Improvement of the corrosion behavior of AISI 304L stainless steel by deep rolling treatment under cryogenic cooling

The effects of deep rolling parameters, particularly, work speed and cooling conditions (dry and cryogenic) on the surface integrity of AISI 304L machined samples and their further impact on uniform and localized corrosion behavior in chloride environment were experimentally investigated in this work. The electrochemical behavior of machined and deep rolled samples was assessed using cyclic potentiodynamic polarization tests in synthetic seawater. It was found that the corrosion behavior of AISI 304L deep rolled components is related to combined factors: surface roughness, recrystallized grains, strain-induced martensite, microhardness and residual stresses. Findings of this study exhibit that grain refinement generated in the surface layers leads to improved corrosion behavior of deep rolled specimens with regard to machining state. In addition, samples deep rolled at a speed of 25 m/min, without cooling, showed better corrosion resistance than those processed under cryogenic cooling. However, the application of cryogenic deep rolling at speeds of 75 and 120 m/min significantly enhanced the electrochemical behavior of mechanically treated specimens. Despite of high amounts of strain-induced martensite that can deteriorate the electrochemical behavior, it was shown that specimens deep rolled under these conditions, presenting better surface characteristics, depicted an improved corrosion resistance.


Introduction
It is known that environmental conditions, as well as interrelated geometrical, metallurgical factors and surface conditions resulting from manufacturing processes, strongly affect the corrosion behavior of stainless steels [1]. Hence, mechanical surface treatments such as thermo-mechanical treatment [2], shot and ultrasonic peening [3], nano peening [4] and laser shock processing [5,6] have been used to enhance the electrochemical behavior of these steels. It should be noted that surface modifications (microstructure, grain size, surface roughness, microhardness, residual stress, etc.) generated by these mechanical processes are considered as the key factors for modifying their corrosion behaviors. Toppo et al. [2] studied the effects of a thermo-mechanical surface treatment, combining conventional shot blasting with subsequent laser surface heating, on the electrochemical behavior of the AISI 304 stainless steel. It has been reported that this surface treatment leads to recrystallized fine grains and strain-induced martensite in the affected layer. Authors noted that the grain refinement, the dispersion and the redistribution of alumina inclusions are the main factors controlling the corrosion resistance of thermo-mechanically treated AISI 304. Lee et al [3] pointed out that shot and ultrasonic peening of AISI 304 produce nano-sized grains of 20-30 nm on the surface layer resulting in the increase of surface microhardness by about 50-100 HV comparatively to the untreated surface. In addition, authors found that the ultrasonically peened sample with lower surface roughness and higher volume fraction of plastic-induced martensite (48%) exhibits better general and localized corrosion resistance than the shot peened sample. Saada et al. [4] mentioned that severe shot peening of AISI 304L induces a nanocrystallized surface layer of 150-μm thickness that consists of ferrite grains enriched with Cr and Mo contents. Thus, the improvement of the corrosion resistance is attributed to the Cr diffusion on the protective film. Lu et al. [5] showed that the corrosion behavior improvement of AISI 304 after the application of massive laser shock peening (LSP) is attributed to refined surface layer and deep compressive residual stress distribution that reaches a depth of around 900 μm. Meanwhile, the processing parameters of these surface treatments must be selected with care as the induced surface modifications by these treatments can deteriorate the corrosion resistance of the treated materials. Indeed, Pałka et al. [7] reported that increasing the burnishing load up to 3000 N reduces the surface roughness by over 80% and increases the yield stress from 230 to 450 MPa, whereas the AISI 304 electrochemical behavior in citric acid solution decreases accordingly. Other studies [8,9] mentioned that plasma nitriding at temperatures ranging between 230°C and 450°C leads to the formation of S-phase that significantly enhances the corrosion properties of austenitic stainless steels. However, it has been reported that the presence of 27% of α ′ -martensite in the austenite matrix deteriorates the corrosion resistance of the AISI 304L nitrided at a temperature above 460°C. Wei et al. [6] pointed out that single LSP produces refined grains in the surface layers, compressive residual stress of about −235 MPa and 24% of α ′ -martensite. It has been stated that the combined effects of grain refinement and compressive residual stress significantly improve the corrosion behavior of the AISI 304 stainless steel in acid chloride solution. However, it was found that double LSP increases the surface hardness from 225 to 322 HV and generates a volume fraction of strain-induced martensite of 32% that decreases the corrosion resistance of treated samples. Consequently, it can be stated from the results of the above literature that it is important to understand the overall effects of the surface modifications (surface roughness, grain size, microstructure, residual stress, etc.) resulting from the surface treatment process of the austenitic stainless steels to explain and predict their corrosion behavior.
In recent years, many investigations have been focused on studying the effects of the environment under which the mechanical surface treatments are applied. Attention was given to the application of coolant to reduce the harmful thermal effects that may occur during these treatments. For example, cryogenic cooling has been used in roller burnishing [10][11][12] to enhance material characteristics. Pu et al. [10] studied the effect of cryogenic burnishing on the corrosion enhancement of AZ31B Mg alloy in the NaCl solution. The reported improvements are attributed to the reduction of grain size and basal-textured grain orientation. Yang et al. [11] reported that cryogenic burnishing improves the surface integrity of Co-Cr-Mo alloy due to grain refinement, work hardening and strain-induced phase transformation. Caudill et al. [12] showed that cryogenic burnishing drastically enhances Ti-6Al-4V alloy properties in comparison with dry and floodcooled burnishing. The authors concluded that cryogenic cooling controls the detrimental effect of the thermal softening. Within this framework, deep rolling (DR) treatment has proven its effectiveness to enhance the surface characteristics, i.e. surface smoothing, introduction of compressive residual stresses and surface strengthening. Nevertheless, the application of deep rolling under cryogenic cooling to improve the corrosion behavior of AISI 304L components in chloride environments has not been studied, yet.
The current study aims at evaluating the potentialities of deep rolling treatment and cooling mode to enhance the resistance to uniform and localized corrosion in synthetic seawater environment. Moreover, limits of the corrosion resistance improvement and its dependency on the process parameters are investigated. For this purpose, experiments were performed to assess the effects of deep rolling parameters; namely, work speed and cooling method on the resulting surface integrity evaluated by means of surface topography, microstructure, surface work hardening and residual stress distributions. Moreover, the corrosion properties of machined and deep rolled AISI 304L surfaces are determined by electrochemical tests.

Work material
The material used in this investigation was the AISI 304L austenitic stainless steel supplied as solution annealed (RS) bars. The chemical composition and mechanical properties at 20°C are listed in Tables 1 and 2, respectively. The asreceived Vickers hardness of the tested material is around 197 HV 0.1 , and its microstructure is shown in Fig. 1. The average grain size of the work material is around 50 μm.

Machining
Specimens of 30-mm diameter and 20-mm thickness (Fig. 2) were machined by turning under lubricated condition using  29.5 mm at a feed rate of 0.4 mm/rev and a cutting speed of 100 m/min in order to standardize initial deep rolling conditions. A rake angle γ of 0°, a clearance angle α of 5°and a nose radius R ε of 1.2 mm characterize the insert geometry.

Deep rolling tool
In this study, a deep rolling tool with an interchangeable roller was designed and manufactured to achieve high processing forces. A schematic representation of the deep rolling tool is illustrated in Fig. 3. The tool consists of two hardened steel parts in which the foremost part is a mandrel ⑦ of hardened steel that holds a roller ⑨, while the supplementary part ⑥ of square cross section is a firmly clamped shank on the lathe's tool post. The roller holder can accommodate a heat-treated roller with a contact width of 4 mm. The active part of the roller is toroidal where the radius of the profile curvature is 2 mm (Fig. 3). The roller can be easily removed from the holder for changing or cleaning by unscrewing the pin ⑧. A ball bearing ⑩ was used to reduce rotational friction when the roller is pushed against the sample. Needle roller thrust bearings ⑫ were used to sustain high axial loads and to prevent peak loads when operating. The AISI D2 roller was subjected firstly to an austenitization at 1050°C (vacuum atmosphere) for 15 min followed by air-cooling and, secondly, to a double tempering at 550°C for 2 h each time followed by cooling with brewed nitride at temperature T = 24°C and pressure P = 5 bar. The measured average hardness and the surface roughness (Ra) of the roller, after polishing, were about 62 HRc and 0.01 μm, respectively.

Procedure of deep rolling
The deep rolling experiments were carried out on a lathe equipped with a liquid nitrogen delivery system as shown in Fig. 4. The deep rolling process was performed directly after turning operation without removing the workpiece from the lathe chuck in order to maintain the same turning alignment and to avoid roundness errors. For dry conditions, no cooling was used; while for cryogenic conditions, liquid nitrogen was delivered to the tool-workpiece interface. The nitrogen was applied through a nozzle with a circular section of 50 mm 2 under a pressure of 0.3 MPa. The roller was free to rotate when contacting the workpiece due to the frictional forces developed at the interface between the roller and the workpiece. Even though previous studies have identified five parameters controlling the deep rolling process (depth of penetration DoP, speed V, feed rate f, number of passes N r and cooling method), only speed and cooling mode were considered in this investigation. The speed range was established based on the works of Nikitin et al. [13] and Ben Moussa et al. [14]. The depth of penetration and feed were set to 0.08 mm and 0.05 mm/rev, respectively. These values were selected according to the studies of Yang et al. [11] and Caudill et al. [12]. The experimental conditions are summarized in Table 3.
In the current study, the range of speed and cooling conditions were varied to evaluate correlations between deep rolling parameters and the surface integrity. Moreover, combinations of processing conditions improving the corrosion behavior of machined AISI 304L components were established.

Testing methods
The deep rolling force components were measured using a piezoelectric transducer based-type dynamometer (Kistler 9257B). A RayCam CA1886 infrared thermo-camera was used during the experiments to record the whole thermal field. The surface roughness was evaluated using a MITUTOYO Surftest type SJ 301. Scanning electron microscope (SEM) examinations of machined and deep rolled surfaces were performed using Carl Zeiss SUPRA 5 VP. The surface hardening was characterized by microhardness measurements using a Wolpert Wilson Micro Vickers hardness tester 402 MVD device under a load of 100 gf. Three measurements were performed at each depth, and then averaged to take into consideration the measurement errors. Samples were cut from the machined and deep rolled workpieces for metallurgical examinations. Oxalic acid solution was used as an etchant to reveal the grain structure after polishing using emery papers (Grades 1000, 1600 and 2400). Optical microscope and SEM examinations were used to reveal the machined and deep rolled microstructures. Phase analysis was conducted using a Philips PANalytical X-ray diffractometer with Cu-Kα radiation (λ =1.54184 Å, K α1 /K α2 = 0.5) from a source operated at 45 kV and 40 mA. The X-ray diffractometry was used to evaluate the quantitative measurement of the strain-induced martensite by evaluating the integrated intensity of each peak. The volume fraction of α ′ -martensite (V α 0 ) on the surface layer is calculated using the following equation [3]: where n, I and R represent the number of peaks corresponding to each phase, the integrated intensity of the reflecting plane and the material scattering factor, respectively. The residual stress profiles were determined using a GNR X-ray diffractometer under the conditions reported in Table 4. For in-depth measurements, metal removal was carried out by the electrochemical etching technique. Residual stress measurements were then corrected using Moore and Evans approach in order to take into consideration electropolishing layer removal. This correction is based on the following formula [15]: where σ, σ m , z, H and z 1 are the corrected stresses, the measured stresses, the thickness of the specimen, the original thickness of the specimen and the distance from the lower surface to the point of interest, respectively. The electrochemical reactivity of deep rolled specimens in artificial seawater was evaluated by cyclic polarization tests using a PGZ301 Voltalab 40 potentiostat at ambient conditions (T = 25°C ± 1) with a potential scan rate of 2.5 mV/s. In the standard 3-electrode corrosion cell, a saturated calomel electrode (SCE) was used as a reference electrode and a platinum (Pt) gauze as a counter electrode. The deep rolled samples were covered with an O-type resin ring (test area 10 mm 2 ), then immersed in the synthetic seawater solution (20 ml). The synthetic seawater was prepared in the laboratory according to ASTM D 1141-81 standard, and its pH was adjusted to 8.2 using 0.1 mol.L −1 NaOH solution.

Results
The surface integrity of the AISI 304L samples, deep rolled under the experimental conditions listed in Table 3, is evaluated in terms of surface roughness, microstructure, phase transformation, microhardness and residual stress. Indeed, the literature review presented in Sect.on 1 indicated that these surface modifications are the main factors controlling the corrosion behavior of metallic components. It is important to notice that, in this work, attention is accorded particularly to the effects of cryogenic cooling on the deep rolling process, the resulting surface integrity and its further impact corrosion behavior.  Relationships between the radial and axial deep rolling force components, F r and F a , and work speeds (V) for the investigated cooling modes are shown in Fig. 5a,b respectively. It can be seen that deep rolling under cryogenic cooling generates higher radial and axial force components when compared to those generated under dry condition. Moreover, it is observed that the highest levels of deep rolling force components are obtained when processing under conditions with low speed. Furthermore, it can be stated that radial force differences between dry and cryogenic conditions become low at the highest speed level (120 m/min).

Deep rolling temperature
The temperature map during the deep rolling process under cryogenic cooling is shown in Fig. 6 that depicts the processinduced temperature variations along the axial direction of the workpiece. Cryogenic cooling by liquid nitrogen significantly reduces the deep rolling temperature at the roller-workpiece contact point (Fig. 7). In addition, it is observed that the highest temperature level is recorded at the contact zone between the roller and the workpiece and decreased smoothly with increasing distance toward the end of the sample. It is also remarked that the maximum temperature value at the contact region is reached when processing with a high work speed. According to Manimaran et al. [16], liquid nitrogen, sprayed on the interface between the deep rolling tool and the workpiece, forms a gas/fluid cushion reducing the friction force at the contact point and provides an efficient absorption of heat. The temperature reduction is considered as the major reason for the favorable effects of cryogenic cooling.
3.2 Effect of the cryogenic cooling on the AISI 304L deep rolled surface integrity

Surface topography
SEM micrographs of AISI 304L machined surfaces depict machining grooves (Fig. 8a) that are completely removed after deep rolling under cryogenic cooling (Fig. 8c). This  can be explained by the repetitive action of the roller on the workpiece resulting in the increase of surface homogeneity and, thus, a better surface finish is generated. Since the AISI 304L stainless steel has a low thermal conductivity, the frictional forces increase during deep rolling process and so, the temperature increases at the contact zone (Fig.  7). In the case of dry condition, surface asperities generated by the machining process are quite difficult to be completely removed by deep rolling (Fig. 8b). However, the liquid nitrogen applied to the deep rolling zone highly reduces the material softening and so, the surface asperities can be readily eliminated (Fig. 8c).
The micro-geometrical quality of the deep rolled surface is characterized by the arithmetic (R a ) and the total (R t ) roughness parameters (Column A of Table 5). The R a and R t values of each sample are the average of three measurements. It can be seen that the application of deep rolling significantly improves the surface micro-geometrical quality of AISI 304L machined surfaces. Indeed, the mean roughness R a is reduced from 4.04 to 1.03 μm, and the total roughness R t decreased from 18.53 to 5.34 μm when processing under dry condition at a speed of 25 m/min. Figure 9a,b shows that the application of liquid nitrogen leads to a considerable improvement of surface roughness comparative to dry condition. For cryogenic condition, the maximum decreases of surface roughness measurements are obtained at a work speed of 120 m/min. Under this condition, the mean roughness R a is reduced by 92%, and the total roughness R t is decreased by 89% comparative to the machined state.

Microstructure
The optical micrograph of the turned AISI 304L cross section (Fig. 10) reveals a deformed structure characterized by slip    [17]. The bulk material microstructure is close to the as-received state that consists in equiaxed grains with an average size ranging from 30 to 50 μm. However, metallographic examinations of the cross section of deep rolled samples (Fig. 11) depict a clear interface between a recrystallized layer with indistinct grain boundaries formed at the topmost surface and a subsequent transition layer characterized by the pattern of sweeping grains. In fact, a large amount of intersecting and irregular multiple-slip bands marked by a high density of deformation twins (Fig.  12) is observed near the topmost surface as shown in Fig. 11. With an increasing depth into the matrix, single slips progressively substitute multiple-slip bands located in the nearsurface regions. It can be stated that the deep rolled surface microstructures consist in austenite grains γ, which have partially transformed to α ′ -martensite grains due to severe plastic deformation. The γ → α ′ transformation is attributed to the nucleation of α ′ -martensite at the intersecting twins due to the relatively low intrinsic stacking fault energy of austenitic stainless steels [18]. Many investigations [2,3] reported similar results for stainless steels type 304.
The topmost deep rolled surface layers consist in ultrarefined grains which are not clearly visible (Fig. 11). These results highlight the fact that the successive action of the deep rolling tool results in severe plastic deformation (SPD) that further leads to grain refinement [19,20]. Indeed, the deep rolling process generates high-energy impacts at high deformation rates, which consequently creates dislocations. Therefore, the number of dislocations progressively increases due to the repetitive impacts. Moreover, it was observed that  Fig. 9 Effects of the cooling mode on the surface roughness of the AISI 304L deep rolled surfaces the thickness of refined layers becomes smaller at high deep rolling speeds (Fig. 11c,d) as the refined grains in the subsurface will grow larger [21]. Figure 13 shows X-ray diffraction patterns of the machined and deep rolled samples. It is known that the austenite phase in type 304L stainless steel is unstable when deformed at room temperature and tends to transform into strain-induced martensite α ′ [22]. The XRD spectra of the machined specimen consist of γ-austenite peaks, while both γ-austenite and α ′martensite peaks are observed in the case of deep rolled surfaces. Actually, α ′ peaks exhibit the existence of remnant martensite which has failed to transform completely into austenite along with the increased intensity of the austenite peak resulting from surface recrystallization. During the deep rolling treatment, α ′ -martensite induced in the plastically deformed layers tends to transform back into austenite, whereas the complete transformation γ → α ′ is not achieved due to the short interaction time of the deep rolling treatment [23]. Besides, the martensitic transformation occurs with the formation of intersecting twins resulting from the successive stacking fault overlap on the (111) γ plane during plastic deformation [3]. Considering the existence of mainly two phases in the processed samples, the (111), (200), (220), (311) and (222) reflection planes for γ-austenite as well as (110), (200), (211) and (220) reflection planes for α ′ -martensite are used to estimate the volume fraction of each phase by means of Eq. 1.

Phase changes
The calculated volume fractions of the strain-induced martensite in the machined and deep rolled workpieces are reported in column B of Table 5.
It clearly appears that deep rolling under cryogenic cooling produces higher amounts of plastic-induced martensite than those generated at the dry condition. This result agrees with  [22] reporting that high deformation rates and low temperatures promote the excessive formation of α ′ -martensite. According to Fig. 13, the highest volume fractions of strain-induced martensite, obtained after deep rolling at a work speed of 25 m/min under dry condition and cryogenic cooling, are 40% and 21%, respectively. However, no significant difference is found between the volume fractions of α ′ -martensite generated after deep rolling at 75 and 120 m/min for each cooling condition.

Surface work hardening
Surface work hardening induced by machining and deep rolling process was performed as described in Section 2.3. Measurement results are shown in column C of Table 5 and Fig. 14. This figure clearly reveals the existence of workhardened layers induced by plastic deformation under both conditions: dry and cryogenic cooling. The surface hardening rate is increased by 94% for the turned surfaces with regard to the bulk material (197 HV 0.1 ). It can be seen that the highest surface hardness values are recorded when processing at a deep rolling speed of 25 m/min. At this speed, the hardening increasing rate is about 127% for deep rolled surfaces under dry condition compared to the turned state. Moreover, cryogenic cooling is seen to introduce higher surface work hardening where the average increase of the surface hardness is 142%. These results are in a good agreement with the deep rolling forces generated at this speed. Concerning depths of the hardened layers, profiles of Fig. 14 show that the turning process generated a work-hardened layer of about 400 μm. Meanwhile, depths of these layers are about 650 μm and 800 μm for deep rolled surfaces under dry and cryogenic cooling, respectively. For deep rolling speeds of 75 and 120 m/min, no significant differences of hardening, at the nearsurface layers, were highlighted. The increased AISI 304L surface hardness results from ultra-fined grains generated in the near-surface regions [11] and plastic-induced martensite (Column B of Table 5) [18].

Residual stress
The surface residual stress measurements in the longitudinal (σ 11 ) and the transverse (σ 22 ) directions for machined and deep rolled samples are shown in Column D of Table 5. It is observed that the machining process generates tensile residual stress distribution in the transverse direction (360 MPa), whereas the axial residual stress fields are compressive (−394 MPa). These stresses are generated by the thermal and mechanical effects that take place at the material-tool interfaces during the material removal by the turning process [24].
The corrected near-surface residual stress profiles of AISI 304L deep rolled samples under both dry and cryogenic conditions are shown in Fig. 15. It should be noted that XRD measurements were limited to a depth around 800 μm. Beyond this depth, the measured residual stress levels are considered unreliable because of the coarse grain size [25]. It can be seen that compressive residual stress distributions in both directions are obtained after the application of deep rolling under dry and cryogenic conditions. The compressive residual stresses resulting from the mechanical effects of deep rolling forces (Fig. 4) are formed in order to create an equilibrium state between the surface and the near-surface layers [26]. The surface residual stress levels in the longitudinal direction are much higher than in the transverse direction. Moreover, the application of liquid nitrogen increases significantly the surface compressive residual stress amplitudes in  Table 5). Analysis of the residual stress profiles of the deep rolled surfaces under both dry and cryogenic conditions reveals that the highest levels of the compressive stresses are reached at depths ranging from 50 to 100 μm (Fig. 15). Under cryogenic cooling, the maximum levels measured for each work speed rate are σ 11  Accordingly, the highest levels of the compressive residual stresses are obtained when processing at a work speed of 25 m/min using cryogenic cooling for which a content of 40% of strain-induced martensite was evaluated. This result is in good agreement with the findings of Wojciech et al. [27] that reported that the martensitic transformation from metastable austenite leads to volume increase that makes the stresses in the martensite highly compressive. The measured surface residual stress values are found to be correlated with the corresponding strain-induced martensite volume fractions (Columns B and D of Table 5).

Effect of the cryogenic cooling on the AISI 304L deep rolled corrosion behavior
The corrosion behavior of machined and deep rolled samples is assessed by means of potentiodynamic anodic polarization test in synthetic seawater. The corrosion current density (Icorr) and the corrosion potential (Ecorr) are determined from polarization curves shown in Fig. 16. The pit potential (Epp), passivation potential (Epass) and passivation current density (Ipass) are evaluated from cyclic polarization curves given in Fig. 17. Results are reported in Table 6. The Icorr and Ecorr values are calculated using the Tafel extrapolation method. It can be clearly seen from Table 6 that the application of deep rolling after machining generates lower values of Icorr and Ecorr. Ennobled potentials Epp and Epass and, likewise, lower values of Ipass are obtained. It is known that low values of Icorr and ennobled values of Ecorr depict higher resistance to uniform corrosion [2,3]. Similarly, Epp and Epass with more positive values indicate better resistance to pitting corrosion [28]. Besides, low values of Ipass indicate higher stability of the formed passive film [2]. Therefore, these results point out the fact that deep rolling process improves the resistance to uniform and pitting corrosion of machined surfaces. Concerning the effect of the deep rolling environment, no general conclusion can be given. Indeed, Table 6 indicates that the application of the cryogenic cooling generates higher corrosion resistance for high working speeds only (75 and 120 m/min). For low working speed V = 25 m/min, AISI 304L deep rolled under dry condition offers better corrosion resistance. Nevertheless, it can be established that enhanced resistance to uniform and localized corrosion and also the stability of passive film can be achieved when deep rolling treatment is performed at high-speed levels under cryogenic cooling. As shown in Table 6, for the working speed of 120 m/min, the corrosion potential Ecorr shifts from −201 mV/SCE for dry condition to 338 mV/SCE when liquid nitrogen is applied, and an increase of the pit potential (Epp) of about 19% was recorded in this case.
In order to explain these enhancements of the corrosion resistance of AISI 304L machined surfaces, it is essential to investigate the effects of surface integrity modifications resulting from the deep rolling process on the resistance to uniform and pitting corrosion in aggressive environments.

Discussion
According to the potentiodynamic polarization test results (Table 6), it can be stated that the AISI 304L deep rolled samples, under both processing conditions (dry and cryogenic), exhibit an improved corrosion behavior with regard to the machined state. Compared to machined samples, an excellent corrosion resistance is particularly observed when liquid nitrogen is applied. This is related to the favorable effects of low temperature that decreases friction forces and contributes to an efficient absorption of heat. Indeed, cryogenic deep rolling leads to better surface quality due to repetitive passes of the roller that eliminates machining marks (Fig. 8).
Consequently, it is shown that mean roughness is considerably reduced from 4.04 to 0.31 μm, and total roughness decreased from 18.53 to 2.05 μm (Fig. 9). This mechanical treatment also produces an ultra-refined surface layer with high density of deformation twins in the near-surface regions (Figs. 11 and 12), whereas only stretched grains are depicted in the turned sample microstructure (Fig. 10). Besides, it increases the surface work hardening by 142% with regard to machining (382 HV 0.1 ). Cryogenic deep rolling introduces high compressive residual stress amplitudes that reach up to −1128 MPa in the longitudinal direction. Moreover, surface residual stress levels in the transverse direction are shifted from tensile (+360 MPa) to compressive (−683 MPa). These surface characteristics' modifications result from severe strain occurred during deep rolling and depend on controlling process parameters.
In the following, the discussion is focused on the impact of dry and cryogenic deep rolling on the AISI 304L electrochemical behavior in synthetic seawater. Here, the effects of the deep rolling environment must be detailed as controversial results depending on the deep rolling speed when no cooling and liquid nitrogen are considered. This issue is discussed based on the modification of surface topography, microstructure, strain-induced martensite, surface hardening and residual stresses owing to material straining performed at different temperatures.
In this work, it is shown that deep rolled samples at 75 and 120 m/min under cryogenic cooling exhibit good corrosion resistance. As reported in Column B of Table 5, it is seen that an increase of α ′ -martensite amounts improves the electrochemical behavior of AISI 304L. This enhancement is reflected in the measured values of corrosion indicators (Ecorr, Icorr, Epp, Epass and Ipass) representing the formation of more protective passive film. The beneficial effect of straininduced martensite, observed within a certain range, is in accordance with the study of He et al. [29]. In addition, the reduction of surface roughness (Column A of Table 5) inhibits pit activation and growth by reducing the number of   metastable pit sites. The beneficial effect of surface quality enhancement on corrosion behavior was reported for the cases of peening [3] and polishing [30]. It is also found that the increase of surface work hardening (Column C of Table 5) results in the improvement of corrosion behavior. In fact, microhardness measurements are related to the density of dislocations and stacking faults occurring by cold work hardening. These promote Cr diffusion to the material surface leading to the creation of passive film rich in Cr that improves the corrosion resistance [6]. Regarding the contribution of compressive residual stresses (Column D of Table 5), an increase of about 300 MPa seems to increase significantly the corrosion resistance. Surface compressive stresses decrease interatomic spacing and thicken the chromium atoms which results in the growth and stability of the passive film. Indeed, the favorable effect of compressive residual stress on corrosion resistance was confirmed by Takakuwa et al. [31] for the case of polishing treatment. Therefore, it can be concluded that the electrochemical behavior of AISI 304L is attributed to the decrease of surface roughness, the increase of surface strengthening and compressive residual stress distributions. Nevertheless, it is shown that the AISI 304L samples deep rolled at 25 m/min under cryogenic cooling has the weakest corrosion resistance. This can be attributed to the highest volume fraction of strain-induced martensite (40%) which can be detrimental due to the galvanic effect occurring between two different phases (austenite and martensite). In fact, many investigations [3,6] have confirmed the negative effects of the increased volume fraction of strain-induced martensite exhibited in other mechanical processes applied to other stainless steels. This deep rolling condition generates a slight reduction of the surface roughness (Ra = 0.99 μm) comparatively to dry condition (Ra = 1.03 μm) which has no significant impact on corrosion resistance. Moreover, the additional surface work hardening produced at the surface (around 56 HV 0.1 ) alters the corrosion behavior. Indeed, the high density of dislocations and stacking faults induced by severe plastic deformation can increase the sensitivity to pitting corrosion [32]. Furthermore, it seems that the increase of surface compressive residual stress by 200 MPa does not give the same beneficial effect on the corrosion behavior as that showed for deep rolling at speeds of 75 and 120 m/min. Consequently, the electrochemical behavior under this condition is mainly influenced by the high amount of strain-induced martensite that nullifies the favorable effect of compressive residual stress fields. It can be deduced that the deterioration of corrosion resistance is worsened by the increase of surface work hardening.
According to these findings, it can be concluded that the combination of surface roughness, strain-induced martensite, surface work hardening and residual stresses can significantly improve the corrosion behavior of AISI 304L deep rolled samples under cryogenic cooling. Thus, the application of cryogenic deep rolling treatment at high speeds can significantly increase the corrosion resistance of stainless steel mechanical components subjected to aggressive environments.

Conclusions
An experimental investigation was performed to study the effects of deep rolling parameters, namely the work speed and the cooling mode, on the electrochemical behavior of AISI 304L components in chloride environment. The main findings of this study are summarized as follows: a. Refined grains in the surface layers are produced by deep rolling process either with or without the application of liquid nitrogen. This grain refinement improves the corrosion behavior of deep rolled samples with regard to machined state. b. An interrelated correlation between many factors such as surface roughness, strain-induced martensite, surface microhardness and residual stresses is found to affect the corrosion behavior of deep rolled specimens. c. At a speed of 25 m/min, cryogenic deep rolled samples exhibit the weakest electrochemical behavior. This deterioration is mainly attributed to: & The high amount of strain-induced martensite (40%) resulting in galvanic effect. This nullifies the favorable effect of compressive residual stresses introduced in the near-surface regions. & The high plastic deformation, which leads to strain hardening, and thus, a high increase of the surface microhardness that reaches up to 562 HV 0.1 .