Behavior of Ionic Liquids Under Nanoconfinement Greatly Affects Actual Friction

Ionic liquids (ILs) are organic salts consisting of anions and cations that exist as liquids at room temperature. ILs exhibit many attractive properties such as negligible volatility, low flammability, and relatively high thermal stability. These properties can be varied in a controlled fashion through systematic changes in the molecular structure of their constituent ions. Some recent studies have aimed to use ILs as new lubricant materials. However, the behavior of ILs as lubricants on the sliding interfaces has not been elucidated. In this chapter, we describe the nanoand macrolubrication properties of some ILs with different types of anions using resonance shear measurement (RSM) and conventional ball-on-plate-type tribotests, respectively. This study reveals that the properties observed by RSM for nanoscale systems can provide important insights for the study of the friction coefficients (macrolubrication properties) obtained by tribotests.


Introduction
Ionic liquids (ILs) are expected to be promising candidate materials for new lubricants [1][2][3][4][5]. In particular, their stability under severe conditions, such as ultrahigh vacuum [6,7] and high temperatures [8], has attracted increasing interest. To choose ILs suitable for use as lubricants, it is important to understand the characteristics of the target materials. However, currently, the details of the lubrication mechanism of ILs are not clearly understood.
The tribological properties of ILs have been studied using a macroscopic tribotester. Most previous research has focused on the lubricating behavior of ILs in the boundary lubrication regime at high loads of several GPa and on their tribochemical reactions with solid surfaces [9][10][11][12][13]. The formation of tribochemical layers on metal surfaces from ILs containing a halogen such as fluorine under sliding conditions has been studied by X-ray photoelectron spectroscopy (XPS) [9][10][11][12], scanning electron microscope with energy dispersive X-ray spectroscopy (SEM-EDS) [9][10][11], and time-of-flight secondary ion mass spectrometry (TOF-SIMS) [12]. Therefore, ILs have also been used as additives for the formation of a tribochemical layer under high loads of several GPa [14][15][16]. When used, these layers have been considered to contribute to the reduction of friction in the system.
On the other hand, due to the interest in applying ILs as lubricants, the properties of ILs confined in a nanoscale space have also been studied by atomic force microscopy (AFM) [17][18][19] and the surface force apparatus (SFA) [20,21]. The oscillating forces observed by both AFM and SFA demonstrated the layered structure of ILs in narrow gaps. However, the relationship between these nanoscale properties and the macrolubrication properties is still not fully understood even for the same surface and ILs. Motivated by this problem, we have recently used resonance shear measurement (RSM) to show that some ILs form a layered structure in the nanoscale space created by the sliding surface [21,22]. We also revealed that a nanostructure consisting of only several IL layers had a large influence on macroscale friction.
In this chapter, we describe the nano-and macrolubrication properties of some ILs ( Table 1) with different anions by using RSM and a conventional ball-on-plate-type tribotester, respectively. This study reveals that ILs with different structures form different nanolayered structures and that their nanoscale behaviors are correlated with their macroscale tribology. In addition to providing information related to the lubrication mechanism of ILs, we also describe the principles for choosing an IL as a lubricant.

RSM for nanoscale properties of ILs
RSM was performed using an in-house resonance shear system based on an SFA [22], as shown schematically in Figure 1. The experimental setup and procedures for RSM are described in detail in a previous publication [21]. Silica sheets used as samples were prepared following the procedure reported by Horn et al. [24]. The root mean square (RMS) roughness value measured by AFM (Toyo Corporation, Agilent 5100 AFM/SPM Microscope) over an area of 5 × 5 μm 2 for the silica sheets was 0.31 nm. Using RSM, the resonance curve between the molecularly smooth silica sheets was measured across IL films at a surface separation D with a resolution of 0.1 nm. The value of D was determined from the fringes of equal chromatic order (FECO) analysis. In brief, two back-silvered silica sheets (thickness of ca. 2-4 μm) were glued onto cylindrical quartz lenses (with a radius of curvature (R) of ca. 20 mm) and mounted onto the RSM system. The RSM system was composed of the upper surface unit suspended by a pair of vertical leaf springs and the lower surface unit mounted on a horizontal leaf spring. The upper surface unit was connected to a four-section piezo tube. In this case, it can be laterally oscillated at various frequencies (ω) by an application of a sinusoidal input voltage (U in ). The deflection (Δx) of the leaf spring was detected as an output voltage (U out ) by a capacitance probe (Microsense 4830, Japan ADE Ltd.). Then, resonance curves were recorded at various D values as the normalized amplitude (U out /U in ) as a function of the frequency ω. RSM was conducted at room temperature (295 ± 0.5 K) and at a humidity of less than 25% maintained by placing silica gel in the sample chamber. The RSM system measured the surface force and resonance shear response by continuously changing the thickness of the liquid film confined between two solid surfaces with a nanometer resolution. The liquid thickness was controlled and determined using interferometric methods in the surface force apparatus. The shear response via resonance method provided a sensitive method for detecting the tiny changes in the liquid properties between the substrates, allowing us to evaluate the viscosity change associated with liquid structuring, frictional/lubricational property, as well as other properties, by simply changing the liquid film thickness. Figure 2 shows the typical resonance shear curves for different surface separations. The resonance curves for the two reference states of separation in air (AS) and for silica-silica contact (SC) were measured prior to the RSM of a liquid (Figure 2). In the absence of a liquid, the system parameters (mass, damping parameter, and spring constant for the apparatus) determined the AS and SC resonance curves. In the presence of a liquid, the resonance curve at long distances showed a peak at a low frequency that was almost the same as the AS peak.
The peak intensity U out /U in was lower than the AS peak intensity, corresponding to the energy dissipation due to the bulk viscosity of the liquid. With decreasing D value, the amplitude of the resonance peaks first decreased, and then, the peak disappeared with a further decrease in the distance. For a further decrease in the surface separation, broad resonance curves were observed at intermediate frequencies between the AS and SC frequencies, and the peak frequencies started to shift toward the SC peak because of the weak coupling of the upper and lower surfaces mediated by the confined liquid [25]. When the applied load (N) was further increased by decreasing the surface separation, the peak frequency shifted further toward the SC peak and the amplitude increased, until finally, for a surface separation of 0 nm, the amplitude was almost identical to that of the SC peak. This means that the liquid was completely removed from the gap between the silica surfaces [26]. Similarly, when the applied load (N) was further increased while the surface separation was fixed at a certain distance, the peak frequency also shifted further toward the SC peak frequency, while the amplitude increased. This means that the liquid remained between the silica surfaces at a certain surface separation due to the stronger coupling of the upper and lower surfaces with the increasing applied load. The RSM system described above was used to study the nanolubrication properties of ILs between smooth silica surfaces. Figure 3

(a) and (b) shows the resonance curves for [DEME] [TFSI] and [DEME][BF 4 ] ILs confined between silica surfaces at various separation distances.
For reference, the resonance curves for AS and SC were measured prior to the resonance shear measurement of the ILs (Figure 3). In the absence of the ILs, the system parameters (mass, damping parameter, and spring constant for the apparatus) determined the AS and SC resonance curves. In the presence of [DEME][TFSI], the resonance curve at D = 228.1 nm showed a peak at a frequency of 185 rad s −1 that was almost the same as the AS peak. The peak intensity U out /U in was lower than the AS peak intensity, corresponding to the energy dissipation due to the bulk viscosity of the IL. With decreasing D value, the amplitude of the resonance peaks started to decrease at 10.2 nm and disappeared at D = 4.9 nm. In the range from D = 4.9 to 2.5 nm, broad resonance curves were observed at intermediate frequencies between the AS and SC frequencies, and the peak frequency started to shift toward the SC peak because of the weak coupling of the upper and lower surfaces mediated by the confined [DEME][TFSI] [21]. When the applied load (N) was increased further, the surface separation did not change and remained at 2.5 nm (Figure 4); however, the peak frequency shifted further toward that of the SC peak and the amplitude increased. This means that [DEME][TFSI] remained between the silica surfaces at the separation of 2.5 nm, and the coupling of the upper and lower surfaces became stronger with the increased applied load. Resonance curves for AS and SC are also shown [22]. Solid lines denote the best fit curves to a physical model [27].
The resonance shear behavior of [DEME][BF 4 ] (Figure 3(a)) was similar to that of [DEME] [TFSI], except in the small D region below 2.5 nm (Figure 3(b)). The resonance curves at 186 rad s −1 did not change in the D range from 220.6 to 95.5 nm, and their amplitudes started to decrease at D = 20.8 nm. The resonance frequencies of these curves were almost the same as the frequency of the AS peak. The amplitude of the resonance peaks gradually decreased with decreasing D from 20.8 to 11.4 nm and disappeared at D = 9.4 nm. This distance is higher than the disappearance distance (D = 4.9 nm) for [DEME] [TFSI]. In the range from D = 9.4 to 2.1 nm, broad resonance curves were observed at intermediate frequencies, and the peak frequencies started to shift toward the frequency of the SC peak because of the weak coupling of the upper and lower surfaces mediated by the confined [DEME] [BF 4 ]. When the applied load (N) was further increased, the surface separation remained at 2.1 nm (Figure 4), indicating that [DEME] [BF 4 ] remained trapped between the silica surfaces.  Figure 5 plots the relative intensity (I r ) of the resonance peak, that is, the ratio of the peak intensity for confined ILs is divided by the SC peak intensity as a function of applied load. This intensity ratio is a measure of the lubrication behavior of the confined ILs, with smaller values indicating better lubrication behavior. With an increase in the applied load from 0. 21 4 mN).

Dynamics of ILs by changing the load applied to the two friction surfaces
The plateau of the relative intensity under applied loads (>0.4 mN) indicated that the IL layer confined between the silica surfaces maintained its lubricating properties.

Effective viscosity (η eff ) of the confined ILs measured to quantitatively measure lubrication performance
We analyzed the resonance curves using a previously developed physical model [27] to obtain a quantitative understanding of the properties of the confined ILs. The details of the analytical procedure are described in the literature [21,27]. Figure 6 plots the effective viscosity (η eff ) obtained for the ILs using the model versus the separation distance between the silica surfa-    We suppose that the rapid increase in the effective viscosity with decreasing distance observed in Figure 6 corresponds to the distance at which the formation of a solid-like structure due to the confinement is initiated. For [DEME] [BF 4 ], the sharp viscosity increase began at a clearly shorter distance than that for [DEME][TFSI], indicating that this ammonium salt restructures more readily than the TFSI salt, or in other words, that the BF 4 salt is more easily crystallized. This consideration is also supported by the results of the crystallization temperature measurement using differential thermal calorimetry (DSC) [28]. As indicated by the DSC measurements, [DEME][BF 4 ], for which the viscosity increases at a relatively long distance, shows a distinct crystallization temperature, whereas the TFSI salt only shows a glass transition temperature but not a crystallization temperature. The same trend is found for aromatic ILs.  In the RSM, the facing silica surfaces were completely separated to avoid partial contact and to allow the analysis of the boundary lubrication property of the confined liquid at a certain separation distance. In this configuration, an increase in the viscosity of the lubricant layer under confinement directly leads to increased friction at the sliding interface.

Reciprocating-type tribotests for evaluation of macroscopic properties
Friction measurements were carried out using a conventional reciprocating tribotester, TRIBOGEAR TYPE 38 (Shinto Scientific Co. Ltd., Tokyo), using a glass ball of 10 mmφ and a glass plate. To obtain a clean surface, the glass ball and glass plate were treated in fresh nitric acid at 373 K for 75 min. The RMS roughness values measured by AFM for the glass ball and plate were 9.9 and 1.2 nm over an area of 5 × 5 μm 2 , respectively. A schematic of the tribotester is shown in Figure 8. The measurements were performed at a movement distance of 10 mm, with a sliding velocity ranging from 5.0 × 10 −4 to 3.0 × 10 −2 m s −1 under an applied load of 196-980 mN at room temperature (295 ± 0.5 K). The friction force was measured by an all-in-one load converter from a gauge attached to the sample holder and was recorded as a function of time. The friction coefficient (μ) was calculated as the friction force divided by the normal load. Friction coefficients obtained for at least five trials were recorded and averaged. The Stribeck diagram of the friction behavior is used to explain the rubbing phenomena occurring in lubricated contacts [30]. A schematic representation of the Stribeck diagram is shown in Figure 9. For high values of ηV/N, the friction coefficient is linearly ascending due to fluid film lubrication, and the friction is related to the viscous dragging forces in the fluid film. When the load increases or fluid viscosity and/or velocity decreases, the ηV/N factor falls. Then, the fluid film becomes thinner, and consequently, the friction coefficient decreases down to the minimum value. For even smaller ηV/N values, the fluid film thickness is further reduced and solid-to-solid contact starts to occur, leading to an increase in the friction coefficient as the ηV/N factor decreases. Such a rise in the friction coefficient is also related to the fluid viscosity increase in some regions of the contact area under high contact pressure. These phenomena characterize the mixed lubrication regime. Further reduction in the ηV/N factor strengthens the solid-to-solid contact, and the film thickness becomes smaller than the height of the surface asperities, leading to the transition of the boundary lubrication regime. We studied the boundary lubrication properties of ILs between a glass ball and a glass plate by using a macroscopic tribotester.  4 ] from mixed lubrication to boundary lubrication. As a control, the friction coefficient was measured between the glass ball and glass plate without an IL and was found to be ≈0.7 at the sliding velocity of 1.0 × 10 −3 m s −1 and under a normal load of 196 mN. Thus, the presence of ILs at the glass-glass interface dramatically reduced the friction coefficient to less than 20% of the value obtained without ILs.

Comparison of measurement results revealing a correlation between the macroscale friction phenomena and the physicochemical properties of ILs in nanospace
For a confined IL, boundary lubrication is dominant, and the contribution of hydrodynamic lubrication due to the change in lubricant thickness is not effective.  We now discuss the possible structural origin for the observed viscosity effects. Canova 4 ]. Based on these values, we determined that in both systems, three layers of ILs were trapped between the silica surfaces. The dramatic reduction in the friction coefficient in "the boundary lubrication region" with ILs was presumably due to the presence of the three layers of ILs between the silica surfaces under an applied load of 196 mN.
These results indicated that the tribotester macroscopic tribological properties correspond to the nanoscale lubrication properties obtained from nanoscopic RSM.

Conclusion
We performed RSM and reciprocating-type tribotests to evaluate the friction properties of lubrication systems consisting of some types of ILs between the silica surfaces.