Generation of deformation-induced martensite when cryogenic turning various batches of the metastable austenitic steel AISI 347

Cryogenic turning of metastable austenitic steels allows for a surface layer hardening integrated into the machining process, which renders a separate hardening process obsolete. This surface layer hardening is the result of a superposition of strain hardening mechanisms and deformation-induced phase transformation from austenite to martensite. The activation energy required for the latter depends on the chemical composition of the metastable austenitic steel. It can hence be expected that the austenitic stability of the workpiece material varies depending on the batch and that differences in the metallurgical surface layer properties and thus also in the microhardness result after cryogenic turning. Therefore, in this paper, various batches of the metastable austenitic steel AISI 347 were turned utilizing cryogenic cooling with the same machining parameters. The thermomechanical load during the experiments was characterized and the resulting subsurface properties were investigated. The content of deformation-induced α′-martensite was quantified via magnetic sensor measurements and the distribution was examined using optical micrographs of etched cross-sections. It was found that similar amounts of deformation-induced α′-martensite were generated in the workpiece surface layer for all batches examined. Furthermore, the workpieces were analyzed with regard to the maximal hardness increase and the hardness penetration depth based on microhardness measurements. A significant surface layer hardening was achieved for all batches. This shows that surface layer hardening integrated in the manufacturing process is possible regardless of batch-dependent differences in the chemical composition and thus varying austenite stability of the metastable austenitic steel.


Introduction
The surface layer properties of a component directly influence its operational behavior [1]. Component surfaces that are exposed to high tribological loads, such as shaft counter surfaces for radial shaft seals, must have high hardness in the surface layer and at the same time a sufficient bulk toughness to guarantee their functionality [2]. For non-brittle materials like stainless austenitic steel, a high hardness in the workpiece surface layer also results in a longer service life of cyclically stressed components, because the crack initiation is hampered and the crack propagation rate is reduced [3][4][5]. Usually, workpieces are post-processed after machining by a subsequent heat treatment or a mechanical hardening process in order to increase the hardness of the workpiece surface layer. Specifically, workpieces made of austenitic stainless steel are usually mechanically hardened by deep rolling, hammering or shot peening [6,7]. These post-treatment processes plastically deform the workpiece surface layer and increase the hardness in the near-surface area by superimposing different hardening mechanisms [8]. The increase in hardness results from grain refinement, the formation of twin grain boundaries and an increased dislocation density 1 3 [9,10]. However, the additional post-treatment process leads to prolonged process chains and thus also to an increase in production costs.
In recent years, a cryogenic turning process has been developed to harden the workpiece surface layer of metastable austenitic steels already during the machining process and thus to render the additional post-treatment process obsolete [11]. For a large number of materials, a hardness increase in the workpiece surface layer was already observed after cryogenic machining, which results from the hardening mechanisms mentioned above [12]. When turning metastable austenitic steels under cryogenic cooling, a phase transformation from γ-austenite into α′-and ε-martensite occurs in the workpiece surface layer [11]. It is therefore possible to superimpose the strain hardening mechanisms with the deformation-induced martensite formation, thus enabling more effective surface hardening. The hardened surface layer of cryogenically turned workpieces results in a higher wear resistance [13] and fatigue resistance [14] and thus improves the components application behavior. Deformation-induced martensite generation and strain hardening mechanisms are both favored by high passive forces and low workpiece temperatures during the turning process [15]. By adapting the CO 2 mass flow rate, nozzle position, machining parameters and cutting edge geometry, the thermomechanical load in the turning process can be controlled. This also enables the control of the martensite content in the workpiece surface layer as well as the hardness [15][16][17][18][19]. However, the materialspecific properties of the workpiece and its effect on the resulting martensite content and subsurface properties after cryogenic turning has not been researched yet. The chemical composition and grain size have a significant effect on the austenite stability [20,21]. The thermodynamic stability of the austenitic phase increases in particular with increasing proportions of the alloying elements nickel (Ni), chromium (Cr), manganese (Mn), molybdenum (Mo), silicon (Si) and especially carbon (C) and nitrogen (N) [22,23]. Due to the effect of the alloying elements on the austenite stability, differences in the resulting α′-martensite content after cryogenic turning are to be expected when turning different batches. In this paper, different batches of the metastable austenitic steel AISI 347 were turned under identical conditions to investigate the influence of the chemical composition on the resulting subsurface properties.

Phase transformation of metastable austenitic steels and investigated materials
In metastable austenitic stainless steels, a phase transformation from paramagnetic fcc γ-austenite to ferromagnetic bcc α′-martensite occurs due to deformation or cooling. The thermodynamical background of this phase transformation is schematically shown in Fig. 1. At the temperature T 0 γ-austenite and α′-martensite are in thermodynamic equilibrium as their free energies have the same values. Below T 0 the phase transformation from γ-austenite to α′-martensite is possible, but occurs only, if a minimum difference of free energy ΔG � min is provided. At the martensite start temperature M s the difference in free energy ΔG � min due to the temperature decrease is sufficient to induce the transformation (energy is provided by ΔG chem ). Between M s and T 0 , a phase transformation can only occur if an additional amount of energy difference ΔG mech is provided by deformation; the sum of ΔG mech and ΔG chem has to reach the level of ΔG � min . The M s and M d temperatures can therefore be used as indicators for characterizing the austenite stability (see Fig. 1). Since the M d -temperature is difficult to determine experimentally, the M d30 -temperature was introduced by Angel et al. [23]. The M d30 -temperature is the value where 50% martensite is formed when 30% plastic strain is applied. The M s -temperature is primarily used to estimate the austenite stability during the purely thermally induced phase transformation, whereas the M d30 -temperature is a proven indicator for the austenite stability with respect to the deformation-induced phase transformation. Based on the chemical compostion, the M s temperature according to Eichelmann [22] and the M d30 -temperature according to Angel [23] can be calculated by the following equations: 1 Correlation between Gibbs free energy G and temperature T from [15] according to [24] Higher M s,Eichelmann -and M d30,Angel -temperatures indicate a more metastable austenitic steel, i.e. at identical process conditions, higher α′-martensite contents are formed. Beside the regression formulas according to Eichelmann and Angel, there are numerous other empirical calculation models for the M s - [25,26] and M d30 -temperature [27,28], which weight the impact of the chemical elements differently. The M s and M d30 temperatures should therefore not be considered as an exact measure of austent stability, but rather as an approximation.
The influence of the grain size on the metastability of autenitic stainless steels is discussed controversary in literature. Raman et al. found an increasing martensite content in coarse grained AISI 304LN [29], whereas Shrinivas et al. found a higher volume fraction of α'-martensite with a decrease in grain size [30].
The investigated materials are five batches of the metastable austenitic stainless steel AISI 347 (X6CrNiNb1810). All materials were provided in solution-annealed state and have a fully austenitic structure prior to machining. The chemical composition of the batches is given in Table 1. Table 2 lists the M sEichelmann -and M d30,Angel -temperatures of the batches, according to Eqs. 1 and 2. Furthermore the average grain diameter including twins is given. As can be seen, all batches have the same AISI grain size number, but their chemical composition and, accordingly, their M s,Eichelmann -and M d30,Angel -temperatures differ significantly.

Experimental setup
The workpieces were turned on a CNC lathe with a feed travel of 18 mm to a final diameter of 14 mm. A thin layer with low martensite content is formed in the near surface area during pre-processing (dry cutting to the initial diameter of 14.4 mm; see Table 2). However, its depth is much smaller than the depth of cut applied in the subsequent cryogenic turning experiments. Hence, this thin layer is completely removed and does not influence the martensite formation of the actual cut. To ensure sufficient statistics, four workpieces were turned out of each batch (B1-B5). A new insert was used for each workpiece in order to minimize the influence of wear. The tool geometry and cutting parameters were kept constant. The machining parameters, taken from previous investigations, are listed in Table 3.
To maintain the necessary low temperatures in order to achieve phase transformation, a CO 2 solid-gas-mixture was applied as coolant via two external nozzles at approx. − 78 °C [12]. One nozzle supplied the CO 2 from flank face direction directly into the contact zone. The second nozzle served for precooling of the workpiece from rake face direction (see Fig. 2). The mass flow rate per nozzle was 1.75 kg/min. To ensure uniform test conditions, the CO 2  cooling system was activated automatically with the NC program.
The workpiece temperature was determined with an infrared thermography system located outside of the CNC lathe. As the contact zone is not accesible for thermographic measurements, the workpiece surface temperature was measured after a 3/4 rotation of the workpiece in order to evaluate and compare the thermal load for the different batches. The process forces were recorded with a piezoelectric three-component dynamometer.
The phase transformation from γ-austenite into α′-martensite results in a change of the magnetic permeability since α′-martensite is ferromagnetic, while γ-austenite is paramagnetic. Therefore, it was possible to evaluate the content of the deformation-induced α'-martensite generated in the turning process nondestructively. This was done by means of a magnetic sensor, type Ferritscope™, which detects ferromagnetic phase fractions up to a depth of approx. 3 mm below the workpiece surface, see [15]. However, ε-martensite cannot be detected by the magnetic sensor, as it has paramagnetic properties like austenite. Furthermore, only the integral α′-martensite content can be determined with the Ferrit-scope™ and no statements on the exact martensite distribution can be made. However, it is an easy and reliable method to detect the phase transformation as an integral value and hence to evaluate metastabilty of the different batches. A multiplication factor of 1.7 was used to convert the feritscope reading (FE-%) into the α′-martensite content according to [31].
Microhardness measurements were carried out with a vickers indenter to determine the hardness profiles in the workpiece subsurface. Indentations were carried out from 20 to 400 µm below the surface at intervals of 10 µm with a load force of 100 mN (HV 0.01) and a loading time of 10 s. Three microhardness profiles of one respective workpiece of each batch were determined and averaged.

Thermomechanical load
The mean values of the feed force F f , passive force F p and cutting force F c are shown in Fig. 3. As the cutting depth a p = 0.2 mm was significantly smaller than the cutting edge radius r β = 1.6 mm, the workpiece material was exclusivly removed by the corner of the insert. The low depths of cut, in connection with the chamfer, resulted in an almost constant, negative effective rake angle in the entire cutting cross section, which led to high passive forces. This was intended, as high passive forces are essential for the generation of deformation-induced martensite in the workpiece subsurface regions.
The mean surface temperatures recorded during the turning experiments are shown in Fig. 4. The differences in chemical composition of the different batches had no significant influence on the thermomechanical loads. Since the process forces were similar for the different batches, the  heat generated in the nearly adiabatic primary shear zone and thus also the measured temperatures were similar as well and their differences where within the standard deviation. The temperature at the workpiece surface is approx. 20-30 °C lower than the M d30,Angel -temperature and significantly higher (90-180 °C) than the M s,Eichelmann -temperature.

Deformation-induced α′-martensite
The integral contents of α′-martensite in the workpiece surface layer after cryogenic turning are shown in Fig. 5 as a function of the M d30,Angel -temperature. Despite the differences in the chemical composition and the associated austenite stability, the magnitude of deformation-induced α′-martensite content of the batches B2-B5 was in the range of the standard deviation. Only batch B1 showed a slightly higher α′-martensite content. Since the chemical composition of B1 was similar to that of B2 and B4 (see Table 1), this difference in α′-martensite content cannot be attributed solely to the differences in the chemical composition. Figure 6 exemplarily shows metallographic sections perpendicular through the surfaces of the cryogenically turned workpieces etched with Beraha II. The materials show segregations due to their chemical inhomogenitiy. It can be seen that all batches have a martensitic needle structure in the surface layer colored in black on the right side of the images. Underneath this surface layer, an austenitic structure is visible.

Microhardness
The percentage standard deviation of the microhardness profiles was between 3.88% (B5) and 8.63% (B4). Regressions for the microhardness profiles were calculated on the basis of the gaussian distribution function, which are shown in Fig. 7. Depending on the batch, values between 0.94 (B1; B3) and 0.98 (B2) were calculated for the coefficient of determination.
Near the surface, this function first shows a progressive decrease, which changes into a degressive decrease in microhardness and then approaches the initial hardness HV 0 0.01. The initial hardness HV 0 0.01 was measured in the bulk material, which was not modified by the cryogenic turning process. The fitting parameters a and b of the regression were determined according to the method of least squares   Table 4). Parameter a describes the difference between the initial hardness HV 0 0.01 and maximum hardness. The exponent b determines the gradient of the function. In general, for otherwise constant fitting parameters, small values for b imply a slow flattening of the function resulting in a high hardness penetration depth. The hardness penetration depth represents the threshold between the hardened subsurface and the unaffected base material. It was quantified as the distance from the surface in which the hardness corresponds to 120% of the initial hardness.
A significant increase in hardness was observed in the workpiece subsurface region of all five batches. For all experiments, the maximum hardness occurred immediately below the workpiece surface. In relation to the initial hardness of the individual batches, the increase in hardness was between 61% (B4) and 87% (B5). The hardness penetration depth was between 154 µm (B3) and 209 µm (B1). A distinct correlation between the increase in hardness as well as the hardness penetration depth and the α'-martensite generated in the cryogenic turning process could not be determined (see Fig. 8).
In recent experiments [18], a rise in maximum hardness and hardening depth was also observed when the α′-martensite content increased. In those experiments, the increase in the α′-martensite content was caused by an increase in passive force via an increase in feed and cutting edge radius. Thus, the plastic deformation in the workpiece subsurface increased and, in addition to the phase transformation, also favoured the strain hardening mechanisms. Note that these experiments were only carried out on one batch (B2), therefore the initial state of all workpieces was the same.
In the experiments presented in this paper, the influence of different hardening mechanisms can vary, depending on the initial material state and the respective changes of them during the turning process. Thus, it can be assumed, that the total increase in hardness in the subsurface region cannot be exclusively attributed to the formation of deformation-induced α′-martensite. Formation of ε-martensite and/ or stacking faults, as well as increased dislocation density, grain refinement and formation of twins also contribute to hardening. To quantify these influences, the exact phase distribution of α′-and ε-Martensite and other material defects would have to be detected e.g. via X-ray diffraction and high-resolution microscopy.

Conclusions and outlook
In this paper results of cryogenic turning experiments on five batches of the metastable austenitic steel AISI 347 are presented. Cryogenic turning was performed with the same process conditions (tools, coolant supply, process parameters), in order to examine the influence of chemical composition and, hence, metastability of the workpiece In the experiments, no significant differences in process forces and temperatures were measured depending on the batch. After cryogenic turning, deformation-induced α′-martensite was detected in the workpieces of all investigated batches. The integral contents of α′-martensite were in a similar range for the various batches, varying from 3.6 to 5 vol.%. When comparing the optical micrographs of the different batches, there were some minor differences in the martensite distribution. However, in all batches, the α′-martensite only occurred up to a depth of 250 µm below the surface generated by cryogenic turning.
For all batches a pronounced surface layer hardening was observed after cryogenic turning, slightly varying in its magnitude among the batches. However, no distinct correlations between α′-martensite content and hardness increase or hardness penetration depth could be determined. Thus, the surface layer hardening during cryogenic turning cannot be exclusively traced back to the volume of α′-martensite generated.
Cryogenic turning proved to be suitable to realize a robust surface layer hardening integrated in the manufacturing process, even at a different austenite stability of the workpiece material. The quantification of the influence of phase distribution, grain refinement, dislocation density and residual stresses on the total increase in hardness and the hardness penetration depth via in-depth analyses is to be driven forward in the future. Furthermore, fatigue tests will be carried out on cryogenically turned specimen from different batches in order to investigate the correlation between the surface integrity and the fatigue behavior.