1 Introduction

In recent times, the transportation sector has witnessed a significant increase in the adoption of electric vehicles (EVs), which include battery electric vehicles (BEV), hybrid electric vehicles (HEV), plug-in hybrid electric vehicles (PHEV), and fuel cell electric vehicles (FCEV). For many applications, the intent is to replace traditional internal combustion engine (ICE) vehicles. Since the ICE has largely driven lubricant requirements, its removal from the transportation sector opens new opportunities for optimizing new lubricants to meet the new demands for EVs, improving energy efficiency, sustainability, and cost reduction [1]. EVs in particular, require efficient lubrication, as about 57% of the energy supplied to an EV is utilized to overcome friction, including rolling resistance and acceleration and braking inertia [2]. To meet the specific lubrication needs of EVs, practical solutions like water-based lubricants (WBLs) and water-containing lubricants (WCLs) have been studied to provide battery and motor cooling and gear lubrication simultaneously. WBLs contain at least 50% by weight water and WCLs contain less than 50% water. Furthermore, recent EV trend reports indicate the interest of OEMs into these kind of fluids [3]. However, as the industry pushes to improve EV range and performance by moving toward higher power density for mechanical components, harsher tribological requirements for lubricants arise [1, 3, 4]. These industrial changes accentuate the need to acquire the rheological and tribological behavior of WBLs to assess whether they can play a meaningful role in helping to meet the challenges facing EVs today and into the future. In response, there has been a rise in basic research involving WBLs and WCLs.

For instance, Matta et al. studied the micropitting behavior of a commercially available hydro-lubricant showing promising results [5]. A review by Rahman et al. explored different water-based lubricant additives and summarized their properties and performance, reporting that WBLs also have environmental benefits over petroleum-based lubricants [6]. Water-containing lubricants have been shown to achieve superlubricity in rolling–sliding elastohydrodynamic (EHL) contacts while still forming lubricant films [7,8,9]. Additional research on superlubricity of water-based polar molecules in boundary lubrication has also received significant attention [10,11,12].

Alongside water-soluble polyalkylene glycols (PAGs) [7, 10, 13, 14], glycerol is a commonly employed water-soluble liquid [8, 15,16,17]. Both types of base stocks exhibit friction coefficients below 0.01 in the superlubricity range as demonstrated in tribometer-based friction measurements [18]. Yilmaz et al. found slightly reduced film thickness for water-containing PAGs compared to polyalphaolefin oils (PAO), attributed to higher density, but did not provide water content information [9]. In a similar study, they also identified superlubricity in gears [19]. Sedlmaier et al. tested lubricants from Yilmaz et al. on a BEV test rig, revealing up to 74% power loss reduction compared to PAOs [9, 19, 20]. Investigating water-soluble PAGs, Wang et al. found coefficients of friction as low as 0.0023 for mixtures with 50 wt% water in a reciprocating ball-on-disk tribometer setup [13]. Liu et al. explored water-containing PAGs of different molecular weights and water concentrations to study superlubricity, observing that achieving stable superlubricity depended on molecular weights and concentration as the phenomenon depends on the binding state between the polymer structure chains and water [21]. Higher molecular weights raised the threshold water concentration for superlubricity, as demonstrated in the studies of Wang et al. [13]. Burbank et al. conducted ball-on-disk friction and film thickness tests on water-containing PAGs under rolling–sliding conditions, uncovering extensive superlubricity [14]. Interestingly, applied load showed minimal impact in the fluid film lubrication regime. Wang et al. studied the tribological behavior of PAG aqueous solutions in combination with layered double hydroxides nanosheets, showing friction coefficients around 0.2 initially and achieved superlubricity values after different periods of time [22]. The underlying mechanisms leading to superlubricity with water-containing lubricants in EHL contacts are still unclear [23]. Several authors explained superlubricity in fluid film regime as arising from surface interactions with water-containing PAGs, forming unbonded hydrate layers with low shear resistance within the fluid film [13, 16, 21]. Ge et al. proposed a stable tribofilm formation by water-containing PAGs, allowing parallel orientation of fluid molecules within the film [10].

Numerical research by Habchi et al. explored glycerol's water content impact on friction and film thickness in EHL contacts [24]. It was found that fluid friction coefficients decrease with up to 40 wt% water, leading to superlubricity for all conditions [24]. However, at 40 wt% water, film thickness was approximately an order of magnitude lower compared to pure glycerol. A similar pattern emerged from Shi et al.’s experimental study on a ball-on-disk tribometer using glycerol [8]. Higher water content, up to 50 wt%, resulted in limited film thickness and a shift from fluid film to mixed and boundary lubrication with higher friction coefficients. Film thickness-inferred pressure–viscosity coefficients decreased with rising water content. Tamayo et al. investigated glycerol containing 5 wt% water and a mixture of 5 wt% water and 30 wt% glycol, achieving similar film thickness in boundary lubrication [17]. Both lubricants reduced friction and wear compared to PAO.

Several studies highlight glycerol’s impressive ability to reduce friction, especially in boundary and elastohydrodynamic lubrication [24,25,26]. An optimal water content improves friction reduction in both lubrication scenarios. Glycerol solutions surpass rapeseed oil in reducing friction by more than three times in elastohydrodynamic lubrication and 30–50% in boundary lubrication. However, adding water to glycerol accelerates wear on steel surfaces, with wear volume increasing as water content rises. The study also notes that the central film thickness of glycerol–water mixtures correlates logarithmically with entrainment speed at water contents below 30%, showing full film elastohydrodynamic lubrication achieved at an entrainment speed of 1 m/s. Beyond this water content threshold, film thickness decreases, transitioning to mixed or boundary lubrication [8]. These observations suggest that viscosity, especially under high contact pressures may be influencing lubrication.

Water-containing lubricants have been reported to exhibit significantly lower pressure–viscosity coefficients compared to traditional ones [8, 17, 23, 27]. These data are largely derived from inference of viscosity via contact tribological testing rather than direct high-pressure and high-temperature–viscosity measurements. With calculated values for 40 wt% water-containing PAG of 3.01 GPa−1 at 50 ℃, for 30 wt% water in Glycerol 4.0 GPa−1 at 25 ℃ [8, 23]. For comparison, the pressure–viscosity coefficient of conventional PAO oil is ~ 23 GPa−1. Water’s kinematic viscosity dependency on pressure is nearly absent, possibly leading to much lower effective viscosity in EHL contact zones compared to conventional oils [28]. This likely contributes to low shear resistance and superlubricity [24]. Research on superlubricity in EHL contacts with water-containing lubricants is active, though systematic investigations on the water content's influence on friction and film thickness are limited. Most studies examined water content for a given PAG or glycerol, resulting in viscosity variations of up to two orders of magnitude among considered water-containing lubricants [8, 16, 24].

Given the pressure–viscosity coefficient values reported in the literature under high pressure are derived from film thickness measurements and the large variation observed, the accuracy of such inferred methods is questionable [29]. This is important because the properties of the fluid are not being directly measured and changes in the conditions can affect the measurements, increasing variability in the results. As such, high-pressure (and high temperature) viscometry is of key importance to untangle the rheological behavior of the fluids under high pressures and higher water contents since they promote lower films. By obtaining the pressure–viscosity coefficients through high-pressure measurements, highly accurate film calculations can be obtained, enhancing the understanding of the sliding tribological behavior observed in WBLs and further assess their suitability for the automotive sector [23, 30,31,32].

For the most part, the literature regarding WBLs sliding tribological behavior with bearing steels is limited [33]. Viscosity properties of WBLs have mostly been inferred from rolling contact experiments and bulk property measurements. Many available research articles focus on lubricants with water contents below 50 wt%, also known as water-containing lubricants (WCLs), not WBLs, lubricants with water contents of at least 50 wt%, which have far greater cooling capacity. Since high water contents severely alter the rheological, tribological, and chemical behavior of the lubricants, the wear mechanisms for the WBLs are not entirely understood. Though some recent studies have focused on evaluating the tribological performance of fully formulated WBL products or on mitigating the assumed limitations of WBLs via nanotechnology or additives, the properties of the neat WBL fluids have been largely unexplored [6, 22, 34, 35]. In short, significant knowledge gaps remain that inhibit the engineering assessment of WBLs as primary lubricating fluids for rolling and sliding mechanical components.

The aim of the proposed work is to fill some of these knowledge gaps by characterizing the high-pressure rheological and tribological behavior of several WBLs. The properties of simple (water mixed with a single addition) WBLs are characterized to better understand their behavior and help explain the puzzling behavior observed in the literature regarding EHD lubrication. The tests include conventional atmospheric and high-pressure rheological experiments at high temperatures yielding directly measured viscosity properties enabling correlation with the lubrication and tribological response observed. Sliding contact tests (ball on flat) are used to assess tribological behavior. Post-test analyses of the tribological specimens are performed via interferometry and SEM/EDX to characterize the tribofilms and surface morphology. Finally, comparisons of the behavior of WBLs to traditional lubricants are made to attain a better understanding of WBLs and the potential they may have for EVs and other tribological applications.

2 Experimental

The WBLs studied were prepared using 50 wt% dilutions of deionized water and either glycerol, ethylene glycol (MEG), polyethylene glycol (PEG), or one of two polyalkelyne glycols (PAG) with base viscosities of 20 and 220 cSt. The 500 mL dilutions were mixed at 50 ℃ with a magnetic stirrer at 500 rpm for 45 min in a sealed container to prevent water evaporation.

The temperature–viscosity relationship was assessed using an Ares G2 rheometer. The parameters selected were a temperature sweep using a conical concentric cylinder geometry with a constant shear stress of 10 Pa and a temperature ranging from 25 to 80 ℃ at a rate of 2 ℃/min. Figure 1a shows a schematic of this rheometer. The kinematic viscosity of the lubricants at 40 and 60 ℃ was obtained using an automatic Houillon capillary viscometer in accordance with the ASTM D7279 standard (see Fig. 1b) [36]. With this instrument, fluid samples under the force of gravity, pass through calibrated glass tubes. Measured flow rates are used to calculate viscosity values.

Fig. 1
figure 1

Rheology testing schematic: a Ares G2 Rheometer with conical concentric cylinder geometry, b Houillon capillary viscometer, and c high-pressure falling body viscometer

Full size image

To characterize the viscosity in highly loaded contacts, a custom high-pressure viscometer was utilized [32]. The high-pressure falling body viscometer yields values for viscosity via modeling based upon Stokes flow. This correlates dynamic viscosity of the sample to the velocity of the falling body due to gravity. Thus, the viscosity of the fluid is inversely proportional to the sinker’s velocity. This correlation was established via calibration with standard fluids with an accuracy of ± 3%, and data reproducibility of 1% [32]. This viscometer can be heated as well as pressurized. The temperature selected for the viscosity measurements was 60 ℃ and the pressure was incrementally increased from one atmosphere (0.1 MPa) to 350 MPa. Figure 1c depicts the high-pressure viscometer.

To quickly assess the corrosion and tribological behavior of the lubricants in sliding point contact, a PCS Instruments High Frequency Reciprocating Rig (HFRR) was used according to the ASTM D7594 standard (see Fig. 2a) [37]. The boundary lubrication regime was achieved with a stroke length of 1 mm, a 50 Hz oscillation frequency, a load of 2 N for a total number of cycles of 225,000, a temperature of 60 ℃, and 2 mL of each lubricant [37]. The maximum Hertzian contact pressure was 0.52 GPa. The lower and upper specimens in the tribometer were made of hardened 52,100 steel and the upper specimen’s sphere diameter was 6 mm.

Fig. 2
figure 2

Sliding test schematics: a PCS instruments HFRR, and b CSEM Pin-on-Disk

Full size image

A CSEM pin-on-disk instrument was used to study the sliding behavior of the fluids under boundary lubrication regime according to the ASTM G99 [38]. Figure 2b depicts the pin-on-disk setup. The test conditions selected were 5 mL of test fluid, 0.05 m/s speed, 2 N load, a temperature of 60 ℃, a total sliding distance of 100 m, and a maximum contact pressure of 0.5 GPa. The specimens were manufactured in hardened 52,100 steel and the upper specimen’s sphere diameter was 6.35 mm.

Analysis of the wear data was performed using a scanning white light interferometry (SWLI) microscope, Zygo Newview 7300. The lower and upper specimens from the HFRR and Pin-on-Disk tests were studied using this instrument. This allowed for the quantification of the wear volumes from the tribological experiments. Optical characterization was performed using an optical microscope (OM). Scanning electron microscopy (SEM) characterization and local composition analysis via energy-dispersive X-ray spectroscopy (EDX) were performed using a Tescan Lyra 3 XMU microscope. An acceleration voltage of 20 eV and a working distance of 10 mm were used according to the ASTM E986-04 standard [39]. All tests were performed in triplicate.

3 Results and Discussion

Initially, temperature ramp tests were conducted using the Ares G2 rheometer to observe the viscosity response of the fluids to temperature variations. Figure 3 illustrates the viscosity–temperature relationship for all tested oils. In this test, traditional lubricants like a mineral ISO 68 oil and a polyalfaolefin 10 oil (PAO 10) were tested to compare their behaviors with the WBLs. Results showed the mineral oil exhibited the highest viscosity values, followed by the WPAG 220. In the lower section of the plot, it becomes apparent that the rest of the WBLs displayed a similar temperature–viscosity response to the PAO 10 oil.

Fig. 3
figure 3

Temperature sweep performed in Ares G2 rheometer to the water-based and traditional lubricants

Full size image

Table 1 presents the viscosity values obtained through Houillon viscometry for the different fluids. The densities of the fluids were obtained at room temperature (25 ℃) and assumed constant for the dynamic viscosity calculations. The WBLs showed dynamic viscosity values ranging from 1.4 to 3.1 centipoise at 80 ℃ (1.3 to 3.0 cSt). This range was lower than the traditional lubricants used in EVs currently with values that range from 4 to 6 cSt at 100 ℃ [1]. Notably, the viscosity values (~ 15 cSt) obtained for WPAG220 were exceptionally high for EV applications, making them unsuitable for further study.

Table 1 Rheological properties of the studied fluids obtained using the high-pressure and Houillon viscometer
Full size table

Turning to the high-pressure-viscosity measurements, an intriguing pattern was observed in Fig. 4. All the WBLs exhibited increased viscosity values as pressure increased. This is a crucial finding as previous literature suggested that these fluids were not expected to display such significant behavior due to the water [28]. Rather, it was expected that pressure would not have much of an effect on viscosity. However, it should be noted that the increase was not as significant as observed for traditional oils like PAO 10.

Fig. 4
figure 4

High pressure–viscosity values of the lubricants studied up to 350 MPa at 60 ℃

Full size image

This trend is evident when deriving the pressure–viscosity coefficient. A widely used method to predict the viscosity–pressure response is the Barus equation (see Eq. 1).

$$\mu \, = \,\mu_{0} \, \times \,e^{\alpha p}$$
(1)

where \(\mu\) is the dynamic viscosity at a certain pressure in Pa∙s, \({\mu }_{0}\) is the dynamic viscosity at atmospheric pressure in Pa∙s, \(\alpha\) is the pressure–viscosity coefficient in GPa−1, and \(p\) is the pressure in GPa. However, this method is known to be inaccurate for many lubricants and conditions. Another model that has shown a higher accuracy is the Roelands model (see Eq. 2):

$$\eta \left( p \right)\, = \,\eta_{p} \, \times \,\left( {\frac{{\eta_{0} }}{{\eta_{p} }}} \right)^{{\left[ {\frac{{p_{p} - p}}{{p_{p} }}} \right]^{Z} }}$$
(2)

where \(Z\) is called the pressure–viscosity index, \({\eta }_{p}\) and \({p}_{p}\) are the viscosity and pressure at the intersection of the pressure–viscosity isotherms when extrapolated to negative pressures (\({\eta }_{p}=6.31 \times {10}^{-5}\) Pa∙s and \({p}_{p}= - 0.196\) GPa), \(\eta\) is the viscosity at the desire pressure (p), and \({\eta }_{0}\) is the viscosity at atmospheric pressure and reference temperature. However, this model also showed inaccuracy for higher pressures. Thus, Bair and Quereshi proposed a method to approximate the α, compensating for the non-linear viscosity–pressure relationship observed in different fluids. This equation can be seen in Eq. 3 [30, 32]:

$$\alpha_{i} \, = \,\frac{{\ln \left( {{\raise0.7ex\hbox{${\mu_{i} }$} \!\mathord{\left/ {\vphantom {{\mu_{i} } {\mu_{i - 1} }}}\right.\kern-0pt} \!\lower0.7ex\hbox{${\mu_{i - 1} }$}}} \right)}}{{p_{i} - p_{i - 1} }}$$
(3)

where \(i\) is the data point starting from ambient pressure (\(i\)= 0), \(\mu\) is the dynamic viscosity in Pa∙s, and \(p\) is the pressure in Pa. To calculate the \({\alpha }_{0}\), Bair and Qureshi found that by assuming a linear variation of \(\alpha (p)\) over a short interval of pressure (from p = 0) so that \(\alpha \left(p\right)= {\alpha }_{0}+mp\), an accurate approximation was obtained. The value of \(\alpha \left(p\right)\) is then approximated by the secant slope.

$$\alpha_{j} \left( {{\raise0.7ex\hbox{${p_{j} }$} \!\mathord{\left/ {\vphantom {{p_{j} } 2}}\right.\kern-0pt} \!\lower0.7ex\hbox{$2$}}} \right)\, = \,\frac{{\ln \left( {{\raise0.7ex\hbox{${\mu_{j} }$} \!\mathord{\left/ {\vphantom {{\mu_{j} } {\mu_{0} }}}\right.\kern-0pt} \!\lower0.7ex\hbox{${\mu_{0} }$}}} \right)}}{{p_{j} }}$$
(4)

If now the first two elevated pressure datapoints are used, the \({\alpha }_{0}\) can be obtained by linear extrapolation according to Eq. 5 [29, 32].

$$\alpha_{0} \, = \,\frac{{p_{2} \alpha_{1} - p_{1} \alpha_{2} }}{{p_{2} - p_{1} }}$$
(5)

where \({\alpha }_{1}\) and \({\alpha }_{2}\) are obtained via Eq. 4. Table 1 presents the pressure–viscosity coefficients for the WBLs ranging from 4.1 up to 14.1 GPa−1. These values resemble those of traditional lubricants, especially for the WPEG with the highest value. Regarding PAG, the values are three-fold higher than the ones reported in literature even considering the higher temperature of the present study [23]. For Glycerol, the \(\alpha\) coefficients disagree with the literature values at 25 ℃ containing 30 wt% water [8]. Authors observed that the \(\alpha\) coefficient decreased with increased water content. Glycerol has shorter molecular chains, limiting the entanglements that could elevate the pressure–viscosity coefficient and thus displays a lower pressure–viscosity coefficient [40,41,42]. The results of the present study showed that the \(\alpha\) coefficient was less sensitive to increases in temperature and water concentration, a contrary behavior [8].

The differences between the present results and those reported in the literature may be rooted in the nature of the measurements [29]. In the present work, the \(\alpha\) coefficient was based on the analysis of direct measurement of viscosity using a falling weight viscometer, whereas the available literature values for WBLs are inferred from film thickness measurements followed by the use of the Hamrock–Dowson equation and the entrainment speed [29, 32, 43]. The analysis of the direct measurement is based on a simpler experimental setup with fewer uncertainties and may be providing more accurate values in addition to the Hertz stress limitations of the EHL film thickness test rigs, only allowing up to 0.8 GPa due to the strength of the glass [29]. In any case, additional measurements of the pressure–viscosity coefficients may enable the development of targeted additives that could yield new fluids that exhibit enhanced film formation capabilities compared to traditional lubricants.

To assess the lubricant properties as well as its corrosion behavior, HFRR tests were performed according to ASTM D7594 [37]. Figure 5 shows the friction coefficient versus the number of cycles. A declining trend in friction coefficient was noticeable for the WBLs, all of which displayed lower friction coefficients compared to PAO 10. Furthermore, after a critical number of cycles, all the friction values decreased below the friction observed for PAO 10 and in certain cases decreased to values near superlubricity, i.e., WGlycerol with an average value of 0.07. The other average values were 0.13, 0.12, and 0.19 for WMEG, WPEG, and WPAG 20, respectively, compared to a value of 0.14 for PAO 10. This may be explained by the run-in period, decreasing the local surface roughness and as such, increasing the relative film thickness. A tribochemical reaction with an initial low reaction rate could also be occurring. This can be evidenced by the film sensor data, measuring the resistivity between contacting surfaces seen in Fig. 6. Higher resistivity values suggest better lubricating properties with more separation between the contacting surfaces. As the roughness of contact area decreases, the separation of the surfaces increases. To delineate whether this behavior is due to a tribochemical reaction or a polishing effect, further analysis is required. The best film values were observed in WPAG and WPEG, followed by WGlycerol with values of 32%, 20%, and 9%, respectively, compared to 82% for PAO 10. These results are crucial in explaining the observed differences in wear rates as higher asperity to asperity contacts are to be expected for lower film values.

Fig. 5
figure 5

Coefficient of friction measurements of the studied fluids in the High-Frequency Reciprocating Rig at 60 ℃

Full size image
Fig. 6
figure 6

Film measurements of the studied fluids in the High-Frequency Reciprocating Rig at 60 ℃

Full size image

Subsequently, the wear volume values, obtained via white light interferometry analysis, are presented in Fig. 7. These values present the average volume of material removed from the surface. The lowest wear rates were observed in PAO, WPAG, and WGlycerol. To further characterize the wear track, the maximum wear track depth versus width was plotted to identify any interesting trends. WMEG and WPEG performed the worst, while WGlycerol and WPAG promoted shallower wear track depths. However, the track widths showed significant variability and were not statistically different from the other fluids, except for WMEG with the highest values.

Fig. 7
figure 7

Wear quantification of the studied fluids in the High-Frequency Reciprocating Rig at 60 ℃

Full size image

Figure 8 displays the wear tracks and visual appearance of the specimens. WGlycerol and WPEG exhibited minimal to mild corrosion, while WPAG and WMEG displayed more severe corrosion. The extent of wear damage confirms boundary or mixed lubrication conditions. The low corrosion rates and tribological response observations affirm that the fluids can be tested as-is.

Fig. 8
figure 8

Wear scars of the disks for the studied fluids in the High-Frequency Reciprocating Rig at 60 ℃

Full size image

To further characterize the tribological behavior of the fluids, pin-on-disk tests were performed. Figure 9 shows the coefficient of friction measured for the different fluids. Multiple trends were evident for the WBLs. Friction coefficients were generally higher compared to PAO, except for WPEG which exhibited the lowest friction coefficient, around 0.1.

Fig. 9
figure 9

Coefficient of friction measurements of the studied fluids in the Pin-on-disk rig at 60 ℃

Full size image

In terms of wear, Fig. 10 shows that both PAO and WPEG displayed the lowest values, with PAO having the lowest wear rate overall. This suggests that WPEG provided sufficient lubricity to reduce wear rates and reduce friction in the boundary regime. These results show that despite the literature indicating that water-containing PEG fluids promote high friction wear rates [44], higher water contents do not promote significantly higher wear rates in the case of WPEG. It is hypothesized that due to the lower corrosion rates observed for this fluid in addition to the higher film-forming capabilities, lower wear rates were promoted. Furthermore, authors found that PEG and PAG’s film-forming mechanism has two contributions, the hydrodynamic component, and an adsorbed boundary film [45, 46]. The higher corrosion rates, despite the high film formation capabilities observed, help explain the higher wear rates observed for WPAG.

Fig. 10
figure 10

Wear quantification of the studied fluids in the Pin-on-disk rig at 60 ℃

Full size image

Examining the wear scars presented in Fig. 11, the same trend was observed. Lower wear rates were noticeable for WPEG and PAO, followed by WMEG, WGlycerol, and finally WPAG, further suggesting that tribochemistry played a vital role in wear mechanisms in addition to film formation capabilities.

Fig. 11
figure 11

Wear scars of the disks for the studied fluids in the Pin-on-disk rig at 60 ℃

Full size image

To further understand how these lubricants provide sufficient lubrication and confirm that tests were run in the boundary lubrication regime, the central film thickness is calculated using the Moes–Venner equation (see Eq. 6) [43]. The Moes–Venner formula offers an approach to quantify film thickness by incorporating rigid and elastic characteristics, encompassing both isoviscous and piezoviscous behaviors. Hamrock–Dowson is only valid for piezoviscous lubricants and thus, it is not applicable to the present system. This formula facilitates the assessment of film-forming attributes in various fluid scenarios [47]. The formula presented by Moes–Venner is presented in Eq. (6) [48]:

$$H_{cen} \, = \,\left[ {\left\{ {H_{RI}^{3/2} \, + \,\left( {H_{EI}^{ - 4} \, + \,0.1\lambda^{4} } \right)^{{{\raise0.7ex\hbox{${ - 3}$} \!\mathord{\left/ {\vphantom {{ - 3} 8}}\right.\kern-0pt} \!\lower0.7ex\hbox{$8$}}}} } \right\}^{2s/3} \, + \,\left( {H_{RP}^{ - 8} + H_{EP}^{ - 8} } \right)^{ - s/8} } \right]^{1/s}$$
(6)
$$s\, = \,\frac{3}{2}\left( {1\, + \,{\text{e}}^{{ - 1.2H_{EI} /H_{RI} }} } \right)$$
(7)

For point contact geometry, \({\uplambda }_{Moes}=1\), then H can be calculated using the following terms:

$$H_{RI}^{ } \, = \,41.34 M^{ - 2}$$
(8)
$$H_{RP}^{ } \, = \,0.9089 L^{2/3}$$
(9)
$$H_{EI}^{ } \, = \,2.419 M^{ - 2/15}$$
(10)
$$H_{EP}^{ } \, = \,1.247 M^{ - 1/12} L^{3/4}$$
(11)

where \(M\) is the Moes dimensionless load parameter obtained as \(M= W/(2 U)^{0.75}\), and \(L\) is the Moes dimensionless lubricant parameter obtained as \(L=G {(2U)}^{0.25}\). U is the dimensionless speed parameter from Hamrock’s equation (\(U= \frac{{U}_{e} \times \eta }{2 \times {E}{\prime} \times {{R}{\prime}}^{2}})\) and G is the dimensionless material parameter (\(G= \alpha \times {E}{\prime})\). As such, the central film thickness values obtained via Moes–Venner equation using the viscosity values obtained through Houillon viscometry, and the pressure–viscosity coefficients from high-pressure viscometry at 60 ℃ are presented in Fig. 12. This was done to ensure the tribological tests were performed in the boundary regime.

Fig. 12
figure 12

Minimum film thickness and lambda calculation for the selected lubricants under the Pin-on-disk test conditions

Full size image

When observing the central film thickness values obtained with the Moes–Venner equation (see Fig. 12), it can be seen that PAO forms the thickest film, followed by WPAG and WPEG. These film thickness values can be used to confirm that tests were within the boundary regime, evidencing significant asperity contact and thus, suggesting that the chemistry of the fluids is also responsible for the differences in the tribological behavior observed. For instance, several authors stated that surface interactions with water-containing PAGs, formed unbonded hydrate layers with low shear resistance within the fluid film [13, 16, 21]. Other authors found that PEG and PAG’s film-forming mechanism have two contributions, the hydrodynamic component, and an adsorbed boundary film [45, 46]. Another observation was the increased corrosion rates in the wear track for the WPAG and WMEG, this could lead to differences in wear rates and further analysis is required. This is especially interesting as previous literature reports that water contents above 40 wt% for PAG result in poor film formation capabilities and enhanced wear rates [23]. The film thickness observed for these fluids is 44 and 56% lower than PAO and all fall in the boundary lubrication regime. Based on the film thickness calculations and the friction and wear behavior, the best boundary lubricants were the PAO, followed by WPEG, WGlycerol, WPAG, and WMEG.

Figure 13 shows the SEM images of the five wear scars and elemental composition analysis inside the wear track. It revealed distinct differences in the chemical reactions occurring among the different fluids, displaying different concentrations of carbon, oxygen, and iron within the wear track. In particular, it was observed that the main differences were found between WPEG and the other test fluids, suggesting the WPEG fluid did not form any tribochemical compounds and its main lubrication mechanism was the film formation. Furthermore, it can be seen that one of the most active fluids were WMEG and WPAG with the higher oxygen contents inside the wear track, confirming the results observed in the optical images and corrosion behavior. Similarly, WGlycerol seems to be reacting with the steel, evidenced by the higher carbon and oxygen contents in the wear scar. This finding highlights the importance of tribochemistry in influencing the wear mechanisms of the different lubricants. Further studies are likely necessary to better characterize the formation of tribofilms in these kinds of fluids.

Fig. 13
figure 13

SEM images and elemental analysis of the disks’ wear scars after the Pin-on-disk tests

Full size image

4 Conclusions

This study aimed to investigate the viscosity response, friction behavior, and wear characteristics of various fluids, including WBLs and traditional oils, under different conditions.

  • Kinematic viscosity measurements provided valuable insights into the dynamic viscosity values of the studied water-based lubricants, which ranged from 2 to 8.3 centipoise, a range that fulfills the low viscosity needs of EV applications. Notably, the WPAG 220 water-based lubricant exhibited very high viscosity values, rendering it unsuitable for EV applications.

  • Under high-pressure conditions, all WBLs displayed an increase in viscosity values, indicating their ability to form fluid films in highly loaded contacts. Although the increase in viscosity was not as significant as observed in traditional oils like PAO 10, this characteristic remains vital in understanding their lubricating capabilities.

  • Higher pressure–viscosity coefficient values were observed for PAG and Glycerol compared to literature values that were obtained indirectly via inference from measured lubricant film thickness experiments. These relatively high-pressure–viscosity coefficient values help explain the good lubricating properties observed for these fluids and suggest their potential to promote films in rolling element bearings.

  • Sliding friction coefficient analyses reveal that WBLs generally exhibited lower friction coefficients compared to PAO 10, with WPEG showing the lowest coefficient near 0.1. The film sensor data demonstrated that most WBLs exhibited unstable films until a higher number of cycles were reached. However, WPAG and WPEG demonstrated better film values, indicating their potential in reducing wear rates.

  • In terms of wear, PAO, WPEG, and WGlycerol had the lowest wear rates. The wear track analysis also revealed that WGlycerol and WPEG promoted lower wear track depths. SEM/EDX analysis of wear scars further supported the importance of tribochemistry in wear mechanisms, and differences in elemental composition were evident among the different fluids.

  • It is hypothesized that the corrosion products due to the high water content are promoting higher wear rates, thus minimizing corrosion may have a beneficial effect on wear for some of the fluids like PAG and MEG.

  • Overall, the best boundary lubricants were the PAO, followed by WPEG, WGlycerol, WPAG, and WMEG. Our findings suggest that WBLs have the potential to be considered as alternatives to traditional oils for certain engineering applications. While they may exhibit different friction coefficients, their lower wear rates and good lubricity properties make them promising candidates for specific conditions.

  • Further research is warranted to delve into the chemistry of these fluids and their tribochemical interactions to fully understand their wear behavior and optimize their performance for various industrial applications.