**3. Results**

#### *3.1. Identification of Critical Breakdown Voltage*

Figure 2a shows the partial voltage and current waveforms during the pre-treatment process of aluminum wires. In addition, dielectric breakdown of the oxide films when the potential was increased to approximately 350 V is shown in the figure. Within *t*r, a rise of potential results in an abrupt increase in current due to the active capacitive effect of the electrolyte load. After reaching a peak, the current decreases exponentially to a relatively stable level (within *t*s). When potential is lower than 340 V, stable levels are almost equivalent to 0.2 A. Above 350 V, the stable level increases with potential with an oscillating current waveform. This probably indicates that dielectric breakdown of oxide films occurred. Before dielectric breakdown, the applied voltage is not high enough to cause discharge and the load keeps in a stable high impedance. So, the current waveform is relatively smooth.

**Figure 2.** (**a**) Partial voltage and current waveforms during pre-treatment process with voltage sweep at 8 V/s and (**b**) their voltage-current characteristic curve.

To better understand this phenomenon, appropriate RMS value of the current and voltage approximations were calculated for each step. Stable value of current (within *t*s) was chosen to improve approximation accuracy for the voltage-current characteristics shown in Figure 2b.

#### *3.2. Electrical Characteristics of Single Pulse Anodizing*

To further understand the influence of pulse voltage on electrical characteristics of single pulse anodizing of aluminum micro-electrode, a group of single pulses with different voltages were applied for 100 μs. Figure 3a depicts the current transients due to these pulses. A peak current (*I*p) was observed within *t*p and increases with the rise of the pulse voltage. Next, a sharp decrease is seen for the current and it reaches a stable and nearly constant current (*I*c) for different level, from 0.2 to 1.2 A, flows within *t*c. Peak current due to capacitive load effect produces a large reactive component of the current. Magnitude of *I*p and current transition time *t*p were related to pulse potential. A higher potential resulted in a larger *I*p and longer *t*p. An increase in the value of *I*c was also observed with an increase in the applied pulse voltage. However, transient behavior was different at 375 V. Current oscillations with a large amplitude were observed and could be attributed to the tiny flashing sparks.

**Figure 3.** Electrical characteristic curves of single pulse anodizing under different potentials, (**a**) current transients and (**b**) *V*–*I* curve.

Figure 3b shows the dynamic *V*–*I* characteristic curves of single pulses, which is characterized by a loop from A to D along the marked direction. An abrupt current increase occurred due to a rapid initial voltage rise (path A-B), which was followed by a decrease in current to a lower level until the voltage reached a target value (path B-C). Next, a gradual and subtle change in voltage and current values were observed (path C-D). Finally, a sudden drop in the current decreases to zero was observed with voltage decreasing at a lower rate (path D-A). Shape and scope of these curves were affected by magnitude and change rate of the pulse potential. Higher potential resulted in longer loop line and larger loop area.

Similar current transients were observed for pulse voltages ranging from 325 to 525 V with a pulse duration of 500 μs (shown in Figure 4). With an increase in the pulse duration, differences among these pulses were prominent. For the pulse potential lower than 375 V, current transient was relatively smooth and decreased almost linearly with time. When the pulse potential was higher than 375 V, oscillation due to spark discharge was observed in the current transient due to dielectric breakdown of the coatings. For the same pulse potential, the effective value of the pulse current is inversely proportional to the pulse duration. This suggests the existence of a more reactive component in the overall current corresponding to the shorter pulses, due to the initial capacitor charging process.

**Figure 4.** Current transient curves for single pulse anodizing under different potentials.

#### *3.3. E*ff*ects of Pulse Parameters on Surface Morphology*

Surface morphology after single pulse anodizing varies for different pulse voltages. This reflects voltage-dependent discharge characteristics. Figure 5a shows that the surface of the micro-electrode before single pulse anodizing was relatively flat and only several knife marks were visible at a higher resolution. Surfaces of micro-electrodes after single pulse anodizing are shown in Figure 5b–d. Figure 5b shows the surface after anodization using 375 V/100 μs pulse. Similar surfaces were observed in Figure 5c,d, which show anodization surfaces for 425 V/100 μs and 475 V/100 μs, respectively.

**Figure 5.** Scanning electron microscope (SEM) images of micro-electrode surfaces, (**a**) before single pulse anodizing, (**b**) anodizing with 375 V/100 μs pulse, (**c**) anodizing with 425 V/100 μs pulse, (**d**) anodizing with 475 V/100 μs pulse.

Figure 6 shows the surface SEM images of the specimens after applying a 400 V pulse for different pulse widths. When shorter pulses of 100 μs and 500 μs were applied, circular discharge channels

with opened pores were found in addition to the groove-like channels, as shown in Figure 6a,b. These surface morphologies are similar to those shown in Figure 5. However, after breakdown, surface morphology for longer duration pulses was found to be significantly different, as shown in Figure 6c,d.

**Figure 6.** SEM images of micro-electrodes obtained in different conditions, (**a**) anodizing with 400 V/100 μs pulse, (**b**) anodizing with 400 V/500 μs pulse, (**c**) anodizing with 400 V/2 ms pulse, (**d**) anodizing with 400 V/5 ms pulse.

#### *3.4. Characteristic Parameters of Discharge Event*

Figure 7a shows the raw voltage and current waveforms during anodizing with 500 V/50 ms pulse. Multiple peaks and low baseline were observed in the current waveform. Each peak in the current was accompanied by a decrease in voltage, which recovered during the periods of low current, until the next series of peaks in the current. A magnified section of the current transient is shown in Figure 7b. The discrete nature of individual peaks in the current is evident here. In addition, some cascades appear as the superposition of two current peaks and these probably represent the occurrence of a second discharge event with the first one still ongoing.

**Figure 7.** Voltage and current waveforms acquired during anodizing with 500 V/50 ms pulse, at (**a**) low resolution (period of 40 ms) and (**b**) high resolution (period of 5 ms).

## **4. Discussion**

An inflection point can be clearly seen in the voltage-current characteristic curve, as shown in Figure 2b. According to our previous work [26], the voltage value of the inflection point can be considered as the critical breakdown voltage (*V*b). Similarly, the corresponding current is the critical breakdown current (*I*b). Increase in the current values with increasing voltage suggests breakdown of the anodic oxides occurs continuously for the whole duration. Although breakdown voltage is dependent on layer thickness and some other characteristics, this voltage-current characteristic curve provides a feasible method to identify breakdown voltage of growing oxide film during PEO.

There is a noticeable peak current at the beginning of every single pulse, which was not associated with discharges. This phenomenon appears to be consistent with a hypothesis of active–capacitive load behavior for the PEO process, which was studied systematically in literature [18]. The simple equivalent circuit of PEO load was composed of series-parallel capacitor and resistance, which can have a physical interpretation as the equivalent capacitance of the oxide film and the equivalent resistance introduced by the current passing through the discharges, respectively. The sharp decrease of current was due to the sudden increase of load impedance after the capacitor charging process. In Figures 3a and 4, current oscillations observed at 375 V reflected intense changes of load impedance, caused by dielectric breakdown of oxide film along with abundant randomly flashing sparks. In Section 3.1 and our previous work [26], the initial critical breakdown voltage of oxide film was about 350 V. So, we speculate that the voltage region for these oscillations is among 350–400 V.

It is assumed that electrons cannot flow through the oxide, so the field across it builds up as the applied voltage is raised, and it may reach the breakdown level for the oxide, at which point a discharge occurs [21]. So, the key issue is whether the electric field across the residual oxide film reaches the level necessary for dielectric breakdown. During PEO, the breakdown voltage of oxide film increases with growing layer thickness. In other hand, PEO process can be divided into several stages according to the evolution of the discharges [27]. At the early stage of the process (step 1), the growing oxide film

breaks down due to an increase in the applied voltage. This stage is characterized by sparks flashing randomly all over the aluminum alloy surface. Following this stage, sparks progressively change to micro-arcs (step 2). Finally, after the end of the rapid micro-arcs decrease, step 3 is characterized by a few strong remaining arcs that appear on the surface. The average lifetime of the discharges increased with increased voltage and processing time in the main period of PEO coating growth [8]. So, we try to establish the correlation between electrical characteristics and physical phenomena, including the oxide film and discharges. In the single pulse anodizing process, all voltage pulses were applied to the same thin initial oxide film. Pulse energies, which depend on the voltage, current and pulse duration, impact the dielectric breakdown of oxide film and discharge shape significantly. The higher pulse energy results in stronger dielectric breakdown strength and more powerful discharge behavior.

A comparison of the surface morphology in Figure 5a with Figure 5b–d reveals some distinct changes on the surface of micro-electrodes. Dielectric breakdown due to the single voltage pulse produced a number of isolated discharge channels on this anodic oxide film. Circular opened discharge channels, tens to a few hundreds of nm in size, were found to be surrounded by a band of evidently thicker film material. In addition, large surface regions without discharge pores were also observed. The quantity and size of opened discharge pores increased with an increase in potential. This occurs because with higher electric field strength applied to the oxide film, a larger driving force is available for tunneling ionization. Furthermore, some groove-like discharge channels, which consist of numerous opened and closed discharge pores, were found on the surface, as shown in Figure 5d. Both pores and discharge channels were randomly distributed on the metal surface. This implies the irregular formation of barrier-type anodic oxide films. When numerous narrow pores overlapped in a small region, a groove-like discharge channel was formed. In some case, a few discharge pores were found to be coated with oxide produced by the repeating plasma discharge events during pulse anodizing, resulting in closed pores.

The surface morphology was also found to be dependent on pulse width to some extent. As shown in Figure 6c,d, it can be seen that the entire surface was covered with a thicker layer of anodic oxide film, modified by dielectric breakdown. The oxide surface was rough and a number of bright spots with sizes of ~1 μm or less were observed. Electric field strength between the electrolyte and metal increased with time due to continuous electron transfer. This higher stored energy makes it easier to breakdown the film. Moreover, temperature increase in the groove-like discharge channel regions could result in transforming amorphous alumina to crystalline, which usually has a higher chemical stability [22]. The closure of opened discharge pores appears to be more obvious with increasing pulse width in comparison to pulse magnitude. For both 2 and 5 ms pulse widths, numerous discharge channels were found on the surface. The size of these discharge clusters (circled in Figure 6c,d) on the oxide surface is proportional to the pulse width and an increased size was observed for longer pulse widths.

Some characteristic parameters can be obtained from Figure 7. However, di fficulty arises with the identification of the start and end of each discharge event. Using the method mentioned in literature [16], we hypothesized that a peak current (*I*p) and a bottom current (*I*b) represent a start and end of discharge event respectively. Two current thresholds were set to estimate the total number of discharge events, 0.2 and 0.15 A. When the current exceed 0.2 A, it can be considered as a peak current of a discharge event. When the current value is lower than 0.15 A, it can be considered as the baseline current of a discharge event. Discharges were considered to initiate when the current reaches the peak and terminates when it reaches the baseline current.

Current level (*I*event), duration (*t*d), diameter ( ϕ), and spatial distribution ratio (Dn) of discharge can be estimated using Equations (1)–(6):

$$\mathbf{N}\_{\text{total}} = \frac{1}{2} (\mathbf{N}\_{\text{P}} + \mathbf{N}\_{\text{b}}) \tag{1}$$

$$I\_{\text{event}} = \sum\_{\mathbf{n}=0}^{N\_{\text{total}}} \frac{1}{\mathbf{N}\_{\text{total}}} (I\_{\text{P}} - I\_{\text{b}}) \tag{2}$$

$$t\_{\mathsf{gap}}^{\mathsf{n}} = \left| t\_{\mathsf{P}}^{\mathsf{n}} - t\_{\mathsf{b}}^{\mathsf{n}} \right| \tag{3}$$

$$t\_{\rm d} = \sum\_{n=0}^{\mathcal{N}\_{\rm total}} \frac{1}{\mathcal{N}\_{\rm total}} t\_{\rm gap}^n \tag{4}$$

$$\mathbf{D}\_{\rm n} = \mathbf{N}\_{\rm total} / \left( \mathbf{S}\_{\rm area} \cdot t\_{\rm pulse} \right) \tag{5}$$

$$\varphi = 2\sqrt{\frac{S\_{\text{area}}}{(\mathbf{D\_n} \cdot \boldsymbol{\pi})}} \tag{6}$$

In Equations (1)–(6): Np, Nb and Ntotal represent the number of *<sup>I</sup>*p, *I*b and total discharge events respectively; *<sup>t</sup>*gap represents the time interval between adjacent peak time (*t*p) and bottom time (*t*b); *S*area represents the superficial area of sample; *<sup>t</sup>*pulse represents the pulse duration.

To eliminate errors due to experimental deviation, a sample size of ten for the same pulse anodizing was carried out. All the characteristic parameters (shown in Table 1) were obtained by calculating the average value of the results of these experiments.

**Table 1.** Estimated characteristic parameters of individual discharge events.


Although the characteristic parameters listed in Table 1 are similar to those results reported in section '1. Introduction', experimental methods and monitoring techniques need to be improved to acquire more accurate data. In addition, a larger sample size of experiments should be carried out to confirm these results.
