Introduction

The solidification microstructure of austenitic stainless steel has always been the interest of researches in academia and industry because it determines the castability, weldability, hot workability, mechanical properties, and corrosion resistance [14]. In austenitic stainless steels, a three-phase reaction region (L + δ + γ), which can be either eutectic or peritectic, exists for compositions of over 15 wt% Cr and 10 wt% Ni according to the Fe–Cr–Ni ternary phase diagram [5, 6]. Therefore, the solidification microstructure, which mainly depends on both composition and cooling rate, is complex as a result of the complicated three-phase reaction [5, 7]. Suutala [8] investigated the solidification conditions on solidifying sequence of a range of AISI 300 series steels by autogenous gas tungsten arc (GTA) welding and concluded that the composition was of primary importance while the cooling rate was only of secondary importance. However, a given austenitic stainless steel with composition passing through the Cr-rich part of the three-phase region can solidify with primary δ ferrite or primary γ phase under different cooling rates [810]. The different solidifying sequences would result in change of elements redistribution path, and thus may alter the type of three-phase reaction and the solidification microstructure. Ma et al. [11] and Fu et al. [12, 13] investigated the detailed microstructural evolution process in directional solidified 304 austenitic stainless steel under estimated cooling rates of 3.3, 1, and 4 °C s−1, respectively. They concluded that eutectic reaction (L → γ + δ) which resulted in the formation of coupled structure occurred among the dendrite arms after primary δ ferrite precipitated from liquid. Liang et al. [14] observed a different phenomenon in 301 austenitic stainless steel at cooling rates of 4–25 °C s−1 under non-directional solidification condition using differential thermal analysis (DTA). They found that peritectic reaction and eutectic reaction coexisted in the microstructure of the sample cooled at 25 °C s−1. In addition, many other researchers [1517] investigated the solidification microstructure of various austenitic stainless steels by sorts of welding methods which have much higher cooling rate than casting and directional solidification. However, the three-phase reaction mechanism is still unclear and it was not directly observed in the above-mentioned literatures owing to the limitations of test method. Confocal scanning laser microscope (CSLM) enables the in situ observation of phase transformation at high temperature, as shown in references [1820]. Huang et al. [19] and McDonald et al. [20] observed the δ/γ interface and microstructure evolution during the peritectic reaction at a cooling rate of 0.05 °C s−1 and constant undercooling degree, respectively. Nevertheless, the effect of larger cooling rate, which may be confronted in many types of conventional casting processes, on the microstructure evolution has not been studied, nor has been the in situ observation.

AISI 316LN steel, a type of nitrogen-alloyed ultralow carbon (<0.02 wt%) stainless steel, is used as the material of main pipelines in AP1000 pressurized water reactor (PWR) which is about to commercially serve in China. The pipes are manufactured by integral hot forging and the ingots for hot forging are obtained by mold casting or electro slag remelting (ESR) which has typical cooling rates of 0–100 °C s−1 [17]. Therefore, in this regard, the studies of phase transformation in 316LN during solidification under different cooling rates would be of interest.

In this paper, two 316LN stainless steels with different compositions were used to investigate their solidification sequence and microstructure evolution. The three-phase reaction was in situ observed in CSLM under different cooling rates and the variation of chemical composition in different phases was discussed.

Experimental

Two ingots of AISI 316LN stainless steel with different values of Creq/Nieq were produced by vacuum induction furnace (VIF) and the casting was carried out at a constant superheat of 50 °C in a cast-iron ingot mold. The ingots were forged at 1200 °C and then solution treated at 1100 °C for 4 h. The chemical compositions are given in Table 1.

Table 1 Chemical compositions of AISI 316LN stainless steel

The cooling rate of this type of mold casting was estimated to be ~2.00 °C s−1 according to the relationship of secondary dendrite arm spacing (SDAS, λ2) to cooling rate (ε): λ2 = 68 × (ε)−0.45 [17, 21, 22]. In order to directly observe the microstructural evolution under different cooling rates, a CSLM (VL2000DX, Lasertec, Japan) was used. The principle of this equipment is described in many literatures [19, 20]. The disc samples with dimensions of φ7 × 3 mm for observation were machined from the solution-treated materials, and then they were mechanically polished with diamond paste. Subsequently, they were put in an alumina crucible which was installed into a furnace at the focal point of the sample chamber. After the chamber was filled with argon, the samples were heated to the melting points and kept for 60 s, and next they were cooled down to 1100 °C with cooling rates of 2, 10, and 100 °C s−1, respectively (see Fig. 1).

Fig. 1
figure 1

Schematic of heating and cooling curves of specimens during in situ observation

After the samples were cooled down to room temperature, they were prepared by conventional process for metallographic observation. The etchant solution contains 0.5 g K2S2O5, 20.0 g NH4FHF, and 100 ml distilled water (Beraha’s etchant modified by Lichtenegger [16]), whose pH value was maintained at about 2.5 by addition of NH4OH or HNO3. Electron probe microanalyzer (EPMA) (JEOL JXA-8230, Japan) was employed to characterize the variation of chemical composition. The volume fractions of primary δ ferrite and primary austenite during in situ observation were measured by image analysis using software of Image Tool Version 3.0.

Results

As-cast microstructure of mold casting ingots

The as-cast microstructures of mold casting 316LN-1 and 316LN-2 are shown in Fig. 2. Beraha’s etchant makes Cr-rich areas show light color contrast and Cr-depleted areas darker (blue or green) color contrast [16]. In other words, ferrite remains unaffected showing as whiteness, while austenite exhibits a range of colors depending on its Cr concentration. Therefore, the white island-like phase which lies in interdendritic areas in Fig. 2a is ferrite. This indicates that austenite is primary phase and δ ferrite forms at the final stage of solidification in alloy 316LN-1. While in 316LN-2, skeletal δ ferrite exists in dendrite branches (see Fig. 2b), which means that δ ferrite precipitates from liquid firstly and austenite forms later. The solidification behaviors during mold casting process may approximate the equilibrium process because the cooling rate of ~2 °C s−1 is relatively small in the range of conventional casting processes. When the cooling rate becomes high enough, the solidification behavior will deviate from the equilibrium process, and different solidification microstructures may appear.

Fig. 2
figure 2

As-cast microstructure of 316LN stainless steel: a 316LN-1, b 316LN-2

Solidified microstructures of 316LN under different cooling rates

Figure 3 presents the solidified microstructures of alloy 316LN-1 under different cooling rates. When the steel was cooled at 2 °C s−1, δ ferrite exists in both interdendritic areas and dendrite branches, as shown in Fig. 3a and b. However, the interdendritic δ ferrite is predominant. The dendrite branch shows light color contrast which means that its Cr concentration is higher than adjacent area (see Fig. 3b). This characteristic is not consistent with the feature of elements segregation in Fig. 2a, which may be associated with the fact that the steel was not fully melted owing to the limitation of the equipment as the temperature reached the peak point (see Fig. 4). The unmelted parts, which are richer in Cr element as a result of redistribution of elements during the reaction of L + δ + γ → L + γ on heating, become dendrite branches as the solidification goes on. When the temperature decreases, the dendritic δ ferrite appears owing to the higher Cr concentration.

Fig. 3
figure 3

In situ observation and metallurgical micrographs of the solidified microstructure in alloy 316LN-1 under different cooling rates. a, b 2 °C s−1; c, d 10 °C s−1; e, f 100 °C s−1

Fig. 4
figure 4

High temperature microstructure of alloy 316LN-1

As the cooling rate increases, δ ferrite still mainly forms in interdendritic areas which can be seen in Fig. 3c–f, but the amount of dendritic δ ferrite is reduced. The shape of ferrite changes from island-like to lacy-like with increasing of cooling rate due to the shortening of diffusion path [23]. Furthermore, both the dendrite arm spacing (DAS) and the mean free path between δ ferrite decrease owing to the inhibition of element redistribution and quicker heat rejection under higher undercooling [17]. Overall, in view of the larger variances of content changes at 10 °C s−1, contents of Cr and Mo can be thought to decrease slightly in both δ ferrite and γ phase while Ni increases in γ phase and decreases in δ ferrite in a small scale with the increasing of cooling rate, as shown in Fig. 5. The larger variances at 10 °C s−1 may be caused by the bigger scattering of measurement points.

Fig. 5
figure 5

Effect of cooling rate on the variations of composition in 316LN-1

The solidified microstructures of alloy 316LN-2 under different cooling rates are shown in Fig. 6. δ ferrite precipitates in dendrite branches under cooling rate of 2 °C s−1, and a coupled growth microstructure of δ + γ was observed when cooled at 10 and 100 °C s−1. In particular, δ ferrite becomes coarser due to the suppression of time for solid-state transformation from δ to γ when cooling rate increases. In pace with this coarsening behavior, the element contents in δ ferrite and γ phase change regularly, as shown in Fig. 7. Cr and Mo increase and Ni decreases in δ ferrite, while in γ phase they change in a reverse direction with the increasing of cooling rate. This phenomenon is thought to be caused by the inhibition of element diffusion during δ → γ transformation as the cooling rate increases [11, 17]. It can be seen that the contents change trend in 316LN-2 is more obvious than that in 316LN-1 and their variation trend of Cr and Mo in δ ferrite is opposite. This can be understood based on that the δ ferrite in 316LN-1 generates through the reaction of L → δ + γ at the final stage of solidification, as will be discussed in next section. As the cooling rate increases, the amount of remaining liquid between primary austenite decreases and therefore the amount of δ ferrite decreases. Thus, the contents of Cr and Mo in δ ferrite decrease, while they increase in 316LN-2 because its δ ferrite is primary phase and the outward diffusion rate of Cr and Mo from δ ferrite was suppressed by increasing cooling rate.

Fig. 6
figure 6

In situ observation and metallurgical micrographs of solidified microstructure in alloy 316LN-2 under different cooling rates. a, b 2 °C s−1; c, d 10 °C s−1; e, f 100 °C s−1

Fig. 7
figure 7

Effect of cooling rate on the variations of chemical composition in alloy 316LN-2

Discussion

Solidification mode

From the above results, the precipitation sequences and morphologies of ferrite can vary, depending on the composition and cooling rate. This can be illustrated by the concept of solidification mode. The solidification mode of austenitic stainless steel can be divided into four types according to the value of Creq/Nieq [8, 24]:

$$ \begin{array}{ll} {{\rm{F mode}}:{\rm{L}} \to {\rm{L}} + \delta \to \delta \to \delta + \gamma }&{{\rm{C}}{{\rm{r}}_{{\rm{eq}}}}/{\rm{N}}{{\rm{i}}_{{\rm{eq}}}} > {\rm{2}}.00}\\ {{\rm{FA mode}}:{\rm{L}} \to {\rm{L}} + \delta \to {\rm{L}} + \delta + \gamma \to \delta + \gamma \to \gamma }&{1.50 < {\rm{C}}{{\rm{r}}_{{\rm{eq}}}}/{\rm{N}}{{\rm{i}}_{{\rm{eq}}}} < 2.00}\\ {{\rm{AF mode}}:{\rm{L}} \to {\rm{L}} + \gamma \to {\rm{L}} + \delta + \gamma \to \gamma + \delta \to \gamma }&{1.37 < {\rm{C}}{{\rm{r}}_{{\rm{eq}}}}/{\rm{N}}{{\rm{i}}_{{\rm{eq}}}} < 1.50}\\ {{\rm{A mode}}:{\rm{L}} \to {\rm{L}} + \gamma \to \gamma }&{{\rm{C}}{{\rm{r}}_{{\rm{eq}}}}/{\rm{N}}{{\rm{i}}_{{\rm{eq}}}} < 1.37} \end{array} $$

where A and F refer to austenite and ferrite, respectively. The values of Creq and Nieq can be calculated using the following Hammar and Svensson equations [24]:

$$ {\text{Cr}}_{\text{eq}} = {\text{ Cr}} + 1. 3 7 {\text{Mo}} + 1. 5 {\text{Si}} + 2 {\text{Nb}} + 3 {\text{Ti}} $$
(1)
$$ {\text{Ni}}_{\text{eq}} = {\text{ Ni}} + 2 2 {\text{C}} + 1 4. 2 {\text{N}} + 0. 3 1 {\text{Mn}} + {\text{Cu}} $$
(2)

For the alloys 316LN-1 and 316LN-2 studied here, the ratios of Creq to Nieq equal 1.47 and 1.55, respectively. Hence, alloy 316LN-1 falls into AF mode while alloy 316LN-2 follows FA mode. This indicates that austenite is the primary phase in 316LN-1 and the precipitation of δ ferrite occurs first in 316LN-2, then three-phase reaction (L + δ + γ) comes about at the terminal stage of solidification in both alloys. Subsequently, solid-state transformation of δ → γ continues below solidus lines [12, 25]. Therefore, the results predicted by Hammar and Svensson equations work fairly concordant with the solidified microstructure of mold casting in Figs. 3 and 6.

In respect of the effect of cooling rate on the solidification mode, the results in Figs. 3 and 6 show that this effect can be almost ignored in the range of 0–100 °C s−1, except that a small region of primary ferrite which co-existed with primary austenite was found in alloy 316LN-2 when it was cooled at 100 °C s−1 (see Fig. 8). That is to say, the solidification mode of FA type in 316LN may change to AF mode if the cooling rate is greater than 100 °C s−1. In this case, the available time for segregation in front of dendrite tip becomes less on account of the high cooling rate [17]. Consequently, the dendritic growth transforms into cellular growth, which results in the achievement of critical value of undercooling degree for metastable austenite formation [8, 17, 26]. While in alloy 316LN-1, its solidification mode may change to A mode under large enough undercooling conditions because the precipitation of interdendritic ferrite will be suppressed by high cooling rate, whereas it does not occur under current conditions.

Fig. 8
figure 8

Solidified microstructure of alloy 316LN-2 cooled at rate of 100 °C s−1

Solidification process of 316LN

Figure 9 presents some representative micrographs of phase formation during in situ observation in alloy 316LN-1 when it was cooled at 2 °C s−1. Based on AF solidification mode, the solidification process can be described as below: when the melt undercooling degree reached a critical value, primary γ phase appeared in liquid and grew in the form of dendrite with the temperature decreasing (see Fig. 9a, b). During this process, Cr and Mo were rejected to liquid [16] and the austenite liquidus line was also lowered down, which both favored for the formation of δ ferrite. Subsequently, a eutectic reaction of L → γ + δ was observed among the primary austenite dendrite when the temperature dropped to 1464.9 °C (see Fig. 9c). As temperature reached 1349.3 °C, the liquid phase disappeared and the solid-state transformation of δ → γ occurred. As discussed in the above section, the solidification mode and δ ferrite’s distribution in 316LN-1 did not change as the cooling rate increased, so did the solidification process. Figure 10 shows the phases evolution during in situ observation in 316LN-1 at 100 °C s−1, and we can see that the characteristics of phases formation are similar to that in Fig. 9 except the corresponding time and temperature.

Fig. 9
figure 9

In situ observation of phase formation during solidification in alloy 316LN-1 at 2 °C s−1. a, b Primary austenite appeared and grew in liquid; c eutectic reaction occurred among the dendrite; d liquid disappeared at 1349.3 °C

Fig. 10
figure 10

In situ observation of phase formation during solidification in alloy 316LN-1 at 100 °C s−1. a Primary austenite grew in liquid; b eutectic reaction occurred among the dendrite

In alloy 316LN-2, different features of phase evolution were observed when it was also cooled at 2 °C s−1. Figure 11a and b show that primary δ ferrites precipitated along impurities and grew into liquid with the temperature decreasing. As this process went on, Ni was partitioned to the remaining liquid while Cr and Mo were absorbed into δ ferrites [13]. Subsequently, γ phase formed at the boundaries of δ ferrite and liquid through the peritectic reaction of L + δ → γ, as shown in Fig. 11c, d. At temperature of 1377.90 °C, liquid phase disappeared and the solid-state transformation from δ ferrite to austenite occurred (see Fig. 11e). Concurrently, Cr was rejected and Ni was accepted by austenite, which resulted in the increasing of austenite’s relative stability. However, as a result of the limitation of elements diffusion on cooling, the transformation from δ ferrite to austenite was uncomplete, and therefore, skeletal ferrite, which is enriched in Cr and Mo while depleted in Ni, was retained in the core area of dendrites (see Fig. 11f). Particularly, it should be mentioned that the impurities were mostly retained around the boundaries of δ ferrite and interdendritic austenite because the solidifying interfaces can push the impurities if the critical radius for pushing or engulfment transition of non-metallic is larger than inclusion size [27, 28]. Liang et al. [29] also observed this phenomenon in their work and thought that δ ferrite can absorb the inclusions moving near the δ-L interface.

Fig. 11
figure 11

In situ observation of phase formation during solidification in alloy 316LN-2 at 2 °C s−1. a, b Primary ferrite appeared and grew in liquid; c, d peritectic reaction occurred at ferrite/liquid boundaries; e liquid disappeared at 1377.9 °C and solid state transformation occurred; f skeletal ferrite was retained in core area of dendrites

The biggest difference between the solidification process of alloy 316LN-1 and 316LN-2 is the precipitation sequence of ferrite and austenite, which has been discussed by many authors [16, 17, 30, 31]. In addition, the alloys behaved in different types of reaction in three-phase region, i.e., eutectic reaction in 316LN-1 and peritectic reaction in 316LN-2, as shown in Figs. 9c and 11c. Figure 12 shows the equilibrium phase diagrams calculated by JMatPro software (version 6.1) and illustrates the positions of 316LN-1 and 316LN-2 in vertical section of Fe–Cr–Ni ternary diagram at constant Fe content. Based on the above results of observation and analysis, it is certain that the melt undercooling degree before solidification process which started did not reach the difference between the liquidus temperature of austenite \( T_{L}^{\gamma } \) and the eutectic trough temperature (T E , indicated by point E in Fig. 12b) in 316LN-1 and that between \( T_{L}^{\delta } \) and T E in 316LN-2. Therefore, when the critical undercooling degree for austenite nucleation (ΔT γ ) in 316LN-1 and that for δ ferrite nucleation (ΔT δ ) in 316LN-2 were reached, austenite or δ ferrite firstly precipitated from liquid phase, respectively. However, the volume fraction of primary δ ferrite (φδ = 85.4 % in Fig. 9c) in 316LN-2 was more than that of primary austenite (φγ = 50.2 % in Fig. 11c) in 316LN-1, because of the about 100 times larger diffusivities of Cr and Ni in ferrite than that in austenite [32] and the more nucleation sites in 316LN-2. Consequently, the remaining liquid phase in 316LN-1 when three-phase reaction began was more than that in 316LN-2. Moreover, the difference between \( T_{L}^{\delta } \) and the temperature of three-phase reaction beginning (T t ) in 316LN-2 was calculated to be 29 °C, which is much larger than that between \( T_{L}^{\gamma } \) and T t , 5.1 °C, in 316LN-1. Therefore, the bigger temperature drop and less remaining liquid phase in 316LN-2 facilitated the peritectic reaction owing to the more amount of primary phase and shorter diffusion distance between δ and L. While in alloy 316LN-1, the longer diffusion distance between primary γ and L, combined with the fact that the cooling line of 316LN-1 in the ternary phase diagram is nearer to eutectic trough than that of 316LN-2 (austenite and δ ferrite almost precipitate simultaneously in 316LN-1 according to Fig. 12a), prompted the eutectic reaction.

Fig. 12
figure 12

Equilibrium phase diagram calculated by JMatPro software (a) and vertical section of Fe–Cr–Ni ternary diagram at constant Fe (b)

While cooling rate increases, the type of three-phase reaction in 316LN-1 does not change. However, under the conditions of 10 and 100 °C s−1, the peritectic reaction transforms into eutectic reaction in 316LN-2, which results in the appearance of a coupled growth microstructure of δ + γ. The main reason for this transformation is that the melt undercooling degrees at the beginning of solidification become higher than the difference in values between \( T_{L}^{\delta } \) and T E . Therefore, δ + γ microstructure formed directly without the occurrence of primary δ ferrite, as shown in Fig. 13.

Fig. 13
figure 13

In situ observation of coupled growth microstructure of δ + γ in 316LN-2 cooled at 10 °C s−1

Conclusions

In situ observation with CSLM and color metallographic method were carried out in two 316LN stainless steels with different compositions in order to investigate their solidification sequence and microstructure evolution during solidification process. The effects of cooling rate on variations of microstructure and chemical composition were also studied. These experimental investigations allow the following conclusions to be reached:

  1. (1)

    During mold casting process, alloy 316LN-1 follows the solidifying path of AF solidification mode, while 316LN-2 solidifies in FA mode. Both the results are consistent with the prediction of Hammar and Svensson equations. 

  2. (2)

    As the cooling rate increases in the range of 0–100 °C s−1, the solidification modes of both alloys do not change on the whole. However, the ferrite in 316LN-1 changes from island-like to lacy-like and it becomes coarser in 316LN-2 with cooling rate increasing. In alloy 316LN-1, the contents of Cr and Mo decrease slightly in both δ ferrite and γ phase while Ni increases in γ phase and decreases in δ ferrite in a small scale with the increasing of cooling rate. On the other hand, Cr and Mo increase in δ ferrite and decrease in austenite, while Ni has a contrary variation trend in alloy 316LN-2.

  3. (3)

    Eutectic reaction occurs in the three-phase region of 316LN-1, while peritectic reaction occurs in 316LN-2 when it is cooled at 2 °C s−1. The bigger diffusivities of Cr and Ni in primary δ ferrite than that in primary austenite, as well as the positions of alloys in phase diagram were thought to be the main reasons accounting for the difference in the type of three-phase reaction. Peritectic reaction transforms into eutectic reaction in 316LN-2 when it is cooled at 10 and 100 °C s−1.