Age-Hardening Behavior in High-Nitrogen Stable Austenitic Stainless Steel

Abstract

Age hardening in stable austenitic stainless steel wires with a chemical composition of Fe–18%Cr–12%Ni and different N contents was investigated to clarify the role of N. Age hardening could be enhanced by increasing the drawing ratio and N content. The highest age-hardening effect was observed at 800 K in the N-bearing specimens. In addition, severe drawing induced low-temperature age hardening at 450–600 K. Differential scanning calorimetry (DSC) analysis revealed that age hardening at 800 K may be attributed to particle dispersion strengthening by Cr₂N. Meanwhile, the exothermic peaks observed at 450–600 K were controlled by the pipe diffusion of N. However, nitride precipitates and clusters could not be detected by 3D atom probe analysis, which implies the formation of atomic-scale N products, such as I-S pairs, at the dislocations.

 

This Paper was Originally Published in Japanese in J. Jpn. Soc. Heat Treat. 61 (2021) 112–118. Result and Discussion are slightly modified.

Fig. 2 Variation in hardness as a function of aging temperature in solution-treated and 30%, 50%, and 80% cold-drawn 0.2N steel aged at different temperatures for 1.8 ks.

1. Introduction

Austenitic stainless steels possess not only good corrosion resistance but also high work hardening ability due to their low stacking fault energy and thus, cold-worked austenitic stainless steels are widely applied in high-strength structural parts in corrosive environment. In particular, the work hardening ability of metastable austenitic steels, such as SUS304 and SUS301, are significantly enhanced by deformation-induced martensite formed during cold working.^1–3) Therefore, severely cold-rolled or drawn products can be used as extremely hard stainless steels, such as springs,⁴⁾ wire ropes,^5,6) and other hard architectural materials.⁷⁾

However, non-magnetic characteristics are sometimes required in structural materials related with power generation,⁸⁾ superconductivity,⁹⁾ and medical devices¹⁰⁾ which are used in the environment where a magnetic field is controlled. In such cases, metastable austenitic steels with body-centered cubic (bcc) martensite cannot be applied; instead, stable austenitic steels must be employed. To strengthen stable austenitic stainless steels that do not form deformation-induced martensite, their work hardening ability should be enhanced. In general, the work-hardening rate of austenitic stainless steels can be enhanced by the addition of interstitial elements, such as C and N. For example, it was reported that the addition of C and N increased the work-hardening rate of 18%Cr–12%Ni steels via deformation twinning¹¹⁾ and development of planar dislocation arrays,^12,13) respectively. Furthermore, C and N addition induces age hardening ability in these steels; thus, it is expected that the strength of work-hardened steels can be further increased by subsequent aging treatment. Establishment of a strengthening method combining work hardening and age hardening is desired in industrial applications, but such studies have not been conducted on stable austenitic stainless steels though some researches were done for metastable austenitic steels containing deformation-induced martensite.^14–16)

In this study, the work-hardening behavior during wire drawing and the subsequent aging phenomena were investigated in stable austenitic stainless steel (Fe–18%Cr–12%Ni alloy) wires with different N contents in order to evaluate the effect of N from the technical point of view. The mechanism of age hardening was then discussed in terms of the formation of nitrides/clusters/I-S (Interstitial-Substitutional atoms) pairs and their activation energies.

2. Experimental Procedure

Austenitic stainless steels with N contents of 0.103 and 0.198 mass% (referred to as 0.1N and 0.2N steels, respectively) were produced by vacuum melting; their chemical compositions are listed in Table 1. The ingots were hot-forged to produce bar specimens with a diameter of 12.0 mm and then solution-treated at 1373 K for 1.8 ks followed by water cooling to produce as-solution-treated specimens. The diameter of the specimens was decreased to 11.5 mm by surface polishing and pickling. The specimens were then cold-drawn to reduce area by 10%–80%. To prepare an additional specimen with a higher N content, the 80% cold-drawn 0.1N steel with a diameter of 3.0 mm was solution-nitrided¹⁷⁾ at 1473 K for 162 ks in a 0.1 MPa N₂ atmosphere, followed by water cooling. The resultant specimen contained 0.52 mass% N (0.5N steel). This specimen was subsequently cold-drawn to reduce its cross-sectional area by 30%–80%. The cold-drawn wire specimens with different N contents were subjected to aging at 473–1123 K for 1.8 ks. In addition, other plate specimens (chemical composition of Fe–17.9%Cr–11.9%Ni and 3 mm thickness) were prepared to investigate the age-hardening mechanism at low temperatures. N was added to the plate specimens by solution-nitriding at 1473 K for 144 ks in a 0.1 MPa N₂ atmosphere, followed by water cooling. The resultant specimen contained 0.46 mass% N (0.5N steel). The plate specimens thus obtained were subjected to cold rolling at 80% (1st rolling), aging at 473 K for 1.8 ks (1st aging), cold rolling up to 15% (2nd rolling), and aging at 473 K for 1.8 ks (2nd aging).

Table 1 Chemical compositions of steels used in this study (mass%).

The phase fractions of the cold-drawn specimens were estimated by X-ray diffraction (XRD) analysis. XRD measurements with a Cu-Kα radiation source were carried out at 40 kV and 40 mA (Rigaku, RINT2100). Age-hardening behavior in the fabricated specimens was investigated by differential scanning calorimetry (DSC; NETZSCH, DSC404F1 Pegasus) at heating rates of 5–20 K/min. Specimen microstructure was observed using a transmission electron microscope (TEM; JEOL, JEM-3200FSK) operating at an accelerating voltage of 300 kV to identify the nitrides precipitated during aging. The specimens for TEM analysis were prepared by twin-jet electropolishing using an electrolyte solution of 10 vol% HClO₄ and 90 vol% CH₃COOH. A 3D atom probe (3DAP; Ametek, LEAP4000HR) was used to evaluate the distribution of Cr and N atoms. Needle-like specimens for 3DAP were cut from the center of cold-drawn specimens using a scanning electron microscope (Quanta3D) with focused ion beam (FIB) milling. The measurement temperature, pulse fraction, pulse rate, and detection speed used for 3DAP analysis were 50 K, 20%, 200 kHz, and 1%, respectively.

3. Results and Discussion

3.1 Changes in hardness by cold drawing and aging

Figure 1 shows the variation in the hardness of wire-drawn specimens as a function of the drawing ratio (reduction in area). By performing X-ray diffraction, we confirmed that there is no deformation-induced martensite in any of the cold-drawn specimens. The hardness of the as-solution-treated specimens plotted on the vertical axis increases with an increase in N content due to solid-solution strengthening. As the drawing ratio increases, the hardness of the specimens increases via strain hardening; however, there still existed a difference in the hardness values of different N-added steels, which suggests that solid-solution strengthening is effective even after severe cold working. A maximum hardness of 450 HV is observed in 0.5N steel subjected to 80% cold drawing. The solution-treated and cold-drawn specimens of 0.2N steel were then aged at different temperatures and the hardness was also measured (Fig. 2). It can be noted that the hardness of solution-treated specimens hardly changes during aging, irrespective of the aging temperature. In contrast, the drawn specimens exhibit significant age hardening with a peak hardness at ∼800 K; however, beyond 900 K, rapid softening occurred in the specimens due to recrystallization. Because the stored strain energy increases with a reduction in area (by cold drawing), the initial temperature for recrystallization decreases as the drawing ratio increases. The increase in hardness upon aging becomes more significant as the drawing ratio increases; the 0.2N steel exhibits a maximum hardness of 490 HV at a drawing ratio of 80%. This result indicates that the deformation structure, probably dislocations, plays an important role in the formation of precipitates or clusters during aging. In addition, it should be noted that the effect of age hardening appears not only at the peak-aging temperature of 800 K but also at much lower temperatures of 450–600 K.

Fig. 1

Variation in the hardness of wire-drawn specimens as a function of area reduction.

Fig. 2

Variation in hardness as a function of aging temperature in solution-treated and 30%, 50%, and 80% cold-drawn 0.2N steel aged at different temperatures for 1.8 ks.

Because a high cold-drawing ratio is necessary for maximizing the effect of age hardening (Fig. 2), the optimal drawing ratio was set at 80% and the effect of N content on age hardening was investigated. Figure 3(a) and (b) show the changes occurring in the hardness of 0.1, 0.2, and 0.5N steels aged for 1.8 ks and the difference in their hardness with respect to as-drawn specimens as functions of the aging temperature, respectively. Figure 3(b) illustrates that hardening at the peak-aging temperature of 800 K increases with an increase in the N content. Similarly, low-temperature hardening at 450–600 K becomes more prominent as the N content increases.

Fig. 3

(a) Variation in the hardness of 0.1N, 0.2N, 0.5N steels aged for 1.8 ks and (b) difference between the hardness of 0.1N, 0.2N, and 0.5N steels aged for 1.8 ks with respect to as-drawn specimens as a function of aging temperature.

3.2 Investigation of factors promoting age hardening

To understand the chemical reactions resulting in age hardening at elevated temperatures, DSC analysis was performed. Figure 4 shows the DSC curves of 0.2N steels solution-treated and drawn by 30% and 80%, which were subjected to continuous heating from ambient temperature to 1123 K. Two scan cycles were conducted on each specimen, and the difference in exothermic heat between the first and second scans was plotted as a function of temperature. Because the second scan did not result in reactions such as precipitation, the first exothermic reaction can be clearly illustrated by using the result of the second scan as the background. In all specimens, including solution-treated specimens, marked exothermic peaks can be detected at ∼850 K. These peaks correspond to the precipitation of chromium nitrides, mainly CrN and Cr₂N, which are reportedly found in various high-N austenitic stainless steels.^18–20) This was confirmed by the TEM analysis of 0.5N steel aged at 833 K for 1.8 ks after 80% cold rolling as shown in Fig. 5. Here, 0.5N steel was used because it was difficult to observe the nitrides in 0.2N steel. The low-magnification BF-TEM image (Fig. 5(a)) reveals that high-density dislocations remain. From the results of the high-magnification observation (Fig. 5(b)–(e)), the diffraction pattern (Fig. 5(e)) of Cr₂N with a hexagonal close-packed (hcp) structure could be observed. In addition, the DF-TEM image (Fig. 5(c)) obtained from the circled spot in Fig. 5(d) shows Cr₂N particles with sizes of several tens of nanometers. The hardening observed at 800 K (Fig. 2) corresponds to precipitation hardening by these nitrides. However, the nitrides tend to precipitate coarsely at austenite grain boundaries in the solution-treated specimen,²¹⁾ and thus, cold working before aging treatment would be necessary for age hardening. Meanwhile, the two cold-drawn specimens (30% and 80%) exhibit exothermic peaks at ∼500 K, corresponding to the low-temperature age hardening observed in Fig. 2 and 3. However, the peaks observed at ∼500 K could not be explained, even by TEM observations. At such a low temperature, the rate of lattice diffusion of atoms in austenite is very low. For example, the diffusion distance of an N atom at 473 K calculated as $\sqrt{Dt} $ (D: diffusion coefficient (m²/s) and t: time (s)) is just 8.6 × 10⁻⁷ nm/s.²²⁾ Therefore, it is unlikely that nitride precipitates that are observable by TEM are formed in the austenite matrix. In addition, the 80% cold-drawn specimen exhibits an additional peak at 980 K, which is attributed to the recrystallization of deformed austenite.

Fig. 4

DSC analysis of 0.2N steels with 0% (as-solution-treated), 30%, and 80% drawing reduction. The specimens were continuously heated from the ambient temperature to 1123 K at a heating rate of 10 K/min.

Fig. 5

Low (a) and high (b) magnification BF-TEM images, DF-TEM image (c), diffraction pattern (d), and key diagram (e) of 0.5N steel aged at 833 K for 1.8 ks after 80% cold rolling. DF-TEM image (c) was observed at the same region of BF-TEM image (b).

To understand the factors responsible for hardening, the activation energy of each exothermic reaction was estimated by the Kissinger method,²³⁾ which defines the relationship between peak temperature T_p, heating rate β, and activation energy E in DSC analysis as follows,   

\begin{equation} \ln \left(\frac{T_{p}{}^{2}}{\beta} \right) = \frac{E}{RT_{p}} + \ln \left( \frac{E}{AR} \right) \end{equation} (1)

where A and R denote the frequency factor and gas constant, respectively. Figure 6 shows the DSC curves of 80% cold-drawn 0.5N steel at different heating rates. There are three exothermic peaks at temperatures that were the same as the peak temperatures in 0.2N steel (Fig. 4), and they increase slightly as the heating rate increases. The temperatures indicated by the arrows represent peak temperatures and were used for calculations. The relationship between ln(T_p²/β) and 1000/T_p, namely the Kissinger plot, is illustrated in Fig. 7 for nitride precipitation at ∼850 K (Fig. 6A) and the low-temperature (∼500 K) exothermic reaction (Fig. 6B). The corresponding activation energies can be read from the slope of the linear plot. In the case of nitride precipitation, the estimated activation energy is 220 kJ/mol, which is almost similar to the energy required for the lattice diffusion of Cr atoms in face-centered cubic (fcc) iron (243–264 kJ/mol).²⁴⁾ Thus, it is reasonable to explain the reaction rate-controlled by chromium nitride precipitation. Meanwhile, the activation energy estimated for the low-temperature reaction is 77 kJ/mol, which is much lower than that required for the lattice diffusion of constituent elements such as Fe, Cr, and N. Even interstitial N has a significantly higher activation energy of 196–213 kJ/mol.²²⁾ This implies that the low-temperature reaction is not due to nitride precipitation in the austenite matrix; instead, N might cause some reactions via rapid diffusion along densely distributed dislocations. In general, the activation energy for solute atom diffusion along dislocations (pipe diffusion) is believed to be 40%–70% of that of the energy required for lattice diffusion in metallic materials.²⁵⁾ In the current case, the activation energy measured for the low-temperature reaction is approximately half of the energy required for the lattice diffusion of N. Although they could not be observed by TEM, some atomic-scale N products, such as clusters, may have formed on the dislocations.

Fig. 6

DSC analysis of 0.5N steels with 80% drawing reduction. The specimens were continuously heated from the ambient temperature to 1123 K at 5, 10, and 20 K/min.

Fig. 7

Kissinger plots corresponding to exothermic reactions at 850 K (Fig. 6A) and 500 K (Fig. 6B) in 80% cold-drawn 0.5N steel.

To analyze atomic-scale N products formed by low-temperature reactions, 3DAP was conducted on aged 0.5N steel. Figure 8 shows the N and Cr distributions evaluated by 3DAP in solution-nitrided 0.5N steel (Fig. 8(a)) and 0.5N steel aged at 473 K for 1.8 ks after cold drawing at 80% (Fig. 8(b)). In both specimens, no precipitates or clusters are observed; Cr and N atoms appear to be distributed uniformly even in aged specimen where the hardness increased obviously after aging treatment. To further analyze N and Cr distribution, frequency-distribution analysis²⁶⁾ was conducted on the obtained data. The ions detected by 3DAP were divided into subdomains with 100 ions and the relationship between the concentration of target ions in the subdomain and the existence frequency of the subdomain was evaluated. Figure 9(a) and (b) show the results of frequency-distribution analysis of Cr and N atoms, respectively, in the 473 K-aged specimen (Fig. 8(b)). Binomial distribution curves corresponding to a random distribution are also plotted and these curves obtained for both elements are almost the same as the measured results, which means that from a statistical point-of-view, Cr and N are randomly distributed in the 473 K-aged specimen. However, considering that age hardening occurred at 473 K, we should consider that the clusters may be too small to be detected as particles by 3DAP. If a cluster consisted of few atoms, such as I-S pairs, it would be difficult to observe by 3DAP. In particular, in 3DAP analysis, the spatial resolution for interstitial light elements such as N and C is considered to be lower than that of metallic atoms.²⁷⁾ Thus, detecting I-S pairs that may have formed at low temperatures might be beyond the capability of the current 3DAP instrument.

Fig. 8

Cr and N distribution maps generated by 3DAP in 0.5N steel. (a) Solution-nitrided and (b) aged at 473 K for 1.8 ks after cold drawing at 80%.

Fig. 9

Frequency of blocks against (a) Cr and (b) N compositions measured by 3DAP in 0.5N steel aged at 473 K for 1.8 ks after 80% cold drawing.

3.3 Indirect evaluation of I-S pairs formed during low-temperature aging

Based on the results described above, it is hypothesized that some atomic-scale N products, which are difficult to detect by 3DAP, were formed along dislocations through the pipe diffusion of N, which resulted in hardening during low-temperature aging. As such, an I-S pair is the only candidate. In high-N stainless steel, the formation of I-S pairs is believed to increase yield stress and develop dislocation planarity,²⁸⁾ and Cr and N would be paired with each other due to their high chemical affinity. However, as there are few reports on the direct observation of I-S pairs, it is unclear whether Cr and N are present in a ratio of 1:1 or if multiple N atoms are gathered around one Cr atom. In any case, only minute structural changes would occur at the atomic level. Monte Carlo simulations revealed that N atoms tend to be located in the most adjacent octahedral sites with respect to Cr when N can freely diffuse in austenite.²⁹⁾ Therefore, in the current investigation, the same phenomenon might have occurred near the dislocations, where N diffusion is relatively easy.

Because the bonding force of I-S pairs is not high, they can be cut by moving the dislocations.²⁸⁾ Hence, cold working of aged specimens is expected to release dislocations pinned by I-S pairs, leading to work softening. To confirm the existence of I-S pairs using this perspective, as shown in Fig. 10, the hardness of 0.5N steel was measured in specimens aged at 473 K for 1.8 ks after pre-cold rolling at 80% (1st aged specimen: circle symbols) and cold–rolled again at up to 15% after 1st aging (2nd cold-rolled specimen: square symbols). The 2nd cold-rolling step after low-temperature aging reduces the hardness by 10–20 HV, and the value remains constant after 5% cold rolling. In addition, a 2nd aging step (diamond symbols) at 473 K for 1.8 ks after 2nd cold rolling increases the hardness to a value as high as that of the 1st aged specimen. These observations can be explained as follows. Initially, work softening occurred when I-S pairs formed in the 1st aging step were cut or isolated from dislocations during the 2nd cold–rolling step; subsequently, age hardening occurred again due to the formation of I-S pairs along the dislocations during the 2nd aging step. This behavior could be demonstrated by DSC analysis, as shown in Fig. 11. In the 0.5N steel after 1st cold rolling at 80% (Fig. 11(a)), an exothermic peak related to the pipe diffusion of N could be observed at 450 K, similar to the case in Fig. 4, while in the specimen subjected to 1st aging at 473 K for 1.8 ks (Fig. 11(b)), no exothermic peaks appears below the aging temperature, suggesting that I-S pairs were already formed. The occurrence of an exothermic peak in 2nd cold-rolled specimen (Fig. 11(c)) indicates the re-formation of I-S pairs. These characteristics correspond well with the changes in hardness (Fig. 10). Based on such indirect evidence, it may be concluded that low-temperature age hardening in high-N steels occurs via pinning of dislocations by I-S pairs formed by the pipe diffusion of N.

Fig. 10

Hardness of 0.5N steel after 1st aging at 473 K for 1.8 ks (circles), 2nd cold rolling at up to 15% (squares), and 2nd aging at 473 K for 1.8 ks (diamonds).

Fig. 11

DSC analysis of 0.5N steel after (a) 1st cold rolling at 80%, (b) 1st aging at 473 K for 1.8 ks, and (c) 2nd cold rolling at 15%. Heating rate = 10 K/min.

4. Conclusions

1. (1)    Age hardening in nitrogen-bearing austenitic stainless steels increases with an increase in the pre-drawing ratio and nitrogen content. The largest hardening effect was observed when nitrogen-containing specimens were aged at ∼800 K. Particularly, in the case of 0.5N steel with 80% pre-cold drawing, its hardness increased to 550 HV when aged at 833 K for 1.8 ks. In addition, severe drawing induces low-temperature (450–600 K) age hardening.
2. (2)    The highest age-hardening effect observed at ∼800 K may be ascribed to the precipitation of Cr₂N, which is controlled by the lattice diffusion of chromium. Meanwhile, age hardening at 450–600 K is caused by pinning of dislocations by nitrogen products formed on dislocations via pipe diffusion. It is thought that the formed nitrogen products are I-S pairs, as their size is too small to be detected by 3DAP.
3. (3)    Cold rolling the specimen aged at 450–600 K reduces its hardness, that is, the work softening takes place. This phenomenon can be explained by considering that I-S pairs formed at the dislocations are cut or isolated from dislocations upon cold rolling.

REFERENCES

• 1)   N.  Tsuchida,  Y.  Morimoto,  T.  Tonan,  Y.  Shibata,  K.  Fukaura and  R.  Ueji: ISIJ Int. 51 (2011) 124–129. doi:10.2355/isijinternational.51.124
• 2)   J.P.  Bressanelli and  A.  Moskowitz: Trans. Am. Soc. Met. 59 (1966) 223–239.
• 3)   T.  Masumura,  K.  Fujino,  T.  Tsuchiyama,  S.  Takaki and  K.  Kimura: Tetsu-to-Hagané 105 (2019) 1163–1172. doi:10.2355/tetsutohagane.TETSU-2019-062
• 4)   T.  Koizumi: Trans. JSSR 65 (2020) 7–13.
• 5)   M.  Topic,  R.B.  Tait and  C.  Allen: Int. J. Fatigue 29 (2007) 656–665. doi:10.1016/j.ijfatigue.2006.07.007
• 6)   T.  Fukumaru,  H.  Hidaka,  T.  Tsuchiyama and  S.  Takaki: Tetsu-to-Hagané 91 (2005) 828–833. doi:10.2355/tetsutohagane1955.91.11_828
• 7)   N.R.  Baddoo: J. Construct. Steel Res. 64 (2008) 1199–1206. doi:10.1016/j.jcsr.2008.07.011
• 8)   Y.  Ikegami and  R.  Nemoto: ISIJ Int. 36 (1996) 855–861. doi:10.2355/isijinternational.36.855
• 9)   T.  Ogata: J. Cryo. Soc. Jpn. 26 (1991) 18–29. doi:10.2221/jcsj.26.18
• 10)   M.  Talha,  C.K.  Behera and  O.P.  Sinha: Mater. Sci. Eng. C 33 (2013) 3563–3575. doi:10.1016/j.msec.2013.06.002
• 11)   M.  Yoshitake,  T.  Tsuchiyama and  S.  Takaki: Tetsu-to-Hagané 98 (2012) 223–228. doi:10.2355/tetsutohagane.98.223
• 12)   T.  Masumura,  Y.  Seto,  T.  Tsuchiyama and  K.  Kimura: Mater. Trans. 61 (2020) 678–684. doi:10.2320/matertrans.H-M2020804
• 13)   T.  Masumura,  Y.  Seto,  T.  Tsuchiyama and  K.  Kimura: J. Jpn. Soc. Heat Treatment 59 (2019) 222–229. doi:10.14940/netsushori.59.222
• 14)   S.H.  Lee,  J.Y.  Choi and  W.J.  Nam: Mater. Trans. 50 (2009) 926–929. doi:10.2320/matertrans.MRP2008416
• 15)   S.  Yamasaki and  K.  Takano: Mater. Sci. Forum 879 (2016) 2164–2169. doi:10.4028/www.scientific.net/MSF.879.2164
• 16)   K.S.  Kim,  J.H.  Kang and  S.J.  Kim: Mater. Sci. Eng. A 712 (2018) 114–121. doi:10.1016/j.msea.2017.11.099
• 17)   T.  Tsuchiyama,  T.  Ito,  K.  Kataoka and  S.  Takaki: Metall. Mater. Trans. A 34 (2003) 2591–2599. doi:10.1007/s11661-003-0018-z
• 18)   I.  Woo,  M.  Aritoshi and  Y.  Kikuchi: ISIJ Int. 42 (2002) 401–406. doi:10.2355/isijinternational.42.401
• 19)   J.W.  Simmons,  D.G.  Atteridge and  J.  Rawers: J. Mater. Sci. 27 (1992) 6105–6115. doi:10.1007/BF01133757
• 20)   T.-H.  Lee,  S.-J.  Kim,  E.  Shin and  S.  Takaki: Acta Crystallogr. Sec. B 62 (2006) 979–986. doi:10.1107/S0108768106034173
• 21)   J.W.  Simmons: Scr. Mater. 32 (1995) 265–270. doi:10.1016/S0956-716X(99)80048-X
• 22)   A.  Kühl,  D.  Bergner,  H.J.  Ullrich,  M.  Schlaubitz and  P.  Karduck: Mikrochim. Acta 107 (1992) 295–302. doi:10.1007/BF01244484
• 23)   H.E.  Kissinger: Anal. Chem. 29 (1957) 1702–1706. doi:10.1021/ac60131a045
• 24)   A.F.  Smith: Met. Sci. 9 (1975) 375–378. doi:10.1179/030634575790444270
• 25)   R.  Kishore,  R.N.  Singh,  T.K.  Sinha and  B.P.  Kashyap: J. Mater. Sci. 32 (1997) 437–442. doi:10.1023/A:1018569803900
• 26)   M.P.  Moody,  L.T.  Stephenson,  A.V.  Ceguerra and  S.P.  Ringer: Microsc. Res. Tech. 71 (2008) 542–550. doi:10.1002/jemt.20582
• 27)   B.  Gault,  F.  Danoix,  K.  Hoummada,  D.  Mangelinck and  H.  Leitner: Ultramicroscopy 113 (2012) 182–191. doi:10.1016/j.ultramic.2011.06.005
• 28)   X.W.  Zhou and  M.  Grujicic: Calphad 20 (1996) 257–272. doi:10.1016/S0364-5916(96)00029-6
• 29)   M.  Grujicic,  X.W.  Zhou and  W.S.  Owen: Mater. Sci. Eng. A 169 (1993) 103–110. doi:10.1016/0921-5093(93)90604-D