**3. Results**

#### *3.1. Electrical Characteristics of the Plasma Electrolytic Oxidation Process*

Figure 2a shows that the current waveform changes by following a well-defined trend: When the voltage reached the corrosion potential of the substrate (Figure 2b, about 150 V), the current increased slightly due to the dissolution of the precursor film. Upon the growth of the oxide film, the plasma electrolytic oxidation system tended to stabilize and the current slowly dropped. When the voltage reached 330 V, the current abruptly increased (Figure 2c) and this might result in the dielectric breakdown of the oxide film. At this point, the current increased sharply when the voltage was further increased, implying that higher voltages led to more pronounced discharge events. Figure 2d shows the trend of the oscillating current for voltages higher than 450 V. In these conditions, powerful discharges might occur, causing destructive effects in the sample as reported in previous literature [6,11]. In this work, an initial *Vb* equal to 330 V and a *Vu* value of 450 V were chosen.

**Figure 2.** (**a**) Voltage and current waveforms of the initial test pulses and (**b**–**d**) partial waveforms.

Figure 3 depicts the waveforms of the pulses used in the test experiment and during the voltage adjustment process. When the value of the real-time feedback current was lower than half the value of *Ib*, a new series of test pulses (330–350 V) was applied. According to the current waveforms of the test pulses, new values of *Vb* (335 V) and *Ib* (0.6 A) were determined and they correspond to the inflection point of the current curve. At this point, a new value of the amplitude (340 V) was assigned to the voltage pulses.

**Figure 3.** Voltage and current waveforms of pulses during a typical self-adaptive adjustment process.

Figure 4a shows the voltage and the current as a function of the measurement time: The voltage increased linearly in the 330–450V range and the current remained rather stable. These observations show that the current overshoot process, which occurs in the traditional voltage-control mode [6], was effectively suppressed. The current termination rule adopted in this work more effectively maintained a steady voltage in time, when compared to the more commonly used fixed-time methods. Figure 4b shows the output power curve, which was obtained experimentally: The power increased as a function of the processing time, reflecting that a higher pulse energy should be applied during the coating growth process to ensure the dielectric breakdown and a series of discharges.

**Figure 4.** (**a**) Voltage–time curve, current–time curve, and (**b**) power–time curve during the self-adaptive plasma electrolytic oxidation processing of aluminum alloys.

#### *3.2. Coating Microstructure*

The surface morphology of the coating exhibits (Figure 5a) regions with elongated open pores and lighter gray areas. Open pores are generally observed for short processing times or when a low-voltage amplitude is applied to low-thickness coatings [15]. The cross-sectional image of the sample (Figure 5b) reveals that the interfaces between the substrate and the pores and between the inner and the outer layer of the coatings were characterized by a wavy profile. The estimated thickness of the coating measures 23 μm and this value was lower than the previously reported ones (100 μm) [15,24]. Figure 5c shows that the coating was mostly composed of Al, O, and Si. Moreover, the phases of the coating were mainly composed of the α-Al2O3 and γ-Al2O3 (Figure 5d), which could enhance the coating microhardness and are formed under a soft-sparking regime.

**Figure 5.** *Cont*.

**Figure 5.** (**a**) Surface morphology, (**b**) cross-section, (**c**) chemical compositions, and (**d**) phases of the plasma electrolytic oxidation coatings.

#### *3.3. Energy Consumption*

The specific energy consumption (*Qc*) of the plasma electrolytic oxidation process can be calculated by using Equation (4):

$$Q\_{\mathcal{E}} = \frac{P \cdot t}{S \cdot \delta} = \frac{1}{S \cdot \delta} \int\_0^t u(t) \cdot i(t) dt,\tag{4}$$

where, *P* represents the power consumption, *t* is the total processing time, *S* corresponds to the superficial area of the sample, and δ is the coating thickness.

The value of *Qc* estimated by using Equation (4) is 1.8 kW h m<sup>−</sup><sup>2</sup> μm<sup>−</sup>1: This value is similar to that obtained in a previous study [24] (2.5–2.7 kW h m<sup>−</sup><sup>2</sup> μm<sup>−</sup>1) and considerably lower than other results (26.7 kW h m<sup>−</sup><sup>2</sup> μm<sup>−</sup>1) reported in the literature [28].

## **4. Discussion**

The oxides of aluminum substrate have large band gaps. Such band gaps might be associated with an oxide structure with highly stable thermodynamic properties. It is well established that a dielectric breakdown commonly occurs across a thin oxide film located on a substrate [1]. Hence, an anodic oxidation process was performed before the plasma electrolytic oxidation processing of the sample, with the aim to promote discharge formation and enhance the dielectric breakdown strength. In order to ensure the electrons do not travel through the oxide film, the final voltage of the anodic oxidation process was set as 300 V, which is lower than the initial breakdown voltage.

The band gaps influence the electric field that tended to build up across the oxide film. A dielectric breakdown strength was normally expressed as a critical breakdown field, at which point, a discharge occurred. Hence, in order to assure enough dielectric breakdown strength, the applied voltage magnitude should be higher than the breakdown voltage. During the plasma electrolytic oxidation process, the electric fields were affected by the dielectric constant, which represents the capacity of the oxide to store electric charge. The dielectric constant increased with increasing thickness of the coating. Thus, the applied voltage should also be incremental. While the increasing voltage might show nonlinear behavior. Therefore, it is necessary to establish the relationship between the breakdown voltage and the coating thickness. The real-time pulse test technique provides a feasible approach to realize the identification of the feature information related to the coating growth.

The repeated formation of discharges on the surface of the sample is the key characteristic of a plasma electrolytic oxidation process. An individual event consisted of a complicated process such as micro-discharges, plasma channels, melting, evaporation, ejection, atomization, ionization, chemical reaction, cooling down, and overgrowth. According to previous research [1,5,10,15,20], the discharges change following a well-defined trend: The first spark, generating a discharge phenomenon, appeared by dielectric breakdown through a "weak site" in the anodic oxide film. The number of weak sites reduced with increasing thickness of the coating. With increasing voltage, the discharge color turned to orange–red, along with the emergence of more intense acoustic noise during the plasma electrolytic oxidation process. The individual discharges become less frequent but more intense when the increasing thickness of the coating, due to the reduced number of discharging sites through which the higher applied voltage needed to o ffer supplementary energy. These discharges have a strong tendency to occur repeatedly at particular locations—for example, they occur in 'cascades' that typically consist of hundreds of individual discharges [11]. Moreover, these discharges become more energetic and more dispersed—in terms of time and location—as the thickness increases. This is also verified by the evolving microstructure of the coating, particularly the pore content and architecture [20]. In a word, with an increasing applied voltage, the size of micropores, discharge channels, and overgrowth protrusions increased, which are mainly attributed to the di fferent discharge energy supplied by high voltage. However, when the applied voltage is too high, negative e ffects on high-quality and good-performance coatings are observed. Hence, the discharge energy and coating microstructure are dependent on the electrical parameters of the process.

These results indicate that the discharge events, which occur during the self-adaptive plasma electrolytic oxidation process, mainly belong to the first three stages described in the literature [15]: Stage 1, a thin oxide film was formed and dielectric breakdown was observed; stage 2, many white sparks were evenly distributed on the entire surface of the sample; stage 3, the sparks were gradually replaced by more intense micro-discharges with yellow or orange appearance. All of the discharges mentioned above are maintained in the soft-sparking regime. When a low *Vu* value is chosen, the discharges occur in the soft-sparking regime and this prevents the formation of cracks and other destructive e ffects on the coating, but this influences the choice of the coating thickness. The solution of such a compromise lies in the use of a novel soft-sparking regime based on the bipolar pulse mode, which occurs only when the ratio between the anodic and cathodic charge is lower than one [19–21,29].

It was estimated [32] that an individual discharge energy was ~1 mJ and that the conversion rate between discharge energy and resultant volume of the coating was ~10<sup>13</sup> J m<sup>−</sup>3, which can also be expressed as about 3 kW h m<sup>−</sup><sup>2</sup> μm<sup>−</sup>1. The specific energy consumption of the self-adaptive plasma electrolytic oxidation process was only 1.8 kW h m<sup>−</sup><sup>2</sup> μm<sup>−</sup>1. Such improvement might be related to the use of the pre-anodized precursor films and the self-adaptive control of several process parameters, such as the voltage, the current, and the processing time. Recently, the precursor anodic porous films prepared by conventional anodizing were demonstrated to reduce the energy consumption and to increase the coating microhardness via promotion of the soft-sparking regime [24,25]. Meanwhile, the energy associated with each discharge may be optimally controlled via real-time precision adjustment of the process electrical parameters.
