Electrical bearing failures in electric vehicles

In modern electric equipment, especially electric vehicles, inverter control systems can lead to complex shaft voltages and bearing currents. Within an electric motor, many parts have electrical failure problems, and among which bearings are the most sensitive and vulnerable components. In recent years, electrical failures in bearing have been frequently reported in electric vehicles, and the electrical failure of bearings has become a key issue that restricts the lifetime of all-electric motor-based power systems in a broader sense. The purpose of this review is to provide a comprehensive overview of the bearing premature failure in the mechanical systems exposed in an electrical environment represented by electric vehicles. The electrical environments in which bearing works including the different components and the origins of the shaft voltages and bearing currents, as well as the typical modes of electrical bearing failure including various topographical damages and lubrication failures, have been discussed. The fundamental influence mechanisms of voltage/current on the friction/lubrication properties have been summarized and analyzed, and corresponding countermeasures have been proposed. Finally, a brief introduction to the key technical flaws in the current researches will be made and the future outlook of frontier directions will be discussed.


Introduction
The first concept electric car was exhibited in the 1830s, and commercial electric cars came out at the end of the 19th century [1]. However, compared with traditional internal combustion (IC) engine vehicles, electric cars showed disadvantages in limited mileage ranges, heavy batteries and difficulties to refuel [2][3][4]. Thus, electric vehicles were not successful once they came out. In terms of advantages, electric vehicles possess high reliability, high power density, high efficiency and the ability to start immediately [5]. At that time, a successful application of electric vehicles was the electric trolley bus powered by catenaries. Nowadays, the global energy crisis is becoming tougher, and the fossil energy reserves as the fuels of IC engines are limited and non-renewable [6,7]. Moreover, the pollution of greenhouse gases (carbon dioxide) and other exhaust gases generated by IC engine vehicles is becoming increasingly more serious [8]. Along with the potential in fields of goods distribution and intelligent transportation systems, these problems make electric vehicles, which were once considered uncompetitive, an attractive role [9][10][11]. In addition, the technical field of electric vehicles has been greatly expanded in the past decades, and great progress has been made in tackling the key issues such as batteries, electric motor drives, automotive technology and system integration [12][13][14][15][16]. In the case of batteries, there have been unprecedented advancements in battery life, energy density, charge capacity, voltage output, energy efficiency, charging systems, etc. [17,18]. As a result, automobile manufacturers have spared no effort in the field of electric vehicles to meet growing market demands. Meanwhile, the market share of electric vehicles is gradually increasing, and electric vehicles have begun to replace IC engine vehicles in China, European and USA [19][20][21]. So far, electric vehicles can be divided into five types: traditional battery electric vehicle (BEV), hybrid electric vehicle (HEV) or plug-in hybrid electric vehicle (PHEV) (equipped with both IC engine and electric motor), fuel cells battery electric vehicle (FCEV), solar battery electric vehicle (SEV) and electric vehicle powered by supply lines [1,22,23]. Although there have been great innovations in propulsion systems (different power sources and converters), these electric vehicles have similar motor drive systems: A DC/DC converter for reducing the voltage, an inverter for driving the motor, and an electric motor (Fig. 1). Consequently, the reliability of electric vehicles is greatly limited by the stability of the motor system. Motors used in electric vehicles include DC motor, induction motor, permanent magnet motor, PM brushless DC motor and switched reluctance motor, etc. [26][27][28]. Among them, the threephase induction motor is by far the most widely used prime mover [29,30]. Regardless of the motor type, there will be shaft voltages and currents generated during rotation, which was first discovered in the 1920s [31,32]. Moreover, the common application of the inverter, which is used to convert the DC voltage of the battery into an AC voltage in the electric vehicles, exacerbates this problem [33][34][35].
The induced shaft voltages and currents can cause premature failure problems in a series of components such as bearings, seals, pads, and gears, and they can also give rise to electromagnetic interference (EMI) and radio frequency interference problems making the motor unstable [36][37][38][39]. Even with the performance improvement of the adjustable speed inverters, the lifetime of the electric motor is further reduced. In comparison, the premature failure problem of the bearings is the most serious, and it was reported that over 40% of motor failures were attributed to bearing failure [40,41]. In recent years, many lubrication failure problems owing to shaft voltages and bearing currents have been reported, and these problems will lead to instability, vibration and noise of a bearing, and consequently, more serious mechanical failures [42][43][44][45][46]. It is reasonable to expect that the electrical 6 Friction 8(1): 4-28 (2020) | https://mc03.manuscriptcentral.com/friction failures in bearings will gradually emerge with the popularity of electric vehicles in the near future. Under these circumstances, in order to ensure the long-term stability of the bearings in electric vehicles, it is of great industrial and scientific significance to deepen and update the influences of an electric environment on the bearing-lubrication system. This paper provides a detailed review of the key research progresses on the electrical bearing failures. Specifically, the review consists of four stages: Firstly, the origin of electric failures: the generation of shaft voltages in electric vehicles will be summarized in Section 1; the electrical model of a motor and various bearing current responses under different shaft voltages will be reviewed in Section 2. Secondly, different appearances of electrical bearing failures will be discussed in Section 3; Thirdly, the fundamental researches on the lubrication performance in an electrical environment will be provided in Section 4. Fourthly, the solutions to suppressing or avoiding electrical bearing failures will be introduced in Section 5. Finally, brief comments on the current research progress will be given and a future prospect of failure researches in an electrical environment will be proposed.

The origin of electrical bearing failures:
shaft voltages and bearing currents

The electric source-shaft voltages
In electric motors, electrical failures such as electric discharge machining (EDM) caused by electromagnetic and electrostatic effects are unavoidable. To further understand these electric phenomena, a basic understanding of shaft voltages (the source of electrical failures) is needed. Here, the shaft voltages are divided into three parts according to their generation: magnetic flux asymmetry, electrostatic effects, and inverter-induced voltage effects.

Magnetic flux asymmetry
Magnetic asymmetry arises due to the deviations in the magnetic pole distribution or shaft position during design, manufacture or installation ( Fig. 2(a)). Specifically, the reasons include asymmetrical windings, rotor eccentricities, casting defects, uneven permeability, unbalanced voltage signal generated by the inverter, etc. [47][48][49][50]. As a result, compared with the symmetrical operating state ( Fig. 2(b)), the shaft will form a voltage during the rotational process of cutting the magnetic induction line (Fig. 2(c)). These voltage/current waves are usually sinusoidal and of low frequency [51].

Electrostatic effects
The triboelectrification effect is due to the contact/ friction behavior of dissimilar materials, especially on the surfaces of dielectric materials, where charges tend to accumulate and persist for a while [52]. In electric vehicles, in order to reduce weight, composite materials and polymer materials of dielectric nature are widely used, for example, body structural parts made of carbon fiber reinforced plastics (CFRP), the cooling systems made of high thermal conductive insulating polymers and sealing rings made of rubber. In these components, triboelectrification can induce significant charge separation between the surfaces and accumulate considerable electrostatic charges [53]. When the electrostatic field reaches the breakdown strength of air or the lubricant, the accumulated charges are released to form discharge currents ( Fig. 2(d)), which is known as the forming process of EDM current.

Inverter-induced voltage
In modern electric vehicles, the pulse-width-modulation (PWM) inverters with fast switching devices such as metal-oxide-semiconductor field-effect transistor (MOSFET) and insulated gate bipolar transistor (IGBT) are widely used in electric motors to achieve variablespeed-control [54]. The high-frequency switching rate of the inverters will induce high-frequency common mode voltage (CMV), which is defined as the voltage between the motor neutral and the stator core (ground). As shown in Fig. 2(e), a three-phase induction motor is driven by a typical adjustable speed drive which contains a three-phase inverter [55], and the CMV (voltage at N in Fig. 2(e)) is equal to one-third of the vector sum of the voltages in three individual phases. If the voltage output from the inverter is a symmetrical sinusoidal signal, the CMV will remain at zero (Fig. 2(f)). However, the inverter often uses a pulse waveform to simulate a sinusoidal waveform, the series of pulse waveforms of a three-phase inverter are asymmetrical, and thus forming a stepwise CMV ( Fig. 2(g)). Similarly, although there is no neutral point in delta-winding-connected electric motor, the unbalanced output of the inverter still causes "common mode" effects. Further, shaft voltages are induced by capacitive and magnetic coupling at the step positions of the CMV waveform ( Fig. 2(h)). In terms of control, the higher the switching frequency, the more the CMV conforms to the sinusoidal waveform, as well as lifting efficiency [56]. In terms of shaft voltage, the high switching frequency means the high dv/dt in CMV waveform, introducing the high-frequency harmonic distortion in the motor system, inducing the high-frequency shaft voltage with a large amplitude [57]. Such a high-frequency shaft voltage could be harmful because it can pass through many interfaces even if they are insulated.

The generation of bearing currents
In a typical induction motor, a complex capacitor system exists due to the presence of air gaps, insulation coating, and lubricants, including the aggregate stator windings to the stator frame (C ws ), the aggregate stator winding to the rotor (C wr ), the rotor the frame (C rf ), the bearings (C bf , C be ), the stator to the rotor (C sr ), the rotor to ground (C rg ), etc. [58,59]. Moreover, when a capacitor breaks down, it turns into resistive. Figure 3(a) shows the simplified physical schematic and corresponding circuit models of an induction electric motor, and the possible current paths caused by different voltage sources are also very complicated. As for the current sources, bearing currents can be divided into "circulating" and "non-circulating" [60,61].

"Non-circulating" currents
The "non-circulating" types of currents include the dv/dt related currents and "electrical discharge machining (EDM)" current pulse ( Fig. 3(b)). It is termed "non-circulating" because these currents pass through the bearings unidirectionally from the rotor to the stator [60]. The dv/dt related currents can be conductive and  capacitive [62]. The conductive currents mainly occur when the motor is running at a low speed. Since an effective insulating film cannot be maintained at low rotational speeds, the internal contact of the bearing is metallic. Thus, the current flowing from the stator winding appears to be conductive in the bearing. At normal rotational speeds, the insulating lubricate oil/grease film electrically works as a capacitor [63]. When switching occurs in the IGBT, as long as the switched voltage does not cause a breakdown, the capacitor of the bearing is charged or discharged. Specifically, the high-frequency (corresponding to the switching frequency) current passes through the stator winding to the rotor, then to the bearing, and finally to the frame (Figs. 3(a) and 3(b)). However, the magnitude of dv/dt related currents are small, and the capacitive ones are only 5-10 mA, while the conductive ones do not exceed 200 mA [64]. Therefore, dv/dt current components are often harmless and only account for a small percentage of all bearing currents. EDM bearing currents occur when the bearing voltage exceeds the threshold voltage of the insulating lubricating film, and the energy of the capacitor is released by destructive currents and arcing [65]. As mentioned above, the bearing works as a capacitive voltage at normal rotational speeds ( Fig. 3(d)). Bearing voltage (V b ) can be estimated from CMV (V com ) and bearing voltage ratio (BVR) [66]: Generally, BVR ranges from 3% to 10% [58]. In consideration that the voltages of AC supplies in electric vehicles are at least 300 V, the peak of bearing voltage can reach ~30 V. The lubricant film thickness in bearings commonly ranges from 0.1 to 1.4 microns, which can withstand voltages ranging from 1.5 to  [67,68]. Thus, the electrical breakdowns are likely to occur, giving rise to the generation of the EDM currents. With the nature of short-circuit current, the amplitude of the EDM currents in a 1.5 kW induction motor can range from 0.2 to 1.4 A [69]. The path of the EDM currents is "shaft-inner race of the bearing-rollers-outer ringframe" [70]. Different from dv/dt currents, which happen at switching moments, the occurrence of EDM currents is not directly related to the rise time of the CMV wave. In actual working conditions, the lubricant film thickness is affected by speed, load, and lubricant viscosity. Moreover, the surface roughness at the bearing interfaces, the uniformity of lubricant distribution, the history of discharge, and the fluctuations of mechanical operation also contribute to the discharge. For these reasons, although high CMV can result in an increased probability of the EDM currents, the occurrence of the EDM currents is more likely to be random [71].
In special cases, when the rotor-to-ground resistance is smaller than the stator-to-ground resistance, part of the bearing current will flow to the ground through the shaft, following "stator -bearing -rotor shaftground" (Figs. 3(a) and 3(c)).

"Circulating" current
Compared to the "non-circulating" type, the generation of the "circulating" type current is far more complicated, involving magnetic induction, inductive coupling and capacitive coupling [60]. The earliest "circulating" current was found in the sine-waveoperated AC motor, which is also known as the classical inductive bearing current [31]. This kind of "circulating" current mostly occurs owing to magnetic flux asymmetry, and the frequency corresponds to the shaft speed. As shown in Fig. 3(e), when the induced shaft voltage is strong enough to break through the oil film/grease in a bearing, the bearing current would circulate in a conductive loop, i.e., "stator -drive end bearing -rotor shaft -non drive end bearingstator" [72]. In recent years, the classical induced bearing current is no longer considered to be a main current component due to the improvement of manufacturing and assembly processes. As reported by Maki-Ontto and Muetze, another kind of "circulating" current was derived from the CMV of the inverters: The parasitic capacitances between the stator winding turns and the frame are excited by the steep edges of high-switching-frequency CMV, generating a highfrequency common mode current which flows through the core of stator [73,74]. Therefore, the current flowing into the stator winding is higher than the current flowing out. By taking a Gaussian surface inside the stator core perpendicular to the axis which encloses all the windings, the enclosed current is not zero due to the loss. In other words, there is an axial net current in the stator winding. According to Gauss' Law, there must be a net flux surrounding the shaft. As shown in Fig. 3(f), the high-frequency net current generates a high-frequency tangential magnetic flux (Φ com ) along and around the motor shaft, inducing the generation of the shaft voltage. Meanwhile, the induced bearing currents share similar formation processes and circulation to the first kind of the current. However, unlike the first kind, the frequency of this current can be as high as several megahertz, and the peak amplitude of this type of current is generally 0.5-20 A [60,74].
In addition, there are some differential-mode bearing currents reported in the literature. Nippes et al. studied the water vapor droplets induced frictional electrification and discharge [37]. The grounding properties of different components also lead to different current paths and amplitudes [75]. Thus, many researches have been carried out to model the bearing currents under different electric parameters [76][77][78]. In comparison, the EDM currents and the high-frequency "circulating" current are the largest, accounting for the highest bearing current ratios, while other types of bearing currents receive less attention.

Electrical failures of bearings
Under a complicated electric environment, failure would occur on all contact/sliding surfaces of components, in which bearings are the most vulnerable. Thus, the bearing could be a typical example for illustrating such failures. The traditional bearing failure modes can be grouped into the following categories: fatigue spalling, fretting, smearing, skidding, abrasive | https://mc03.manuscriptcentral.com/friction wear, corrosion, cracks, true or false brinelling, etc. [79]. In addition to these failure modes, the presence of shaft voltages and bearing currents can cause new failure modes [80]. Generally, the electric-related failure modes can be mainly divided into two parts: morphological damages and the influence on the lubrication process, corresponding to static results and dynamic processes, respectively (Fig. 4).

Morphological damage
Morphological damage due to shaft voltages and bearing currents can be classified into five types: frosting, fluting, pitting, spark tracks, pitting and welding.

Frosting
Frosting originates from weak but dense discharges. Owing to the satin-like appearance, a frosted surface is difficult to distinguish with the naked eye ( Fig. 5) [32]. However, microscopically, this sand-blasted surface is composed of small "craters", and each "crater" indicates a melting effect when an EDM current occurs. Meanwhile, the surface area to volume ratio of a frosted bearing race increases, enhancing the subsequent chemical corrosion.

Pitting
Similar to frosting, pitting is also composed of "craters". However, the discharge is more intense. As a result of the single discharge pulse which has a larger current amplitude and lasts longer, the "craters" in pitting have a larger size (Fig. 6). In terms of distribution, the corroded pits corrosion appears as a random pattern, and they are sparser than frosting [88]. Komatsuzaki et al. reported that the key factor of the electrical pitting process was bearing current rather than shaft voltage, and a 90 mA current was sufficient to cause pitting [89]. Chiou et al. investigated the formation mechanism of electric pitting: at a constant film thickness, the interface power increases with the increasing bearing currents, and the relationship between the pitting area A p (in unit of 10 3 μm 2 ) and the interface power P (in W) is cubic function [90]: In consideration of voltage (V), current (I) and the thickness of oil film (h), the pitting area can be expressed as [90]:

Fluting
As a result of the periodic currents, fluting is the most common damage, manifested as flute burnt scars evenly distributed along the circumference of the bearing race. Under the microscope, a fluting pattern occurs like a washboard with ''dark stripes'' and ''bright stripes'' ( Fig. 7(a)). However, under scanning electron microscopy ( Fig. 7(b)), the fluting pattern is composed of scratch marks [91]. Prashad et al. studied the relationship between the resistance of the contact area and the damage mode, and they found that low-resistance contacts caused electrochemical decomposition/corrosion of grease and gradually  | https://mc03.manuscriptcentral.com/friction leaded to fluting, while high-resistance contacts were more likely to induce pitting [92]. By analyzing the fluting surface and wear debris, Liu et al. suggested that the nature of fluting was a three-body abrasive wear caused by the vibration of bearing rollers, while the vibration was excited by the bearing current [91]. This dual effect of mechanical and electrical discharge can also explain the appearance of fluting in heavyload areas. In addition, fluting is also a critical stage in bearing tests.

Spark tracks
The initial appearances of spark tracks are irregular scratches askew to the direction of rotation (Fig. 8).
Although looking like a mechanical scratch, the spark track also has an electric nature. The bottom of the track is sometimes melted, and the corners of the scratch are sharp, which should be more rounded as a result of mechanical scraping [32]. In addition, the depth of a spark track is consistent with its entire surface. The cause for the spark tracks is proposed as the debris blasted out of the surface by electrical discharges.

Welding
Welding mainly occurs in the housing splits, pads and seals of the bearing, and it is attributed to the thermal effect when a large amount of current passes the bearing ( Fig. 9) [32]. However, as discussed in the previous section, the usual EDM process cannot generate such a large current flow, and hence welding is mainly caused by direct momentary contact between the stator and the rotor. This phenomenon has characteristics such as spot welding that can be easily distinguished. According to different working parameters, such as load, rotate speed, bearing type, roller and interface material, and lubricant conductivity, the appearance of electrical damages varies widely. Didenko and   The most severe case of electrical bearing failure: welding. The pads were welded into its retainer. Reproduced with permission from Ref. [85], © IEEE 1991.
Pridemore observed a different fluting pattern in the roller, in comparison with the outer race of a Tri-Lobe roller bearing [94]. As shown in Fig. 10, the anode (roller) has wider and shallower "craters", while the cathode has narrower and deeper ones. Using an electrical pitting wear tester, Raadnui et al. provided a corrugated parameter model, containing load, current, housing temperature and test duration [95]. As for the pitting wear debris, long test duration and a large current will enhance the production of black spherical wear particles. In the case of weak bearing currents of no more than ~mA, Xie et al. found pitting marks on the inner ring using a lubricant with a higher conductivity than an insulated lubricant [96].

Lubrication failures
In addition to the morphological damage, lubrication failure under charged conditions is also a key issue that has attracted a lot of attention in recent years [97,98]. Improper lubrication can lead to increased friction and wear, unstable operation, resulting in a sharp reduction of bearing lifetime [45].

Degradation of lubricant
In a bearing, the lubricant should be capable to reduce friction and wear, evacuate the heat, inhibit corrosion, and clean the surface, etc. [99][100][101]. In a well-lubricated running bearing, the two counterfaces are completely separated by a thin lubricant film, avoiding direct contact. Therefore, the traditional mode of bearing failure mode can be improved by suitable lubrication strategies. Although lubricants are generally chemically inert, shaft voltage and bearing current provide the potential and energy required for chemical reactions, accelerating the degradation process [102]. The free radicals generated by electrical excitation will rapidly react with oxygen to form peroxide groups, which in turn induce the formation of new radical groups, and such a chain reaction will eventually form carboxylcontaining products [82]. Commonly, there will be oxidation of base oils, antioxidants, and thickeners, which will produce acidic and highly viscous products, resulting in the loss of lubricity. Along with this process, lubricating additives in the lubricant, such as molybdenum disulfide, are found to be separated from the lubricant and agglomerated onto the raceway, and these lubricants additives also degrade with the application of an electric field [103,104]. Moreover, the thermal effect of the discharge process would also cause the evaporation of the oil component, which was considered as the main factor for the grease failure in

Microbubble effects
The generation of microbubbles in charged lubricant films was first discovered by Luo et al. with the relative optical interference intensity technique [106]. Under charged condition, abundant microbubbles appeared around the lubricated contact area (Fig. 11), which was proposed to be attributed to local overheating [83,107]. Correspondingly, when moving to the outer region, the bubble tended to be unstable and the coalescence sometimes occurred (Fig. 11). The generation and collapse of microbubbles can destabilize the lubrication, which can lead to additional noise and vibration in a bearing [107]. In addition, lubricants containing microbubbles are more susceptible to electrical breakdown. An anomalous phenomenon is that the generation of microbubbles was more intense after the electrode was coated with an insulating layer, and the required input electrical energy was much smaller compared with that of the uncoated electrode [108]. The bubble generation was also found to be closely related to the frequency of the AC electric field. Similarly, the difference in the threshold voltage and current caused by the frequency was also enhanced by the interfacial dielectric properties. Based on a series of experimental and theoretical studies, a detailed model containing the formation and motion of microbubbles in nanoconfined liquid films was established [109]: As a result of competition for the microbubble generation and the liquid refilling, large bubbles appeared in the lubricants with a high surface tension or a high liquid viscosity. In terms of motion, microbubbles were driven by the pressure gradient in the contact zone, dielectrophoresis force and viscous drag.

Interfacial stress induced by electric field: Electrowetting
The contact angle of a droplet on a surface can be regulated by a voltage between the electrodes, which is called the electrowetting effect [110]. This electromechanical mechanism dominates the microfluidic behavior by modifying the interfacial tension [111]. More relevant to the bearing working conditions, it is observed that the nonpolar dielectric lubricant in the contact area of a steel ball and a metallic layer will spread along the surface under the action of an electric field [112]. In the case of emulsion, a typical two-phase immiscible liquid system used as the lubricant, the stable dispersion state of the two-phase system would be destroyed by the electrostatic pressure and the surface tension due to the difference in dielectric properties [113][114][115]. Thus, the lubrication properties could be unstable due to the coalescence of emulsion droplets. Adopting a simplified needle-plate barrier discharge experiment, Lee et al. investigated the effect of discharges on lubrication stability [116]. A deformation of an oil film was observed under a high voltage, and a surface fluctuation occurred when the voltage was switched. Furthermore, this deformation was related to the voltage polarity and the temperature, and the negative voltage polarity and the high temperature helped to deform for silicone oil droplets [117].
In addition to the lubrication failure modes mentioned above, the shaft voltage and bearing current also have some other effects. For instance, the interfacial electric field enhances the electrostatic interaction between the rollers and the raceway, generating extra electrostatic pressure. Moreover, a reduction in the lubricant flow of a confined liquid film was also observed under an electric field [118].

Fundamental researches on tribological performance in an electrical environment
As mentioned above, the presence of shaft voltage and bearing current has brought out an urgent need to improve the lubrication state of the bearing. In a properly designed lubrication system, morphological damages and lubrication failures caused by the electrical environment can be weakened or eliminated [119,120]. In order to achieve the suppression of electrical damage and even use the electrical environment to promote lubrication, an in-depth understanding of the relationship between the lubrication properties and the electric field/current/charges is needed [121].
In the past decades, researchers have made great efforts to promote fundamental researches on the tribological performance in an electrical environment. The electrical failure is decomposed into several scientific subjects concerning the effect of electric field on friction and wear, in which lots of remarkable results revealing the underlying physical and chemical nature have been demonstrated. These results help to fill the blank in the theory of lubrication under charged conditions, as well as provide guidance for the design of lubrication systems in practice. It has been confirmed by a large number of experiments that the lubrication/friction properties can be tuned by an external electric field [122][123][124]. However, unlike specific electrical damages that have common appearances, the electrical responses of the lubrication properties are significantly different, and multiple mechanisms are involved. As shown in Fig. 12, there have been four main influence mechanisms of the effects of the electric field/charge on the lubrication performances so far [125][126][127][128][129][130]: (1) electrostatic interaction; (2) structural change/transfer film formation; (3) changes in physical/chemical properties; (4) carrier/ charge distribution. In actual situations, these mechanisms are often used collectively, while their contributions vary in different material systems and under different working conditions.

Electrostatic interaction
Electrostatic interactions are common at all frictional interfaces, and the classification here refers only to the category of the electrostatic forces acting directly on the friction interaction. The stronger the electrostatic interaction, the more severe the friction and wear. Weaker electrostatic interaction promotes lubrication.
Firstly, due to the difference in the Fermi levels of different materials, electron transfer occurs when dissimilar materials contact, to reach a balance of the potential at the interface, thereby generating the contact potential (i.e., self-generated potential) [131]. As a result, the positive and negative charge centers are formed on both sides of the interface, giving rise to the electrostatic force at the interfaces. In addition, the surface static charges and the transient polarization charges formed by the triboelectric effect further enhance the electrostatic interaction. By offsetting the self-generated potential by the application of an external electric field, Yamamoto et al. achieved the friction reduction [132]. The friction forces of 34 kinds of metallic friction pairs in open circuit, short circuit, zero current (i.e., the external voltage offsets the selfgenerated potential), and constant current (forward and reverse) conditions were compared by Chen et al. [133,134]. It was found that for the Fe, Co, Ni, Ti, Cr and Cu systems, the friction forces in the open circuit state and the zero current state were smaller than those in other electrical states, and the friction at zero current was the lowest [133,134]. Conversely, it is also possible to strengthen the charge separation at the interface by an external electric field. Further, the friction and wear mechanism of the stainless steel changes with interfacial potential differences, and it is proposed that adhesive wear dominated under the low potential difference and abrasive wear dominated under the high potential difference [135]. Thus, it was believed that slightly increasing the interfacial potential difference during the running-in phase could effectively shorten the running-in period, while in other friction stages, it was necessary to reduce the potential difference reasonably to reduce friction and wear. The applied electric field not only affects the conductor/ conductor friction pair, but also applies to any metalcontaining friction pair. The electric filed between carbon black rubber and aluminum to cause the electrostatic attraction between the contact peaks was energized by Hurricksa, and a significant friction 16 Friction 8(1): 4-28 (2020) | https://mc03.manuscriptcentral.com/friction increase was observed [136]. Similarly, the applied electric field can also enhance the friction in the stainless steel/ice [137,138], ferroelectric materials/ stainless steel [139,140] systems.
In the micro/nanoscale friction experiments on the basis of atomic force microscopy, the interfacial adhesion force can also be effectively tuned by applying an electric field between the AFM tip and the sample surface. However, so far, the researches on molybdenum disulfide [141], InAs nanowires [142], and silicon/silicon dioxide materials [143,144] have generally suggested that the electric field enhanced friction and wear. It is worth mentioning that Liu et al. studied the friction response of the Langmuir-Blodgett monolayer films under the actions of a DC voltage and an AC voltage [145]. It was found that the DC voltage enhanced friction, while a friction reduction was observed under an AC voltage of certain frequencies, owing to the vibration effect of the fluctuating electrostatic force [145].

Structural change/transfer film formation
The structural change/transfer film formation mechanism is mainly embodied in the friction pair containing an interfacial structure with electrical responsiveness. With the application and removal of the interfacial electric field, the contact surface exhibits different molecular structures or transfer film orientations, corresponding to different lubrication properties.
In the case of dry friction, Csapo et al. studied the dynamic electrical contact friction behavior of graphite-graphite under argon atmosphere, and found that the friction coefficient increased significantly after applying a current [146]: Specifically, the graphite particle crystals at the contact position recombined, so that the basal plane was parallel to the sliding surface, resulting in a decrease in the friction coefficient. When the current passed, the basal plane of the graphite particles was transformed to be perpendicular to the sliding surface to enhance the interfacial conductivity. This change increased the contact points per unit area, enhancing friction and wear. In the controlled atmosphere, the graphitegraphite and graphite-copper friction pairs showed reduced friction and increased wear under the passage of the current owing to the formation of the oxidative transfer film.
Lavielle et al. studied the friction between ternary polyethylene film (pin) and steel (disc) under different electrical conditions [147]: when a forward voltage was applied, the carboxyl group on the polymer surface was repelled by the steel surface, which mainly showed the lubrication of the alkyl group; when a reverse voltage was used, the adhesion of the carboxyl group to the steel surface was enhanced to increase the friction. Similarly, the friction properties of graphene oxide [148] and self-assembled monolayer [149] also differed under forward and reverse voltages.
Such a mechanism is more widely used in liquid environments. Sweeney et al. achieved the control of the surface adsorption state by controlling the potential of the gold surface in the perchlorate/sulphate solution, and thereby controlling the lubrication properties of the surface [127]. Herminghaus et al. controlled the lubrication performance and the bearing capacity by controlling the molecular brush shape of the polymer electrolyte through the potential [150]. Similar studies have expanded to more liquid-phase environmental systems in recent years, such as long-chain alkanes [151], polyols [152], various ionic liquids [153,154] and hydrogels [155].
The structural change/transfer film formation mechanism can achieve two-way (enhancing and weakening) lubrication performance control, because it can control the interface morphology, and it is easier to achieve lubrication than other mechanisms.

Changes in physical/chemical properties
As compared with the former mechanisms, this influence mechanism mainly emphasizes the chemical reaction and the physical absorption at the interface. With the application of an electric field, polar molecules, anions and cations in the lubricant will be physically adsorbed to the charged interfaces. Chemical reactions occur on the contact surfaces when the external electric potential meets the electromotive force (EMF). These effects change the physical/chemical properties of the original surface and thus change its frictional properties.
For water-based lubricants or organic lubricants with reactive functional groups, discharges induce the hydroxide or reactive groups to accumulate on the metal surface. The representative work is a series of metal/ceramic friction pairs developed by Meng et al., and they found that the applied voltage increased the friction, because the hydroxide ions formed by the water decomposition at the metal electrode aggregated and reacted with the metallic surface to form a phase film [124,156]. Zhai et al. found that the saponification reaction of GCr15/45 interface in the aluminum stearate solution was varied by the applied voltage, and different adsorption characteristics of the saponified film corresponded to different friction properties [157].
For dry friction, Paulmier et al. explored the friction properties of the graphite/XC48 carbon steel friction pair under the passage of the current [158]. It was found that the steel surface formed an oxide film after the application of the current in the atmosphere, and the friction was reduced by 35% when the steel was the cathode. Moreover, similar effects could apply to the graphite/copper [159], graphite/graphite [160], and CN x film/aluminum [161] friction pairs in the atmosphere.

Carrier/charge distribution
In terms of the frictional energy dissipation process, the electronic excitations and creation of electron-hole pairs enhance the energy loss. The rearrangement of carrier/charge distribution under the electric field | https://mc03.manuscriptcentral.com/friction could influence these processes, leading to the changes in friction.
A systematic study on silicon pn junctions was carried out by Park et al., and it was revealed the feasibility of electronically controlled friction [129,162]: As compared to the n region, the p region shows a higher friction force under the external electric field. It was proposed by the authors that the strong accumulation of carriers in the p region produced a large ohmic loss and increased the friction. By applying an external electric field, Wang et al. regulated the charge distribution between graphene layers and tuned the interlayer friction [130]. In addition, the stress in the contact region could produce a series of changes in the electrical property change: the induced bending and carrier dispersion affected the interface dislocation mobility, and the formation of local quantum dots in the contact region promoted the electron-hole pair recombination.
In Fig. 13, the detailed material systems in the fundamental researches and the proportion that each mechanism works are summarized. It can be concluded that the electric field can be used to tune the lubrication performances. Particularly, the use of polar lubricants or additives in the liquid phase can effectively maintain the lubrication performance under charged conditions. However, more efforts need to be made to bring these experimental results to industrial applications.

Solutions to electrical bearing failures
The purpose of studying electrical failure is to prolong the bearing service life and achieve long-term stability. In the following part, some effective solutions to the electrical bearing failures will be discussed in the following part.

Reasonable grounding and minimizing the electric field
A typical patented solution is the use of a grounding ring composed of conductive microfibers, which is installed on the shaft outside the bearing [163]. With the conductive brushes connected to the shaft, the ground ring works as a diverter, directing the shaft voltage to ground and bypassing currents that would otherwise flow through the bearing (Fig. 14) [164,165]. It has been proven that this technique works well for the EDM currents and the high-frequency circulating currents. Fig. 13 The material systems in the researches on the relationship between the electric field and the lubrication properties. The percentages represent the proportion of different mechanisms involved in literature: 39% of the studies involve electrostatic interactions, 59% of the studies involve structural changes, 17% of the studies involve physical/chemical properties change, and 13% of the studies involve carrier/charge distribution. The red parts represent the proportion of the lubrication effect in each mechanism, the corresponding material systems are also marked red. One reasonable idea is to suppress CMV from the source. Simply reducing the switching frequency of the inverter can reduce the electrical damages, and however, it will limit the performance of the speed control system. An active approach is to add CMV filters between the inverter and the motor [166]. This method will divert the CMV away from the motor, and the currents will be directed back to the inverter or to the ground [167]. The filter design has been well developed in recent years. Pairodamonchai et al. suggested a hybrid output EMI filter to eliminate the high-frequency CMV components [168], and the researches on the optimization of filters have also been reported [169,170]. Moreover, an advanced inverter design can also help to reduce CMV: The active zero state PWM (AZSPWM) method [171] and the near state PWM (NSPWM) method [172] effectively suppress the neutral voltage of a motor.
Another strategy is to shield the electric field [173]. Busse et al. evaluated a modified induction motor, i.e., electrostatic shielded induction motor, and the PWM induced shaft voltages could be effectively suppressed by constructing a Faraday shield in the air gap between the stator and the rotor [174]. Similarly, by shielding the wires between the inverter and the motor, the capacitively coupled current was effectively weakened [175].

Improve the insulation performance of the bearing
A classic way to suppress the high-frequency bearing current is to build an insulation layer on the bearing [176,177]. For example, hybrid/full ceramic bearings have been used in commercial EV [51]. The purpose of this method is to raise the impedance between the bearings and the ground. Thereby, electric discharges can be prevented by the insulating layer. Circulating currents can be significantly suppressed by the ceramic or hybrid bearings, while the EDM currents are less affected [178]. However, an insulated bearing limits the dissipation of the heat flow from the rotor. Furthermore, the insulation method sometimes transforms the discharge process and changes the proportion of the different components of the currents. Therefore, enough attention should be paid to the current composition.

Enhance the conductivity of the lubricated interface
In contrast, the solution on the basis of enhanced conductivity also works in some systems. Although it seems to be the opposite of the insulating methods, it has been experimentally proved that a conductive grease can prevent the fluting [88]. Suzumura suggested that due to the formation of electrical channels, the electric current density of the rolling contact area with conductive greases was lower than that with non-conductive greases [88]. In terms of energy consumption, the insulated interfaces can be easily corroded during discharging when a large amount of energy is released instantaneously in a confined area, and the interfaces with an excellent conductivity accumulate less energy. Therefore, suppressing the interfacial resistance, which is related to nonconducting lubricants, insulating surface layers and asperity contacts, could be effective [179]. However, the method is closely related to the lubricant components, and simply adding metal particles to the grease could increase mechanical wear [180]. Thereby, a welldesigned electric contact lubricant is required to ensure the lifetime and reliability of the system. Zhang et al. added carbon black into traditional overbased calcium sulfonate complex grease and lithium, enhancing their conductivity, friction-reduction and anti-wear properties [181]. Conductive greases based on lithium salts (LiBF 4 , LiPF 6 , LiNTf 2 ) and their related ionic liquids also showed excellent lubrication properties. Typically, the use of ionic liquids, which are composed of weakly coordinated anions and organic cations, has attracted a lot of interest. In addition to the characteristics satisfying the needs of a lubricant, e.g., non-volatile, non-flammable, and low melting point, | https://mc03.manuscriptcentral.com/friction the dipolar structure of ionic liquids (ILs) helps form a boundary lubricant film, and meanwhile, the high conductivity suppresses arc discharges [120].
Transition metal powders (gold, silver and copper) provide excellent electric conductivities, and however their lubrication properties and resistance to degradation are poor [182]. Thus, these materials have been used as the surface or coating materials in the form of functional composites, e.g., AgSnO 2 , AgI, Ag/C, CuW, Cu/C, carbon nanotube film and graphene [183][184][185][186][187]. However, there is still a lack of long-life industrial lubricants with outstanding interfacial conductivities and desirable lubricities. The use of additives such as silver is too expensive for industrial applications, and some ILs are corrosive. In addition, this method only weakens the effect of the current, which does not avoid the influence of the electrical environment. A large number of new materials are still in the experimental stage. To meet the high-speed, high-load, complex vibration conditions and achieve commercialized mass production, great efforts need to be made.
In conclusion, because different inverter-motor systems induce different bearing currents, specific strategies can vary dramatically. Some technologies such as voltage filters, although effective, are not suitable for industrial uses because of their high costs and difficulty in installation.

Summary and outlook
The researches on the premature electrical failure of bearings are one of the key bottlenecks in electric vehicles at present and in the forthcoming decades. In this paper, an overview on electrical bearing failure in electric vehicles has been presented. Relevant topics such as common mode voltages, bearing currents, electric discharge machining and lubrication instability have been regrouped to get a comprehensive and systematic perspective on the phenomena. The generation and composition of shaft voltage and bearing current and the appearance of electrical bearing failure, and then to fundamental researches on lubrication behaviors under charged conditions, and finally feasible ways to solve the problem, are discussed. In terms of the depth and breadth of current studies, considerable efforts have been made to classify and quantify bearing voltages and currents, and however the lurking lubrication problems have received less attention. Nevertheless, the recent progress of fundamental lubrication researches has continuously improved the theoretical systems of the lubrication behaviors under charged conditions, which could guide the design and the protection of electrical contact interfaces.
Based on this study, more researches on relevant directions are still urgently needed, which are enumerated below: (1) A commonly used bearing current prediction model for various motor systems is still lacking. This paper reviews the most typical shaft voltage and bearing current modes of the induction motors. However, the motor and inverter control systems of different electric vehicles are not completely consistent, and the design of the overall electrical system, as well as the type of the motor, can vary greatly. For instance, in an electric vehicle driven by wheel-hub motors, the compositions of shaft voltages and bearing currents are more complicated. More work is needed to detect, classify and quantify the electrical environments in these systems.
(2) A bridge is needed to closely relate the study of the lubrication failure under charged conditions to the actual bearing failures. Although many fundamental mechanisms have been revealed, the electrical environment in which the bearing actually works is much more complicated than that in the experiment, and how these mechanisms correlate with each other is still an open question. Furthermore, the electrical failure of bearings is mainly studied on the basis of the damage morphology. The running state of the bearing (e.g., lubrication instability) will be potential and compelling directions.
(3) More comprehensive, flexible, low-cost solutions that can meet industrial needs are desirable. The foregoing parts have summarized several strategies to suppress the electrical bearing failures, and while turning these technologies into industrial applications is still challenging. Recent development of new materials offers a broader range of possible solutions, e.g., self-lubricating and self-healing materials, smart surface structures with electrical responses. Hence, exploring the application of new lubricating materials in motor bearings could be an integrated part of future researches.   (1996). Prof. Luo has been engaged in the research of thin film lubrication and tribology in nanomanufacturing. He has been invited as a keynote or plenary speaker for 20 times on the international conferences.