (25Cr-5Ni-1Mo-2.5Cu-0.18N) has been developed to achieve lower material cost than conventional duplex stainless steels and super duplex stainless steels by decreasing Mo content, and the new steel S82551 can be utilized in the as-welded condition in slightly sour conditions.^7,8,9) The characteristic of the alloy design of S82551 is to use Cu, instead of Mo, as an alloying element to ensure corrosion resistance and strength. In addition, S82551 is expected to be less sensitive to sigma phase precipitation during single or multi-pass welding compared with conventional duplex stainless steels and super duplex stainless steels due to the significant decrease in Mo content. In order to utilize the new material S82551 by welding in a similar manner to S31803 and S32750,¹⁰⁾ the prevention of sigma phase precipitation is an important subject. A lot of research into conventional duplex stainless steels and super duplex stainless steels about the effect of chemical composition on sigma phase precipitation has been performed.^{11,12,13,14,15,16,17,18,19)} The effects of Mo and Cu contents on sigma phase precipitation have not been investigated previously in S82551. In this study, in order to clarify the difference in sigma phase precipitation behaviour between S82551, S31803 and S32750, Time-Temperature-Precipitation (TTP) diagrams of sigma phase were produced by experimental work. To complement the experimental work, a C-curve of sigma phase precipitation was calculated based on the kinetic model using the classical nucleation theory and diffusion controlled spherical growth model by Ogawa.^20,21) The effects of Mo and Cu contents on the nose time and temperature of the C-curve of sigma phase precipitation for S82551 is discussed.

2. Experimental

2.1. Materials

The materials used for this research to generate the TTP diagram were laboratory melted duplex stainless steel plates with different Mo and Cu contents based on the new material – UNS S82551 chemical composition. The plates (LM0~LM8) were melted with a laboratory electric furnace and were fabricated into 15 mm thick plates by hot forging and hot rolling followed by solution treatment. The chemical composition of the plates is given in Table 1. S82551 grade (LM1) contains 1%Mo which differs from conventional duplex grade containing 3%Mo. Mo content is decreased to reduce the alloying cost albeit limiting the application environment of S82551 to slightly sour environment, and maintaining good phase stability in terms of the precipitation of sigma phase. In addition, Cu instead of Mo is added to ensure the corrosion resistance in slightly sour environments and strength. It has been reported that Cu is not effective on pitting corrosion resistance in chloride solution, but Cu can improve the resistance to sulphide stress corrosion cracking (SSC) in sour conditions because of the formation of stable sulphides of Cu.⁸⁾

Table 1. Chemical composition of materials investigated.

Material	Mark	Chemical composition (mass%)							Notes
		C	Mn	Cu	Ni	Cr	Mo	N
Plate	LM0	0.016	0.95	2.33	4.92	24.8	<0.01	0.19	Mo free
	LM1	0.015	0.92	2.43	4.91	24.8	1.01	0.19	S82551
	LM2	0.015	0.94	2.34	4.94	25.1	2.03	0.20	2.0%Mo
	LM3	0.015	0.95	2.38	5.08	24.8	3.00	0.21	3.0%Mo
	LM4	0.015	1.0	< 0.01	5.00	25.0	1.00	0.20	Cu free
	LM5	0.015	1.0	1.5	5.00	25.0	1.00	0.20	1.5%Cu
	LM6	0.015	1.0	3.5	5.00	25.0	1.00	0.20	3.5%Cu
	LM7	0.019	0.51	< 0.01	5.00	22.0	3.00	0.20	S31803
	LM8	0.018	0.50	< 0.01	7.00	25.1	3.91	0.29	S32750

2.2. Isothermal Heat Treatment Test

For producing the TTP diagram, the specimens were machined to cubes of 10 mm sides and heated to various constant temperatures ranging from 650 to 900°C for various times followed by water quenching. These heat treatment temperatures were carefully selected based on the equilibrium phase diagram for S82551 chemical composition calculated using Thermo-Calc software (Fig. 1). As shown in Fig. 1, the formation of different phases is influenced by the temperature. When in the temperature range between 650 and 900°C, δ-ferrite phase has a much lower proportion in the equilibrium condition. Instead, enrichment of sigma phase is confirmed in this temperature range as well as nitride, Cu and carbide. The amount and size of sigma phase vary with the holding time at temperature. Therefore, the exposure times 100 to 36000 s were applied to the test. After the heat treatment, the specimens were ground to #1200 grit, polished to a 1 μm finish and electrochemically etched in 10% oxalic acid and 40%KOH. The sigma phase precipitation of each specimen was detected with light microscopy at 500x magnification. Image analysis with image analyser was carried out to measure the area fraction of sigma phase precipitation.

Fig. 1.

Phase diagram for LM1 (UNS S82551) calculated using Thermo-Calc. (Online version in color.)

3. Results and Analysis

3.1. Sigma Phase Precipitation

Figure 2 shows the microstructure of LM3 containing 3%Mo aged for 1000 s, 5000 s and 10000 s at 850°C. This is a typical microstructure of duplex stainless steel where the dark phase is ferrite (δ) and the bright phase is austenite (γ) and the precipitation is considered as sigma (σ) phase. As can be observed in Fig. 2, the amount of sigma phase is increasing with increasing holding time, although ferrite content is decreasing. Ferrite phase transformed into sigma phase and secondary austenite by long holding time at high temperature. This transformation is clarified by the phase diagram shown in Fig. 1. A small amount of sigma phase can be observed in Fig. 2(a), typically, precipitated along ferrite /austenite phase boundaries. The amount and the size of the sigma phase precipitation are increasing and growing with increasing the time at 850°C. The area fraction of sigma phase precipitation measured by image analyser were 5.3% for 1000 s, 19.9% for 5000 s and 24.1% for 10000 s.

Fig. 2.

Microstructure of LM3 containing 3%Mo aged for (a) 1000 s, (b) 5000 s and (c) 10000 s at 850°C.

3.2. TTP Diagram by Experimental Work

Figures 3 and 4 show the TTP diagrams obtained from the experimental data of sigma phase precipitation in the duplex stainless steel plates with different Mo and Cu contents based on UNS S82551 chemical composition. As shown in Fig. 3, it is clearly confirmed that lower Mo content supresses the sigma phase precipitation. For LM3, which contains 3%Mo, sigma phase precipitation started from around 100 s. With decreasing Mo contents, the time for precipitation of sigma phase tended to be delayed. For LM1 containing 1%Mo (which is S82551 chemical composition), the time to precipitate sigma phase is obviously slower than that of LM2 and LM3 containing higher Mo content. Therefore, decreasing Mo in the material lowers the nose temperature and increases the nose time. On the other hand, the effect of Cu on sigma phase precipitation is not huge compared with that of Mo content as shown in Fig. 4. Increasing Cu in the material slightly reduces the nose time for the sigma phase precipitation.

Fig. 3.

TTP diagrams: (a) LM0 (Mo free), (b) LM1 (Mo: 1 mass%), (c) LM2 (Mo: 2 mass%), (d) LM3 (Mo: 3 mass%).

Fig. 4.

TTP diagrams: (e) LM4 (Cu free), (f) LM5 (Cu: 1.5 mass%), (g) LM1 (Cu: 2.5 mass%), (h) LM6 (Cu: 3.5 mass%).

Better understanding the time and temperature at the nose of the TTP diagrams with the minimum time for precipitation would be desirable in terms of industrial application. The results of the effects of Mo and Cu content on each nose time and nose temperature were plotted as shown in Figs. 5 and 6. Considering the effect of Mo content on sigma phase formation, Mo promotes sigma phase precipitation, whereby the nose of the precipitation start curve moves from about 100 s for LM3 containing 3%Mo to 5000 s for LM1 containing 1%Mo. This suggests that the nucleation rate of the sigma phase formation would drop to around one-fiftieth. In addition, the precipitation start nose for LM3 containing 3%Mo is at around 900°C, and it is moving to around 850°C for LM2 containing 2%Mo and to around 750°C for LM1 containing 1%Mo. The nose time is increasing, and the nose temperature is dropping with decreasing Mo content in duplex stainless steels. In addition, non-sigma phase precipitation was observed in the Mo-free tested material (LM0). In terms of the phase stability, LM0 would be the best material among the test materials. However, Mo is a beneficial alloy element for the superior corrosion resistance by stabilising the passive film through enrichment beneath the film. Therefore, the addition of at least 1%Mo would be better than Mo-free for the corrosion resistance in sour environments. On the other hand, the effect of Cu on sigma phase precipitation is not strong as shown in Figs. 5 and 6. The nose time and temperature do not change too much with increasing Cu content although the nose shape tend to be wider with increasing Cu content.

Fig. 5.

Effects of Mo and Cu content on the nose time in TTP diagram of sigma phase precipitation.

Fig. 6.

Effects of Mo and Cu content on the nose temperature in TTP diagram of sigma phase precipitation.

Figure 7 shows the TTP diagram of LM7_S31803 (22Cr-5Ni-3Mo). Comparing LM7_S31803 with LM3 (25Cr-5Ni-3Mo) (Fig. 3(d)), the main difference in chemical composition is only Cr content. The effect of Cr content on sigma phase precipitation can be clarified by comparison of the TTP diagrams of these two materials. The higher Cr content from 22%Cr to 25%Cr promotes sigma phase precipitation to slightly shorter time: increasing Cr content in the material reduces the nose time of the C-curve. Next, comparing LM7_S31803 with LM1 (25Cr-1Mo-5Ni, which is S82551 chemical composition), the difference of chemical composition is Cr and Mo contents. As discussed above, increasing Cr reduces the nose time of the C-curve. As can be seen in Figs. 7(a) and 3(b), the sigma phase precipitation is more suppressed by decreasing Mo content from 3%Mo to 1%Mo even though LM1_S82551 has higher Cr content than LM7_S31803. From these results, the effect of Mo content on sigma phase precipitation is more influential than that of Cr content. In addition, TTP diagram of LM8_S32750 (25Cr-7Ni-4Mo) containing high alloy elements is shown in Fig. 7(b). Nose time and nose temperature of the C-curve for LM8_S32750 are shorter and higher respectively, compared with LM7_S31803 and LM1_S82551.

Fig. 7.

TTP diagram for (a) LM7_S31803 (22Cr-5Ni-3Mo) and (b) LM8_S32750 (25Cr-7Ni-4Mo).

3.3. TTP Diagram by Calculation

Next, in order to clarify the effects of Mo and Cu and their mechanism, a C-curve of sigma phase precipitation was calculated based on the kinetic model on the classical nucleation theory and diffusion controlled spherical growth model referenced by Ogawa.^20,21) The C-curve was calculated with the experimental data and Johnson-Mehl, Avrami and Kolmogorov (JMAK) approach.^22,23,24) The current implementation focuses on sigma phase precipitation and growth. According to Ogawa, the C-curves corresponding to a progress ratio of precipitation, X, are calculated in the following growth equation of the form (1) and (2):   

$$\text{X}=1-\text{exp}\left[-\frac{V_{ex}}{V_0\omega }\right]=1-\text{exp}\left[-{(k_{p}t)}^{\text{m}}\right]\hspace{1em}\left(\mathrm{m}=5/2\right)$$(1)

Where V_ex is the extended volume of sigma phase which grows spherically, V_o is the volume of sigma phase in the equilibrium condition, k_p is the growth rate constant, ω is the supersaturation, t is the time. The growth rate constant k_p is shown as the following equation.   

$$k_p=K_{00}{\left(\frac{T_{eq}}{\mathrm{\Delta }T}\right)}^{4/5}{D}_{eff}exp\left(-\frac{\mathrm{\Delta }{G}_h}{RT}\right)$$(2)

Where D_eff is the diffusion constant of the chromium, T_eq is the solvus temperature of sigma phase, ΔT is the degree of undercooling (=T_eq – T). Figure 8 shows the comparison of diffusion constants of Cr and Mo. Diffusion constant of Cr and Mo are almost equivalent as shown in Fig. 8. Therefore, in terms of the simplified model applied, the nucleation rate is considered to be controlled by the diffusion of chromium atom, which means the dominant atom in the diffusion process is chromium in the case of sigma phase nucleation.   

$$\mathrm{\Delta }{G}_h=\phi {\left(T_{eq}/\mathrm{\Delta }T\right)}^2$$(3)

  

$$K_{00}={\left[\left(16\sqrt{2}{\pi }^2/15V_0\cdot C_0{N}_0{\omega }^{\frac{1}{2}}{\sigma }^2/{\left(C_p\mathrm{\Delta }H\right)}^2/r_a^4\right)\right]}^{2/5}$$(4)

  

$$\phi =32\pi /15V_0^2{N}_0\cdot {\sigma }^3/{\left(C_p\mathrm{\Delta }H\right)}^2$$(5)

Fig. 8.

Comparison of diffusion constants of Cr and Mo.

Where N₀ is the number of atoms to determine the growth rate, V₀ is molar volume, σ is interface energy, ΔH is the enthalpy of transformation of intermetallic, r_a is the lattice constant, C_p is concentration in intermetallic phase of dominant atom to determine the growth rate and C_o is concentration in steel of dominant atom to determine growth rate. The growth constant k_p at any temperature T is determined by the diffusion constant and the parameters T_eq, ϕ and K₀₀ which are functions of the alloying elements of materials. T_eq was calculated using a thermodynamic database in the Thermo-Calc software. The time tc reaching each progress ratio of precipitation, X, is calculated by Eq. (6) obtained from the above Eqs. (1) and (2):   

$$tc=K_{00}{}^{-1}{\left[-\text{ln}\left(1-\text{Xo}\right)\right]}^{2/5}{\left(T_{eq}/\mathrm{\Delta }T\right)}^{-4/5}/{\text{D}}_{eff}\cdot \text{exp}\left(\frac{\mathrm{\Delta }{G}_h}{RT}\right)$$(6)

The C-curve of the tested materials calculated by the above Ogawa’s approach gives us the prediction of the sigma phase behaviour. Figure 9 shows the effect of Mo content on the calculated C-curve of 0.5% sigma phase precipitation. The values of T_eq, ϕ and K₀₀ listed in Tables 2 and 3 were applied for the calculation. According to Fig. 9, the nose time is increased, and the nose temperature is decreased with decreasing Mo content in tested materials, which shows a similar tendency to the TTP diagram generated from the experimental results. It was confirmed that the calculated TTP curves have a good fit to the experimental results: C-curve was shifted to longer time and lower temperature by the decrease of Mo. Figure 10 shows the driving force of sigma phase precipitation ΔG_m^δ→σ at 850°C of LM0, LM1, LM2 and LM3 calculated by Thermo-Calc. From these results, the driving force of sigma phase precipitation at 850°C for LM1 is significantly lower than that of LM3. Therefore, the suppression of sigma phase by decreasing Mo content is due to the decrease of the driving force for sigma phase precipitation.

Fig. 9.

Effect of Mo content on the C-curve of 0.5% sigma phase precipitation calculated by using kinetic model referenced by Ogawa.^20,21)

Table 2. Constant values used for calculation of growth rate constant K for 25Cr-5Ni-Mo-2.5Cu-0.18N.

Mo	1	2	3
Teq (K)	1143	1193	1238
Deff (m2/s)	2.3 × 10−4
Q (kJ/mol)	239
φ (J/mol)	118	111	104
K00 (1/m2)	1.3E + 11	1.1E + 11	0.8E + 11

Table 3. Constant values used for calculation of growth rate constant K for 25Cr-5Ni-1Mo-Cu-0.18N.

Cu	0	1.5	2.5	3.5
Teq (K)	1126	1137	1143	1149
Deff (m2/s)	2.3 × 10−4
Q (kJ/mol)	239
φ (J/mol)	66	100	118	132
K00 (1/m2)	0.7E + 11	1.1E + 11	1.3E + 11	1.4E + 11

Fig. 10.

Driving force of sigma phase precipitation ΔG_m^δ→σ at 850°C for LM0, LM1, LM2 and LM3 calculated by Thermo-Calc.

In the same way, as the C-curve was calculated with different Mo contents, the effect of Cu content on the C-curve of 0.5% sigma phase precipitation was calculated as shown in Fig. 11. These calculation results show that the nose time do not change too much with decreasing Cu content but the nose shape of C-curve tend to be wider with increasing Cu content, which is also the same trend as the in TTP diagram produced from the experimental results. It was clarified that the C-curve of sigma phase precipitation shifts to longer time slightly by decreasing Cu content. The driving force of sigma phase precipitation ΔG_m^δ→σ at 800°C of LM1, LM4, LM5 and LM6 was calculated by Thermo-Calc. The driving force of S82551 containing 2.5%Cu at 800°C was a little bit higher than that of LM4 without Cu shown in Fig. 12. Therefore, the effect of Cu addition on sigma phase precipitation is not larger than that of Mo.

Fig. 11.

Effect of Cu content on the C-curve of 0.5% sigma phase precipitation calculated by using kinetic model referenced by Ogawa.^20,21)

Fig. 12.

Driving force of sigma phase precipitation ΔG_m^δ→σ at 850°C for LM1, LM4, LM5 and LM6 calculated by Thermo-Calc.

The effect of Mo and Cu on sigma phase precipitation and its mechanism are considered as follows. As shown in Eqs. (4) and (5), the parameters of K₀₀ and ϕ are a function of the enthalpy ΔH of sigma phase transformation and these parameters increase with decreasing the enthalpy ΔH. The parameter K₀₀ is also a function of the dominant atom contencentration, C₀. The dominant atom contencentration, C₀, corresponds to the constant Cr content of 25% in the materials. Considering the effect of Mo content on the acceleration of growth, the parameters of K₀₀ and ϕ are almost constant in the tested materials due to the constant Cr content of 25% in the materials, which means that the enthalpy ΔH of sigma phase transformation is almost constant even though higher Mo is present. Therefore, it is considered that the reason Mo promotes the sigma phase growth is due to increasing the solvus T_eq of the sigma phase. As shown in Table 2, the solvus T_eq is increasing with increasing Mo content. Higher T_eq leads to larger overcooling resulting in a larger driving force at the temperature with a higher diffusion rate of the dominant atom in the growth of sigma phase, which is the same tendency shown in Fig. 10.

Considering the effect of Cu content, the value of ΔH could be reduced by increasing Cu content due to lower ferrite content which promotes sigma phase precipitation by concentrating Cr and Mo to the ferrite phase. As shown in Table 3, the parameter ϕ is increasing with increasing Cu content, therefore increasing Cu promotes the precipitation. In addition, Cu also raised the solvus T_eq of the sigma phase therefore the growth rate is accelerated and the driving force is increasing with increasing Cu content. However, as described above, the effect of Cu addition on sigma phase is not larger due to lower driving force compared to that of Mo content.

From the TTP diagrams produced by the experimental work and calculation, the sigma phase precipitation is more accelerated by increasing Mo content, and the effect of Cu content on sigma phase precipitation is not large for duplex stainless steel. Furthermore, it was clarified that S82551 is less sensitive to sigma phase precipitation than S31803 and S32750 because of lower Mo content.

4. Conclusion

The following conclusions are obtained from the investigation regarding the TTP curve of sigma phase precipitation in 25Cr-5Ni-Mo-Cu-1Mn-0.18N duplex stainless steel.

• According to the TTP diagrams generated, lower Mo content suppresses sigma phase precipitation, which shifts to lower temperature and longer time.

• TTP curve of sigma phase precipitation is described by using the kinetic model referenced. The suppression of the sigma phase precipitation by decreasing Mo content is due to the decrease of the driving force of sigma phase precipitation.

• Increasing Cu reduces the nose time of the C-curve slightly for the sigma phase precipitation.

• From the TTP diagrams produced by the experimental work and calculation, the sigma phase precipitation is more accelerated by the increase of Mo content, and the effect of Cu content on sigma phase precipitation is not large for duplex stainless steel.

• UNS S82551 has less sensitivity to sigma phase precipitation than UNS S31803 and UNS S32750 mainly because of lower Mo content.

References

• 1)   J. O.  Nilsson: Mater. Sci. Technol., 8 (1992), 685. https://doi.org/10.1179/mst.1992.8.8.685
• 2)   S. D.  Kahar: Int. J. Eng. Res. Appl., 7 (2017), 27. https://doi.org/10.9790/9622-0704042736
• 3)   J.  Charles: Steel Res. Int., 79 (2008), 455. https://doi.org/10.1002/srin.200806153
• 4)   T.  Rogne,  H. I.  Lange,  M.  Svenning,  S.  Aldstedt,  J. K.  Solberg,  E.  Ladanova,  S.  Olsen,  R.  Howard and  R.  Leturno: Proc. Corrosion/02, NACE International, Houston, TX, (2002), 02428.
• 5)   H.  Amaya,  A.  Taniyama,  K.  Ogawa and  M.  Ueda: Proc. Corrosion/08, Curran, Red Hook, NY, (2008), 08100.
• 6)   P.  Woollin: Weld. World, 51 (2007), 31. https://doi.org/10.1007/BF03266598
• 7)   D.  Motoya,  K.  Yamada,  S.  Nakatsuka,  H.  Takabe,  M.  Hamada,  H.  Amaya,  M.  Ueda,  M.  Sagara and  K.  Ogawa: Proc. Eurocorr 2012, European Federation of Corrosion, (2012).
• 8)   M.  Sagara,  D.  Motoya,  K.  Yamada,  H.  Takabe,  A.  Taniyama,  H.  Amaya,  K.  Ogawa and  M.  Ueda: Nippon Steel Sumitomo Met. Tech. Rep., (2015), No. 107, 59.
• 9)   H.  Shitamoto,  M.  Sagara,  H.  Amaya,  N.  Hisamune,  D.  Motoya and  Y.  Watatani: Proc. ASME 2014 33rd Int. Conf. on Ocean, Offshore and Arctic Engineering, ASME, New York, (2014), V06AT04A049. https://doi.org/10.1115/OMAE2014-23694
• 10)   L.  Karlsson: Weld. World, 56 (2012), 65. https://doi.org/10.1007/BF03321351
• 11)   Y.  Maehara,  M.  Koike,  N.  Fujino and  T.  Kunitake: Tetsu-to-Hagané, 67 (1981), 577 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.67.3_577
• 12)   Y.  Maehara,  N.  Fujino and  T.  Kunitake: Tetsu-to-Hagané, 68 (1982), 673 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.68.6_673
• 13)   Y.  Maehara,  Y.  Ohmori,  J.  Murayama,  N.  Fujino and  T.  Kunitake: Met. Sci., 17 (1983), 541. https://doi.org/10.1179/030634583790420457
• 14)   Y.  Ohmori and  Y.  Maehara: Trans. Iron Steel Inst. Jpn., 24 (1984), 60. https://doi.org/10.2355/isijinternational1966.24.60
• 15)   D. M. E.  Villanueva,  F. C. P.  Junior,  R. L.  Plaut and  A. F.  Padilha: Mater. Sci. Technol., 22 (2006), 1098. https://doi.org/10.1179/174328406X109230
• 16)   H.  Sieurin and  R.  Sandström: Mater. Sci. Eng. A, 444 (2007), 271. https://doi.org/10.1016/j.msea.2006.08.107
• 17)   M.  Pohl,  O.  Storz and  T.  Glogowski: Mater. Charact., 58 (2007), 65. https://doi.org/10.1016/j.matchar.2006.03.015
• 18)   N.  Llorca-Isern,  H.  López-Luque,  I.  López-Jiménez and  M. V.  Biezma: Mater. Charact., 112 (2016), 20. https://doi.org/10.1016/j.matchar.2015.12.004
• 19)   G. S.  da Fonseca,  P. S. N.  Mendes and  A. C. M.  Silva: Metals, 9 (2019), 34. https://doi.org/10.3390/met9010034
• 20)   K.  Ogawa and  T.  Osuki: ISIJ Int., 59 (2019), 122. https://doi.org/10.2355/isijinternational.ISIJINT-2018-477
• 21)   K.  Ogawa and  T.  Osuki: ISIJ Int., 59 (2019), 129. https://doi.org/10.2355/isijinternational.ISIJINT-2018-478
• 22)   W.  Johnson and  R.  Mehl: Trans. Am. Inst. Min. Metall. Eng., 135 (1939), 416.
• 23)   M.  Avrami: J. Chem. Phys., 7 (1939), 1103. https://doi.org/10.1063/1.1750380
• 24)   T.  Nishizawa: Thermodynamics of Microstructures, ASM International, Materials Park, OH, (2008), 91.