Effects of Matrix Silicon Content on the Plasma Electrolytic Oxidation of Al-Si Alloys Using Different Power Modes

: The plasma electrolytic oxidation (PEO) of pure Al and Al alloys containing 4, 9, 12, or 15 wt.% Si were investigated under pulsed bipolar current and pulsed bipolar voltage modes, respectively. It was determined that the discharge sparks preferentially occurred on the SiO 2 relative to the Al 2 O 3 during the initial stage of PEO processing regardless of the power mode. Following 30 min of PEO treatment under the two modes, the thicknesses of the layers decreased, whereas their speciﬁc energy consumption increased with increasing Si content in the matrix. The presence of primary Si in the alloy with 15 wt.% Si had a signiﬁcantly negative effect on the PEO process in the pulsed bipolar current mode: The layer thickness decreased by 45%, and its speciﬁc energy consumption increased by 52%, compared with those on pure Al. However, in the pulsed bipolar voltage mode, the layer thickness on the evaluated samples only decreased slightly, and it became much more similar after treatment.


Introduction
Cast Al-Si alloys are commonly used to produce mechanical parts because of their excellent castability, low shrinkage, and high specific strength, among other beneficial features [1,2]. Some eutectic or near-eutectic Al-Si alloys (e.g., AlSi12, AlSi10Mg) have recently been considered as some of the most promising materials for the additive manufacturing of Al alloy products [3,4]. However, the low hardness and poor corrosion and abrasion resistance properties significantly limit the lifetime of Al-Si alloy products [5,6]. Therefore, surface technologies, such as anodizing [7], laser remelting [8], plasma electrolytic oxidation (PEO) [9], and cathode plasma electrolytic deposition [10] have been applied to address these issues.
In particular, PEO has been widely employed for the surface modification of Al, Mg, and Ti alloys because it is an eco-friendly and high-efficiency method [11,12]. It is an effective method for the surface modification of wrought, cast, or foamed Al alloys (e.g., 2024, 6063, 7075, A356), which are often used to fabricate mechanical, automobile, or electronic products [13][14][15][16]. Recently, PEO has attracted increasing interest in terms of fabricating in situ ceramic layers on Al-Si alloys that are used in the manufacture of pistons, wheels, and many other products because the PEO layer exhibits excellent adhesion, antiwear, and corrosion performance and demonstrates significant advantages as an anodic film [9,[17][18][19]. However, the eutectic or primary Si grains in the Al alloys have adverse impacts on the growth and structure of the PEO layer [20,21]. The mechanism governing Si particles had been discussed in some previous studies; for example, it was proposed that Si inhibited the growth of the film during the anodic oxidation stage [18,22]. The micro-arc discharges began at the Al-Si interfaces when the voltage reached a critical value, and then, the discharges gradually spread across the primary Al and Si regions [23]. The Al 2 O 3 and SiO 2 phases mixed together and converted to mullite under the high temperature conditions generated by the discharge sparks [23]. The layer on the bulk Si was much coarser during the early stage of oxidation, but the surface morphology of the layer became more uniform as the PEO process continued [21,23]. In addition, the layer thickness, density, and abrasion resistance all decreased as the Si content increased [21,24]. Several solutions have been implemented to mitigate the adverse impacts of Si or to improve the performance of the PEO layer on Al-Si alloys. For example, a pre-anodizing treatment of the A356 alloy accelerated the layer deposition rate and enhanced the energy efficiency of the PEO process [25]. The fine Si grains in the eutectic Al-Si alloy helped improve the compactness and corrosion resistance of the PEO layer [6,26]. The elimination of Si on the surface of Al-Si alloys reduced the mullite content and enhanced the corrosion resistance of the PEO layer [26,27]. To obtain a uniform oxide layer on the Al-Si alloy, additives such as NH 4 VO 3 could be introduced [28]. Such aluminate-based electrolytes are known to promote the formation of a denser coating on cast Al-Si alloys [29]. In addition, a post-PEO sealing process improved the hydrophobicity, impedance, and anti-corrosion performance of the PEO layer on an A356 alloy [30]. Ceramic-based self-lubricating layers (i.e., PEO/PI/WS 2 ) prepared via PEO and spin-layer methods significantly improved the tribological properties of a ZL109 alloy [31]. Nevertheless, the pre-or post-treatments mentioned above render the surface modification process relatively complex, thus making it undesirable for industrial applications.
The PEO process is very intricate owing to the involvement of thermal, chemical, electrochemical, and plasma reactions at the substrate-electrolyte interface [32]. In addition to the alloying elements and electrolyte composition aspects mentioned above [24,30], electrical parameters (e.g., power supply mode, applied voltage, current density, duty ratio, pulse frequency) are also critical factors in terms of the formation, morphology, and performance of the PEO layer on Al-Si alloys [18,33]. A constant current (-10 to 35 A/dm 2 ) [22,23] or constant voltage (-140 to +550 V) [24,31] mode based on an asymmetric alternating current (AC) power supply is commonly applied for the PEO of Al-Si alloys. The oxidation process and layer characteristics change significantly under different power supply modes, although previous studies regarding the evolution of Si grains during the initial stage of PEO treatment were mostly based on a constant-current mode. Therefore, in this work, pure Al and Al-Si binary alloys with different Si contents were treated by PEO with pulsed bipolar current (PBC) and pulsed bipolar voltage (PBV) modes. The effects of the Si content on the layer growth process, characteristics, and energy efficiency were analyzed under different power modes.

Materials and Methods
Pure Al (>99.7%) and cast Al-Si binary alloys with Si contents corresponding to 5, 9, 12, and 15 wt.% were obtained from commercial sources and used as the substrate material; the Al-Si alloys were labeled as Al-5Si, Al-9Si, Al-12Si, and Al-15Si, respectively. The size of each sample was 20 mm × 20 mm × 4 mm, and they were polished using 1200-grit sandpaper, cleaned via ultrasonic treatment in acetone, and then dried under hot air before the PEO treatment.
The electrolyte was comprised 10 g/L of Na 2 SiO 3 , 2 g/L of NaOH, and 5 g/L of C 6 H 12 N 4 . The PEO layers were prepared on the samples using an AC pulsed power supply (WHD60, Harbin, China) with a maximum power of 60 kW. A schematic diagram of the output current and voltage signals is presented in Figure 1. When using the PBC mode, the applied positive current density was 10 A·dm −2 , and the negative current density was 4 A·dm −2 . When applying the PBV mode, the positive voltage was 450 V, and the negative voltage was 50 V. The pulse frequency was 400 Hz, and the duty cycle was 25% in both power modes. The thickness of each PEO layer was measured using an eddy current thickness gauge (Surfix N, PHYNIX Sensortechnik GmbH, Neuss, Germany). The morphology and elemental composition were analyzed with a field emission scanning electron microscope (Nova NanoSEM 430, Fei Company, Hillsboro, OR, USA) equipped with an energy dispersive spectrometer (EDS INCA, Oxford Instruments, High Wycombe, UK). The crystalline structures of coated samples were characterized via X-ray diffraction (Miniflex 600, Rigaku, Tokyo, Japan) using Cu Kα radiation over a 2θ range of 20-70 • . Figure 2 shows the relationship between the output positive voltage and the PEO processing time. The voltage curves followed a similar trend as in a previously reported case where Al-Si alloys underwent constant-current oxidation treatment [21]. At any point during the oxidation, the positive voltage decreased with increasing Si content in the substrate. However, the extent of deviation was reduced as the treatment time was prolonged. After 30 min of oxidation, the positive voltage deceased from approximately 464 to 442 V when the Si content in the matrix increased from 0 to 15 wt.%.   (Figure 3a,d,g). However, the layers on the Al-9Si and Al-15Si samples were uneven during the initial stage of oxidation. Some coarse oxide was formed in the eutectic Si and primary Si regions, where discharge sparks preferentially occurred (Figure 3b,c). The layers on the Al-9Si and Al-15Si samples became increasingly uniform as the oxidation process continued for~10 or 16 min, respectively (Figure 3e,f). By the end of the 30 min process, the surface morphologies of the three layers were all very flat, and the number of pores on their surfaces increased slightly with the substrate's Si content (Figure 3g-i).      Figure 6 shows the cross-sectional morphologies of the layers on pure Al, Al-9Si, and Al-15Si after 30 min of processing in the PBC mode. The practical layer thicknesses were consistent with the tested values, as shown in Figure 4. However, some large pores with irregular shapes were distributed randomly across these layers, and a dense sub-layer existed near the matrix, which indicated good adhesion between the substrate and the ceramic coatings.  Figure 7 presents the positive currents of different samples over 30 min of treatment in the PBV mode. It took several dozen seconds before the output voltage reached the set value because the power supply avoided exceedingly high currents to protect itself (the membrane resistance was very small during the early stage). A similar phenomenon was observed in the PEO of ZL101 when applying a positive voltage of 530 V [18]. The current increased rapidly during the first 34 s of the oxidation process; however, the current of each Al-Si alloy sample experienced a fluctuation between 20 and 25 s, where their values decreased slightly and then began to climb again. In each case, the current reached a peak when the voltage reached 450 V, attaining maximum values of 13.7 to 26.7 A as the Si content increased from 0 to 15 wt.%. The current decreased gradually as the process continued. In general, the currents of samples containing relatively more Si grains were always higher at all points during the oxidation process.

PEO of Samples Using the PBV Oxidation Mode
The PEO layer on pure Al was essentially uniform throughout the entire process (Figure 8a,d,g). However, the morphologies of the layers on the Al-9Si and Al-15Si samples were uneven when the voltage reached 450 V. The coarse oxides induced by the discharge sparks were mainly generated in the eutectic Si or primary Si regions (Figure 8b,c). The morphologies of these two layers became much more uniform after 50 s of oxidation (Figure 8e,f). The pores derived from the discharge channels were distributed across most of the sample surface. By the end of the oxidation process, the layers exhibited rough surface structures, and the number of pores on the surface increased with the Si content in the substrate (Figure 8g Figure 10 shows the energy consumptions of the layers on different samples following the 30 min oxidation treatment in the PBV mode. The total energy consumption of the samples increased from 0.08 to 0.18 kWh as the Si content in the matrix increased from 0 to 15 wt.%, while the specific energy consumption of the layers also increased (from 2.9 to 8.1 kWh m −2 µm −1 ). Moreover, the total and specific energy consumption both increased rapidly when the Si content of the samples increased from 12 to 15 wt.%.  Figure 11 shows the cross-sectional morphologies of the layers on pure Al, Al-9Si, and Al-15Si after 30 min of processing in the PBV mode. The practical thickness of each layer was consistent with the corresponding measured values (Figure 9). Although there were many pores randomly distributed across the three coatings, their inner layers adopted relatively denser structures.

Mechanistic Analysis of the Impact of Si Grains on the PEO of Al-Si Alloys
The power supply mode plays a key role in governing the layer's characteristics and the energy efficiency of the Al-Si alloys after a 30 min PEO treatment. This is likely related to the oxidation behavior of Si grains under different power input methods. Because it is difficult to observe the oxidation of eutectic Si owing to its small size, analyzing the oxidation process involving large primary Si might better reveal the mechanism whereby the Si phase influences the PEO processing of Al alloys. Figure 12 shows the surface morphology and composition of the PEO layer on/near the primary Si of the Al-15Si sample at the initial oxidation stage. After 270 s of oxidation in the PBC mode, a significant amount of elemental O was detected on the sample surface as SiO 2 and Al 2 O 3 films grew on the primary Si (region 1) and primary Al (region 2), respectively (Figure 12a). The volume expansion of Si to SiO 2 was approximately 120%; such expansion would be hindered by the surrounding Al 2 O 3 , thus generating high stress [23]. As a result, the fragile SiO 2 partially collapsed and left various defects (e.g., pits and cracks) on its surface. The discharge sparks affected the primary Al and Si regions when the oxidation process lasted for 600 s, but the SiO 2 layer was still much more porous than the Al 2 O 3 layer (Figure 12b). Additionally, numerous O, Si, and Al atoms were detected at the boundary between SiO 2 and Al 2 O 3 (region 2), which gradually converted to Al-Si-O compounds under the higher temperature condition caused by the sparks. When the oxidation treatment continued for 960 s, the SiO 2 layer on the primary Si became thicker, and the voids on its surface were significantly reduced (Figure 12c). The interface between SiO 2 and Al 2 O 3 disappeared, and these phases mixed at their boundary, where Al, Si, and O elements were abundant (region 2). However, at the central part of the SiO 2 layer, the Al elemental content was still relatively low (region 1). Still, an SiO 2 layer containing numerous small voids was generated on the primary Si (region 1) after 34 s of oxidation in the PBV mode (Figure 12d). In contrast, only a few voids appeared on the Al 2 O 3 film (region 2) because the discharge sparks mainly impacted the SiO 2 at this point. A few seconds later, the discharge sparks covered the entire sample surface (Figure 12e). The area and number of voids on the SiO 2 layer decreased substantially (region 1), and the SiO 2 and Al 2 O 3 began to mix together (region 2). When the oxidation treatment continued for 300 s, the SiO 2 layer became much thicker and denser (Figure 12f, region 1). The boundary between the primary Al and Si regions disappeared as more SiO 2 and Al 2 O 3 mixed together and converted to Al-Si-O compounds (region 2). The oxidation of primary Si was significantly accelerated in the PBV mode because strong discharges occurred rapidly on its surface, which resulted in a much weaker effect of the Si content on the layer growth rate for the evaluated samples over 30 min of processing. According to the experimental results described above, a growth model for the oxide layer on the Si grains was established and is schematically illustrated in Figure 13. In stage I, an anodic Al 2 O 3 with few defects grew rapidly on the surface of the Al phase, while a thin anodic SiO 2 film with numerous defects (e.g., voids and cracks) grew on the surface of the Si phase. The width of the tunnel barrier for Al 2 O 3 was approximately 127 Å, while that for SiO 2 was 42 Å [33]. Electrons tunneled through the porous SiO 2 thin film, and gas bubble generation was promoted on the surface of oxidized Si grains [33,34]; a significant amount of electrical energy would be exhausted by the side reactions in this case (i.e., water electrolysis). However, electron tunneling through the relatively denser and thicker Al 2 O 3 was limited because appreciable electron tunneling only becomes favorable when the width of the barrier is less than 100 Å [33]. In stage II, the discharge sparks began to occur as the voltage exceeded a threshold value. The theoretical resistance of SiO 2 is approximately 10 10 Ω·cm −1 , which is four orders of magnitude lower than that of Al 2 O 3 [35,36]. Therefore, dielectric breakdown preferentially occurred on the SiO 2 film, rather than on the Al 2 O 3 film. The electrons tunneling through the thin regions of the SiO 2 film still promoted the generation of gas bubbles. Some bubbles could become trapped in the oxides during the chilling process. Thus, many voids existed on the surface of the SiO 2 layer. When the resistance of SiO 2 increased and approached that of Al 2 O 3 , the oxidation process entered stage III, wherein the discharge sparks occurred on the anodic Al 2 O 3 layer as well. The electron tunneling phenomenon was reduced significantly as the SiO 2 layer grew much thicker. The number of voids in the SiO 2 layer also decreased, and the SiO 2 and Al 2 O 3 gradually converted to Al-Si-O compounds. As a result, the SiO 2 had a much weaker influence on the layer growth thereafter. Once the SiO 2 had mostly converted to mullite, the process reached stage IV, wherein the effect of SiO 2 on the PEO process was weak. The oxide layer exhibited a much more uniform morphology containing few voids on its surface. The oxidations of primary Si and eutectic Si to SiO 2 were essentially the same; however, it was difficult for the large SiO 2 to mix with the Al 2 O 3 to produce mullite. In general, the negative effect of primary Si on the PEO of Al-Si alloys would last for a much longer time than in the case of eutectic Si. Moreover, the size and number of Si grains were directly correlated with the Si concentration in the substrate. Therefore, the layer thickness was reduced, and the specific energy consumption increased appreciably as the matrix Si content increased during a 30 min process in either the PBC or PBV mode. Both of these parameters changed more significantly when the Si content increased from 12 to 15 wt.%, because there were numerous bulk Si grains in the Al-15Si alloy. In addition, the oxidation of Si grains was accelerated in the PBV mode relative to the PBC mode. The layer growth entered stage IV quickly, which meant that the effect of SiO 2 on the layer growth weakened within a few minutes. Thus, the layer thicknesses in different samples following 30 min of PBV oxidation treatment were very similar. However, the positive current clearly increased as the Si content increased during the first 5 min of processing (i.e., when the energy consumption contributed a major portion of the overall energy). A greater amount of energy was exhausted by the side reactions or resistance heat for the sample with the highest Si mass. Thus, the specific energy consumption of the final layer increased significantly with the matrix Si content in the PBV mode.

Conclusions
In this study, pure Al and Al alloys containing 5, 9, 12, and 15 wt.% Si were treated via PEO for 30 min in pulsed bipolar current and pulsed bipolar voltage power modes. The main conclusions drawn from the results are as follows: (1) The eutectic Si and primary Si grains had slow oxidation rates when using the pulsed bipolar voltage mode. It took a long time for the SiO 2 growing on the large Si grains to mix with the surrounding Al 2 O 3 to generate Al-Si-O compounds. The negative effect of porous SiO 2 on the PEO process continued for a longer time in the Al alloy containing larger Si grains; therefore, the final layer thicknesses on different Al alloys decreased significantly with increasing Si content in the matrix; (2) The eutectic Si and primary Si grains were rapidly oxidized and converted to Al-Si-O compounds when using the pulsed bipolar current mode. The effect of Si or SiO 2 on the layer growth process weakened after a short period of oxidation, such that the layer thicknesses on different Al alloys were very similar by the end of the pulsed bipolar voltage oxidation process; (3) The specific energy consumption of the layers decreased as the Si content increased, regardless of whether the pulsed bipolar current or pulsed bipolar voltage mode was applied.