Materials Science and Engineering A 435–436 (2006) 139–144

Solute segregation and microstructure of directionally solidified austenitic stainless steel P.L. Ferrandini, C.T. Rios, A.T. Dutra, M.A. Jaime, P.R. Mei, R. Caram ∗ State University of Campinas, C.P. 6122, Campinas 13083-970, SP, Brazil Received 10 February 2006; received in revised form 28 May 2006; accepted 11 July 2006

Abstract An evaluation was made of liquid to solid transformation during the solidification of an austenitic stainless steel based on an investigation of the material’s solidification mode and microstructure. The solidification mode was studied using directional solidification at low growth rates and differential thermal analysis. The alloy analyzed here showed solidification, with the liquid transforming completely into austenite; however, depending on the solute segregation level, it could present the formation of austenite and ferrite. In addition, most of the alloying elements in this steel showed a partition coefficient of less than 1. © 2006 Elsevier B.V. All rights reserved. Keywords: Phase transformation; Segregation; Microstructure; Austenitic stainless steel

1. Introduction The microstructure of an austenitic stainless steel is dependent on a compositional balance between ferrite and austenite stabilizing elements, mostly nickel and chromium contents [1]. It is accepted that the most effective ␦-ferrite (bcc phase) stabilizer is chromium, while the most effective ␥-austenite (fcc phase) stabilizer is nickel. The microstructure also depends on thermal parameters such as cooling rate, growth rate and the thermal gradients applied during the liquid/solid transformation [2,3]. The discussion about the solidification of these alloys is based on the partial phase diagram of Fe–Ni–Cr [2,4,5]. In addition to chromium and nickel, stainless steels may contain several other elements (Mn, C, Cu, N, Co, V, W, Ti, Nb, Al, Mo and Si) that also affect the austenite–ferrite equilibrium. Somehow, all these elements affect the alloy’s solidification and the extension of this phenomenon led to the definition of equivalent contents for nickel and chromium. The concept of nickel and chromium equivalent contents is a convenient way to represent the effect of several elements on the basic crystal structure of austenitic stainless steels. Information reported in the literature allows one to evaluate the influence ∗

Corresponding author. Tel.: +55 19 37883314; fax: +55 19 37883314. E-mail address: [email protected] (R. Caram).

0921-5093/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2006.07.024

of these elements on the transformation of phases. Eqs. (1) and (2), respectively, show the Ni and Cr equivalent contents [6–9]: NIeq = %Ni + a(%Mn) + b(%C) + c(%N)

(1)

Cr eq = %Cr + d(%Si) + e(%Mo) + f (%Ti) + g(%Nb)

(2)

Coefficient values for both equations are shown in Table 1. The effect of alloying elements on the solidification microstructure can be classified into two groups: ␦- and ␥phase stabilizers, as well as the equivalent Ni and Cr contents, from which the different solidification modes can be established. The solidification modes of austenitic stainless steels are controlled by both primary phase solidification and subsequent solid transformations [10]. The solidification mode is related to the alloy’s composition, expressed by the ratio Creq /Nieq as follows [8]. 1.1. Mode A This mode occurs when Creq /Nieq < 1.25. The liquid transforms completely into austenite. Austenite is stable throughout the entire process and the crystalline structure is totally fcc: transformations :

liquid → (liquid + ␥) → ␥

140

P.L. Ferrandini et al. / Materials Science and Engineering A 435–436 (2006) 139–144

Table 1 Coefficient values used for obtaining the equivalent contents of Ni and Cr

liquid → (liquid + ␥) → (liquid + ␥ + ␦) → (␥ + ␦)

phase becomes Ni-rich and the Cr content remains unchanged. These partition coefficient values may change in response to the presence of other alloying elements, possibly causing the Ni and Cr contents of the liquid phase to increase, which may, in turn, lead to the formation of another phase at the end of the process. The solidification of metallic alloys usually results in a dendritic morphology. The dendritic structure is composed of primary, secondary, tertiary and sometimes higher order arms, resulting in a complex system whose dendritic arms contain second phases and precipitated particles. Dendritic growth is very important in the solidification of metal alloys, since this kind of array affects various features such as micro and macro segregation, porosity distribution and the mechanical properties of the end product. In austenitic steels, the dendritic structure depends on the solidification mode. The solidification process around a dendritic arm depends on the partition of solute in the solidifying liquid phase and the occurrence of diffusion within the solid. Therefore, the main purpose of this study was to evaluate the solute segregation and microstructure of an austenitic stainless steel under different solidification conditions.

1.3. Mode FA

2. Experimental methods

Occurs when 1.48 < Creq /Nieq < 1.95. At the end of the process the microstructure contains ferrite and austenite. The solidification begins with the formation of ferrite and the remaining liquid becomes rich in ␥-stabilizer during the solidification, transforming into austenite. Hence, the final microstructure presents vermicular ␦ as a primary phase within the dendritic arms enveloped by the austenite:

Samples were prepared from a commercial AISI 316L austenitic stainless steel, whose composition is shown in Table 2. The transformation temperatures were determined by differential thermal analysis (DTA) using standard alumina crucibles (5.0 mm diameter and 5.0 mm height) under vacuum and argon atmosphere, and applying heating/cooling rates of 5 and 10 ◦ C/min. No evidence of sample oxidation was detected. Directional solidification was performed using a Bridgman device equipped with a vacuum system combined with injection of high purity argon. The samples obtained were 100.0 mm long and 6.5 mm in diameter, and were grown at 1.0, 5.0 and 15.0 cm/h. The temperature gradient was 80 ◦ C/cm and the hot zone temperature was set at 1600 ◦ C, controlled by an optical pyrometer. The directionally solidified samples were sectioned for scanning electron microscopy and prepared for optical microscopy using HCl/HNO3 /H2 O solution. Energy-dispersive spectroscopy (EDS) and X-ray fluorescence were utilized to determine the compositions.

a

b

c

d

e

f

g

Reference

0.5 0.5 0.5 0.5

30 30 30 30

– 30 – –

1.5 1.5 1.5 1.5

1 1 1 1

– – – 2

0.5 0.5 0.5 0.5

[6] [7] [8] [9]

1.2. Mode AF Occurs when 1.25 < Creq /Nieq < 1.48. At the end of the process, the microstructure contains austenite and ferrite. The solidification begins with the formation of austenite and during the process, the composition of the liquid phase may undergo some modification, becoming rich in ␦-stabilizer. This leads to the formation of ferrite at the dendrite boundaries at the end of the solidification process: transformations :

transformations : liquid → (liquid + ␦) → (liquid + ␦ + ␥) → (␦ + ␥) 1.4. Mode F Occurs when Creq /Nieq > 1.95. Solidification leads to the complete formation of ferrite, which may partially transform into austenite depending on the Creq /Nieq ratio. Ferrite is stable during the solidification and transforms into austenite during cooling, which leads to the formation of the Widmanst¨atten structure. In this case, the dendritic structure is not easily observable: transformations :

liquid → (liquid + ␦) → ␦ → (␦ + ␥)

In the above-described transformations, the formation of ferrite and austenite from liquid depends on the different solute partition coefficients, k, of Cr and Ni [2]. According to the literature [2], during the solidification of ferrite, the Ni partition ␦ ) is close to 0.85 while the Cr coefficient (k ␦ ) coefficient (kNi Cr varies from 0.98 to 1.15. In austenite solidification, the solute ␥ ␥ segregation coefficients are kNi ≈ 0.98 and kCr ≈ 0.92. Therefore, austenite solidification causes Ni and Cr to segregate in the liquid phase, whereas during ferrite solidification, the liquid

Table 2 Chemical composition of the austenitic stainless steel (wt.%) Fe Cr Ni C Mn Mo Si Cu P

Balance 16.6 10.1 0.03 1.4 2.2 0.47 0.20 0.001

P.L. Ferrandini et al. / Materials Science and Engineering A 435–436 (2006) 139–144

141

Table 3 Compositions (wt.%) obtained at positions close to the planar/cellular transition and in the dendritic microstructure (positions 2–4, see Fig. 5b and positions 5–7, see Fig. 6)

Fig. 1. Vertical section of the Fe–Cr–Ni phase diagram showing the solidification path of a 70% Fe alloy (wt.%) [7].

3. Results and discussion The DTA plots show some variations in the liquidus and solidus temperatures TL (L
L + ␥) and TS (L + ␥
␥) as the heating and cooling rates changed. It should be noted that

Position

Cr

Ni

Mo

Mn

Cu

Si

P

2 3 4 5 6 7

17.7 30.9 18.8 16.8 18.9 16.8

11.3 6.0 12.1 12.7 13.3 12.4

2.0 8.1 1.6 1.2 2.1 1.2

3.1 4.0 3.4 1.8 2.2 1.8

1.3 1.0 0.9 0.6 0.9 0.8

0.4 0.8 0.5 0.6 0.9 0.6

0.0 0.3 0.0 0.0 0.0 0.0

the solidification mode A comprises the transformations liquid → (liquid + ␥) → ␥. A comparison of the results obtained at cooling rates of 5 and 10 ◦ C/min reveals different temperature transformation values which are caused by the difficulty of measuring equilibrium conditions. Low solidification rates yield more coherent values. Depending on the solidification rate and the solute segregation level, the AF solidification mode may occur, which involves the transformations liquid → (liquid + ␥) → (liquid + ␥ + ␦) → (␥ + ␦). This solidification path is observed at low solidification rates. Very little ferrite and energy are formed during the transformation. Hence, at higher cooling rates, the DTA analysis was unable to detect

Fig. 2. Microstructural evolution during the directional solidification of the Fe–Cr–Ni austenitic stainless steel (longitudinal view): (a) austenite growth presenting plane solid/liquid interface; (b) cellular growth; (c) cellular/dendritic transition and (d) dendritic growth.

142

P.L. Ferrandini et al. / Materials Science and Engineering A 435–436 (2006) 139–144

ferrite formation. Using cooling rate of 5 ◦ C/min, TL was found to be 1434 ◦ C and TS 1387 ◦ C, while cooling rate of 10 ◦ C/min resulted in 1395 ◦ C and 1323 ◦ C, respectively. Fig. 1 depicts a vertical section of the Fe–Cr–Ni phase diagram, which shows the cooling of a 70% Fe alloy that is close to the composition of the Fe–Cr–Ni alloy analyzed here. This diagram allows one to track the solidification path and to check the coherency between phase transformations, microstructures and temperatures obtained by DTA analysis. Fig. 1 allows one to estimate TL , TS and the small amount of ␦ formed during solidification. According to the literature [1], although the Creq /Nieq ratio is almost 1.4, the alloy investigated here presented a solidification mode A. When the Creq /Nieq ratio increases, some interdendritic ferrite may solidify, in which case the solidification mode is AF. Several authors [1,2] concluded that most of the partition coefficients of this alloy are lower than 1 when the solidification mode is A. In this case, the primary phase is austenite and the central part of the dendritic arms is solute-poor, since the alloying elements are segregated in the liquid. Our samples were directionally solidified in order to evaluate their solidification mode, solute segregation and microstructural evolution. Directional solidification experiments allowed for evaluations of the intensity of solute distribution and the morphological evolution, since the directional solidification process allows one to control the direction of heat extraction, the solidification rate and, to a certain extent, the thermal gradient (G ≈ 80 ◦ C/cm). Fig. 2 shows the microstructural evolution of a directionally solidified sample, where fs = x/L, x is the position on the sample and L is the sample’s length. The ingot’s initial part, up to fs = 0.6, was completely austenitic and displayed a few columnar grains, indicating that a planar solid/liquid interface occurred during solidification. As the process continued, segregation of the solute caused the remaining part of the sample to become richer in solute. Our investigation of the solute profile led us to conclude that all the elements in this sample presented a partition coefficient of k < 1, which is in partial agreement with previously obtained data on solute segregation [1]. The results obtained here are not congruent with those presented by Kerr ␥ and Kurz [2], who mentioned that kNi is higher than 1. Fig. 3 shows the distribution of the sample’s elements, indicating that the Cr content changed very little from the beginning to the final portion of the sample, which means that Cr presented k close to but lower than 1 during the formation of austenite. The composition was identified by X-ray fluorescence analysis of 3 mm diameter cross-sections along the sample. With respect to the behavior of Ni, the data found in the literature suggest that this element has a value of k equal to or slightly higher than 1. However, the concentration profiles obtained here clearly show ␥ that, like Cr, kNi is close to but lower than 1. Mo presented a lower partition coefficient than did Cr and Ni, since the distribution profile indicates that the Mo content in the last solidified region of the sample was considerably higher than in the first solidified region. The Mn distribution indicated a slightly lower value of k than those of Cr and Ni. Cu was distributed evenly throughout the sample, leading us to conclude that its partition

Fig. 3. Solute distribution profile after directional solidification at V = 1.0 cm/h.

coefficient was the closest to 1. The distribution profile shows that Si presented k < 1 and was strongly segregated. Since the data reported in the literature suggest that Cr presents k = 0.9, a rough estimate allows us to propose that Si presents k ≤ 0.7. Finally, the distribution of P indicated that k was lower than 1 and that this element was strongly segregated. The distribution of solute in the sample was altered by the directional solidification process and led to a small alteration of the Creq /Nieq ratio, as indicated in Fig. 4. The Creq /Nieq ratio was calculated from Eqs. (1) and (2) and the coefficients given by Rajasekhar et al. [8] are listed in Table 1. The results obtained in this study indicate that the Creq /Nieq ratio was lowest in the initial portion of the sample (Creq /Nieq ≈ 1.4) and highest in the final portion of the sample (Creq /Nieq ≈ 1.5). These Creq /Nieq values may cause some ferrites to solidify in the interdendritic regions, though only austenite was formed while the solid/liquid interface remained smooth. It is worth noting that the Creq /Nieq ratio variation was determined under conditions of near equilibrium,

Fig. 4. Creq /Nieq ratio evolution through the directionally solidified sample at V = 1.0 cm/h.

P.L. Ferrandini et al. / Materials Science and Engineering A 435–436 (2006) 139–144

143

Fig. 6. Directionally solidified dendritic microstructure (fs ≈ 0.7) at a 1.0 cm/h growth rate (positions 5 and 7 indicate dendritic arm cores, while position 6 indicates an interdendritic region). Fig. 5. Magnification of position 1 in Fig. 2c: (a) Cr and Mo carbides forming a eutectic structure; (b) magnified view of (a) showing specific regions where compositions were determined by EDS.

when low growth rates were applied, thus differing considerably from those of a welding process, for example, where high growth rates are the norm. The chemical composition was determined by X-ray fluorescence, so the carbon content could not be checked accurately, although there was evidence of its presence. As the directional solidification takes place, carbon is segregated to the liquid and finally, when its content reaches a certain level, carbides are precipitated in the form of eutectic structures, as shown in Fig. 5. This figure provides a magnified view of position 1 in Fig. 2c. Our EDS analysis revealed higher content of Cr and Mo, both strong carbide generators. In addition, solutes rejected by the freezing solid progressively increased the liquid composition so that it caused constitutional supercooling. Hence, the initially smooth solid/liquid interface became unstable and degenerated. Table 3 lists the contents of several elements in the positions specified in Fig. 5, indicating that the contents of carbide generator elements were considerably higher than the average contents found in austenite. While the Cr content in the austenitic phase was almost 18%, as expected, in the eutectic structure it was higher than 30% (wt.%). Ni, on the other hand, displayed the opposite behavior. The eutectic structure presented a Ni content of almost 6% while the austenitic structure contained almost 12% Ni. The transition from planar to cellular and dendritic growth was detected as the solidification rate rose from 1.0 to 5.0 cm/h. Higher growth rates tended to hinder solute segregation, since less time was available for the segregated atoms to migrate to a position far from the solid/liquid interface. A 5.0 cm/h growth rate almost eliminated the solute’s distribution. Phosphorus was the only exception, segregating to the final portion of the sample even at a high growth rate.

Up to this point, our considerations on solute distribution refer to long-range segregation and may be called macroscopic. When considering a casting or welding process, the most common morphology is dendritic and the growth of each individual arm involves solute segregation, just like the directional solidification process, albeit in a microscopic range. To evaluate the microscopic solute distribution, the compositions around a dendritic arm grown at V = 1.0 cm/h were analyzed. Fig. 6 shows the dendritic microstructure (fs ≈ 0.7) used for the microscopic segregation analysis and Table 3 shows the compositions obtained. Positions 5 and 7 indicate dendritic arm cores, while position 6 indicates an interdendritic region. The compositions identified clearly showed that microscopic solute segregation occurred during dendritic growth and that Cr, Ni, Mo, Mn, Si and P exhibited a partition coefficient lower than 1. Again, Cu did not show a specific trend, as long as the dendritic arm number 5 presented low Cu content while arm number 7 presented a higher Cu content, which was entirely unexpected. Cu probably presented k ≈ 1 during the formation of austenite. The results obtained from the directional solidification experiment carried out at V = 15.0 cm/h are congruent with those reported by other researchers, except when the high solidification rate allowed for growths other than the dendritic. 4. Conclusion The solidus and liquidus temperatures of the Fe–Cr–Ni stainless steel, which were determined by differential thermal analysis, were 1387 and 1434 ◦ C, respectively. The usual solidification path was found to be: liquid → (liquid + ␥) → ␥, but depending on the solute segregation level, the liquid composition was altered and the solidification path was: liquid → (liquid + ␥) → (liquid + ␥ + ␦) → (␥ + ␦). In this case, the formation of ferrite occurred in the interdendritic region. The alloy’s directional solidification at 1.0 cm/h revealed that Cr, Ni, Mo, Mn, Si and

144

P.L. Ferrandini et al. / Materials Science and Engineering A 435–436 (2006) 139–144

P presented a partition coefficient of k < 1, while Cu presented k ≈ 1. We also found that solidification at low growth rates allows one to obtain a planar solid/liquid interface. Solute segregation led to the formation of Cr and Mo carbide in the form of a eutectic structure, which caused planar growth degeneration and cellular growth, followed by dendritic growth.

References [1] [2] [3] [4] [5] [6]

Acknowledgments The authors gratefully acknowledge the Brazilian research funding agencies FAPESP (State of S˜ao Paulo Research Foundation) and CNPq (National Council for Scientific and Technological Development) for their financial support of this work.

[7] [8] [9] [10]

S. Fukumoto, W. Kurz, ISIJ Int. 37 (7) (1997) 677–684. H.W. Kerr, W. Kurz, Int. Mater. Rev. 41 (4) (1996) 129–163. S. Fukumoto, W. Kurz, ISIJ Int. 39 (12) (1999) 1270–1279. T. Okane, T. Umeda, ISIJ Int. 38 (5) (1998) 454–460. S. Fukumoto, W. Kurz, ISIJ Int. 38 (1) (1998) 71–77. N. Suutala, T. Takalo, T. Moisio, Metall. Trans. A 11A (1980) 717– 725. J.A. Brooks, A.W. Thompson, Int. Mater. Rev. 36 (1) (1991) 16–44. K. Rajasekhar, C.S. Harendranath, R. Raman, S.D. Kulkarni, Mater. Charact. 38 (1997) 53–65. T. Takalo, N. Suutala, T. Moisio, Metall. Trans. A 10A (1979) 73– 1181. M.J. Cieslak, A.M. Ritter, W.F. Savage, Weld. J. 61 (1) (1982) 1S–8S.