Understanding Electric Current Effects on Tribological Behaviors of Instantaneous Current-Carrying Pair With Recurrence Plot

Electromagnetic launch [1] is a new technology for linear propulsion, which has attracted much attention in high-energy weapons [2]. An electromagnetic launcher consists of the armature and rail. Unlike traditional electric tribopairs such as a pantograph-catenary system [3], armature–rail electrical contact is an instantaneous current-carrying friction, i.e., a sliding electric-mechanical impact friction [4] and transition-induced [5] arc erosion on a millisecond time scale. The service quality of electromagnetic energy equipment is dominated by the armature–rail tribo-electrical performance [6]. Hence, the instantaneous current-carrying friction behavior and wear mechanism of the armature and rail materials are interesting topics.

Al and Cu and their alloys have excellent electrical properties with high conductivity and good mechanical properties, which are especially suitable for manufacturing high-current current-carrying friction pairs. Hence, Al and Cu and their alloys are often used as the armature and rail materials [7]. Previous studies have primarily focused on contact parameters affecting the friction wear behaviors of Cu-based or Al-based materials. For experimental research, Wang et al. [8] selected Al 7075 and CuCrZr as the armature and rail materials and conducted the block-on-pin current-carrying wear tests with 40 and 80 min. They found that the friction coefficient decreases, and the material transfer intensifies with the friction time. Balic and Blanchet [9] believed that the armature–rail interface transition negatively correlates with the contact load. Yang et al. [10] performed the Cu-C current-carrying friction tests on a ring-on-block tribometer under high-speed and large-current conditions (350 km/h, 250 A, and 10–60 min). The results indicated that arc discharge is a crucial cause of delamination wear. Siopis and Neu [11] observed that the melt wear on the current-carrying friction surfaces of Al 6061 and Cu C11000 became severe with the increasing load and velocity. Wu et al. [12] found that the contact resistance of CuCrZr alloy material decreases with increasing current and load. Lin et al. [13] conducted the Cu-Al current-carrying friction tests with 20 s on a pin-on-disk tribometer at 50 and 150 A currents. They confirmed that the surface roughness and friction coefficient are lower at 150 A current, and the intermetallic compounds promote Al adhesion. For numerical simulation research, the reports have primarily focused on current-skin effect [14], velocity-skin effect [15], and low-load transition [16]. Research has confirmed that the current level dominates the armature–rail contact load and velocity [17]. Therefore, the current is a crucial factor affecting armature–rail electrical contact behaviors. However, it is unclear how the electric current affects instantaneous current-carrying friction behavior and wear mechanism with the armature and rail materials at multifriction cycles. Specifically, in previous studies, the time scale of current-carrying friction tests based on pin-on-disk tribometer, etc., was usually minutes or hours. However, the results on a millisecond time scale were lacking. The armature–rail sliding electrical contact duration is typically milliseconds (about 10 ms), which is determined by the millisecond-level current loading duration. Unlike the long and steady friction of conventional sliding electrical contact systems, the armature–rail instantaneous current-carrying friction process is typically high impact, and the friction interface exhibits transition, arc erosion, and material melting phenomena. However, armature–rail tribological test methods still utilize prolonged friction (time scales of minutes and hours) by the conventional current-carrying tribometer. Therefore, focusing on tribological test results of instantaneous current-carrying pair on a millisecond time scale makes sense. Besides, there were few reports on the influence of electric current, and the multifriction cycle tests when considering multilaunch conditions.

The friction-induced signals and the wear topographies are essential information for revealing friction behavior and wear mechanism, respectively. The friction behavior results can not only guide surface design but also support data-driven online health assessment of tribopairs [18,19], such as wear state recognition and prediction based on the friction vibration, friction force, etc. The rail wear faults are incredibly harmful. Therefore, the predictive maintenance of the rail is as essential as the surface design for extending the service life of the electromagnetic launcher. The previous research on rail wear prediction was mainly based on the highest or average value of the friction-induced signals (contact resistance [20], voltage [21], etc.). However, the friction-induced signals inevitably exhibit nonlinear characteristics due to their strong dependence on chaotic tribological systems [22,23]. Hence, the linear analysis methods based on friction-induced signals may be unreliable for predicting wear behavior [24]. To our knowledge, it is more effective to apply the characteristic quantifiers of the nonlinear dynamic system state to identify or predict the wear state, after reconstructing the friction-induced signals into high-dimensional phase space [25–27]. This fact motivates us to further reveal the system state of the friction-induced signals under instantaneous current-carrying in this study. The results enrich the friction behavior results and provide a new reference for the predictive maintenance of the rail.

The challenge to reveal the system state of the friction-induced signals under instantaneous current-carrying is twofold: the nonlinearity [22] and short-time nature [28] of signals on a millisecond time scale. Correlation dimension [29], Lyapunov exponent [30], and Kolmogorov entropy [31] are often used as the dynamic quantifiers of the nonlinear system. They have been preliminarily applied for wear state monitoring [25] and so on. However, the aforementioned quantifiers have significant calculation errors for the short-term signals [32]. A recurrence plot (RP) [33] is a graphical tool for analyzing nonlinear time series and can intuitively represent the state of complex systems. Especially for the short-time signals, RP can accurately detect the characteristic patterns hidden in the high-dimensional phase space [34]. Recurrence quantification analysis (RQA) [35] quantifies the system state reflected by the RP structures. Thus, RP and RQA quantizers have been widely applied to analyze the nonlinear system state of short-term mechanical signals. The results have been successfully used to predict the angular contact ball bearing damages [36], evaluate surface polishing performance [37], and so on. Here, we used RP and RQA quantifiers to reveal the system state of the instantaneous current-carrying friction signals.

The wear topography evaluates the wear surface performance. Traditional statistical parameters, such as Sa and Sq, are often used to characterize the wear topography [38]. However, the aforementioned parameters are scale dependent, i.e., they depend on the sampling length and resolution of the instruments [39]. Besides, the nonlinear properties [40] of the wear topography may also affect the calculation results of the aforementioned parameters. The wear topography has been proven to have fractal characteristics [41]. The multifractal spectrum [42] is a scale-independent fractal method; that is, it does not vary with the sampling length and resolution, and hence, it is widely used to analyze complex wear behavior [43,44]. In this study, we described the topography nonlinear characteristics via the multifractal spectrum.

This study aimed to reveal the current (50–300 A) effects on instantaneous current-carrying friction behavior and wear mechanism at multifriction cycles. First, we developed a unique instantaneous current-carrying tribometer and conducted relevant tests. Second, after analyzing the frictional sound pressure (FSP) time-domain variations with the increasing current, the corresponding system state evolutions were further revealed via RP and RQA quantifiers. Finally, the wear mechanism transition with the current was analyzed, and the wear topography was comprehensively characterized by the multifractal spectrum and surface roughness Sa. The correlation between the RP patterns and RQA quantifiers and the wear damage state was established.

We developed a unique tribometer to imitate the instantaneous current-carrying friction, see Fig. 1(a). The tribometer comprised a ball-plate instantaneous friction pair, DC power supply, and monitoring system. The ball-plate instantaneous friction pair contained a ball, weight, rotating rod, bearing assembly, and plate. The rotating rod had a weight and a ball stabilized by the ball holder at one end. The other end of the rotating rod was connected to the bearing assembly. The plate was mounted on the disc surface by a plate holder insulated with a base.

Figure 1(b) depicts the working principle. First, the gravity of the weight drove the rotating rod to carry the ball and accelerate its rotation downward from a stationary state (stage I). Then, the instantaneous current-carrying friction and transition arc occurred between the ball and the plate (stage II) on a millisecond time scale. Finally, the friction ended when the ball and plate were separated (stage III). The rotating rod was fixed temporarily by the snap before stage Ⅰ and after stage III. According to Hooke's law, we could adjust the friction load F_n by changing the spring size and material, as well as presetting the spring compression amount at stage II. The equation for friction speed was $v=2gh$⁠, where g = 9.8 m/s² was the gravitational acceleration, and h was the initial lifting height between the ball and plate before stage Ⅰ. Hence, we could control v by adjusting h.

A DC power supply could provide a controllable current from 1 to 500 A during ball-plate instantaneous contact. The current-carrying circuit sequentially passed through the positive pole of the DC power supply, conductive base, plate, ball, rotating rod, bearing assembly, and negative pole.

A monitoring system consisted of the AWA14425 noise sensor, NI 9234 high-frequency acquisition card, and daq software. The friction coefficient was complex to measure accurately on a millisecond time scale because friction load was usually measured by a contact force sensor. For contact measurement, high-frequency signals might experience severe attenuation when considering factors such as sensor installation and large mechanical structures. However, the FSP signal was often measured in a contactless way. This measurement way could accurately reflect the high-frequency motion information of tribopair in a wide frequency response range.

There was a positive correlation between the FSP and the friction coefficient. Ding et al. [32] measured the friction coefficient and friction-induced sound pressure level (FSPL) of GCr15 and AISI 1045 during wear. They found that the friction coefficient and FSPL are positively correlated, i.e., they follow a simultaneous time-varying pattern. According to the mathematical relationship between sound pressure level (SPL) and sound pressure (SP), i.e., SPL = 20log₁₀(SP/P_e), where P_e = 20 μPa is the reference SP, it is clear that SPL and SP have a strictly positive relationship. Therefore, the friction coefficient and FSP are positively correlated. The friction process can be understood as the behavior of relative contact, collision, and shearing of several asperities on two surfaces. Theoretically, two asperities in a crash can generate shock waves of equal size and opposite directions. The superposition of many shock waves creates a fluctuating friction force, which excites friction vibration, and then generates friction noise that can be quantitatively characterized by FSP. Therefore, when the surface is rougher, the friction system may excite more significant friction vibration and friction noise, i.e., higher FSP, accompanied by a higher friction coefficient. Hence, the positive correlation between friction coefficient and FSP has an explainable physical reason. In this work, it is clear that the instantaneous current-carrying friction surface may be jointly supported by asperities and molten metals with different proportions (see Figs. 8 and 9). From the aforementioned analysis, the positive correlation between the friction coefficient and the FSP applies to the asperity contact part. For the molten metal loading part, the correlation still applies. This is because the molten metals reduce the friction interface shear, which reduces the friction coefficient and friction vibration and subsequently reduces friction noise. Thus, we acquired the FSP to characterize the friction coefficient.

The ball (armature) and plate (rail) materials are Al 1060 and Brass H62, respectively. The performance of the sample materials is listed in Table 1, where IACS refers to the international annealed copper standard. The diameter of the ball was 12 mm, and the plate was a cuboid with dimensions of 180 × 80 × 3 mm. The initial roughness of all samples was Ra = 0.2 ± 0.05 μm.

This work aimed to reveal the electrical current effects on instantaneous current-carrying tribological behaviors at multifriction cycles. The test arrangements are listed in Table 2. The current level was set as 50 A, 100 A, 150 A, 200 A, 250 A, and 300 A. When pre-experiments were carried out based on the aforementioned currents, the instantaneous current-carrying wear surfaces showed typical features of armature–rail friction, such as transition-induced arc erosion and deposited layer (see Figs. 8 and 9). Therefore, we established 50–300 A currents. The friction load and speed were adjusted according to the working principle in Sec. 2.1. The spring was made of 65Mn steel, with wire diameter × pitch diameter × original length = 1.6 × 10 × 23 mm. We set that the spring original length was compressed to about 21 mm when the ball and plate made contact. The h was set as 850 mm. Hence, the load and speed were approximately 37 N and 4 m/s, respectively, which were constant for all tests. When considering multilaunch conditions, 10 and 20 friction cycle tests were performed, respectively. Note that stages I–III in Fig. 1(b) are treated as one friction cycle between the ball and the plate. Multifriction cycles refer to the repeated operation of one friction cycle. In addition, the armature was replaced after each launch in practice, but the rail was reused. Therefore, we set that the ball was replaced, but the plate was not after each friction cycle. All tests were carried out at 20 ± 5 °C room temperature and 50 ± 4% relative humidity.

To entirely obtain the FSP signal during ball-plate instantaneous current-carrying friction at each friction cycle, the time-domain for signal acquisition was set from stages I–III (Fig. 1(b)), with a sampling frequency of 51.2 kHz. Only stage II could generate the FSP signal reflecting the ball-plate friction. The ball and plate were noncontact at stages I and III, meaning that the FSP signals of these two stages were ineffective. Therefore, the FSP signal, acquired from stages I–III at each friction cycle, was intercepted to retain only the stage II signal. The interception range was set to the time-domain of signal mutation. We stipulated that the intercepted stage II signal was considered the FSP signal for each friction cycle. The pretests indicated that the FSP signal length at each friction cycle was about 1500 points, and the corresponding friction time was approximately 30 ms, thus meeting the test requirement. For the convenience of analysis, for any group of tests in Table 2, the FSP signal at each friction cycle was spliced chronologically, and the signal after splicing was treated as the FSP signal of this group of tests. Therefore, the signal lengths of 10 and 20 friction cycles were 15,000 and 30,000 points, respectively.

The tested plates were washed in anhydrous ethanol. The worn topography and surface roughness Sa were obtained by VK-X1000 laser scanning microscope. The elements of the worn surface were analyzed by Gemini 500 thermal field emission scanning electron microscope.

References [34,37,47] described a detailed relationship between the RP structures and the system states. The main microscopic structures of RP contain black dots, white dots, and diagonals parallel to the main diagonal, respectively. The higher the distribution uniformity of black-white dots, the weaker the nonstationarity of the system state. The number of black dots is often negatively correlated with the system state complexity. The number of black dots and the distribution uniformity of black-white dots should be comprehensively considered. A diagonal structure is displayed in RP when the phase-space trajectory of the time-domain signal visits the same phase-space area at different times. The longer and more diagonal lines, the stronger the certainty and the weaker the unpredictability of the system state.

The three primary macroscopic RP patterns contain homogeneous pattern, drift pattern, and disrupted pattern. A homogeneous pattern comprises many isolated and uniformly distributed black-white dots, meaning the prominent system state complexity and weak nonstationarity. Simultaneously, the isolated dots often cause diagonals to break, increasing the system state unpredictability. When the system state is progressive, RP is a drift pattern with the evolution of the dot-line structures along the main diagonal. The drift pattern represents that the system state may be shifting, implying that two types of RP patterns are transitioning. A disrupted pattern RP shows a concentration of the white or black areas, representing a nonstationary system state. Besides, the aggregation of dot structures may lead to more extended diagonal structures, reducing unpredictability and complexity. The system state should be comprehensively analyzed via the microscopic and macroscopic structures of RP.

The FSP signals under different currents at ten friction cycles are exhibited in the left column of Fig. 2. It can be observed that the FSP amplitude is relatively high at 50 and 100 A. As the current enhances from 150 to 300 A, the FSP amplitude decreases gradually. A positive correlation exists between the friction coefficient and the FSP [32]. Thus, it can be confirmed that a higher current helps reduce and stabilize the friction coefficient; that is, higher currents exert a more significant lubricating effect at the instantaneous current-carrying friction interface. The aforementioned results also appeared at 20 friction cycles.

The FSP absolute value under different currents was obtained, and the corresponding probability distributions were calculated, see the right column of Fig. 2. It can be seen that the FSP absolute value is concentrated in the range of 0–10⁻³ Pa. Figure 3 displays the FSP absolute value probability between 0 and 10⁻³ Pa under different currents. Whether it is 10 or 20 friction cycles, the probability first increases rapidly with increasing current, and then the increasing trend slows down. Note that the higher the probability, the more the part of absolute values of FSP is in the range of 0–10⁻³ Pa, and the less the part is above 10⁻³ Pa, representing a smaller FSP amplitude. FSP and coefficient of friction have a positive correlation. Hence, the probability results quantitatively confirm that a higher current promotes the interface lubrication effect.

For the instantaneous current-carrying friction, the rise in interface temperature is dominated by arc heat. As the current increases, the frequency and energy of the arc enhance, resulting in higher arc heat. Thus, the friction surface materials are gradually softened or melted with increasing current. Based on Sec. 5.1, for the lower current levels (50 and 100 A), the softening and partial melting of the interface materials leads to a weakness in the fracture resistance of asperities, thereby reducing the FSP amplitude and friction. As the current increases (150–300 A), more molten metal further enhances lubrication at higher currents by acting as an interface lubricant. The molten material at the interface may lead to a solid–liquid–solid (ball–molten metal–plate) contact state, which reduces the friction interface shear and subsequently reduces the friction coefficient and enhances the lubrication effect. However, the current lubricating effect may not be unlimited because of the more significant electrical wear in a larger current, such as arc erosion pits, molten topographies, Al adhesion layers, and oxide layers (as confirmed in Figs. 8 and 9).

According to Fig. 3, whether 10 or 20 friction cycles, the current-induced lubrication effect at the friction interface increases rapidly before 200 A. Then, the increasing trend slows as the current increases from 200 to 300 A. According to the wear topographies in Figs. 8(a) and 9(a), it can be found that the friction interface lubrication effect may be caused by a combination of a large area of current-induced softened asperities and a small area of molten aluminum under the smaller current of 50 A. These softened asperities and molten aluminum reduce the interfacial shear during friction, resulting in antifriction and lubrication effects. As the current continues to enhance up to 200 A, the amount of molten aluminum increases rapidly to the point where it nearly dominates the instantaneous current-carrying friction interface. These melts may further reduce the interfacial shear; thus, the lubrication effect at 50–200 A rapidly increases. When the current ranges from 200 to 300 A, although more molten aluminum continues to be generated, their ability to promote interface lubrication may be limited. When the friction interface is filled with abundant molten aluminum, the continued improvement in the lubrication effect may depend more on the molten material properties (such as viscosity) rather than quantity. Therefore, the increasing trend of lubrication effect slows down from 200 to 300 A.

Figure 3 also illustrates that compared to 10 friction cycles, the current lubrication effect at 20 friction cycles is more potent at 50 and 100 A. Mechanical wear may dominate the surface state under low currents, and more friction cycles enhance the adaptability of asperities, similar to the running-in effect. The lubrication effect at 10 and 20 friction cycles is close when the current varies from 150 to 300 A.

RPs of the FSP signals were constructed using Eqs. (1)–(3) to further reveal the evolution of the system state with the current. The FSP signals at 10 and 20 friction cycles were resampled by 1/5 and 1/10, respectively. Figure 4 shows RPs under different currents at ten friction cycles. The coordinate lengths of RPs for each group of tests are different due to the difference of the m and τ.

RP at 50 A is a disrupted pattern, as shown in Fig. 4(a). RP has black dots with high aggregation (region A1). The distribution of black-white dots is highly uneven. The black-white area boundary is evident. It indicates that the phase trajectory of the FSP signal varies dramatically, and the corresponding friction system has strong nonstationarity. In addition, it can be observed that RP has many long diagonals (solid lines with arrow, partially displayed) parallel to the main diagonal (dashed line with arrow). This structure indicates the strong certainty and weak unpredictability of the system state. Besides, many black recurrence dots suggest a more similar system state. It can be concluded that under 50 A current condition, the nonstationarity of the system state is strong, and the unpredictability and complexity are weak.

Figures 4(b) and 4(c) depict that the RPs at 100 and 150 A are drift pattern. Under 100 A current condition, there is a significant difference in the black-white dot clustering and distribution between the bottom-left and top-right corners of RP, as shown in Fig. 4(b). Besides, region A2 is a disrupted pattern, but region B1 is a nearly homogeneous pattern. The RP structures of region A2 tend to evolve to the region B1. Compared to 50 A current condition, the dispersivity and uniformity of black-white dots in RP at 100 A are improved. Thus, the clarity of the black-white area boundaries decreases. In addition, the length and number of diagonals and the number of black recurrence dots are reduced. As depicted in Fig. 4(c), the aforementioned RP structures at 100 A are further inherited and enhanced accordingly in RP at 150 A. The aforementioned results indicate that compared to 50 A current condition, the nonstationarity of the system state at 100 and 150 A is weakened, and the unpredictability and complexity continue to enhance.

RPs at 200–300 A present a homogeneous pattern, as shown in Figs. 4(d)–4(f). Figure 4(d) shows that the length and the number of the diagonal of RP at 200 A are significantly reduced compared to 50–150 A. Thus, the diagonals are discretized into more dot structures. The aforementioned results further enhance the dispersivity and uniformity of black-white dots. The black-white border is particularly blurred. Meanwhile, the number of black recurrence points is significantly reduced. The RP structures at 200 A are further amplified correspondingly in RPs at 250 and 300 A, as depicted in Figs. 4(e) and 4(f). In short, when the current varies from 200 to 300 A, the system state tends to stabilize gradually, and the complexity and unpredictability are further enhanced.

Figure 5 shows the RPs under different currents at 20 friction cycles. Figure 5(a) shows that RP at 50 A is a disrupted pattern, such as high clustering black dots in area A1. Figure 5(b) indicates the RP at 100 A is a drift pattern. The disrupted pattern in region A2 gradually shifts to the homogeneous pattern in region C1. It can be observed that RP at 150 A still presents a drift pattern, as shown in Fig. 5(c). A smaller area (region A3) is a disrupted pattern, while a larger area (region C2) transitions toward a homogeneous pattern. Figures 5(d)–5(f) illustrate that RPs are a homogeneous pattern with the current variation from 200 to 300 A. The aforementioned results imply that as the current increases, the nonstationarity of the system state weakens, and the complexity and unpredictability enhance. The evolution law of the system state with the current is consistent for 10 and 20 friction cycles.

The aforementioned analyses confirm that as the current increases from 50 to 300 A, RPs exhibit disrupted, drift, and homogeneous patterns in turn. This change signifies that as the current increases, the nonstationarity of the system state weakens, and the unpredictability and complexity enhance. The various system states, generated under different currents, can be effectively identified by RPs. Meanwhile, the wear damage of the plate intensifies with the current (as confirmed in Sec. 5.1). The results imply that the RP patterns correlate with the surface damage state.

RQA, a method for quantifying the nonlinear system state, was conducted by calculating RR and DET of RP using Eqs. (4) and (5), respectively. As shown in Fig. 6(a), RR decreases monotonically as the current increases from 50 to 300 A at 10 and 20 friction cycles. This change indicates that the complexity of the system state is gradually enhanced with the current. Figure 6(b) shows DET under different currents. DET monotonically decreases with the increasing current at both 10 and 20 friction cycles, meaning an enhanced unpredictability of the system state. Besides, compared to 10 friction cycles, RR and DET at 20 friction cycles are lower under the same current, indicating higher unpredictability and complexity. To sum up, as the current increases from 50 to 300 A, the complexity and unpredictability of the friction system increase monotonically. The result validates the aforementioned analysis results of RPs from a quantitative perspective.

An interpretable physical connection exists between the system state and the wear behavior. A higher current promotes increased electrical wear characteristics and surface roughness (as confirmed in Sec. 5.1). When rough surfaces engage in friction motion, the interface electric field strength variability is triggered by the dynamic properties of the gap, pressure, and shear between asperities. Thus, arc behavior may be stochastic, such as the randomness of the arc root, arc energy, and arcing rate. The above factors may lead to the uncertainty of the wear surface damages. The randomness of the arc behavior and damage is usually positively correlated with the surface roughness [50]. Besides, the system state in the friction-induced signals is closely related to the wear surface state [25]. Hence, more significant randomness, unpredictability, and complexity of the system state were exhibited with increasing current. In addition, high current generates abundant molten Al, enhancing interface lubrication and friction-reducing effects. This may be the reason why the nonstationarity of the system state weakens with the current.

The aforementioned analyses show that as the current increases from 50 to 300 A, RR and DET monotonically decrease, and the plate surface wear intensifies (as confirmed in Sec. 5.1). In other words, the wear surface damage negatively correlates with RR and DET. It can be concluded that the RP patterns, RR, and DET can characterize the instantaneous current-carrying wear damage state. The conclusion may provide new driving data for the rail online wear prediction and identification. In addition to analyzing the FSP signals in the time-domain, this section further reveals the evolution law of the system state with increasing current, which enriches the instantaneous current-carrying friction behavior results.

Figure 7 depicts the scanning electron microscope (SEM) and the energy dispersive spectroscopy (EDS) images of the plate surface before and after test 2. Before testing, the smooth surface contained much and little oxygen, as shown in Fig. 7(a). After testing, abundant molten Al occupies the surface, and the oxygen content increases to 17.5%, as depicted in Fig. 7(b). At the same time, it can be confirmed that the main oxidized object is Al because of the high coincidence of Al and oxygen EDS images. The results indicate that the ball melts during the instantaneous friction, and the molten Al materials are condensed and deposited as tribolayers (Al adhesion and oxide layers) on the plate surface.

Figure 8 presents the wear topographies of the plates under different currents at ten friction cycles. Under 50 A current condition, Fig. 8(a) shows that the upper left corner of the surface has a slight Al adhesion with a large area and small thickness. In other regions, Al adhesion layers are small and dispersed. There are also many pits and grooves that cross the view field. When the current is 100 A, it can be seen that the area and thickness of Al adhesion layers further increase, and the slight black oxidation traces occur on the lower right surface region, as depicted in Fig. 8(b). Figure 8(c) shows that many arc erosion pits begin to appear on the surface at 200 A. In addition, the area of the oxide layers continues to increase. The area and thickness of Al adhesion layers further increase to almost cover the entire surface. After entering 300 A, the wear topography has significantly changed, as shown in Fig. 8(e). Delamination, a manifestation of fatigue wear, can be observed to occur on the Al adhesion and oxide layers of the right wear surface. More arc erosion pits and molten Al cover the left area of the surface. The thickness and area of Al adhesion layers continue to increase to cover the entire surface. Besides, there are many randomly distributed cracks on the surface.

Figure 9 shows the wear topographies of the plates under different currents at 20 friction cycles. The wear topography at 50 A is relatively smooth, with only slight Al adhesion and grooves, as shown in Fig. 9(a). As the current increases to 200 A, the surface electrical wear characteristics, such as Al adhesion, oxidation, ablation pits, and cracks, continue to intensify in degree and coverage area, as shown in Fig. 9(c). Under 300 A current conditions, delamination occurs on the surface based on inheriting and boosting the electric wear topography at previous currents, as depicted in Fig. 9(e). The variation law of the wear topography with increasing current is consistent at 10 and 20 friction cycles. In addition, under the same current, the electrical wear characteristics are more significant at 20 friction cycles.

The aforementioned results confirm that the wear topography changes significantly as the current increases from 50 to 300 A. This change signifies the wear mechanism transition from abrasive wear and slight adhesive wear to arc ablation, fatigue wear, and severe adhesive wear. The discussion on the wear mechanism transition is presented in Sec. 5.2.

The surface roughness Sa is calculated by selecting stochastically four regions on the plate wear surface after each group of tests in Table 2, as shown in Fig. 10. It can be found that as the current enhances from 50 to 300 A in general, the value of Sa gradually increases whether it is 10 or 20 friction cycles, implying intensified wear damage. The Sa results are consistent with the wear topography variation with increasing current. Arc heat may be the primary cause of surface damage. As the current is enhanced, the frequency and energy of the arc at the friction interface increase, leading to higher arc heat. This may increase the probability of surface damage, such as arc ablation, adhesion, and material melting, which leads to an increase in Sa.

Whether 10 or 20 friction cycles, Sa shows a gradual upward trend overall with the increasing current. However, a decrease in Sa at 250 A compared to 200 A disrupts the overall trend of Sa. This may be attributed to the randomness of the instantaneous current-carrying friction system. The recurrence analysis of the FSP shows that the unpredictability of the instantaneous current-carrying friction system state increases with the increasing current, i.e., the randomness increases. Particularly at higher currents of 250 A and 300 A, DET is less (see Fig. 6), which implies higher randomness of friction system state. One of the reasons for the higher randomness of the friction system at high currents may be the randomness of the arc behavior (see Sec. 4.3), and the second may be the instantaneous current-carrying friction process with impact. For example, the friction interface at 250 A is filled with a large amount of molten aluminum (see Figs. 8(d) and 9(d)). When impact friction is applied, the molten aluminum may splash or rupture during the loading process with a random behavior. Given this randomness, interfacial molten aluminum may fill previous ablation pits, intensify surface ablation, or both. Therefore, the wear topographies at higher currents may exhibit stronger randomness. This may be the reason for the decrease in Sa at 250 A compared to 200 A.

In addition, at 20 friction cycles, although Sa at 200 and 300 A have overlapping errors, the surface topography has some differences. The interfacial molten aluminum may have a random distribution due to the instantaneous friction with impact. After the molten aluminum cools and solidifies, the wear topography may exhibit ablation pits, covering deposits, or both, all of which lead to changes in surface height. This may be the reason why Sa is similar at 200 and 300 A, but the wear topography has some differences.

Wear topographies of the balls after the first friction at ten friction cycles are shown in Fig. 11. Figure 11(a) illustrates that the ball surface at 50 A has slight arc ablation features, such as minor protrusions and pits. When the current increases to 100 and 200 A, more severe ablation topography, including larger craters and protrusions, is observed in Figs. 11(b) and 11(c). Under 300 A condition, it can be found that the ball surface has significant ablation craters and a more comprehensive range of protrusions, i.e., the damage is more severe. The averaged Sa of any four areas of the worn ball surface at 50 A, 100 A, 200 A, and 300 A is 33.9 μm, 36.8 μm, 51.4 μm, and 57.4 μm, respectively. The ball damage is more severe than the plate, possibly because the ball (Al 1060) possesses lower mechanical properties (see Table 1), such as hardness, than that of the plate (Brass H62).

Figure 12 shows the schematic of the wear mechanism transition with increasing current. The wear mechanism transition may be attributed to the rise in interface temperature. The arc heat dominates the interface temperature rise of the instantaneous current-carrying friction. The higher the current, the higher the friction interface temperature. The materials near the friction interface soften or partially melt at 50 and 100 A (as confirmed in Figs. 8 and 9). Under the shearing action of multifriction cycles, abrasive wear is caused by the material fragments. The softer material properties of Brass H62 and Al 1060 may determine the mildness of abrasive wear. Under the instantaneous impact load, the contact stress between asperities may be greater than the material flow strength. Therefore, the Al material undergoes mechanical transfer to the plate surface under lower currents (50 and 100 A), resulting in slight adhesive wear. In addition, the smaller currents also cause slight electrical wear and Al adhesion. In short, the wear mechanism at lower currents is mainly slight adhesive wear and abrasive wear.

When higher currents (150–300 A) are applied, the flash temperature phenomenon becomes more significant at high arc energy. Molten Al significantly increases at the friction interface, manifesting as a larger area and thicker Al adhesion layer. Thus, adhesive wear intensifies with the current. On the other hand, large currents exacerbate the surface oxidation. Under the shear action of multifriction cycles, the brittle oxide layers are prone to delamination and fragmentation. In addition, the mechanical and physical properties of the friction materials deteriorate in a high-temperature environment. High thermal stress promotes the initiation and propagation of the cracks. The aforementioned factors may further exacerbate the delamination and fragmentation of tribolayers. Hence, the higher currents induce fatigue wear. Delamination, fragments, and cracks are critical causes of wear failure. In addition, for enhanced current conditions, the high arc energy and temperature increase leads to many erosion pits and splashing topographies on the contact surface. The aforementioned analyses confirm that arc erosion, fatigue wear, and severe adhesive wear dominate the wear surface damages at larger currents.

The scale-independent multifractal spectrum was utilized to further describe the nonlinear characteristics of the wear topography. The multifractal spectrum has an inverted “U” shape. The topography has multifractal characteristics when the multifractal spectrum presents a unimodal convex shape. The opening width of a multifractal spectrum is referred to as the spectrum width (Δα) [51]. The Δα quantifies the degree to which a fractal object is unevenly distributed on a probability measure. Here, we used the Δα to describe the fluctuation and irregularity of the wear topography. The greater the Δα, the stronger the fluctuation and irregularity of the topography [52,53].

Based on the wear profile, the multifractal spectrum is often calculated using box-counting method [52]. The weight factor is generally set to −80 to 80 with an interval of 1. For the wear topography after each group of tests in Table 2, four wear profiles were randomly selected, and their multifractal spectra are calculated. Figure 13 shows the multifractal spectrums for one of the four wear profiles under different currents. The unimodal convex shape multifractal spectrums indicate that the instantaneous current-carrying wear topographies have multifractal characteristics. Figure 13(a) shows the multifractal spectra under different currents at ten friction cycles. It can be observed that the spectrum width at 50 A is smaller. After that, as the current enhances from 100 to 300 A, the spectrum width continues to increase. Similarly, the spectrum width gradually increases with the current at 20 friction cycles, see Fig. 13(b). The aforementioned results qualitatively illustrate that the irregularity and fluctuation of the topography increase with the current.

The Δα under different currents is shown in Fig. 14. The evolution law of the Δα with the increasing current is consistent for both 10 and 20 friction cycles. As the current increases from 50 to 300 A, the Δα increases monotonically. This change indicates that the topography fluctuation and irregularity continually grow with the current. The quantitative results of the Δα are consistent with the qualitative analysis of the multifractal spectra. The reason is that the higher the current, the higher the arc energy at the friction interface. Thus, more ablative pits, oxide layers, and Al adhesion layers cover the wear surface. The factors may lead to severe the topography fluctuation, increase the surface roughness Sa, and enhance the topography fluctuation and irregularity. The Δα is positively correlated with the Sa, i.e., positively correlated with wear surface damage.

In addition, the Δα at 20 friction cycles is wider than that at ten friction cycles at the same current. The result indicates that the more friction cycles, the stronger the topography irregularity and fluctuation. After comparing Figs. 10 and 14, it can be found that the Δα error is less than that of the Sa at the same current. This is attributed to the scale independence of the multifractal spectrum.

In this study, we developed a unique instantaneous current-carrying tribometer. The instantaneous current-carrying friction tests were performed with Brass H62 and Al 1060 under 50–300 A current conditions at multifriction cycles. The friction behavior evolution with increasing current was revealed via the FSP signal, RP, and RQA quantifiers. In addition, the wear mechanism transition with the current was analyzed. A multifractal spectrum was applied to characterize the wear topography. The key findings are as follows:

1. As the current increases from 50 to 300 A, the decreasing FSP signal amplitude indicates that higher currents exert a stronger lubrication effect at the instantaneous current-carrying friction interface. Recurrence analysis of the FSP signals shows that RP presents disrupted, drift, and homogeneous patterns in turn with the current. Together, RR and DET decrease monotonically. The changes indicate that the nonstationarity of the system state is weakened, and the complexity and unpredictability are enhanced with the current. The aforementioned law is more evident at more friction cycles. The randomness of arc behavior and its damage, as well as the lubrication effect of molten metal, may lead to the system state evolution.

2. The electrical wear characteristics (ablation pits, delamination, cracks, adhesion, and oxide layers) of the plates become more remarkable with increasing current, but the grooves decrease. This change signifies a wear mechanism transition from abrasive wear and slight adhesive wear to arc ablation, fatigue wear, and severe adhesive wear. The increasing surface roughness Sa with the current indicates an intensification of wear damage. In addition, the instantaneous current-carrying wear topographies have multifractal characteristics. The Δα increases monotonically with the current, which implies enhanced wear topography irregularity and fluctuation. The aforementioned law is more evident at more friction cycles.

3. The wear damage state is correlated with the RP patterns, negatively correlated with the RQA quantifiers (RR and DET), and positively correlated with the Δα. In other words, the RP patterns and RQA quantifiers can characterize the wear damage state under instantaneous current carrying. The results are significant for guiding the antiwear design and data-driven online degradation tracking of electromagnetic launcher rail.

• The National Natural Science Foundation of China (Grant No. 52375178).

• The Natural Science Foundation of Anhui Province (Grant Nos. 2308085ME157 and 2308085ME135).

There are no conflicts of interest.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.