GB2372041A - Electrochemical surface treatment of metals - Google Patents

Abstract

A method for treating a metal comprises subjecting the metal to electrolysis in the presence of an electrolyte using alternating pulses of voltage and/or current, said alternating pulses being of opposite polarity. The method is effective in improving the surface properties of the metal, such as hardness, friction and wear properties. The metal treated is preferably austenitic stainless steel and the electrolyte is preferably aqueous sodium nitrite. The electrolyte may also be non-aqueous. The metal may be subjected to heat treatment after the electrolysis.

Description

Patent Application Electrochemical Surface Treatment of Metals and Metallic Alloys The subject of this application for a patent is the electrochemical treatment of metals and metallic alloys in aqueous or non-aqueous electrolytes involving the cyclic variation of the voltage and/or the current at the metal surface in order to achieve improvements in surface properties (such as hardness, friction, wear, corrosion resistance or other properties) through modification of surface and near-surface material, for example.

through formation of compounds and/or modification of the metallic structure. cr By way of example we append a paper entitled"Electrochemically induced annealing of stainless-steel surfaces" (nine pages), a report entitled"Improving the tribological

properties of stainless steel surfaces by novel electrochemical processing" (thirty-nine c pages), and another report entitled"Tribological enhancement of stainless steels by electrochemical surface treatment" (twelve pages). 1. Introduction A new surface treatment involving electrolysis in aqueous solutions has been developed to modify surfaces of austenitic stainless steels in order to overcome a galling problem. It is known that the surfaces of austenitic stainless steels can be transformed to manensite by the strain through mechanical deformation. Our experiments have shown that the strain-induced martensite within the austenite phase can be relaxed and removed by appropriate electrochemical processing. The hardness of the surface increases after the same electrochemical treatment despite loss of the manensite. Sliding wear tests have revealed that friction force during low-speed sliding is effectively reduced by the surface treatment for a certain period. The electrical contact resistance of the treated surface against a sliding steel ball is remarkably high during the low friction regime. The results indicate that gallingresistant film has been produced on the surface by the electrochemical processing. We conclude that this novel electrochemical procedure induces an improvement in the tribological properties of austenitic stainless steel surfaces, while eliminating the brittle martensite phase from the surface.

Austenitic stainless steels are notoriously susceptible to salling. i. e.. high z friction. the formation of strong adhesive bonds, and extensive surface damage when sliding against themselves in the absence of effective lubrication [1]. Localised

material transfer or removal occurs between two sliding metals under high loads. and c severe gulling can cause seizure. This poor tribological performance limits many c A--c applications of austenitic stainless steels where they would otherwise be favoured for their corrosion resistance and ductility.

Various methods have already been developed for improving the mechanical properties of the material surface, including electroplating such as chromium plating [2]. thermochemical diffusion processes (e. g. nitriding. boronizing. and carburizing) [3]. PVD and CVD processes to produce hard non-metallic coatings [4, 5]. Recent development of nitrogen ion implantation at appropriate temperatures is also promising to enhance the galling resistance of austenitic stainless steel surfaces [6]. However, it is known that chromium plating carries with serious health and environment hazards [7], and other methods described above cannot readily or economically be applied to relatively large substrates. A new electrochemical surface treatment in aqueous solutions at low temperature is very appealing in that it may potentially overcome these problems, and it is also expected to make iii sitii surface modification of real structures possible.

The passivity of stainless steels originates from the very thin oxide film on the surface. The thickness of the passive film is typically ca I nm at equilibrium in air at room temperature. However, disruption of the film occurs readily when the asperities of two surfaces make sliding contact, and consequently bare metals contact each other.

This phenomenon is responsible for their strong susceptibility to galling [8]. In other words, galling can be effectively reduced by increasing the film thickness.

The objective of this project is to study the feasibility of novel ideas to modify the austenitic stainless steel surface in order to produce a galling-resistant film through electrolysis in aqueous solutions. Results from the tribological tests of the treated surfaces were fed back to modify and refine the electrochemical process. The electrochemical and tribological processes must have complemented each other in order to tackle the objective. The present communication describes some preliminary results, which are, ~however, believed to pave the way for further development of applications where surface stability and hardness are important.

2. Experimental 2. 1 Elecrrochemical Cell The specimens used for the present measurements were cut into I cm squares from 304L stainless steel sheet (1.2 mm thickness). Analysis of the metal by energy dispersive X-ray (EDX) analysis gave Cr 18.8, Ni 7. 8. Mn 1. 3, Si 0. 87, Fe bal.

Carbon and sulphur were not analysed. Each sample was carrying a thin shank from one corner to allow electrical contact. The surfaces of each were ground and polished to a 0.25 m finish with successive grades of silicon carbide paper and diamond paste.

gh-purity graphite (99. 997%) was used as the counter electrode. A aqueous 0 c solution of 8M NaNO ; was prepared from the analytical grade reagent and doubledistilled water.

The electrochemical treatment involved electrolysis using a two-electrode cell.

A schematic of the electrochemical cell is illustrated in Figure 1. The cell voltage between the stainless steel specimen and the graphite electrode was applied by a bipolar high power supply (KEPCO, Model BOP 100-2M). The solution temperature was set at 80'C by a home-built temperature controller with a thermistor and a tape heater (containing a wound nichrome wire). The stainless steel working electrode was then subjected to anodic or cathodic side of electrolysis, depending on the cell voltage applied. A function generator (Hewlett-Packard, model 8844) was employed to control the bipolar power supply to generate a series of anodic/cathodic voltage pulses.

The electrode potential was monitored using a saturated calomel reference electrode The electrode potential was mon (SCE), which was placed in a separate holder outside the test cell and was connected to the system by a Lutin probe and a salt bridge. The electrode potential was fed via a high impedance (1010 #) home-built buffer amplifier (of unit gain) to a digital voltmeter (Hewlett-Packard, 3457A). The cell voltage or the electrode potential was recorded on a computer via the digital voltmeter at a sampling rate of 3. 5 Hz if necessary.

After electrochemical treatment, the specimen was withdrawn and washed thoroughly. X-ray diffraction (XRD) using CuK radiation. Auger electron spectroscopy (AES) and EDX analyses were then carried out to gain some insight of the electrochemically treated surface.

2. 2 Friction tests Friction tests were performed using a low-speed sliding wear apparatus. shown schematically in Figure 2. The specimen was fixed on the sliding stage, which

-1 baU (AISI 52100 was slid in the directions of the arrow driven by a motor. A steel ball (AISI 52100 stainless steel. Diameter : 6. 35 mm) was clamped beneath a pivoted weight-stand. which allowed dead weights to place on it to gain a desired normal force. The specimen thus slid repeatedly over the same track against the stationary ball with a constant normal force. The friction force was sensed through a transducer which was mounted on the pivoted weight-stand. Details of the algorithm to calculate the friction coefficient in the present study are described in APPENDIX 1. All friction tests were carried. out under the same conditions, with a normal force of 2. 15 N. a sliding

distance Of cal 6 mm and each sliding time of 1 s. All measurements were made in air without lubrication at ambient temperature (175"C).

Electrical contact resistance between the specimen and the steel ball was measured simultaneously. A small constant current I (1 mA) was applied between them during sliding tests using a home-built electronics device. The contact resistance

(AV7) was thus obtained readily by measuring the potential difference (AV) between c them. The applied current is regarded low enough to eliminate the influence on sliding wear process.

3. Results 3. 1 Disappearance ofManensiric Sn'lIctlire XRD analysis of 304L stainless steel surface polished to a 0. 25 m finish is shown in Figure 3 (solid line). The XRD pattern shows the presence of the austenitic structure (y) together with small reflections from the martensite structure (ex'). It is known that plastic deformation at room temperature can lead to the formation of martensite in metastable austenitic steels [9-16]. We can thus anticipate that the surface of 304L stainless steel was rendered partly martensitic structure during the mechanical surface preparation. The specimen surface was then subjected to cathodic treatment in an aqueous solution of 8 M NaNO2 for 6 h at a solution temperature of 80 C. The cathodic treatment was carried out using a two-electrode cell with a cell voltage of-4. 1

V. The XRD pattern of the surface after the cathodic treatment is also shown in Figure cr 3 (broken line). It has been reported that cathodic charging in some aqueous electrolyte systems can result in a considerable amount of transformation to martensite because hydrogen atoms reduced from water during cathodic treatment may be

absorbed into austenitic steel surface, consequently leading to the lattice strain [17-20]. z As presented in Figure 3. however, no discernible increases in peak intensity of L reflections from α' martensite were observed in the XRD patterns.

It should be noted that details of the XRD data reveal that each peak position is shifted slightly after the cathodic treatment. This implies that the lattice is distorted due to absorbed hydrogen into the metal. Table 4 lists changes in lattice parameter ( (Aala) xl00) after electrochemical treatments in different electrolytes. Each data shows the mean value of the changes in five peaks from the austenitic structure'. A remarked feature is that the cathodic treatments in potassium carbonate and potassium hydroxide give rise to the lattice expansion, whilst those in sodium nitrite lead to the lattice contraction (the data after the pules treatment is discussed in 4. 1). In general, cathodic charging results in a considerable lattices expansion due to introduction of

hydrogen atoms into the lattice of the metal [36]. The origins of the lattice contraction c

'Y t (111). (200). (220), (311). and (222) planes. served in nitrite solutions are discussed in 4. 1.

Table 4 Changes in lattice parameter after electrochemical treatments in different electrolytes

(#a/a) x 100 Electrolytes Electrochemical Treatment Conditions (%) (%) 8M K. CO, Cathodic,-4. 5 V (-0.7 A cm'2). 80'C, 6 h +0. 102 8M KOH Cathodic,-5. 0 V (-0.7 A cm 80 C.6 h +0.109 8M N NO, Cathodic-4.1 V (-0.7 A cm' 6h. 80'C-0. 050 20M Cathodic,-5.0 V (-0.7 A cm-2), 6h. 100'C-0. 083 NaNO2 8M NaNO. Pulse.-4. 1V/+1.2V (-0.7 A cm-2), 80 C, -0. 213 3h(*)

This pulse condition is described later.

Figure 5 shows the XRD pattern obtained of 304L stainless steel after 0 cathodic/anodic pulse treatment in 8 M NANO, at a temperature of 80"C together with ce that from the same specimen polished to a 0. 25 urn finish before the pulse treatment.

The pulse treatment comprised 33 cycles of the cell voltages which alternated between - 4. 1 V for 277 s (cathodic) and +1. 2 V for 56 s (anodic). Total processing time was 3 h. An intriguing feature shown in Figure 5 is that the reflections from a martensite disappeared completely (at 26=44. se, 64.5 , 82. 8 ) whilst those from the austenite

were unchanged. The manensitic phase, which was originally generated by the strain c during the surface preparation, was removed after the electrochemical pulse treatment one may speculate that the martensite generated on the surface may have c dissolved away during anodic pan of the pulse treatment. Thus, we attempted to treat the steel surface by anodic treatment alone. The specimen surface was subjected to anodic treatment in 8 M NaNO2 under a cell voltage of +1 : 2 V, which was the same as that of anodic part of the pulse treatment. The resulting XRD patterns are presented in

Figure 6. Since the a' (110) plane gives rise to the most prominent reflection at L-e 26=44. 5" amongst those from martensite structure [17]. XRD patterns around this peak are shown in the following context. Figure 6 shows no loss of martensite even after 2 h anodic treatment Note that total anodic processing time of the pulse treatment in Figure 5 was 30 min. Thus, the martensite removal through the pulse treatment was not simply caused by dissolution during its anodic processing. Clearly the martensite disappearance was a consequence of the application of both the anodic and cathodic potential pulses, rather than from the individual anodic or cathodic part of the electrochemical signal alone.

An example of pulsed cell voltage and electrode potential responses is given in Figure 7. The cell voltage alternated between-4.3 V for 277 s (cathodic) and +1. 2 V for 56 s (anodic). The electrode potential at the cathodic part of the pulse treatment was ca-2. 0 V (SCE), whilst that at the anodic part is ca +0.28 V (SCE) for the first anodic pulse and decreases slightly with consecutive pulses, showing M +0. 12 V (SCE) at steady state. It is to be noted that these anodic potentials are lower than that of transpassive dissolution of chromium for 8 M NaNO ; (+0. 337 V (SCE)) [21].

It is known that the stress-induced martensite can be removed by heat treatment over 700uC in the absence of the electrochemical pulse [22]. Figure 8 sh ows the XRD pattern of the 304L stainless surface after thermal annealing at 750'C for 20 min in argon, together with those before and after the electrochemical pulse treatment (which

are the same data as shown in Figure 5). Thermal annealing procedure also gave the c 0 loss of the martensite, although a very small martensite peak is still visible. The electrochemical pulse treatment has the same effect as thermal annealing, so that we termed the martensite disappearance by this means"electrochemically induced annealing" [23].

As described above, the surface of austenitic stainless steel can be transformed to martensite by mechanical deformation. More severe plastic deformation leads to greater formation of martensite [24]. The effect is demonstrated in Figure 9, which

li shows the XRD patterns from steel surfaces after different finishes and cold-rolling. The surface ground to a 2500 grit finish is rougher than that polished to a 0. 25 J. lm finish, and is expected to suffer from more deformation. Specimens of the same steel were also cold-rolled to produce a 25% reduction in thickness. Martensite must have been induced throughout the structure by the cold-rolling [24]. The XRD pattern from the cold-rolled specimen shows clearly that amplitude of the reflection from a' (110) plane increases significantly with the degree of mechanical deformation (Figure 9). In

particular, the far greater amplitude of the martensite peak from the cold-rolled sample c is due to the fact that the rolling treatment generated strain-induced martensite throughout the structure. z Since the surface finish by 0. 25 pm mechanical polishing is very fine one. one may consider that subtle changes in diffraction intensity observed in Figures 6 and 8

can be due to any increase in surface roughness during the electrochemical treatment, z which subsequently affects the XRD measurement, and therefore may not be due to the loss of the strain-induced martensite. As explained later, the application of optical profilometry showed that the surface roughness was approximately unchanged after the electrochemical pulse treatment. We then applied electrochemically induced annealing process to the cold-rolled specimens (the electrochemical treatment was the same as described for the data in Figure 5). As shown in Figure 10. the electrochemical pulse treatment reduced the peak from martensite remarkably over a

period of 9 h. Although the procedure took longer to reduce the martensite peak c substantially, we nevertheless observed a significant reduction even after 3 h. The electrochemically induced removal of martensite clearly occurs to some significant depth beneath the metal surface. and is not confined to the surface atoms alone2.

The phase transformations between y austenite and a'martensite induced by mechanical, thermal, and electrochemical processes observed in the present section are summarised in Figure 11.

3.2 Microhardness The electrochemical annealing process described above gives rise to mechanical properties of the surfaces that are different from those achieved by thermal annealing.

This is shown in Figure 12, which presents the hardness of the surfaces measured using a Vickers microhardness indenter (with a 50 g load). Error bars stand for the standard deviation of between three and five independent measurements. After polishing to a 0. 25 m finish, the stainless steel surface had an initial harness of 205 Hv (Figure 12 (a) ). The thermal annealing at 750. C for 20 min reduced the hardness to 196 Hv (Figure 12 (b)). consistent with the loss of the strain-induced martensite. The annealed specimen was again polished to a 0. 25 pm finish, resulting in rehardening to the initial hardness (Figure 12 (c) ).

After the electrochemical annealing procedure described above. the hardness was in contrast still high, at 235 Hv, despite the loss of the martensite as shown in Figure 12 (d). Moreover, when the surface had been electrochemically annealed and subsequently thermally annealed at the lower temperature of 550'C for 20 min. the hardness remained high (252 Hv as presented in Figure 12 (e)). significantly higher than that was achieved by thermal annealing alone.

The hardness resulting from the electrochemical pulse treatment might at first glance be thought to be induced by the penetration of hydrogen into the metal matrix : this possible explanation was examined experimentally by applying a cathodic treatment alone, without the anodic pulse using a -more negative voltage of -4.5 V and a far longer time of 6 h. Under these conditions, hydrogen is generated cathodically on the metal surface by reduction of water. Note in comparison that the total cathodic component of the annealing pulse treatment was only 2. 5 h. The hardness generated under this more severe exclusively cathodic treatment was 238 Hv as shown in Figure 12 (t). However this was removed b Z the subsequent 550 C heat-treatment (Figure 11(g)) in contrast to the corresponding hardness after the pulse treatment which

* The depth at which X-ray penetrates into the 304L stainless steel surface, is calculated to be 6. 9 Am tor the reflection from ex' (110) plane at 26=44. 3* assuming (1/1o) =0. 9 (where I is we intensity z transmitted and 10 is the original beam intesity) [50]. c remained high lifter 550"C thermal annealing (Figure 12 (e)).

We conclude that the electrochemical annealing procedure engendered by pulse c treatment induces relatively irreversible and stable changes into the stainless steel surface involving loss of the martensite phase and retention of surface hardening.

3. 3 AES and EDX Anal. v ! ; es Figure 13 shows Auger spectra from 304L stainless steel surfaces with no z c electrochemical treatment and after electrochemical pulse treatment The voltage pulse comprised 216 cycles each out 4. l V for 50 s (cathodic) and +0. 42 V for 50 s (anodic) during 6 h, equivalent to electrode potential pulses between ca-2. 1 V (SCE) and ca. 0. 1 V (SCE). The effect of the electrochemical pulse treatment is clear ; the peaks from 0, Cr and Ni increased after the treatment whilst that from Fe decreased. In particular. the increase in Ni content and the decrease in Fe content are prominent. The alloying elements such as Ni and Cr are known to increase pitting corrosion resistance in aggressive solutions containing chloride ions [25]. No discernible N peak was observed in either trace .

Enrichment of Ni and 0, and depletion of Fe after pulse treatment were also observed in EDX spectra (Figure 14). The electrochemical pulse procedure was the same as described in Figure 5. The change in Cr peak (583 eV) is not seen clearly, presumably because it is masked by the greater increase in 0 peaks. No N peak was observed (389 eV) in Figure 14; this is because the soft x-ray window used in the EDX system cuts off the signals around 340 eV as another EDX measurement demonstrated that no N peak was detected even for a zirconium nitride specimen. It is generally known that EDX gives information on much deeper depth in the surface. whilst AES detects only a few tens of atomic layers (approximately equivalent to a few nanometers in depth) [26]. The depth of the X-ray emitting layer at an acceleration voltage of 5 keV is estimated to be the order of 0.5 Lm. We can thus conclude that the enrichment of Fe and 0 and the depletion of Ni occur to some considerable depth

through the electrochemical pulse treatment.

3. 4 Hydrogen Penetran"on and Release It is well known that hydrogen is penetrated into austenitic steel surfaces by c cathodic charging in aqueous electrolyte [27-30. 37]. By applying subsequent anodic treatment, hydrogen absorbed in the metals is envisaged to be released into the c c electrolyte. Figure 15-shows polarisation curves of 304L stainless steels in 8 M The N peak in AES is positioned at 389 eV [26).

NaNO, at a sweep rate of 10 mV s''at ambient temperature, immediately after cathodic treatments at different electrode potentials for 20 min in the same electrolyte. In the tests, the potential was swept from the electrode potential of each cathodic treatment.

This type of cathodic treatment is referred to as"prior"cathodic treatment in the following context. Two interesting features are observed in Figure 15; first, we see ramps in the passive states just above the corrosion potentials. Second, the passive current density increases with decreasing the electrode potential applied during the prior cathodic treatment. Another experiment demonstrated that the polarisation curve swept from-1. 0 V (SCE) at a seep rate of 10 mV s t in the absence of prior cathodic treatment showed (i) no ramp in the passive state and (ii) lower passive current density than any of those observed after the prior cathodic treatments. Therefore, we deduce that these ramps were caused by release of hydrogen absorbed during the prior cathodic treatment; more hydrogen is likely be absorbed into the metal at lower potentials, thereby generating higher passive current density in subsequent anodic polarisation.

The anodic charge density q that has flowed for a time period from the

corrosion potential Ecorr to a given potential E ( > Ecorr) is given by

q = f-djEx-- (1) E. de

where dt/dE is the reciprocal sweep rate. The integral represents the area under the polarisation curve which was obtained by graphical integration. Figure 16 presents the anodic charge density integrated between the corrosion potential and-0.20 V (SCE) (the time period is ca 60 s), as a function of the applied potential in prior cathodic treatment. It is shown clearly that the charge density increases with increasing the electrode potential. Assuming that all the charge is expended only by re-oxidation of hydrogen in the metal, we can estimate the number of hydrogen ions released into the electrolyte. For instance, the charge density of 4.0 mC cm' at an applied potential of1.9 V (SCE) (Figure 16) indicates that 2.5 x 1016 hydrogen atoms ( !) per square centimetre are re-oxidised and the resulting ions are released into the electrolyte by the anodic treatment.

3. 5 Sliding Wear Figure 17 shows a trace of the friction coefficient measured from 304L stainless steel polished to a 0. 25 pm finish under a load of 2. 15 N in the absence of electrochemical surface processing, together with one of the electrical contact resistance between the specimen and the sliding steel ball measured simultaneously. The time zero in the figure represents the commencement of the sliding test. Three gimes can be distinguished in the variation of the friction coefficient. In regime I, the friction coefficient raised rapidly after the sliding wear test started and attained the highest values around 1.2-1. 6 at ca 15 s. The friction coefficient dropped shortly and showed the lowest values around 0.6-0. 7 at ca 60 s (regime II). Then, the friction coefficient increased gradually, and the contact resistance started to show discernible values after 200 s. It should be noted that the sliding test time (tel) of 200 s is equivalent to that of 200 sliding cycles. The region, characterised by the slow increase in friction coefficient and the onset of high contact resistance, is termed as

regime III. The boundary between regimens 1. II and III is poorly defined.

The topography of the worn surface in each regime was examined by optical profilometry. The specimen surface was originally polished to a 0.25 m finish.

After sliding wear test was performed for a defined period of time. a three-dimensional (3D) surface profile of the sliding scar was measured. An example of the 3D profiles is depicted in Figure 18 (Ist = 353 s). From these records, two-dimensional (2D) traces across the scars were derived to explore development of their geometry with time of sliding wear.

Figure 19 shows a series of scar profiles measured after given periods of the

sliding wear tests. As a whole. the scar width increases with time of the sliding wear. Dotted curves in the figure represent a sphere of the sliding steel ball employed for the tests. At very initial stase of sliding wear Irsl = 5 s). the scar was shallow and the As the sliding shape is similar to the geometry of the sliding steel ball (Figure 19 (a)). As the sliding wear tests proceeded, the stainless steel surface suffered from more severe plasuc deformation (Figures 19(b)-(e)). A remarked feature shown in Figure 19 (b) is that the scar is scooped out much deeper than that predicted from the sphere of the sliding steel ball. This stage corresponds to regime I, as marked by the high friction coefficient (u= 1. 24) shortly after the start of sliding wear (Tsl = 17 s). Severe surface roughening is observed at the boundary between regimes I and II as shown in Figure 19 (c) (@q1 = 30 s). Figure 19 (d) shows that the scar geometry approaches the sphere of the sliding steel ball. whilst the surface in the scar remains rough (tsl = 63 s). This stage corresponds to regime II. In regime III. the scar shape is almost the same as the sphere of the sliding steel ball and the scar surface becomes less roughened, as shown in Figure 19(e) (tsl = 500 s).

The development of scar geometry described above is also clearly shown in Figure 20. The graph shows scar widths plotted against their depths formed by the sliding wear tests (the arrows indicate the direction of time). Each data point shows the mean of eight independent 2D profiles measured by optical profilometry. The error bars represent the standard deviation in the data. The scar width increases with time of sliding wear whilst the scar depth does not follow in the same manner. Also shown is the relationship between scar width and depth formed by a sliding steel ball (the radius is 3.175 mm), calculated from the geometry depicted on the left hand of Figure 20. It is assumed that the sliding ball is cuning the material and all material displaced by the ball is removed so that the scar curvature is the same as that of the ball (the motion of sliding ball is normal to the paper) [33]. Even though the cutting mode is not relevant to the sliding wear mechanism for the present system4, the comparison between the data and the theoretical relationship may give rise to the following intriguing features. First, the scar depth increases far greater than that from the theoretical relationship in regime I, indicating the formation of grooves in the scar. Second, the error bar of me scar depth is the larges at the boundary between regimes I and II, implying that the most severe surface roughening occurred at this stage (as the error bars represent the standard deviation in the depth data). Third, in regime III the scar width increases with increasing the scar depth and its relationship is close to the theoretical one. surface by sliding wear. It is clear that the-oxygen peak at 0. 52 keV increases with time of sliding wear. particularly at 500 s. whilst other peaks from carbon, iron, and nickel are approximately unchanged. It is to be noted that no increase in oxygen peak was observed at 63 s in regime U. Note too, that a sliding test was interrupted during regime MI (after 500 s) simply by lifting up the pivoted weight-stand (Figure 2), and fragments of wear debris on the worn surface were wiped off swiftly by a cotton wool containing acetone. Then, the sliding test was carried on again. The experiment demonstrated that the amplitude of contact resistance was nearly unchanged after removal of the wear debris whilst the friction coefficient was slightly reduced (approximately 8 %). Therefore, we deduce that the increase in contact resistance in regime III (Figure 17) is due primarily to the oxide formation on the scar.

Figure 22 presents a trace of the friction coefficient from the stainless steel after cathodic treatment in 8 M NANO, with a cell voltage of-5. 0 V for 6 h at a solution temperature of 80'C (solid line). Regime I was removed from the friction coefficient trace by applying cathodic treatment. However, after subsequent thermal annealing at 550"C for 20 min, the regime I reappeared in the friction trace (broken line in Figure 22). The friction trace after the subsequent thermal annealing is similar to that in the absence of the electrochemical process (Figure 17). This change in friction coefficient is consistent with that observed in microhardness (3. 2) ; it may well be that the steel surface was modified by the penetration of hydrogen during the cathodic treatment, resulting in the lower friction force as well as the surface hardening. However, as hydrogen in the metal could be outgassed readily at higher temperature [30], the effects were removed after the thermal annealing.

Figure 23 presents a trace of the friction coefficient from the stainless steel after electrochemical annealing procedure. Also shown is the electrical contact resistance between the surface and the sliding steel ball measured simultaneously. The electrochemical pulse treatment was applied in the same manner as described above for a period of 3 h (Figure 5). Regimbe I at the beginning of friction test was truncated. showing the maximum friction coefficient of 0. 9. The contact resistance was also substantially high over the same period. However, at 30 s after the start of the wear test, the friction coefficient rose sharply and the contact resistance disappeared from the trace ; the contact resistance became lower than a few ohms so that it was invisible on the scale presented (0 to 3 kQ). One may also consider that this may be due to hydrogen penetration into the metal as described above induced by the cathodic part of the pulse treatment. It should be noted that no prominent rise in contact resistance was detected from the specimen after the cathodic treatment alone at the beginning of the friction test. Furthermore, the friction trace that after the cathodic treatment alone shows a gradual increase with time as a whole (Figure 22) without a sudden rise as seen in that after pulse treatment (Figure 23). Thus, the surface after the pulse treatment is apparently different from that after the cathodic treatment alone.

The specimen subjected to the electrochemical pulse treatment was then thermally annealed at 550 C for 20 min. As described above, the effect of hydrogen penetration in the metal, if any, must have been eliminated by the subsequent thermal annealing. The results of the friction coefficient and contact resistance measurements are shown in Figure 24. The friction coefficient was much lowered by applying the subsequent thermal annealing ; it stayed at 0.35 for 180 s and the contact resistance was substantially high over the same period. The friction coefficient then started to increase with intermittent spikes. The transition in friction force is associated with a disappearance in the contact resistance. The phenomena are basically the same as those observed in Figure 23. but the friction coefficient became lower and the period of low friction and high contact resistance was longer. It is intriguing to note that the datum of the scar width and depth formed on the pulsed surface at the low friction regime ( =60 s) fits the theoretical curve as shown in Figure 20 (white square).

3.6 Surface Roughness Similar traits such as low friction force and high contact resistance were also observed for rougher surfaces. Figure 25 shows traces of friction coefficient and contact resistance traces measured from the steel surface ground to a 1200 grit finish.

The sliding motion was normal to the grinding direction. During 40 s after the start of wear test, the friction coefficient retained low (from 0.25 to 0.38) and the contact resistance showed high. One may consider that enhancement of the tribological properties by the voltage pulse treatment is introduced simply by surface roughening ough the electrochemical process. To clarify this argument we investigated surface roughness of the treated specimens by optical profilometry.

Figure 26 shows changes in surface roughness of the steel surfaces after various surface treatments. The r. m. s. roughness (symbol Ru) defined as the root

mean square deviation of the profile from the mean line (Equation (2)) was employed :

, 1 iL R-=-y- (..) d. (2)

where v is the height of the surface above the mean line at a distance x from the origin, and L is the over all length of the profile under examination (as schematically c illustrated in Figure 26). In this analysis L was ca 100 lim. Each data presents the mean of five 2D profile measurements, with an error bar of the standard deviation in the data. As shown in Figure 25, the surface roughness increased slightly after the

electrochemical pulse treatment (Rq= 18. 3 nm, Figure 26 (b)) and the subsequent q c thermal annealing (Rq=21. 5 nm, Figure 26 (c)), from the original surface roughness polished to 0. 25 mfinish (Rq=14. 3 nm, Figure 26 (a)). However. the changes are very subtle and the surfaces are still very smooth compared with that finished to 1200 grit (=164 nm. Figure 26 (d) ). Therefore it can be concluded that the improvement of the tribolosical properties after the electrochemical treatment and the subsequent thermal annealing (Figures 23 and 24) is not caused by the surface roughness generated during the treatments.

4. Discussion

4. 1 Elecrrochemically Induced Annealing The experiments demonstrated that martensitic structure, induced by mechanical deformation on the surface of austenitic stainless steel, can be eliminated by electrochemical pulse treatment in aqueous solution at very low temperature. To our knowledge, this is the first time that such an annealing phase change procured by electrochemical means has been reported. Although the origins of this electrochemical annealing process are not clear at present, some aspects can be rationalised as follows.

First, we discuss the effect of hydrogen. In the cathodic part of the potential pulse, water is reduced by electrolysis to hydrogen. Hydrogen atoms entering the metal produce considerable strain within the lattice, as schematically illustrated in Figure 27 (a). In the anodic part of the potential pulse, hydrogen in the metal is reoxidised anodically and released into the electrolyte as described in 3. 4 (Figure 27 (b) ). The defects generated by ingress of hydrogen will leave vacancies after releasing hydrogen. As a consequence, metal atoms can move over very short range.

This kind of structural relaxation may give the observed annealing effect-The defects . ; at remain would then be available for further penetration of hydrogen atoms in the subsequent cathodic pulse. In this way a deeply affected layer could be induced. We estimate the depth of the annealed layer after 9 h treatment to be of the order of 8 urn.

Second, let us consider the effect of nitrogen. Even though we could not confirm the presence of nitrogen from the AES (Figure 13), the nitrite ion must be reduced to nitrogen as well. The role of nitrogen on the electronically induced annealing is, at first glance, likely to be the same as that ot hydrogen uescnbea above.

The possibility of the reduction to nitrogen cannot be discounted as an explanation for the observed phenomena although it might be expected that nitrite would reduce electrochemically to ammonia rather than to nitrogen dissolved in the metal [34).

Indeed. since the reduction of nitrite to ammonia involves transfer of six electrons. one might expect that nitrogen could be formed as an intermediate reduction step. since reduction to nitrogen involves only three electrons. It is to be noted that nitrogen is one of strong austerute-stabilising elements for Cr-Ni sinless steels [35]. Thus, even if the reduction to nitrogen occurs with only low current efficiency and a very small amount of nitrogen is dissolved in the metal surface, its influence to induce the a'-'y transformation cannot be ignored.

The present study demonstrated another interesting observation that the cathodic treatments in nitrite solutions lead to the lattice contraction whilst those in non-nitrite solutions give rise to the lattice expansion (Table 4). It is commonly recognised that hydrogen dissolved in a metal occupies interstitial sites in the host lattice, thereby resulting in the expansion of the crystal lattice of the host metal [36].

This can explain the present results ontained from non-nitrite solution treatments, but cannot rationalise those from nitrite solution treatments. It was reported recently that the contraction of lattice parameter occurred in metal hydrides such as PdH and NiH at high temperatures (700''C-800'C) and this contraction could be retained at ambient conditions and even after degassing hydrogen [37-40]. The lattice contraction was also observed for aluminium by electrochemical charging in sulphuric acid [41]. The phenomenon is accounted for by the formation of lattice vacancy at the surface accompanying the introduction of hydrogen [37-41]. The penetration of hydrogen into the metal can give rise to either the volume expansion or the contraction. Now. let us discuss the results obtained here. The element which could be reduced at the metal surface by the cathodic treatment in potassium hydroxide solution at the potentials applied. is confined to hydrogen. Thus. the penetration of hydrogen induces the lattice expansion for this system. In contrast. the cathodic treatments in nitrite solutions lead to the lattice contraction. Now a question arises as to the origin of the converse behaviour. Presumably, this is caused by nitrogen dissolved in the metal, simply because no other element can be involved in the process. It is supposed that the tension of nitrogen may induce the formation of vacancies. The changes in lattice parameter due to the cathodic treatments are-0. 050 % and-0. 083 in. for 8M and 20M NaNO2 as shown in Table 4. It should be noted that these changes (absolute values) are greater than that reponed for aluminium by cathodic treatment in H2SO4 solution (

0. 017 %) [41]. A more salient feature is that the lattice contraction due to the pulse treatment (-0. 213 %) in nitrite solution is far greater than those due to the cathodic treatments, even though the processing time of the pulse treatment is shorter. This suggests that repetitive motions of penetration and release of nitrogen enhance the C nitrogen enhance formation of vacancies greatly, leading to the electrochemically induced annealing process.

Lastly. the effect of Ni enrichment and Fe depletion is argued. One plausible idea is that if iron forms a soluble complex with ammonia and it leaves from the surface, nickel may well be relatively enriched in the steel surface5. It is widely known that nickel is the primary element to stabilise austenite structure of stainless steels [35]. The enrichment of nickel is therefore thought beneficial to induce the

electrochemical annealing process6. 4. 2 Enhancement of M ; crohardne, s, s It was shown that the electrochemical annealing process through the pulse treatment engendered surface hardening despite the loss of the martensite phase (3.2).

However, since manensitic phase is far harder than austenitic one. this observation is paradoxical. What kind of mechanisms are involved in the enhancement of surface hardness by the electrochemical pulse treatment ? Typical width of indentation scar of the microhardness tests (with a load of 50 g) is ca 18 gm so that the scar depth is estimated to be order of a few gm. Thus. it can be considered that the origin of the

surface hardening observed is not caused by changes in surface atoms alone. We have already seen that the hardening was not due to the penetration of hydrogen into the metal during cathodic pan of the pulse treatment alone.

We put forward the following explanation : during the potential pulse treatment. hydrogen and possibly nitrogen are dissolved in the metal in the anodic part and released into the electrolyte in the cathodic pan as described above. Repetitive actions of the penetration and release of these elements may generate defects or

vacancies in the metal matrix. Two consequences are plausible due to the engendered

We hnve not contTrned whether this notion is feasible thennodynamicaHy.

.. It should be noted that 420 martensitic stainless steels (14. 2% Cr. 0. 98% Ni and 0. 76% Si) showed no changes in manensite peaks and no new peaks from auslenile on X-ray diffraction patterns after the ekctrochemicat putse treannent.

defect ; one is the electrochemical annealing process as described above, and the other is the surface hardening. We envisage that the detects induced by the pulse c treatment impede dislocation motion along the slip plane. thereby leading to the observed surface hardening. It is to be noted that the depth of the affected layer after 3 h pulse treatment is estimated to be the same order of that of the indentation scar of the microhardness tests.

This notion is considered akin to the mechanism involved in the well-kmown process of neutron irradiation-hardening [42]. Some researchers have reported considerable increase in Vickers hardness due to irradiation for austenitic and terrific stainless steels [43-45]. When metals are irradiated with neutrons, various defects an : known to be formed in the matrixes as a result of collision between neutrons and atoms. It is commonly recognised that the irradiation-hardening stems from the interaction of these defects with moving dislocations [45. 46].

4. 3 mprot-ement of Friction First. we consider plausible origins of the Ici transitions in friction coefficient measured from the steel surface with no electrochemical treatments (Figure 17). Blau [31, 32] described that eight forms for friction coefficient versus time behaviour had been commonly observed in the literature. The Iw transitions observed here may well be categorised into one of these eight forms. He attempted to develop a semi-empirical model capable of generating these common forms of friction transitions. However, sliding wear is known to be complex phenomena and is affected. by a large number of parameters such as material properties, surfaceconditions. and operating environments. [8]. Mechanismus involved are likely to be different under different conditions. For example, severe wear at the initiation of sliding, followed by mild wear with prolonged sliding, has been observed in many cases [47-49]. Some mechanisms were suggested to account for this severe-mild transition. One explained that this was due to oxidation effect (8] whilst anor attributed this to a decrease in surface roughness [32]. Neither notion is, however. relevant to rationalise the 1-U transition for the present system (Figure 16). because no discernible rise in oxygen content was detected in regime n (3. 3) and the surface was severely roughened during the 1-D transition (3. 5). Therefore. it is worth describing possible explanations for the present system as follows.

Rtbimd The sharp rise in friction coefficient indicates that the air-formed oxide film on the stainless steel surface is removed easily immediately after the commencement of the sliding test. As a result, bare metal contacts the sliding steel bull and galling steins. tvletnl transfer occurs from the stainless steel to the sliding seul hall. thereby leading to the formation of the deep grooves on the stainless steel surface (Figures 18 and 19). The friction coefficient shows the highest because of the severe plastic deformation.

Regimbe 11 As a consequence of the surface roughening due to galling. contact area between the stainless steel surface and the sliding steel ball becomes smaller. The smaller contact area gives rise to lower friction force. Moreover. the scar surface is likely to be hardened due to the plastic deformation. Formation of a hard surface layer is known to reduce friction coefficient [48]. The lowest friction coefficient observed in regime n can be attributed to a combination of the two effects.

Regime III With the prolonged sliding, the friction coefficient increases gradually. Following four factors must be considered: (1) The scar size increases and the roughness decreases gradually with prolonged sliding (Figures 18 and 19). thereby generating larger contact area. (2) More fragments of debris are obtained on the worn surface with longer sliding (3. 5). (3) Oxide film grows on the scar due to frictional heat (3. 5). (4) The scar surface is kept to be work-hardened. The factors of (i) and (ii) increase the friction coefficient whereas those of (iii) and (iv) decrease it conversely. Although the quantitative partition of each effect on friction force is unclear, the friction coefficient increases gradually as a result of the sum of these

competing effects. c The friction coefficient is lowered by applying the electrochemical pulse treatment as shown in Figure 22. indicating that sailing is effectively reduced. We suggest two explanations for this observation ; one explanation attributes this to the increase in surface hardness induced by the electrochemical annealing process (3. 2).

It is known that increasing the hardness of surface may lead to a reduction of galling [8]. However, the hardness alone is a poor indicator of the wear resistance and other factors are often much more important [1].

The other is that the electrochemical pulse treatment produced galling resistant film on the surface. For the present study. the latter is thought to be more effective.

The film must be the oxide formed on the surface during anodic pan of the pulse treatment, leading to reduce adhesion at the contacting surfaces. The presence of the oxide film is also anticipated by the high contact resistance over the same period of lower friction coefficient. The oxide film was. however, ruptured at 30 s after the commencement of the sliding wear. and consequently galling started (Figure 22). We have seen that the subsequent thermal annealing is very beneficial to enhance sliding wear resistance of the stainless steel surface (Figure 24).

Now, some questions arise as follows : (i) Does the cathodic part of the pulse treatment confer any effects on the growth of the oxide film ? The metal surface just passivates in accordance with the classical behaviour of stainless steel at the anodic part of the pulse treatment, or more subtle mechanism is involved in the film growth through the pulse treatment? We have discussed the possibility of generating the defects in the metal surface by repetitive penetration and release of hydrogen and possibly nitrogen. Does the process also affect the oxide film growth ? For instance, is there any possibility that a thicker oxide film has been produced by the pulse treatment? (ii) As significant enrichment of Ni in the film is attained by the pulse treatment (3. 3), we expect that the composition of the oxide film generated by the pulse treatment is different from that by the anodic part alone. Nickel itself is known to have a poor galling resistance because of its high ductility. Then. nickel oxide formed by the electrochemical treatment is beneficial to reduce galling ? (ici) The subsequent thermal annealing is shown to be remarkably effective to reduce galling. The properties of the oxide film generated by the electrochemical treatment must be changed by the thermal annealing. Two possibilities are envisaged here. One is that the thermal annealing may decrease adhesion of the oxide film against the sliding steel ball significantly. The other is that the thermal annealing may make the surface film more adherent to the metal matrix so that the film is more resistive to denudation by a sliding object.

Unfortunately these questions cannot be answered at the present For instance, in order to answer the question (i) we need to study the treated surfaces by ellipsometry. Future work should be conducted to clarify them.

5. Conclusions The manensitic structure induced by mechanical deformation on 304L stainless steel surface can be removed by appropriate electrochemical pulse treatment in sodium nitrite solution. This new process observed at very low temperature is termed

elecnocheznicalls indced snnealing as the same effect can be achieved by thermal annealing at 750oC. The hardness of the surface increases after this electrochemically induced annealing process despite loss of the martensite. Sliding wear tests demonstrate that sailing is effectively reduced by the electrochemical processing. It is also shown that the subsequent thermal annealing after the electrochemical processing can much lower the friction coefficient. It is concluded that the new electrochemical treatment induces an improvement in the tribological properties of austenitic stainless steel surfaces, while eliminating the brittle martensite phase from the surface.

Acknowledgements

We are grateful to the EPSRC for financial supports under the ROPA scheme. r

References

1. I. M. Hutchings, Tribology : F/-/c ?/ ? / ? / Wfor c/'Mfn/ ! Mfna.

Edward Arnold. London, 207 (1992).

2. W. G. Wood, in Hannbook of Srai'iless Steels (Eds. D. Pecker, I. M.

Bernstein), McGraw-Hill Book Company, New York. 34-2 (1977).

3. J. R. Davis, in ASM Handbook vol. J, Surface Engineering, ASM International, 741 (1994).

4. S. Rundenja, J. Pan, 1. Odnevall Wallinder, C. Leygraf. P. Kulu, J.

Elect-ocher Soc., 146, 4081 (1999), 5. P. Millet, T. Hang, J. Mat. Sci.. 31. 351 (1996).

6. X. Li. M. Samandi. D. Donnes G. Collines, J. Tendus K. Short, R.

Hutchings, r. Coat. Tech.. 85. 28 (1996).

7. J. Edward. Coating and Surface Treatment Systems for Mewls, ASM International. Materials Park, Ohio, 78 (1997).

8. M. B. Peterson, L. K. Ives. K. J. Bhansali, in Metal Transfer and Galling in /vlnalic Systems (Eds. H. D. Merchant. K. J. Bhansali). TMS of AIIE Warrendale PA. 1 (1987).

9. T. Angel. J. Iron Steellnsr.. 177. 165 (1954).

10. D. C. Lugwi2son. J. A. Berger. J. IW ! fSreel ! st.. 207. 63 (1969).

11. G. B. Osons M. Cohen. Mer. Iisms 6A. 791 (1975).

12. G. B. Oison, M. Rohen et. r/vw.. 13A. 1907 (1982).

13. G. L. Hang, D. K. Mattock, G. Krauss, Met. Trans.. 20A, 1239 (1989).

14. K. Tomimura, S. Takaki, T. Tokunaga./SIJ Intern ., 31. 1431 (1991).

15. S. Takaki. 11. Tomimura, S. LTeda. IS/J Imernar.. 34, 522 (1991).

16. S. F. Petersen, M. C. Malaya, D. K. Matlack. J. A-/r.. 49. 54 (1997).

17. A. P. Bentley, G. C. Smith. Mer Trans.. 17A, 1593 (1986).

18. S. Tdhtinen. J. Kivilahti. H. Hanninen. -m f.. 19. 967 (1985).

19. K. Okada, Y. Hosoi. S. Abe, Corrosion, 26. 183 (1970).

20. N. Ohtani, S. Asano, Y. Fujishima. Y. Yamamasu, J. Japan Insu. Met., 37.

746 (1973).

21. M. Pourbaix. Arias of Electrochemical Equilibria in Aqueous solutions, Pergamon/CEBELCOR. Oxford, 256 (1966).

22. H. Sumitomo, J. Iron Steel Inst. Japan, 77, 558 (1991).

23. K. Sasaki. G. T. Burstein, I. M. Hutchings. bffd M fre, (1999).

..

24. D. T. Llewellyn. J. D. Murry.Iron Steel Inst. London Rept. No. 86, 197 (1964).

2 A. J. Sedriks, Corrosion of Stainless Steels, The Electrochem. Soc. Inc., Princeton, New Jersey, 70 (1979).

26. C. L. Hedberg (Ed), Handbook of Auger Electron Spectroscopy, Physical Electronics, Inc. , Eden Prairie, Minnesota. 32 (1995).

27. A. P. Bentley, G. C. Smith, Met. Trans.. 17A, 1593 (1986).

28. S. T htinen, J. Kivilahti, H. Hannien, Scripta Met., 19,967 (1985).

29. H. Okada. Y. Hosoi, S. Abe, Corrosion, 26, 183 (1970).

30. N. Ohtani, S. Asano, Y. Fujishima, Y. Yamamasu, J. Japan Inst. Met., 37.

746 (1973).

31. P. J. Blau. Wear, 71, 29 (1981).

32. P. J. Blau. J. Tribology, 109. 537 (1987).

33. K. Hokkirigawa. K. Kato, Tribology International. 21. 51 (1988).

34. R. C. Newman, G. T. Burstein, l. Elecrrochem. Soc., 127, 2527 (1980).

35. R. W. K. Honeycomb. H. K. D. H. Bhadeshia, Steels : Microstrlcuare and Properties. London, Edward Arnold. 254 (1995) 36. H. Peisl. Hydrogen in Metals I, Topics in Applied Physics, Vol. 28 (Eds. G.

Alefeld and J. Vlkl), Springer Verlag, Berlin, 53 (1978).

37. Y. Fukai. N. Okuma. 7. J. Appl. Phy., 32. L1256 (1993).

38. Y. Fukui. N. Okuma. Phys. Re\'. Lerr., 73. 1640 (1994).

39. K. Nakamura. Y. Fusai J. // Co/ ? ! .. 231, 46 (1995).

40. Y. Fukai, J. Alloys Comp.. 231, 35 (1995).

41. H. K. Birnbaum, C. Buckle. F. Zeides, E. Sirois, P. Rozenak. S. Sooner.

J. S. Lin,, J. Allo.'I,"s Comp., 253-254. 260 (1995).

42. V. V. Penkovskii, Effect of Ratliarion on Metals, Elsevier Publishing Company, Amsterdam, (1964).

43. H. R. Higgy, F. H. Hammad, J. Ncl. fr.. 55, 177 (1975).

44. R. L. Klueh, J. Nitcl. Marer., 179-181, 728 (1991).

45. G. D. Johnson, F. A. Garer, H. R. Bracer, R. L. Fish, in Effect of Radiation on Mrttfnas. Proceedings of the 10th Inreranonal Sympotm (Eds. D. Kramer.

H. R. Bracer, J. S. Perrin), ASTM. 393 (1980).

46. A. L. Bernent, Jr.. in Procee'. 2/i Infernational Conference on the Strength of Metals and A/loys, ASM. 693 (1970).

47. N. C. Welsh, J. Appl. Phys., 28. 960 (1957).

48. F. P. Bowden, D. Tabor, The Friction and Lubrication of Solid, Part II, Oxford University Press, London. 356 (1964).

49. M. Antler, Wear, 7, 181 (1964).

50. H. P. Klug, L. E. Alexander, X-ray Diffraction Procedures, John Wily & Sons. Inc. , New York. 92 (1954).

51. KaleidaGraph Reference Guide, SYNERGY SOFTWARE. Reading, PA. 172.

APPENDIX 1 Algorithm to calculate the friction coefficient The output from the transducer for the detection of friction force was fed into a home-built amplifier operating at a gain of 5 (Figure 2). The signal from the amplifier was acquired at a sampling rate of 100 Hz by an a-d data acquisition boards installed on a HP computer, and then the program calculated the running mean of the amplified voltage for 0.20 s (20 data points) to improve the signal-to-noise ratio. Figure A. I shows a 30 s section of the running mean voltage. Also shown is the offset of the transducer, which was the running mean for 20 s. Since the offset is not zero. it must be subtracted from the signal. After the subtraction, the program calculated the absolute value and then convened it to the friction force F. As the normal force was constant (2. 15 N) throughout the experiments, the friction coefficient can be calculated readily (u = FIN). Finally. the friction coefficient was filtered to exclude the data near zero using KaleidaGraph Filter Macros (% Error = 30) [51] as illustrated in Figure A. 2. The flow chart of the algorithm is presented in Figure A. 3.

Claims (6)

1. A method for treating a metal comprising subjecting the metal to electrolysis in the presence of an electrolyte using alternating pulses of voltage and/or current, said alternating pulses being of opposite polarity.

2. A method according to claim 1, wherein the metal is austenitic stainless steel.

3. A method according to claim 1 or claim 2, wherein the electrolyte is an aqueous electrolyte.

4. A method according to claim 3, wherein the electrolyte is aqueous sodium nitrite.

5. A method according to claim 1 or claim 2, wherein the electrolyte is a non-aqueous electrolyte.

6. 35mm) 6 Specimen 7 Sliding stage 8 Motor 9 Current source Figure 8 A Polishing to 0. 25 m B Electrochemical pulse treatment C Thermal annealing at 750C for 20 min Figure 9 A Cold-rolled

B 2500 grit finish C 0. 25 um polishing

6. A method according to any preceding claim, which further comprises, after electrolysis, heat treatment of the metal.

7. A metal which has been subjected to a method as defined in any preceding claim.

8. Use of a method as defined in any preceding claim, to improve the resistance of austenitic stainless steel to mechanical degradation.

9. Use of a method as defined in any preceding claim, to remove or transform martensite from austenitic stainless steel. Keys to Figures Figure 1 1 DVM 2 Buffer amplifier 3 Bipolar power supply 4 Function generator 5 Saturated calomel reference electrode 6 Saturated KCI 7 8M NaNO2 8 Temperature controller 9 Thermistor 9 Temperature controller 10 Graphite electrode 11 Condenser 12 Luggin probe 13 Stainless steel specimen 14 Tape heater E Electrode potential V Cell voltage Figure 2 1 Amplifier 2 Transducer 3 Weight-stand 4 Insulator 5 Steel ball ( :