Electropulses induced direct aging and ultrafast precipitation in additive manufactured 15-5 PH stainless steel

Abstract

Periodic electric current pulses, i.e. electropulses, were applied on selective laser melting prepared 15-5 precipitation hardening stainless steel for direct aging. Traditional aging treatment typically takes hours to achieve peak hardening and could induce reverted austenite. Herein, electropulsing treatment on as-built sample can dramatically reduce the processing time to 6 min. Moreover, the nucleation of reverted austenite triggered by the segregation of Ni atoms on the interfaces between Cu-rich precipitates and martensite was not observed. These two unique phenomena were rationalized to the unsynchronized enhancement of diffusivity of Cu and Ni atoms under electropulses due to their different electron configurations.

GRAPHICAL ABSTRACT

IMPACT STATEMENT

Electropulsing treatment selectively boosted the diffusivity of Cu atoms in additive manufactured maraging stainless steel and dramatically reduced the direct aging time from hours to minutes without inducing reverted austenite.

KEYWORDS:

• Electropulsing treatment
• Maraging stainless steel
• Selective laser melting
• Precipitation
• Direct aging

1. Introduction

Precipitation hardening (PH) martensitic stainless steels (SS) with low carbon content are typically strengthened by the precipitation of nanometer-sized intermetallic particles within the martensite matrix [1,2]. Taking the widely used 15-5 PH stainless steel (15 wt.% Cr and 5 wt.% Ni) as an example, it contains approximately 3 wt.% of Cu, and the hardening is mainly attributed to Cu-rich precipitates (CRPs). To obtain CRPs, a solution treatment (typically at ∼1040°C for 1 h) is usually implemented to attain supersaturated Cu content in martensite before the subsequent artificial aging (in the range of 425–625°C slightly below the Ac₁ for various durations). Moreover, direct aging, i.e. without the solution treatment, of the as-wrought blocks or as-machined components of PH stainless steel has also been utilized in the application in view of the reduction of the lead time, manufacturing cost and thermal distortion [3].

Additive manufacturing (AM) techniques have also been implemented on PH stainless steels in recent years and expanded their range of applications. Due to the rapid solidification, AM produced PH stainless steel exhibits short and narrow martensitic laths and contains a fraction of retained austenite [4–6]. This unique microstructure feature results in greater tensile strength than its wrought counterpart [4,5]. However, the presence of retained austenite phase possesses an obstacle in the direct aging treatment, because the higher solubility of Cu in the austenite than that of the martensite contributes to sluggish precipitation kinetics [7,8]. Moreover, the diffusion controlled phase transition from martensite to reverted austenite can also occur during the direct aging, compromising the wear resistance, hardness and strength [9,10]. Since the CRPs formation and austenite reversion are intercorrelated during the ageing treatment [11,12], it is highly desirable to explore a processing route to obtain CRPs while controlling and suppressing the formation of reverted austenite.

Using periodic electric current pulses, i.e. electropulsing (EP) treatment, to stimulate microstructure evolution such as recovery and recrystallization after cold/warm working [13,14], and precipitate formation and dissolution after solution and aging treatment [15–18], has been drawing increasing attention due to its potential to supplant conventional thermal treatment. Moreover, the pulsed electric current scattering on the crystal lattice defects also triggers unique phenomena that cannot be achieved by thermal treatments [19,20]. Most recently, the utilization of EP treatment has been investigated on additively manufactured metallic materials [21,22].

In this study, we implemented EP treatments solely to induce direct aging by facilitating the formation of CRPs in as-built 15-5 PH stainless steel samples fabricated by selective laser melting (SLM). We observed a unique phenomenon that EP treatments resulted in ultrafast formation of CRPs without inducing reverted austenite, which cannot be achieved by traditional artificial thermal aging treatment.

2. Materials and methods

As-built SLM 15-5 PH SS strip specimens with a size of 30 mm (X) × 2 mm (Y) × 1 mm (Z) were prepared and went through EP treatments (Figure 1a). Square wave pulses with the current density of 850 A/mm², pulse duration of 1 ms and frequency of 0.25 Hz were applied. The strip samples exhibit an effective length of 20 mm between the electrodes, as 5 mm at each end contacts the Cu electrodes. The electropulse generator was built with low voltage (60 V) and high capacity (450 F) to produce electric current pulses with steady square waves.

Figure 1. (a) The schematic illustration of experimental setup and (b) the temperature evolution of three locations (points A, B, C) on the strip sample during EP treatment with current density of 850 A/mm². Note that three locations, i.e. point A, B and C were chosen to present.

The variable of the EP treatment in this study was chosen to be the EP cycles, i.e. number of pulses, namely 30, 60, 90, 120, 150 and 180 times. During the pulses, temperature raised due to Joule heating was monitored using K-type thermocouples. By measuring three locations (shown in Figure 1) on the strip sample, we found that the temperature decreased gradually from the center to the edge. This was because of the heat transferred away via conduction to the Cu electrodes [23]. Moreover, the temperature rose rapidly and reached a plateau within 10 EP cycles, then stabilized with no further increment as the Joule heating and heat conduction via the Cu conductors reached a dynamic balance. This indicated that temperature cannot be a variable determining the EP effect once the number of EP cycles is higher than 10 times; and in this study the minimal EP cycles were chosen to be 30 times.

The hardness of the samples before and after EP treatment were determined on a Vickers hardness tester using a load of 0.3 kg and a dwell time of 15 s. The microstructure characterization was carried out using a Zeiss Gemini 450 scanning electron microscope (SEM) with anOxford electron backscatter diffraction (EBSD) detector, FEI-Talos F200X scanning/transmission electron microscopy (S/TEM) with a Super-X energy-dispersive X-ray spectroscopy (EDS) detector and a Rigaku Smartlab X-ray diffractometer.

3. Results

After EP treatments, we first obtained average hardness distribution on the strip specimenby repeating tests on three samples that went through the same processing condition. Figure 2(a) is a heat map demonstrating the hardness distribution after 90 EP cycles. It clearly shows the decreasing trend from center to edge region of the sample. Three regions can be distinguished, namely center area, transition area and edge area. After 90 EP cycles, the hardness of the center area was in the range of 404–427 HV; the transition area was within 360–404 HV; and the edge area was below 360 HV, which was close to the hardness of the as-built sample, 345 ± 3.1 HV. This indicates that, in this case, the hardness enhancement cannot be achieved solely by electropulses, and the Joule heating is essential in this process. Without sufficient temperature, the hardness enhancement cannot be achieved. In addition, the hardness distribution on both sides of the sample is almost symmetrical, and there is no obvious macroscopic difference.

Figure 2. (a) A heat map of the hardness distribution of 15-5 PH SS after 90 EP cycles; (b) the hardness evolution at the center area of the sample with respect to the number of EP cycles; (c) the comparison between the aging time and EP time cost to achieve hardness increment, the data points for aging treatment are from Refs. [10,24–27].

To study the effect of the number of EP cycles on the hardness enhancement, we compared the hardness of the center region after 30, 60, 90, 120, 150 and 180 times of EP cycles. It is worthwhile to note that after around 10 pulses, the temperature of the center region was stabilized at ∼390°C as shown in A point curve of Figure 1(b). Thus we can exclude the effect temperature and only consider the accumulated energy introduced by increased number of electropulses.

From Figure 2(b), we can see that the average hardness of the center area of the strip sample increased ∼84 HV from 345 ± 3.1 HV in the as-built state to 427 ± 4.3 HV after 90 cycles, then decreased gradually with additional EP cycles. This behavior is similar to the transition from under-aged to peak-aged and over-aged conditions in artificial aging treatments. However, conventional heat treatments of the as-printed or solution treated 15-5 and 17-4 PH SS take 1 – 4 h to reach peak-aged state with the hardness increment of ∼80 HV. Meanwhile, EP treatments only require 6 min to achieve the same hardness increment, see Figure 2(c). Finally, regarding the inhomogeneity of the hardness increment on the strip sample, future discussion is given in supplementary materials.

To verify the formation of CRPs, we conducted S/TEM on a specimen that went through 180 EP cycles. The reason why we did not choose the 90 cycles sample was because the peak hardening is contributed by nano-scale Cu clusters which can only be characterized by atom probe tomography (APT) [6,28]. Coarsened CRPs from the over-aged condition can be observed by TEM characterization. Figure 3 demonstrates the existence of CPRs with an average radius of 4.6 nm in the martensite matrix, which verified that the hardness increment was attributed to the EP treatment induced CRPs. Moreover, we did not observe the segregations Ni on the interfaces between CRPs and the martensite matrix.

Figure 3. (a) A BF-STEM image demonstrating the CRPs after 180 EP cycles; (b) the selected area diffraction pattern took from the same region showing martensitic matrix; (c, d) the EDX mapping of Cu Kα and Ni Kα showing the Cu clusters and the homogeneous distribution of Ni.

Another important observation is the suppression of reverted austenite during the formation and even coarsening of CRPs. We first conducted XRD measurements on the center area of one specimen in a quasi-in-situ manner to track the possible phase transformation with respect to increasing EP cycles (Figure 4a). We fitted all XRD patterns and quantitatively estimated the volume fraction of the FCC and BCC structures (the PDF cards of the FCC and BCC structures are #65-5131 and #65-4899, respectively). We found there was a dramatic increase of volume fraction (∼5%) of FCC phase after 180 EP cycles and it can be attributed to CRPs (using PDF card of #47-1405), see supplementary materials and Figure S2.

Figure 4. (a) Quasi-in-situ XRD measurements with respect to increasing EP cycles and the evolution of BCC and FCC phases; (b–e) phase maps acquired from EBSD measurements showing the evolution of martensite and austenite before and after 30, 90 and 180 EP cycles.

To confirm this statement, we also performed EBSD measurements to characterize the evolution of the volume fraction of austenite in the specimens after EP treatments. This is because in the aged 15-5 PH stainless steel, FCC phase reflections on the XRD pattern could be originated from both austenite and CRPs [27,28]. This means in addition to the formation of CRPs, the increase of FCC phase could also be partially attributed to the formation of reverted austenite. Figures 4(b–e) show the phase distribution of martensite and austenite in the center area of the strip specimens after different EP cycles, and the fraction of austenite did not change in 180 EP cycles. Therefore, there was no reverted austenite formation during the EP treatments, yet high density CRPs were harvested. It is also worth noting that 5% volume fraction increment of FCC phase solely due to the formation of CRPs is reasonable according to previous studies [11,29,30]. Finally, to confirm the absence of reverted austenite formation, we performed solution treatment on the SLM fabricated 15-5 PH SS sample to obtain fully martensitic microstructure and then carried out EP treatments. The hardness changed from peak-aging to over-aging state, but reverted austenite phase was not observed (see supplementary materials).

4. Discussion

The importance of the result lies twofold. First, the time (6 min) and the temperature (∼390°C) consumed in achieving peak hardness by the EP treatment was about one tenth of and 100°C lower than the conventional artificial aging treatment (60–240 min at ∼490°C). This demonstrates the high efficiency and low energy consumption of EP treatment. Second, the formation of CRPs without introducing reverted austenite is a very unique phenomenon.

We first discuss the significantly high growth rate of CRPs during EP treatment. From Figure 3, after 180 EP cycles, i.e. 720 s, the average radius of the CRPs was 5.2 ± 0.9 nm. Thus the growth rate under EP treatment is estimated to be 0.43 nm/min. Meanwhile, a recent study showed that the growth rate of CRPs in SLM manufactured 15-5 PH SS via conventional aging treatment was 0.042 nm/min (average radius of 2.5 nm achieved after aging at 482°C for 1 h from Ref. [6]). Since the growth rate of spherical precipitates is determined by the bulk diffusivity of the rate-controlling solute species [31], the bulk diffusivity of Cu atoms in martensite matrix under EP treatment at ∼390°C should be at least 1 order of magnitude higher than that of it under traditional aging treatment at ∼490°C. It is also noteworthy that, because the retained austenite was not removed by solution heat treatment in this study, one would expect Cu atoms diffused into the retained austenite phase during EP treatment and resulted in sluggish growth of CRPs in martensite.

The accelerating effect of electrical current pulses on the precipitates formation in steels has also been observed in other studies [15,32–34]. One possible explanation is that the formation of nuclei with electric conductivity higher than that of the surrounding media could be promoted [35]. Based on this theory, Qin et al. proposed an analytical model demonstrating that the thermodynamics of phase transformation under EP can be affected by the change of electric current associated free energy $\mathrm{\Delta }{G}^E$ when the current density surpasses 10 A/mm² [32,36]. Since CRPs are composed of ∼80 at% of Cu atoms and mainly Fe atoms as the rest substitutional impurities [11], we can use the electrical conductivity of Cu alloys with 30–10 wt.% of Fe to compare with that of martensite, which are 40–80 MS/m (35–70% IACS) [37] and 1.2 MS/m [38], respectively. So, the faster precipitation of CRPs under EP treatment can be attributed to its 20 times higher electrical conductivity than martensite from the perspective of the promoted thermodynamics driving force.

Regarding the formation of reverted austenite, T. Zhou et al.'s recent work demonstrated that after aging at 500°C for 20 h, CRPs with an average radius of 3.1 nm and a shell layer of Ni triggered the formation of lamellar structured reverted austenite [11]. This process is thermodynamically favorable as segregation of Ni on the CRP/martensite interface lowers the interfacial energy [39]. Peng et al.'s work showed that after aging at 550°C for 4 h and once the average diameter of CRPs reached 10 ± 2 nm, strip-shape reverted austenite can be clearly observed [40]. Couturier et al.'s work showed that reverted austenite can be characterized after 505°C for 5 h when CRPs reached an average radius of 2.3 nm [12]. However, in this work, even though the sample that went through 180 EP cycles was in the over-aged condition as the CRPs evolved to FCC structure with an average radius of 4.6 nm, and leaded to the decrease of hardness; Ni segregation around CRPs was not observed in the STEM-EDX mapping and reverted austenite was not characterized from EBSD measurements. This indicated that the bulk diffusivity of Ni atoms should be significantly lower that than of Cu.

The diffusion coefficients of Cu and Ni in ferromagnetic α-Fe were measured to be $D_{\mathrm{Cu}/\mathrm{Fe}}=0.47\cdot \text{exp}\left(-244.3/\mathrm{RT}\right)$ cm²/s [41] and $D_{\mathrm{Ni}/\mathrm{Fe}}=1.4\cdot \text{exp}\left(-245.6/\mathrm{RT}\right)$ cm²/s [42], respectively. It can be seen that the diffusion activation energy for Cu and Ni are very close. This could be due to the fact that the radius of Cu atoms (0.126 nm) is very close to that of Ni atoms (0.124 nm). Therefore, provided that the monovacancy formation enthalpy is the same in martensitic matrix, the monovacancy migration enthalpy, e.g. the enthalpy required for the position exchange between a Cu or Ni atom with a monovacancy in martensite should also be very close. Moreover, a recent study demonstrated that EP treatment could reduce the vacancy formation energy and increase the vacancy concentration [34], but this effect should impose the same extent of facilitation for Cu and Ni atoms diffusion.

One possible reason for the discrepancy between the bulk diffusivity of Cu and Ni could be due to the different electromigration driving force exerted on Cu and Ni atoms. According to Huntington and Grone [43], the electrical force acting on the diffusing atoms is proportional to the effective charge or effective valence $Z^{\ast }$, which is determined by the electrical conductivity and the valence charge of the ion species. However, the effective charge of transition metal ion cannot be accurately assigned due to the simultaneously localized and the itinerant character of the d electrons [44]. Therefore, the electrical force on Cu and Ni atoms cannot be quantitatively compared.

Nevertheless, we still could surmise the influence of the different electronic structure of Cu and Ni atoms, as their respective configurations of the outer electrons are 3d¹⁰4s¹ and 3d⁸4s², on the discrepancy of their diffusivity under electric field. The monovalent Cu atom is with fully occupied d band that lies substantially below the Fermi level, resulting in the dominant role of the conduction electrons on 4s band [45]. Regarding the divalent Ni atom, it is with not fully occupied d band, and the top of its d band lies above the Fermi level [46], leading to the s scattering dominated by the holes of the d band [47]. Moreover, the fact that the relaxation times of the conduction electrons exhibiting different anisotropies over the Fermi surfaces could also induce different influence on the Cu and Ni atoms under electric field [48]. At last, a study investigating the interdiffusion in the Cu–Ni system revealed that the enhancement of interdiffusivity under electric field is directional [49]. A marked increase of the interdiffusivity was observed when the electric flow was from Ni to Cu and an opposite direction of electric flow did not result in any enhancement. This result also indicates that the influence of electric field on the increment of the diffusivity of Cu and Ni atoms isdramatically different.

Therefore the fundamental difference of the electron configuration may result in higher diffusivity of Cu than that of Ni atoms in martensitic under pulsed electric current, and lead to retarded segregation of Ni on CRPs and the inhibition of the reverted austenite formation.

5. Conclusion

In conclusion, we found ultrafast precipitation of CRPs in 15-5 PH stainless steel by electropulsing treatment as the peak hardening was achieved in 6 min, compared to the hours of conventional thermal treatment. This was attributed to the high electrical conductivity of CPRs and the enhanced diffusivity of Cu atoms in martensite. In addition, the absence of reverted austenite indicated that the enhancement of Ni atoms diffusivity in martensite was not synchronized as that of Cu, even though their bulk diffusion activation energy in ferromagnetic α-Fe is close. This could be due to the different electron configuration and effective valence of Cu and Ni atoms. This work paves the way to achieve precise control of the intercorrelated formation of CRPs and reverted austenite phase, and demonstrates the possibility of implying quantum mechanics in physical metallurgy via the EP processing technique.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

YQ acknowledges the National Natural Science Foundation of China (grant number 52101134), Natural Science Foundation of Guangdong Province (grant number 2022A1515010275), and Guangdong Province Science and Technology Department Major Project (2021B0301030005). SC acknowledges the National Natural Science Foundation of China (grant number 52204391).