Pulse current plasma assisted electrolytic cleaning of AISI 4340 steel

Pulse current plasma assisted electrolytic cleaning of AISI 4340 steel

Journal of Materials Processing Technology 210 (2010) 54–63 Contents lists available at ScienceDirect Journal of Materials Processing Technology jou...

2MB Sizes 1 Downloads 28 Views

Journal of Materials Processing Technology 210 (2010) 54–63

Contents lists available at ScienceDirect

Journal of Materials Processing Technology journal homepage: www.elsevier.com/locate/jmatprotec

Pulse current plasma assisted electrolytic cleaning of AISI 4340 steel A. Yerokhin ∗ , A. Pilkington, A. Matthews Department of Engineering Materials, University of Sheffield, Sir Robert Hadfield Building, Mappin Street, Sheffield S1 3JD, United Kingdom

a r t i c l e

i n f o

Article history: Received 16 April 2009 Received in revised form 12 August 2009 Accepted 14 August 2009

Keywords: Plasma electrolysis Cleaning Pulse current Hardness Surface roughness Residual stress Fatigue Electrochemical processes

a b s t r a c t An electrolytic plasma process (EPP) for cleaning AISI 4340 steel was performed in a 10% solution of sodium bicarbonate operated at 70 ◦ C. The effects of the pulse frequency (f) and duty cycle (ı) on the surface morphology, microstructure, mechanical and corrosion properties were investigated. Compared to the conventional DC process, the pulsed EPP cleaning resulted in reduced surface roughness and compressive residual stress at the surface. Minimal reduction in hardness and no reduction in toughness due to hydrogen embrittlement (ASTM F519) were found. At the same time, rotating bending beam fatigue tests indicated a noticeable reduction in fatigue life, which could be offset by a shot peening treatment prior to EPP cleaning at 10 kHz and ı = 0.8. Glow discharge optical emission spectroscopy indicated minimal changes in the surface composition and potentiodynamic corrosion studies revealed a slight ennoblement of the surface attributable to an increased rate of cathodic processes. Optimal process parameters were identified for ı = 0.8 and f = 100–10,000 Hz. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Surface cleaning is an important step in most manufacturing processes. It can be present as a finishing operation, e.g. for ultrahigh vacuum applications (Herbert et al., 1994), but often as a coating pre-treatment to enhance the adhesion and the overall performance. The cleaning pre-treatment is essential in the manufacture of steel components that are prone to formation of mill-scales, pigmented films and rust spots during prior processing and in-process storage. This temporary storage is often accompanied with application of protective barrier films, oil coating or greasing that also introduce surface contamination by organic substances. Well-known treatments used for de-scaling and de-rusting purposes include mechanical techniques, such as abrasive blasting and tumbling, and chemical reactions using molten salt baths, acid etching or pickling. A further enhancement is achieved through electrochemical means by electrolytic acid pickling or alkaline cleaning. In contrast, removal of organic contaminants relies on solvent and emulsion cleaning (ASM Handbook, 1994). In practice however, none of the single-step cleaning processes are proven to be completely effective and various multistage procedures are used.

∗ Corresponding author. Tel.: +44 1142225510; fax: +44 1142225943. E-mail address: [email protected]field.ac.uk (A. Yerokhin). 0924-0136/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2009.08.018

To increase the efficiency of individual cleaning methods, various physical phenomena, e.g. ultrasonic cavitations (Noltingk and Neppiras, 1950; Crawford, 1963) and high-energy fluxes of activated species, e.g. laser beams (Tam et al., 1998) or plasma jets (Tendero et al., 2006; Bardos and Barankova, 2008) are considered to synergistically compliment the main cleaning process. For example, a relatively new electrolytic plasma (EP) cleaning technology relies upon intensification of conventional electrolytic alkaline cleaning by a plasma discharge that is initiated at the surface of a cathodically polarised workpiece due to abundant gas liberation and electrolyte evaporation at high voltages (Yerokhin et al., 1999). Extreme temperature and pressure variations are created at the surface by rapid plasma generation and extinction, which leads to instantaneous melting and quenching together with cavitation effects. These processes enhance the main cathodic reduction mechanism with thermally and mechanically activated removal of mill-scales and organic contaminants from the steel surface (Meletis et al., 2002). The EPP treated surface has an increased corrosion resistance and a micro-textured surface profile that is favourable to promote adhesion of subsequent coatings. The EPP cleaning technology has features essential for high throughput in-line production and provides effective high-speed single-stage cleaning suitable for heavily contaminated surfaces. However, reduction of mechanical properties due to uncontrolled thermal impact from plasma becomes of significant concern, especially for thermally hardened steels. The solution of this problem represents the main challenge in further developing EPP cleaning technology.

A. Yerokhin et al. / Journal of Materials Processing Technology 210 (2010) 54–63

55

2. Analysis of existing work

2.2. Effects of electrolytic plasmas on materials

2.1. Fundamental aspects of electrolytic plasma processing

Complete separation of the metal electrode from the electrolyte solution in the CGDE regime dramatically affects heat exchange conditions at the electrode surface, resulting in heating of the electrode by the passage of electric current. Two associated effects are commonly observed: formation of a crater-like surface topology and thermal degradation of the electrode material. Gupta et al. (2007) discussed topology formation mechanisms comprising instantaneous local melting, eruption and quenching cycles at the metal surface, accompanied by acoustic shock waves and cavitation effects at the plasma–electrolyte interface. The resultant micro-textured surface profile is suitable to promote adhesion of subsequent coatings through anchoring sites in the cavities. However an increased surface roughness has negative effects on both tribological and fatigue performance of some treated materials. This important issue has been disregarded by previous researchers and no effective means of roughness control is currently available. Yet more severe deterioration of mechanical properties can occur due to annealing of thermally strengthened steels exposed to prolonged treatments under CGDE conditions. Kinetic aspects of metal heating by electrolytic plasmas have been comprehensively studied, with key results summarised in monographs by Duradzhi and Parsadanyan (1988) and by Belkin (2005). It was shown that within the initial seconds, the electrode surface temperature increases at a rate of 150–200 ◦ /s and then stabilises at a certain point which is dependent on the applied voltage, electrolyte temperature and flow rate. The temperature distribution across the metal electrode is consistent with conditions of high-rate surface heating, with a slow increase in the depth of the heat affected layer evolving with treatment time. Normally, the steady-state surface temperature decreases with both increased electrolyte temperature and flow rate, whereas the voltage dependence exhibits a maximum at the midpoint. In the Kellog region, the heating effect is promoted by a reduction in the vapor sheath fluctuations with voltage; whilst in the CGDE regime, it is inhibited by electrolyte impingement due to acoustic shock waves and cavitations. Under similar process conditions, the maximum achievable electrode temperature (Tc) is known to be higher, its voltage dependence (dT/dU) to be much stronger and the midpoint voltage to be lower in the cathodic configuration compared to the anodic one. For example, when a 6 mm diameter steel rod was treated cathodically in a 30% Na2 CO3 solution, these characteristics corresponded to Tc ≈ 1200 ◦ C, dT/dU ≈ 20 ◦ /V and Uc ≈ 140 V vs. Tc ≈ 700 ◦ C, dT/dU ≈ 4 ◦ /V and Uc ≈ 200 V for similar anodic treatment (Duradzhi and Parsadanyan, 1988). Effective process control and prevention of thermal degradation of the electrode material are therefore crucial for cathodic EPP cleaning of thermally hardened steels. None of the above process parameters influencing the metal temperature are adequately controllable for these purposes. The literature does not provide reference to alternative effective approaches to enhance the controllability of cathodic EPP cleaning treatments, thus limiting practical exploitation of this process.

Electrolytic plasma cleaning belongs to the group of highvoltage electrochemical processes wherein, at a certain current density, conventional electrolysis is interrupted by the formation of a gaseous phase completely separating the working electrode surface from the electrolyte. Initial observations of such behaviour were made over a century ago during electrolytic reduction of Al from molten salts and termed as the ‘anode effect’. However it was only in 1997 that a satisfactory theoretical explanation of the underlying mechanisms was provided by Vogt (1997), who considered nucleation, growth and detachment of gas bubbles from the electrode surface due to both surface tension and buoyancy forces. Vogt’s theory posits that the electrode wettability by the electrolyte and fluid dynamics within the bubble layer are the major factors influencing the critical current density. These considerations provide explanation for the experimentally observed effects due to electrolyte concentration and flow, electrode and gas bubble sizes, pressure and temperature. Kellog (1950) observed phenomena similar to the anodic effect during electrolysis of aqueous solutions, in both anodic and cathodic configurations. Unlike molten salts, vaporisation of the aqueous electrolyte from the hot electrode surface was suggested to be of significance in this case. Hikling and Ingram (1964) showed a presence of two voltage regions with differential behaviour of the vapor sheath. At lower voltages (Kellog’s region), the sheath oscillates with an amplitude similar to the mean average sheath thickness, which is independent of voltage. Parfenov et al. (2005) used Fourier spectral analysis to estimate that the fundamental frequency of these oscillations lays in the bandwidth of 500–1500 Hz. The mean average current falls slightly with voltage although both these instantaneous characteristics largely fluctuate, causing intermittent optical emission. Around a certain midpoint voltage (Uc), the sheath is stabilised and a steady-state glow discharge is established, with minor fluctuations of instantaneous voltage and current around their mean values. Both sheath thickness and current then increase with voltage, indicating a change in the conductivity mechanisms in the high-voltage region. This regime was termed ‘Contact Glow Discharge Electrolysis’ (CGDE) to differentiate it from the Glow Discharge Electrolysis (GDE), where one of the metal electrodes is located above the electrolyte (Davies and Hickling, 1952). Subsequent studies by Sengupta and Singh (1991, 1994), Sengupta et al. (1997), showed that transition from Kellog’s region to the CGDE occurs at the midpoint voltage as a result of the onset of hydrodynamic instabilities in solvent vaporisation due to Joule heating, rather than electrolytic gas evolution at the electrode surface. This is equivalent to the burnout point in boiling fluid heat transfer, indicating a transition from bubble to film boiling conditions (Rohsenow et al., 1998). The CGDE regime exhibits abnormal evolution of gaseous products (e.g. H2 and O2 ) which substantially exceed Faraday yields. Plasma chemical reactions involving accelerated particles and free radicals occur, with formation of hydrogen peroxide in the electrolyte. The nature of these interactions appears to be similar for GDE and CGDE modes, including either electrolyte decomposition due to bombardment by electrons emitted from a metallic cathode or sputtering by cations present in the plasma, in the case of a metallic anode (Sengupta and Singh, 1994; Cserfalvi and Mezei, 1996). Importantly, the excessive gaseous products not only stabilise the sheath but also provide a relative tolerance to fluctuations in the electrical parameters. This is manifested in hysteresis of the voltage range observed during CGDE for oscillating voltage waveforms (Slovetskii et al., 1986).

2.3. Present approaches to enhance controllability of EPP treatments The technology of plasma electrolysis was reviewed by Yerokhin et al. (1999), where pulsed and pulsed-reversed current regimes were identified to provide more effective control over various electrolytic plasma discharges used in surface treatments and coatings. Khan et al. (2005, 2008) studied oxidation treatments of Al and Ti using a pulsed current regime. It was demonstrated that increased current pulse frequency (f) and decreased duty cycle (ı) suppressed formation of high-temperature oxide phases and promoted the formation of compressive internal stresses in the surface layer. The

56

A. Yerokhin et al. / Journal of Materials Processing Technology 210 (2010) 54–63

application of current pulses in the 1–10 kHz range was also shown to reduce surface roughness and power consumption of electrolytic plasma treatments (Yerokhin et al., 2005). These works provide strong evidence that the pulsed current regime is effective in mitigating thermal impacts on the surface, thus controlling adverse effects on the mechanical properties of the metal substrate. Similar approaches are currently being explored in diffusion treatments (Aliofkhazraei and Roohaghdam, 2007), indicating a feasibility of pulse current modes under conditions similar to EPP cleaning. However there is little data available on the effects of pulsed current plasmas on surface properties of thermally hardened steels. This makes the research and development of pulse current EPP cleaning processes both important and timely. 3. Objectives of the work Based the above analyses the major objectives of the present study were identified as follows: (i) Study effects of the key process parameters voltage, pulse frequency and duty cycle on the surface topology, roughness, chemical composition and stress state of cleaned steel samples.

(ii) Develop a process for pulse current EPP cleaning of high strength AISI 4340 steel test pieces of different geometry. (iii) Evaluate effects of the cleaning treatments on characteristics of mechanical and corrosion performance of the cleaned steel surfaces. (iv) Compare performance of the new pulse current EPP cleaning treatment with that of the benchmark DC process. 4. Experimental 4.1. Pulse current EPP treatment facility The cleaning experiments were carried out on a specially designed and instrumented EPP rig, shown schematically in Fig. 1. The scheme employed allowed for an adjustable inter-electrode gap, controlled sample rotation and reciprocating movement as well as controlled electrolyte flow in the range 2–10 l/min. A separate heater allowed the electrolyte temperature to be adjusted between ambient and 80 ◦ C. The electrical bias was provided by a high-voltage DC power supply rated at 20 kW. The output DC voltage was pulsed by a high-speed solid state switching unit, which allows independent control of frequency and duty cycle from

Fig. 1. Schematic diagram of experimental EPP apparatus for the treatment of: (a) rods; (b) discs; (c) flat surfaces.

A. Yerokhin et al. / Journal of Materials Processing Technology 210 (2010) 54–63

57

Table 1 Matrix of experimental design and parameters of pulse current EPP cleaning process. Sample no. 1 2 3 4 5 6 7 8 9 10

f (Hz)

ı

U (V)

DC 100 1000 10000 100 1000 10000 100 1000 10000

1 0.8

180 180 180 205 180 240 340 220 350 450

0.5

0.2

t (s) 30 45

60

150

4.3. Characterisation and testing

Fig. 2. Typical appearance of plasma discharge during EPP treatment in the arrangement corresponding to Fig. 1c. The optical emission from the discharge is characteristic of electrolyte composition (e.g. Zn2+ in this illustration).

DC to 32 kHz. Voltage and current signals were captured using a data acquisition system based on a Tektronix TDS 430A digital oscilloscope. The voltage was sampled through a Tektronix P5200 high-voltage probe and current signals were monitored using a Tektronix A6303 current probe connected to a TM502A current probe amplifier. The mechanical system was designed to rotate the test piece and traverse the nozzle and is electrically isolated from the plasma power supply. This allowed DC pulsed unipolar or bipolar treatments to be performed. The electrical circuit is made between the test piece (cathode) and the nozzle, which formed the anode in these arrangements. The nozzle assembly was manufactured from graphite and dispensed electrolyte through 3 overlapped rows of 1 mm diameter holes with 2 mm pitch, creating a curtain flow. Fig. 1a shows the arrangement for the treatment of disc specimens, in which samples were mounted in a PTFE holder with an electrical contact made onto the back of the test piece. A set of carbon brushes allowed the sample to be connected to the power supply cathode. The controlled flow of electrolyte impinges normal to the sample rotation axis through the stationary nozzle. Rod or cylindrical components are treated in the horizontal axis shown in Fig. 1b, in which the vertical nozzle is traversed along the workpiece separated by a predetermined electrode gap. Fig. 1c shows the arrangement for treating flat coupons where the nozzle is again stationary and the coupon is traversed. This nozzle arrangement is shown with the plasma discharge in Fig. 2. The electrolyte was supplied to the nozzle through a variable flow valve by a chemically resistant vane pump and recirculated from the sump tank.

The surface roughness (Ra) was measured using a Dektak3 ST surface profilometer fitted with a 12.5 ␮m radius stylus. A further SEM analysis was carried out using a Camscan electron microscope, to observe the surface morphology of the treated surface. Glow discharge optical emission spectroscopy (GDOES) was used to obtain elemental profiles from the surface to the bulk composition in a Horiba GD-ProfilerT HR instrument. Also, conventional –2 XRD scans in 2 range of 10◦ to 120◦ and sin2 studies (Cullity and Stock, 2001) of Fe (3 1 0) peak at 116.3◦ 2 were carried out using a Siemens D5000 X-ray diffractometer (Cuk␣ radiation) to evaluate the phase structure and the stress state in the surface layer. The surface Vickers hardness was evaluated using a Mitutoyo MVK G1 microhardness tester under loads of 25–200 g; five parallel measurements were made followed by statistical analysis to derive mean average values and to evaluate the variance. Fatigue tests were performed according to ASTM E466 using the rotating bending beam technique. Some fatigue test pieces were shot peened using hard steel shot with 200% coverage, before the EPP cleaning process. Shot peening of high strength steels is widely used industrially and is well known to mitigate reductions in component fatigue life resulting from electrolytic deposition processes (Boyer, 1986; Nascimento et al., 2001). Process induced hydrogen embrittlement was investigated according to ASTM F519 using type 1C round notched specimens under a sustained static load. To evaluate electrochemical behaviour of the treated surfaces, potentiodynamic corrosion tests were carried out in a 3.5% NaCl solution, in deaerated conditions, using a Solartron 1286 potentiostat–galvanostat. The scans were performed at a rate of 1.667 mV s−1 in the potential range from −0.7 to 0.3 V vs. SCE; a Pt plate served as a counter electrode and N2 gas was purged through the solution. 5. Results and discussion

4.2. Materials and EPP cleaning procedure 5.1. Process U–I characteristics A full factorial 23 experimental design was employed for the cleaning process on the AISI 4340 steel discs of the following nominal composition (wt%): 0.4 C, 1.8 Ni, 0.8 Cr, 0.25 Mo, balance Fe. The surface condition was ground with a surface finish of Ra = 0.2 ␮m. The test pieces for general characterisation and corrosion testing had a diameter of 30 mm and a thickness of 4 mm, whereas the samples for hydrogen embrittlement and fatigue tests were as per ASTM F519 (type 1C) and ASTM E466-96 (type 2), respectively. No special pre-treatment was given to the samples prior to EPP cleaning. Parameters of EPP cleaning are given in Table 1; the treatments were performed in a 10 wt% solution of sodium bicarbonate (technical grade 99.5%) operated at 70 ◦ C, with a working voltage of 100–350 V and treatment times of 0.5–5 min. The electrolyte flow rate and the inter-electrode gap were fixed at 3 ± 0.5 l/min and 6 ± 1 mm, respectively.

Fig. 3 represents a typical example of voltage–current characteristics in the systems under investigation. The electrolytic process progresses through several stages, including conventional electrolysis (0–Ua), followed by the transient region (Ua–Ub) and the plasma discharge stage which commences at the potential Ub in the range 180–220 V, depending on the inter-electrode gap, electrolyte temperature and flow rate. The corresponding midpoint voltage separating Kellog’s region from the CGDE regime is situated within the 230–270 V range. The phenomena observed are consistent with earlier observations by Hikling and Ingram (1964), and characteristic voltages—with the values reported by Meletis et al. (2002), for DC EPP treatments in a similar electrolyte system. Voltage and current evolution is illustrated in Fig. 4a–c and d–f for the DC and the pulsed current conditions, respectively.

58

A. Yerokhin et al. / Journal of Materials Processing Technology 210 (2010) 54–63

Fig. 3. Example of current–voltage diagram for EPP cleaning treatment of AISI 4340 steel in DC mode.

Fig. 4a shows current and voltage transients typical of conventional electrolytic cleaning corresponding to the region 0–Ua in Fig. 3. Under these conditions, the current is stable indefinitely providing that the electrolyte flow remains constant. In Fig. 4b, the instability corresponding to the region Ua–Ub in Fig. 3 is shown over a 20 ms timeframe. The current periodically falls down to zero over a 3 ms period, corresponding to the breakdown of the conventional electrolysis. During this period the voltage increases and passes through a maximum of 240 V which causes the plasma discharge to ignite. A linear rise in current at 2.5 A/ms is observed with falling voltage, indicating an increase in conductivity of the plasma as temperature and ion mobility increase. This process then becomes effectively an arc discharge and the rate of current change increases rapidly. This corresponds to the point at which the voltage drops from 100 V to the order of a few volts, as the plasma discharge is extinguished and electrolysis then continues the passage of current to a maximum as the electrode gap is cooled by fresh electrolyte. The voltage then starts to increase as the current falls sharply. With increasing voltage to 175–300 V (region Ub–Uc in Fig. 3), the process again becomes stable (Fig. 4c) and the plasma discharge remains controlled, however close examination reveals increased fluctuations in voltage and current transients as the voltage passes through the midpoint Uc ∼ 220–230 V. The waveform for pulsed DC at f = 1 kHz is illustrated in Fig. 4d, which shows sharp-edged voltage pulses of 225 V magnitude and corresponding current pulses varying between 10 and 20 A. Under these conditions, the plasma discharge is stable and the current fluctuates but does not extinguish during the voltage pulse and overshoots during the voltage off time. The 1 kHz pulses are shown in Fig. 4f with a 2 ms timeframe, where the current pulse can be seen to rise rapidly to 8 A and after 0.2 ms increases to 11 A. The increasing current during the pulse is attributable to heating of the electrolyte causing increased mobility of electrolyte ions and ionised gaseous species. The pulse duration is too short to allow the current to progress to a runaway condition and is commutated by the voltage zero. The optical emission from the plasma discharge is characteristic of the excited ionic species present in the electrolyte; in sodium carbonate solutions this varies from dull orange to a bright yellow glow, becoming more intense with increasing discharge voltage for both DC and pulsed conditions (Fig. 4b and c). 5.2. Surface morphology Fig. 5 shows effects of pulsed current parameters on roughness of the EPP treated samples. The Ra values for the treated surfaces ranged from 0.04 to 0.125 ␮m. It was observed that the EPP cleaning treatment leads to a decrease in the roughness value with all the

treated samples, including the DC treated sample (Ra = 0.15 ␮m), having lower Ra values compared to that of the untreated steel (Ra = 0.2 ␮m). The data shows that an increase in frequency and a decrease in duty cycle led to a lower surface roughness. These trends can be associated with changes in the discharge characteristics when pulsed current is applied, except for the range of f = 1–10 kHz and ı = 0.5–0.2 where the discharge became unstable. These parameters caused oxide scale formation which appears to provide a significant contribution to the smoothing effect but the quality of cleaning is compromised. The SEM micrographs in Fig. 6 give an indication of the effect of the EPP treatment parameters on the surface morphology as compared to that of the untreated substrate. For the original surface, ridges were observed characteristic of the grinding finishing process on the steel. Samples 1–5 (f = 100–1000 Hz at ı = 0.2–0.5 and f = 10000 Hz at ı = 0.8) and 8 (f = 1000 Hz at ı = 0.2) all show typical EPP treated surface morphologies with the presence of crater features due to the discharge events occurring during the cleaning process. It was observed that the surface features reduced in size with an increase in frequency and a decrease in duty cycle. The surface of sample 10 (f = 10 kHz; ı = 0.2), for which the plasma discharge condition was unstable, had an obvious blackened surface associated with thermally induced surface oxidation processes. The surface consists of smooth overlapping craters possessing a suitable profile for subsequent coating treatments. Pulse current EPP creates a high-temperature plasma discharge having a high density of small hydrogen bubbles in the thin electrolyte layer on the surface of the workpiece. The discharge causes some surface melting due to the local high power density, and shock waves due to collapsing bubbles impacting on small molten regions of steel. The plasma activated hydrogen is able to reduce iron oxides present at the surface and results in a clean surface (Meletis et al., 2002). The sample treated at f = 10 kHz and ı = 0.8 showed the best surface profile with no direct evidence of fusion and solidification processes having occurred, indicating that the thermal impact from the treatment was minimal. 5.3. Crystalline structure and residual stresses The phase structures of the untreated and EPP cleaned steel surfaces are compared in Fig. 7. The analysis shows that after EPP cleaning the crystal structure remains essentially similar to the untreated sample, consisting predominantly of BCC ␣-Fe; however the peaks become higher, indicating removal of surface oxides and contaminants. Fig. 8 shows the variation of the residual stress with process parameters. The untreated steel had a compressive residual stress of −1440 MPa. In comparison, the DC treated sample had a tensile stress of 627 MPa. The residual stress calculated for the treated samples varied from −1690 MPa for the high frequencies to 778 MPa for low treatment frequencies. It was deduced from the data that increasing ı and decreasing f leads to more positive residual stress values, with the stress progressing from compressive to a tensile state. At low frequency (100 Hz) there is the greatest effect on the residual stress, with all treated surfaces being in a tensile state similarly to that of the DC treated sample. For the combination of the highest studied frequency (10,000 Hz) and duty cycle (ı = 0.5), the internal stress values are closest to that of the untreated steel substrate. The nature and the value of the resultant stress depend on the balance between expansion and contraction components (Khan et al., 2005, 2008) and it is likely that in the case of EPP treatments, the surface stress state is affected by temperature gradients in local discharges. The magnitude of the stresses induced depends upon the spatial density and characteristics of individual discharge events (Yerokhin et al., 2003, 2004) which in turn depend on the treatment parameters. For example, intense discharges may appear during longer current

A. Yerokhin et al. / Journal of Materials Processing Technology 210 (2010) 54–63

59

Fig. 4. Oscillograms of current and voltage signals during different stages of EPP treatment: (a) DC, 0–Ua, (b) DC, transient Ub–Uc, (c) DC, Ub, (d) DC Uc, (e) pulsed DC 1 kHz Uc, and (f) Uc pulsed DC 1 kHz, 2 ms frame.

pulses characteristic for high ı (0.5 and 0.8) and low f. These result in high discharge temperatures and hence in steep temperature gradients at the surface. The strongest effect of thermal stresses should therefore be observed at lower f and higher ı which is consistent with the experimental results obtained. Conversely, the treatments at higher frequencies retain compressive residual stresses characteristic of untreated surface, which is advantageous for subsequent coating deposition processes to be carried out. 5.4. Surface chemical composition The elemental depth profiles of Fe and O obtained from GDOES analyses on the untreated and EPP treated surfaces are presented in Fig. 9. The minor alloying elements were found not to vary

from the bulk composition. The untreated sample shows a nearsurface broad peak oxygen profile attributed to residual oxide scale resulting from the sample manufacturing and storage. The breadth of the O peak and extended tail profile indicated an effect due to surface roughness, where the surface material in valleys was sputtered at a lower rate than the asperities. The EPP cleaned surfaces had a significantly lower sub-surface O profile than the untreated sample but showed a similar high initial surface peak. This is evidence that the EPP process had successfully removed or reduced surface oxides, which together with reduced surface roughness, resulted in a lower near-surface O content. It is also evident from the GDOES O profiles that the pulse current EPP cleaning treatment is as effective in removing surface oxide as the DC process.

60

A. Yerokhin et al. / Journal of Materials Processing Technology 210 (2010) 54–63

Fig. 5. Effects of EPP cleaning on surface roughness (Ra).

Fig. 7. Typical XRD patterns of the untreated and EPP cleaned steel. The peaks correspond to ␣-Fe (JCPDS 06-0696).

5.5. Microhardness Fig. 10 illustrates the effect of the process parameters on the hardness of the treated surfaces. The indentation load variation from 25 to 200 g allowed the hardness to be probed at different depths from the surface, varied from 1 to 3.5 ␮m, respectively. The hardness of the untreated sample varied from 740 to 800 HV as the load was decreased from 200 to 25 g as is typical for hard materials (Jonsson and Hogmark, 1984; Atkinson, 1991) and can be explained by the increased contribution of elastic deformation at low loads. The DC treated sample also showed an increase in hardness values (710–765 HV) with decreasing load. Interestingly a considerable reduction in hardness values is observed for the

DC sample at 50 g load compared to that of the untreated sample, which could be an indication that some degradation of the mechanical properties occurred during the treatment under these particular conditions. In comparison, the samples treated at ı = 0.8 showed increased hardness at a similar load of 50 g. Samples treated with f = 10 kHz and ı = 0.8 showed a minimal degree of mechanical degradation, with the hardness trend following a similar pattern to that of the untreated sample. However, at lower treatment frequencies of 100–1000 Hz, a substantial change in hardness value was observed which clearly indicates the probability of mechanical degradation occurring due to the thermal effects that could be related to longer exposure times during the cleaning process.

Fig. 6. SEM micrographs showing the surface morphologies of EPP cleaned steel compared to that of the untreated and DC treated steel: (a) untreated sample; (b) DC treated; (c) f = 10 kHz; ı = 0.8; (d) f = 0.1 kHz; ı = 0.2.

A. Yerokhin et al. / Journal of Materials Processing Technology 210 (2010) 54–63

Fig. 8. Effect of EPP cleaning parameters on residual stresses in steel substrates.

61

Fig. 11. Fatigue diagram of untreated and EPP cleaned (ı = 0.8) steels. Unfilled data points indicate that the samples have not failed.

sheath straight, probably due to desorption: HCO3 − → H+ + CO3 2− H+ + OH− → H2 O H2 Oaq → H2 Oplasma + + eaq Naaq + → Naplasma +

Fig. 9. Typical example of iron and oxygen profiles in untreated steel compared with oxygen profiles in EPP treated steels, according to GDOES analysis.

5.6. Mechanical properties of EPP treated materials All samples passed the static load hydrogen embrittlement test, indicating that the EPP process did not cause significant hydrogen uptake in the AISI 4340 test pieces. A plausible explanation for this can be provided when considering that the mechanism of plasma sheath formation (as discussed in Section 2 of this article) is dominated by solvent evaporation due to Joule heating and whereas cathodic reactions yield comparatively small amounts of gaseous products. In mildly alkaline sodium bicarbonate solutions, carbonic acid dissociates fully (Bardal, 2004) and the charge through the electrolyte–sheath interface is transferred mainly by water vacancies (Cserfalvi and Mezei, 1996), with metal cations entering the

Because of high pressure, the mean free path of cationic species in the plasma is much smaller than the sheath thickness, so numerous dissociation-recombination cycles are expected. This eliminates a possibility of impact assisted formation of large amounts of free hydrogen that could be adsorbed on the metal surface to affect its mechanical properties. The hydrogen embrittlement test does not discriminate between the treatment conditions, however, and the results of the rotating bending beam tests presented in Fig. 11 are more informative. The S–N data illustrates the relationship between maximum cyclic stress and number of cycles to failure for untreated, EPP treated and shot peened + EPP treated samples (ı = 0.8). It is clear that the EPP process caused some deterioration in low cycle fatigue life under higher stress and a reduction of approximately 20%, in high cycle fatigue as N = 1 × 106 is approached. The EPP treated samples show a steeper gradient of stress with increasing cycles than the untreated case. The DC and 0.1 kHz samples showed the greatest reduction, with the fatigue limit for the DC case reduced from 830 to 670 MPa for N = 2 × 106 cycles. There is a slight improvement demonstrated for the 10 kHz EPP sample only at a stress level of 980 MPa, which may be mainly due to reduced surface roughness (Fig. 5), possibly in combination with process induced residual compressive stress on the surface (Fig. 8). The prior shot peening treatment is effective in increasing the fatigue life of the EPP cleaned test pieces up to 1 × 106 cycles, but for higher cycles no further improvement is evident. This could be due to alteration of compressive residual stress field in the surface layer caused by the cleaning process and partial stress relaxation during the test (Torres and Voorwald, 2002; Capello et al., 2004). It is notable however that the EPP cleaning process has a significantly lower effect on the high cycle fatigue limit than is reported for electroplating processes (Guzmana et al., 2000). 5.7. Electrochemical behaviour of EPP cleaned surfaces

Fig. 10. Effects of pulsed plasma treatment parameters on surface hardness at different loads (samples treated at f = 10 kHz).

Fig. 12 shows the electrochemical behaviour of the EPP treated samples tested in deaerated 3.5% NaCl solution. The untreated steel sample had a corrosion potential of −0.57 V vs. SCE and a

62

A. Yerokhin et al. / Journal of Materials Processing Technology 210 (2010) 54–63

Fig. 12. Potentiodynamic corrosion plots showing the electrochemical behaviour of untreated and the EPP cleaned steel in deaerated 3.5% NaCl solution.

showing that the degradation due to the thermal impact from the treatment is insignificant. 3. The treatment induced relaxation of residual stresses in the steel is hindered at higher frequencies; these conditions are considered to be favourable for subsequent coating deposition processes. 4. The EPP cleaning has no effect on hydrogen embrittlement of the steel. There is however, an adverse effect on fatigue life which is greatest for the DC treatment and tends to reduce with increased pulse frequency. Shot peening prior to EPP is effective to mitigate this effect, indicating that this is related to the residual stress profile in the near-surface regions of the substrate material. 5. Corrosion tests indicate that EPP treatments affect primarily the rate of the cathodic reaction of hydrogen reduction. This is attributed to surface homogenisation and removal of surface oxide layers. Acknowledgements

region featuring partially passive behaviour between −0.5 and −0.3 V. A transpassive region is situated above −0.3 V and characterised by rapid increases in current density with anodic potential. Apparently, the EPP treatments in both DC and pulsed current modes mainly affected the kinetics of the cathodic processes, resulting in displacement of corrosion potentials towards more positive values. This effect was demonstrated most noticeably by the DC treated sample for which the corrosion potential shifted to −0.42 V, whereas pulsed treatments resulted in smaller shifts, e.g. Ecorr = −0.52 V was observed for the sample treated under conditions of ı = 0.8 and f = 10 kHz (Fig. 12). The anodic behaviour of the steel was relatively less affected by the EPP treatments, with only minor changes to the characteristic potential of passivity breakdown and lower overall anodic current densities were observed. In deaerated solutions with a low oxygen concentration, corrosion processes of iron can be described by the following reactions (ASM Handbook, 1987): Anodic reaction : Cathodic reaction : Overall reaction :

Fe = Fe2+ + 2e− 2H+ + 2e− = H2 Fe + 2H+ = Fe2+ + H2

It can be deduced that EPP treatments predominantly facilitated the hydrogen reduction process on the cathodically charged fraction of the metal surface. This can be associated with a reduction of anodic areas and/or with enhanced proton absorption. Evidence to support the former mechanism can be found in the morphology of the treated surfaces (Fig. 6) displaying a higher degree of homogeneity and reduced quantity of non-metallic inclusions due to thermal and hydrodynamic impacts of discharges. The latter mechanism also appears to become increasingly likely due to the removal of surface oxygen, evident from the GDOES analysis; however this should also contribute to the reduction of anodic areas. 6. Conclusions A novel pulsed current EPP technology was developed and showed enhanced capabilities compared to simple DC treatments for cleaning of thermally hardened AISI 4340 steel: 1. According to surface profilometry, SEM and GDOES analyses, the EPP cleaning process is more effective at a higher duty cycle and frequency (preferably ı = 0.8 and f = 10 kHz). 2. Under optimal conditions of the pulse current EPP cleaning, the surface hardness is comparable to that of the untreated material,

Financial support for this work was provided by US DoD SERDP and by the UK EPSRC. The hydrogen embrittlement and fatigue testing facilities were provided by the US Naval Research Laboratory within the SERDP framework. References Aliofkhazraei, M., Roohaghdam, A.S., 2007. A novel method for preparing aluminum diffusion coating by nanocrystalline plasma electrolysis. Electrochem. Commun. 9, 2686–2691. ASM Handbook, Corrosionvol.13. ASM International. ASM Handbook, Surface Engineering, vol. 5. ASM International. Atkinson, M., 1991. Origin of the size effect in indentation of metals. Int. J. Mech. Sci. 33, 843–850. Bardos, L., Barankova, H., 2008. Plasma processes at atmospheric and low pressures. Vacuum 83 (3), 522–527. Bardal, E., 2004. Corrosion and Protection. Springer-Verlag London, Ltd, pp. 315. Belkin, P.N., 2005. Electrochemical Thermal Treatment of Metals and Alloys. Mir, Moscow, pp. 336. Boyer, H.E., 1986. Atlas of Fatigue Curves. ASM, Materials Park, Ohio, pp. 111–113. Capello, E., Davoli, P., Filippini, M., 2004. Relaxation of residual stresses induced by turning and shot peening on steels. J. Strain Anal. 39 (3), 285–290. Crawford, A.E., 1963. A practical introduction to ultrasonic cleaning. Ultrasonics 1 (2), 65–69. Cserfalvi, T., Mezei, P., 1996. Operating mechanism of the electrolyte cathode atmospheric glow discharge. Fresenius J. Anal. Chem. 355, 813–819. Cullity, B.D., Stock, S.R., 2001. Elements of X-ray Diffraction, 3rd ed. Prentice-Hall, New Jersey, ISBN 0-13-178818-3. Davies, R.A., Hickling, A., 1952. Glow-discharge Electrolysis. Part I. The Anodic Formation of Hydrogen Peroxide in Inert Electrolytes. J. Chem. Soc., 3595–3602. Duradzhi, V.N., Parsadanyan, A.S., 1988. Metal Heating in Electrolytic Plasma. Shtiintsa, Chisinau, pp. 216. Gupta, P., Tenhundfeld, G., Daigle, E.O., Ryabkov, D., 2007. Electrolytic plasma technology: Science and engineering – An overview. Surf. Coat. Technol. 201, 8746–8760. Guzmana, C., Diaza, N., Berriosb, J.A., 2000. Fatigue properties of a SAE 4340 steel coated with a Nimet HP autocatalytic nickel deposit. Surf. Coat. Technol., 133–134, 561–571. Herbert, J.D., Groome, A.E., Reid, R.J., 1994. Study of cleaning agents for stainless steel for ultrahigh vacuum use. J. Vac. Sci. Technol. A 12 (4), 1767–1771. Hikling, A., Ingram, M.D., 1964. Contact glow-discharge electrolysis. Trans. Faraday Soc. 60, 783–793. Jonsson, B., Hogmark, S., 1984. Hardness measurements of thin films. Thin Solid Films 114, 257–269. Kellog, H.H., 1950. Anode effect in aqueous electrolysis. J. Electrochem. Soc. 97 (4), 133–142. Khan, R.H.U., Yerokhin, A.L., Pilkington, T., 2005. Residual stresses in plasma electrolytic oxidation coatings on Al alloy produced by pulsed unipolar current. Surf. Coat. Technol. 200, 1580–1586. Khan, R.H.U., Yerokhin, A.L., Matthews, A., 2008. Structural characteristics and residual stresses in oxide films produced on Ti by pulsed unipolar plasma electrolytic oxidation. Phil. Mag. 88 (6), 795–807. Meletis, E.I., Nie, X., Wang, F.L., 2002. Electrolytic plasma processing for cleaning and metal-coating of steel surfaces. Surf. Coat. Technol. 150, 246–256. Nascimento, M.P., Souza, R.C., Pigatin, W.L., 2001. Effects of surface treatments on the fatigue strength of AISI 4340 aeronautical steel. Int. J. Fatigue 23, 607–618. Noltingk, B.E., Neppiras, E.A., 1950. Cavitation produced by ultrasonics. Proc. Phys. Soc. B 63, 674–685.

A. Yerokhin et al. / Journal of Materials Processing Technology 210 (2010) 54–63 Parfenov, E.V., Nevyantseva, R.R., Gorbatkov, S.A., 2005. Process control for plasma electrolytic removal of TiN coatings. Part 1. Duration control. Surf. Coat. Technol. 199, 189–197. Rohsenow, W.M., Hartnett, J.P., Cho, Y.I., 1998. Handbook of Heat Transfer, 3rd edition. McGraw-Hill Professional, pp. 1344. Sengupta, S.K., Singh, O.P., 1991. Contact glow discharge electrolysis: a study of its onset and location. J. Electroanal. Chem. 301 (1–2), 189–197. Sengupta, S.K., Singh, O.P., 1994. Contact glow discharge electrolysis: a study of its chemical yields in aqueous inert-type electrolytes. J. Electroanal. Chem. 369 (1–2), 113–120. Sengupta, S.K., Srivastava, A.K., Singh, R., 1997. Contact glow discharge electrolysis: a study on its origin in the light of the theory of hydrodynamic instabilities in local solvent vaporisation by Joule heating during electrolysis. J. Electroanal. Chem. 427 (1–2), 23–27. Slovetskii, D.I., Terent’ev, S.D., Plekhanov, V.G., 1986. A plasma–electrolyte metal heating mechanism. High Temp. 24, 277–285. Tam, A.C., Park, H.K., Grigoropoulos, C.P., 1998. Laser cleaning of surface contaminants. Appl. Surface Sci., 127–129, 721–725.

63

Tendero, C., Tixier, C., Tristant, P., 2006. Atmospheric pressure plasmas: a review. Spectrochim. Acta B: Atom. Spectrosc. 61 (1), 2–30. Torres, M.A.S., Voorwald, H.J.C., 2002. An evaluation of shot peening, residual stress and stress relaxation on the fatigue life of AISI 4340 steel. Int. J. Fatigue 24, 877–886. Vogt, H., 1997. Contribution to the interpretation of the anode effect. Electrochim. Acta 42 (17), 2695–2705. Yerokhin, A.L., Nie, X., Leyland, A., 1999. Plasma electrolysis for surface engineering. Surf. Coat. Technol. 122, 73–93. Yerokhin, A.L., Shatrov, A., Samsonov, V., 2005. Oxide ceramic coatings on aluminium alloys produced by a pulsed bipolar plasma electrolytic oxidation process. Surf. Coat. Technol. 199, 150–157. Yerokhin, A.L., Snizhko, L.O., Gurevina, N.L., 2003. Discharge characterization in plasma electrolytic oxidation of aluminium. J. Phys. D: Appl. Phys. 36, 2110–2120. Yerokhin, A.L., Snizhko, L.O., Gurevina, N.L., 2004. Spatial characteristics of discharge phenomena in plasma electrolytic oxidation of aluminium alloy. Surf. Coat. Technol., 177–178, 779–783.