Ionic liquids on uncharged and charged surfaces: In situ microstructures and nanofriction

In situ changes in the nanofriction and microstructures of ionic liquids (ILs) on uncharged and charged surfaces have been investigated using colloid probe atomic force microscopy (AFM) and molecular dynamic (MD) simulations. Two representative ILs, [BMIM][BF4] (BB) and [BMIM][PF6] (BP), containing a common cation, were selected for this study. The torsional resonance frequency was captured simultaneously when the nanoscale friction force was measured at a specified normal load; and it was regarded as a measure of the contact stiffness, reflecting in situ changes in the IL microstructures. A higher nanoscale friction force was observed on uncharged mica and highly oriented pyrolytic graphite (HOPG) surfaces when the normal load increased; additionally, a higher torsional resonance frequency was detected, revealing a higher contact stiffness and a more ordered IL layer. The nanofriction of ILs increased at charged HOPG surfaces as the bias voltage varied from 0 to 8 V or from 0 to —8 V. The simultaneously recorded torsional resonance frequency in the ILs increased with the positive or negative bias voltage, implying a stiffer IL layer and possibly more ordered ILs under these conditions. MD simulation reveals that the [BMIM]+ imidazolium ring lies parallel to the uncharged surfaces preferentially, resulting in a compact and ordered IL layer. This parallel “sleeping” structure is more pronounced with the surface charging of either sign, indicating more ordered ILs, thereby substantiating the AFM-detected stiffer IL layering on the charged surfaces. Our in situ observations of the changes in nanofriction and microstructures near the uncharged and charged surfaces may facilitate the development of IL-based applications, such as lubrication and electrochemical energy storage devices, including supercapacitors and batteries.


Introduction
Ionic liquids (ILs) are liquid molten organic salts composed of cations and anions with melting points of < 100 °C. ILs have garnered substantial research interests owing to their exceptional properties, such as low vapor pressure, non-flammability, excellent thermal conductivity, high ionic conductivity, high chemical and thermal stabilities, and wide electrochemical stability windows [1][2][3]. These properties render ILs extremely promising for a range of electrochemical and engineering applications, e.g., energy storage and lubrication [4][5][6][7][8][9]. It is noteworthy that the adsorption/moving dynamics and microstructures of ILs adjacent to surfaces are vital to electrochemical reactions occurring at electrode surfaces, as well as to the effective boundary lubrication of sliding contact surfaces [4,10]. For instance, it was observed that the ionic conductivity of confined ILs in silica nanoparticles decreased compared with that in the bulk, which was likely due to the low ion mobility in the nanoconfined ILs [11]. A smaller diffusion coefficient was recorded for an IL confined between graphite walls compared with that in the bulk [12]. However, both experimentalists and theorists discovered the acceleration of ion diffusion in confined systems [13][14][15][16]. For example, the diffusion coefficient of ions in phosphonium bis(salicylato)borate IL confined in 4 nm Vycor pores is discovered to be 35-40 times greater than that in the bulk [15].
The dynamics of ions in ILs on charged interfaces is more complicated than that on uncharged surfaces. Rajput et al. [17] observed that the diffusion of ions in an IL confined inside a slit graphitic nanopore with a width of 5.2 nm was greater at charged walls compared with neutral ones. A more recent study by Wu et al. [18] indicated an opposite trend, i.e., ions at charged surfaces propagated slower than those at neutral ones, and the ions propagated slower as the charge density increased.
It was reported that the frictionless surface of confined walls exhibited a strong correlation with the liquid diffusion acceleration and reduction [19][20][21]. We previously employed nanofriction measured via atomic force microscopy (AFM) to describe the mobility of supported polymer films [22]. Recently, "nanofriction" has been considered as a good indicator of interfacial ion mobility for describing the dynamics of ILs at interfaces [13,23]. This is because nanofriction is sensitive to variations in the microstructures of ILs at sliding surfaces, e.g., the interfacial IL ordering [24][25][26]. Increased nanofriction is discovered in confined ILs as the number of ion layers decreased [25,27,28]. The higher number of ordered IL layers with a greater concentration of excess ions closer to the confined surfaces contributed to the increased nanofriction [25]. However, a different effect of the ordered IL layers on the nanofriction was also reported in Refs. [24,28,29], in which well-formed interfacial structures of ILs facilitated the formation of a lubricating sliding plane, resulting in enhanced lubricity and lower nanofriction. This clearly contradicts the effect of IL microstructures on the nanofriction at IL-solid interfaces.
The microstructures of ILs can be manipulated by charging the confined surfaces to create an electric field across the nanoscale IL film [30,31]. Both experimental and theoretical findings indicate that nanoconfined charged surfaces enable IL ordering into alternating charge layers [25,[32][33][34]. In particular, the microstructural changes of near-surface ILs, i.e., liquid-to-solid phase shifts, were discovered to be induced by surface charging [35], which was more pronounced for metallic surfaces [36]. In addition, ordered solid-phase ILs resulting from surface charging are expected to decrease nanoscale friction by preventing direct surface-to-surface contact [36]. We previously combined AFM measurements with molecular dynamics (MD) simulations to unravel the microscopic mechanisms that determined the nanofriction response in terms of IL microstructures [13]. In this study, it was discovered via simulation that changes in the nanofriction and microstructure occurred simultaneously at IL-solid interfaces and were associated with each other [13]. However, in situ nanofriction-microstructure changes of near-surface ILs have not been observed experimentally, particularly when the surfaces are charged.
In this study, colloid probe AFM was employed to elucidate in situ nanofriction-microstructure changes of ILs on uncharged and charged surfaces. SiO 2 (silica) and polymethyl methacrylate (PMMA) colloid probes were used instead of standard sharp tips. The tribointerface in IL systems scanned via AFM can be regarded as a complex "black box" (the left panel in Scheme 1(a)) composed of an AFM probe, ILs, and solid surfaces with characteristic roughness, chemical heterogeneities, charges, etc. Two questions arise: How can one consider the contribution of solid surfaces, and how can one detect the simultaneous nanofrictionmicrostructure changes of ILs occurring at IL-solid interfaces? Hence, in this study, we utilized colloid probes to model solid surfaces (the right panel in Scheme 1(a)) to directly detect nanofriction at the IL-solid interface. The torsional resonance frequency www.Springer.com/journal/40544 | Friction is captured simultaneously when the nanoscale friction force is measured at a specified normal load. This frequency is an indicator of the contact stiffness [37,38], which reflects in situ IL microstructural changes. A higher contact stiffness is indicative of more ordered IL layers, as confirmed by our simulation. The two extensively investigated ILs, 1-butyl-3methylimidazolium tetrafluoroborate, ([BMIM][BF 4 ], BB) and 1-butyl-3-methylimidazolium hexafluorophosphate, ([BMIM][PF 6 ], BP), serve as well-defined model systems for investigating in situ changes in the nanofriction and microstructure of ILs on uncharged and charged surfaces.
We employed bare SiO 2 and PMMA probes to simplify the solid surfaces that were in contact with the ILs and simultaneously captured the nanofriction and torsional resonance frequency at the IL-solid interfaces. The nanofriction obtained is a good indicator of the interfacial ion mobility for describing the dynamics of ILs at interfaces and it is vital to electrochemical reactions and boundary lubrication. It is noteworthy that the colloid probe propagates closer to the substrate when the applied normal load increases, thereby resulting in varying confinements for the ILs (the top panel in Scheme 1(b)). Hence, a slit-pore model was used in the MD simulation to mimic the colloid probe approaching uncharged and charged surfaces (the bottom panel in Scheme 1(b)) to investigate the IL ordering at interfaces as well as to confirm the experimentally detected microstructural changes (Scheme 1(c)). A general overview of the interdependencies among nanofriction, confinement, microstructure, IL ordering, and ion dynamics is shown in Schemes 1(b) and 1(c).

Preparation of AFM colloid probes
Dodecenyl succinic anhydride (DDSA) and benzyldimenthylamine (BDMA) were purchased from Sinopharm Chemical Reagent Co., Ltd., China and Araldite Resin CY212 (CY212) was obtained from Sigma-Aldrich, USA. Rectangular AFM PNP-DB cantilevers were purchased from Nanoworld, Switzerland. SiO 2 (silica) and PMMA microspheres (20 μm in dimension) were obtained from EPRUI Biotech Co., Ltd., USA, The AFM colloid probes were prepared by attaching SiO 2 microspheres to rectangular cantilevers (type: PNP-DB, and the cantilever 1 with a nominal spring constant of 0.48 N/m was used) with epoxy glue (BDMA:DDSA:CY212 = 1:10:10 in volume ratio). The surface morphologies of the AFM colloid probes were observed using the scanning electron microscope (SEM; JSM-IT500HR, JEOL, Japan) and AFM. Silica and PMMA colloid probes were glued using a double-sided tape on a mica support, which had been placed onto a steel disc to enable magnetic fixation under the scanning tip in tapping mode with a 240AC-NA cantilever (tip radius < 7 nm, OPUS, Mikromasch Europe, Germany). The focused scanning positions of these colloid probes were monitored using the optical microscope equipped in the AFM system prior to imaging.

Preparation of IL films on mica and highly oriented pyrolytic graphite (HOPG) surfaces
Mica and HOPG were purchased from Electron Microscopy Sciences, USA and MikroMasch, USA, respectively. ILs, 1-butyl-3-methylimidazolium tetrafluoroborate (BB, purity ≥ 98%) and 1-butyl-3methylimidazolium hexafluorophosphate (BP, purity ≥ 97%), were purchased from Sigma-Aldrich, USA. To prepare films on the surfaces of mica and HOPG, the ILs were dissolved in ethanol to yield IL solutions with a concentration of 10 −3 mL IL in 1 mL ethanol. The as-prepared IL solutions (1 μL) were deposited either on mica or HOPG substrates (area = 1 cm × 1 cm); subsequently, the ethanol was allowed to completely evaporate in vacuum at 35 °C for 12 h to obtain IL films. The morphologies of BB and BP on the mica and HOPG surfaces were scanned in our previous study [27] via tapping mode AFM using silicon probes, as shown in Fig. S1 in the Electronic Supplementary Material (ESM). Extended layering structures were clearly observed, and some of the IL drops resided on the layering-assembled IL films, forming drop-onlayer structures.

In situ friction force and torsional resonance frequency measurements
The friction force measurements were performed on a Dimension Icon AFM (Bruker, Germany) in contact mode under ambient conditions. The as-prepared SiO 2 and PMMA colloid probes were employed throughout the friction measurements with a scan rate of 2 Hz under the probe at a 90° scan angle to the cantilever long axis. The lateral calibration of cantilevers supporting the colloid probes was performed based on the method of Liu et al. [39]. The in situ changing torsional resonance frequency was simultaneously recorded by capturing the lateral friction force and it is a useful indicator of the probe-IL contact stiffness. The torsional resonance frequency measurements were conducted by performing a frequency sweep with "high speed data capture" in the Dimension Icon AFM to obtain the amplitude spectra (lateral deflection vs. frequency). Surface charging was performed by applying biased sample voltages (−8 to 8 V) to the HOPG-supported ILs on the AFM sample holder, where the probe was connected to the ground. SiO 2 and PMMA colloid probes were utilized to measure the net adhesion forces of the ILs using the probes based on a continuous approach via AFM. The adhesion forces reported in terms of cantilever deflection vs. the separation distance could be acquired as the force jump during retraction, which represented the pull-off force required to separate the probe after contact.

MD simulation
The slit model was used to investigate the molecular properties of fluids under confinement [40][41][42]. In our typical AFM experiments, SiO 2 microspheres and HOPG were used as the AFM probe and substrate with ILs between them. Hence, we constructed a slit pore model with a SiO 2 slab as the probe and a three-layer graphene sheet as the substrate in the www.Springer.com/journal/40544 | Friction simulation. The ILs, i.e., BB and BP, were confined in the SiO 2 -graphene slit pore (Fig. 1). The SiO 2 slab and graphene sheets were arranged parallel to the X-Y plane and fixed as such with a dimension of 6.20 nm × 6.24 nm, thereby allowing the flow of sufficient ILs to obtain reliable statistical results. Slit pores with varying pore widths of 1, 2, 5, and 10 nm were utilized to mimic the probe approaching the surface at different normal loads [43,44], in which the probe was placed closer to the surface to accommodate a relatively high normal load. We applied the voltages of 8 and −8 V on the ILs confined in a 2 nm SiO 2graphene slit pore along the Z-axis using the constant potential method to determine the effect of surface voltages on the IL microstructures. At the voltages of 8 and −8 V, the electric field was directed toward SiO 2 and graphene, respectively. The number of confined ion pairs (Table S1 in the ESM) was calculated using an effective volume based on the bulk density of the ILs [45,46] in the slit pore, which provided an accurate estimate of the density of the confined ILs. For example, we placed 119, 238, 597, and 1,194 ion pairs of BB inside the modeling systems with slit pore widths of 1, 2, 5, and 10 nm, respectively (Table S1 in the ESM). A three-dimensional (3D) periodic boundary condition was applied to the simulation box. Periodic boundary conditions were applied in all directions, and a 5 nm vacuum gap was set on the top plane of the SiO 2 slab and the bottom plane of the graphene sheets to eliminate image effects in the Z-axis direction.  Table S2 in the ESM.
The refined force field of imidazolium-based ILs developed by Liu et al. [47] was employed to describe BB and BP in our study. This model was modified based on the assisted model building with energy refinement force field and had been proven suitable for simulating the microstructures of ILs at solid interfaces, which was consistent with experimental results [47,48]. It was noteworthy that the Hertzian contact pressure in the graphite system with the SiO 2 probe was estimated to be 5-15 MPa at varying normal loads in the AFM measurements, whereas this pressure was 11-63 MPa at different confinements in the MD simulation. Because the order of magnitude of the pressure during the AFM experiments agreed well with that in the MD simulation, it was inferred that the force field parameters used in our systems were reasonable and acceptable. The force-field parameters for SiO 2 were extracted from Ref. [49], whereas those for HOPG were obtained from Ref. [50]. In the present study, we constructed SiO 2 slit pores without considering the surface chemistry, such as the surface hydroxylation of SiO 2 , as in Ref. [51]. The Lennnard-Jones 12-6 and Coulombic potential models were used to determine the van der Waals (vdW) and electrostatic interactions, and the relevant parameters are listed in Table S2 in the ESM. The Lorentz-Berthelot combining rules were used to determine the vdW interactions for different types of atoms. All equilibrated simulation systems were simulated in an NVT ensemble at 300 K using the Nose-Hoover thermostat with a damping time of 100.0 fs. Noting that an ensemble with three parameters (the number of particles N, the volume V, and the absolute temperature T) is called the NVT ensemble. The motion of atoms was described using the classical Newton equation, which was solved using the velocity-Verlet algorithm with a 1.0 fs time step. The cutoff distance of the vdW and electronic interactions was 1.2 nm, and the particle-particle particle-mesh method was employed to calculate long-range electrostatic interactions [52,53]. All MD simulations were performed using the large-scale atomic/molecular massively parallel simulator package [54]. The simulation system was relaxed and optimized for 20 ns to obtain an equilibrium state, and the trajectory coordinates of the molecules were obtained within 20 ns, with a storage frequency of 100 time steps.

Morphology of AFM colloid probes
SiO 2 and PMMA microspheres with a diameter of ~20 μm were attached at the end of the rectangular cantilevers, as shown in Figs. 2(a) and 2(b), respectively. Enlarged AFM images show the local surface features of the obtained SiO 2 and PMMA colloid probes. A slight difference is observed between the roughness of the SiO 2 and PMMA colloid probes, with the surface of the SiO 2 probe (R q = 34.6±1.7 nm; Rq is the rootmean-squared roughness.) being slightly smoother than that of PMMA (R q = 38.4±1.5 nm).
The nanoscale roughness significantly affects the friction force and the roughness effects depending on the contact stress [55]. In an earlier study, it was discovered that the friction coefficient increased monotonically with the roughness of the contacts mediated by the IL ethylammonium nitrate using AFM colloid probes [56]. Furthermore, MD simulations show that friction at contacts lubricated by the ILs can either increase [57] or decrease [58] with the surface roughness. The measured roughness presented in Fig. 2 shows that the SiO 2 and PMMA colloid probes possess similar surface roughnesses; therefore, any significant differences caused by this parameter are negligible.

Nanofriction of ILs on uncharged surfaces
The friction force (F F ) of ILs BB and BP on the uncharged mica and HOPG surfaces as functions of normal load (F N ) are shown in Figs. 3(a) and 3(b), respectively, using the SiO 2 and PMMA colloid probes. The average friction coefficient μ (cf. Table 1) is determined from the gradient of F F vs. F N based on the modified Amontons' law of friction, where F 0 is the friction force at zero applied normal load. The nonzero intercept, F 0 , depends on the lubricant between the sliding surfaces, as well as the roughness and compressibility of the probe/sample [56,59,60]. Greater friction forces are observed in the ILs on mica ( Fig. 3(a)) compared with those on HOPG ( Fig. 3(b) and probes. This may be because the surface of mica is negatively charged owing to the presence of K + [61], whereas the HOPG surface is neutral. Previously, we performed X-ray photoelectron spectroscopy (XPS) to confirm that the imidazolium ring in BB or BP interacted vigorously with the surface of mica via a covalent bond formed between the N atom of the ring and the Si atom of mica [27]. For ILs on HOPG surfaces, Yokota et al. [62] and Sha et al. [63] speculated the preferential interaction of the imidazolium ring in the cation [BMIM] + with the surface of graphene. Furthermore, it was discovered that the tilt angle of the plane of the ring with respect to the surface normal significantly affected the friction, in which a smoother and better-defined shear plane facilitated friction reduction [64].
As shown in Table 1, the friction coefficient μ measured for BP supported by mica with a SiO 2 colloid probe is 0.23±0.021, whereas that for BB is higher (  = 0.40±0.017). By contrast, as shown in Fig. 3(b) and Table 1, the friction force and the resulting friction coefficient are greater (  = 0.064±0.003) in the HOPGsupported BP than those in BB (  = 0.010±0.001).
Furthermore, the friction force F F was affected by the probe material, in which the values of F F for a specified IL were higher when PMMA colloid probes were used instead of SiO 2 (Fig. 3(a)). The resulting friction coefficients  on the mica surface obtained using PMMA colloid probes are 0.32±0.012 and 0.53±0.021 for BP and BB, respectively, which are higher than those obtained using SiO 2 (0.23±0.021 and 0.40±0.017 for BP and BB, respectively) ( Table 1).

Adhesion forces
The observed higher friction is contributed primarily by the greater probe-IL interactive strength [65] and enhanced IL ordering [25]. Our previous findings combining AFM measurements and the corresponding state theory reveal that stronger liquid-solid interactions result in a higher friction coefficient at the liquidsolid interface [65]. To substantiate this, we captured the first force-distance curve for the colloid probe at the BB or BP surface supported by mica and HOPG to quantitatively examine the interactive force of the colloid probe toward the IL (Figs. 3(c) and 3(d)). The obtained net adhesion force Net ( ) F provided a direct measure of the probe-IL interactive strength. We discovered that the Net F ranked in the following order (Table 1) Table 1.
| https://mc03.manuscriptcentral.com/friction friction is measured by the PMMA probe on the ILs as compared with the SiO 2 probe. The different adhesion forces obtained are primarily caused by vdW, electrostatic, and capillary meniscus forces, which are significantly associated with the surface properties of the probe and substrate, as well as the ILs between them [66][67][68].
It is noteworthy that when the same colloid probe was used on the mica substrates, the measured probe-IL interaction, F Net , was significantly higher in BB than that in BP, which resulted in stronger friction in BB. By contrast, on the HOPG substrates, F Net was higher in BP than that in BB when the same colloid probe was used, demonstrating stronger friction in BP. It is noteworthy that a sharp break occurred in the detachment of the SiO 2 probe from the BB on the HOPG surface, whereas a gradual response was observed from BP. This indicates that BP is pulled away by the SiO 2 probe during detachment owing to the enhanced electrostatic interactions of BP with the SiO 2 probe, stemming from the higher electron densities in [PF 6 ] − anions [69] compared with those in [BF 4 ] − . The gradual detachment behavior is not observed in the colloid probes with ILs on the mica surface, which is likely due to the stronger interactions between the ILs and the mica substrate [27].

Contact stiffness
The contact stiffness can be indirectly assessed [70] from the torsional resonance frequency, which can be simultaneously measured with the friction force [37,38] indicative of the IL ordering at interfaces. The higher contact stiffness associated with enhanced IL ordering contributes partially to the higher friction. Smith et al. [25], An et al. [27], and Cooper et al. [28] discovered that more ordered ILs resulted in a higher friction force. This might be because IL ordering resulted in a greater concentration of excess ions, thus resulting in a stronger "interlayer attraction for the same contact area" [25]. As shown in Fig. 4, the torsional resonance frequency is higher in a specified IL when the PMMA colloid probe is used instead of SiO 2 , reflecting a higher contact stiffness at the PMMA-IL interface. By using the same colloid probe, the torsional resonance frequency was discovered to be higher at the mica-supported BB surface than that at the BP surface.

Substrate effect on in situ nanofriction-microstructure changes of ILs on uncharged surfaces
To further investigate the contribution of the substrate to the in situ nanofriction-microstructure changes of ILs on uncharged surfaces, we investigated HOPGsupported IL systems using SiO 2 colloid probes (Fig. 5). As the normal load increased, i.e., the probe penetrated through an additional ion layer closer to the substrate [44] (cf. Scheme 1(b)), the position of the torsional resonance peak on the HOPG-supported ILs shifted to a higher frequency, indicating a higher contact stiffness. The contact stiffnesses and friction forces at the SiO 2 -BP interfaces (Fig. 5(a)) are higher than those at the SiO 2 -BB interfaces (Fig. 5(b)) under a similar range of applied normal loads. Compared with the mica-supported BB system shown in Fig. 4(b), the contact stiffnesses obtained on the HOPGsupported BB surfaces using SiO 2 colloid probes are lower and hence indicate smaller friction forces ( Fig. 5(b)). It is noteworthy that the colloid probes and the substrates are not in direct contact during the friction measurements, as confirmed by the higher contact stiffness on the IL-coated substrates as compared with the bare substrates (Figs. 4 and 5 and Fig. S3 in the ESM).

In situ nanofriction-microstructure changes in ILs on charged surfaces
It was previously reported that the effect of surface charging on the IL microstructures at interfaces resulted in a considerable variation in the nanofriction forces owing to the surface charging density [71]. However, the correlation between charge-dependent nanofriction and near-surface IL microstructures, i.e., in situ nanofriction-microstructure changes in ILs on charged surfaces, is yet to be elucidated. Using the HOPGsupported ILs (BB and BP) system as an example, the nanofriction and torsional resonance spectra were measured simultaneously using SiO 2 colloid probes under varying negative and positive bias voltages (Fig. 6). It is discovered that nanofriction indicates a monotonic dependence on the surface charges [72], although not for all cases [32,71]. In our case, we observed that the nanofriction of BB and BP on the charged HOPG surfaces increased as the positive-bias voltage increased    At a specified normal load (e.g., F N = 63 nN), the simultaneously recorded torsional resonance frequencies in the BB increase as the positive-or negative-bias voltages increase (458 → 480 kHz when 0 → 8 V (Fig. 6(c)), and 458 → 481 kHz when 0 → −8 V (Fig. 6(e))). This phenomenon is less evident in the BP at a normal load of 112 nN (717 → 727 kHz when 0 → 8 V (Fig. 6(d)) and 717 → 723 kHz when 0 → −8 V (Fig. 6(f))). The torsional resonance frequencies at other normal loads for both BB and BP exhibit the increasing trend when a higher sample bias is applied (Fig. S4 in the ESM). This implies there is a higher contact stiffness when a higher sample bias (B S ) is applied, corresponding to more ordered ILs at the interfaces, which also contributes to the simultaneously observed higher friction force [25].

Ion orientation evaluated using orientational order parameter
To obtain more insights into the changes in contact www.Springer.com/journal/40544 | Friction stiffness on uncharged and charged surfaces, the variation in the orientational order parameter (denoted as S) as the SiO 2 probe approached the ion layers at graphite surfaces was investigated further via MD simulations. The orientational order parameter S is defined as the ensemble average of the second Legendre polynomial, which is expressed as [73,74]: where θ is the angle between the imidazolium ring normal of the cation [BMIM] + and the surface normal (Z-axis). The values of S vary from −0.5 to 1, and S = −0.5 indicates that the imidazolium ring normal is perpendicular to the surface normal (θ = 90°); meanwhile, S = 1 indicates that they are parallel to each other (θ = 0° or 180°), corresponding to the two typical orientation structures of the imidazolium ring (structures (I) and (II)) in Fig. 7(a). In the disordered isotropic bulk phase of the IL, S equals 0. The orienttational order parameter S offers a clear depiction of the orientational preference of [BMIM] + in the X-Y plane.  In the AFM experiments, the colloid probe approached the substrate as the applied normal load increased. Hence, a slit-pore model was employed to mimic the probe approaching the substrate by varying the confinement widths (1, 2, 5, and 10 nm), in which a smaller width corresponded to a higher normal load. The slit pore was composed of a SiO 2 probe and a three-layered graphene substrate. The number density profiles suggest that the ions of ILs prefer to accumulate at the surface of the slit pores to form a compact and ordered structure, independent of the slit width (Fig. S5 in the ESM).
The two-dimensional (2D) contour plots of the imidazolium ring distribution of BB and BP confined in the graphene-SiO 2 slit pores are further illustrated in Figs. 7(b)-7(i), which show different preferential orientations of the imidazolium ring to the surface of the X-Y plane. The blue and red regions indicate that the imidazolium ring normal is perpendicular (structure (I)) and parallel (structure (II)) to the surface normal, respectively, whereas the green region corresponds to irregular orientations, i.e., the disordered isotropic bulk phase. As shown in Figs. 7(b)-7(i), the imidazolium ring normal vector is parallel to the surface normal in all slit widths (1, 2, 5, and 10 nm), indicating that the imidazolium ring is "sleeping" in parallel onto the surface of the slit pore (red, structure II, θ = 0° or 180°). This parallel "sleeping" structure results in the accumulation of ions at the surface of the slit pores, thereby facilitating the formation of more compact and ordered ion layers. The other dominant structure (blue, structure (I), θ = 90°) appears discontinuous within the slit pore, implying that this structure is less favorable for the formation of compact and ordered ion layers. It is clear that the orientational ordering of both BB and BP is more pronounced as the slit widths decrease from 10 to 1 nm. This confirms the higher degree of ordering within the ion layer at a higher normal load and a higher contact stiffness, as shown in Fig. 5.

Angular probability distributions
To further investigate the ion distribution on the surface, the detailed angular probability distributions for the imidazolium ring, butyl chains, and anions on the HOPG and SiO 2 surfaces without biased voltages are presented in Fig. 8 and Fig. S6 in the ESM. On the HOPG surface ( Fig. 8(a) and Fig. S6(a) in the ESM), for both BB and BP, most of the imidazolium rings lie almost flat along the surface (θ ≈ 7° or 170°) for all slit pore widths (1, 2, 5, and 10 nm) because of the π-π interactions between the cations and the graphene  [75]. A parallel orientation of the butyl chains along the graphene surface was observed in both BB ( Fig. 8(b)) and BP (Fig. S6(b) in the ESM). The [BF 4 ] − anions show a tilted orientation of approximately 118° from the HOPG surface normal (Fig. 8(c)), whereas the angle probability distribution between the [PF 6 ] − anion and HOPG surface normal exhibits two pronounced peaks at approximately 53° and 125° (Fig. S6(b)  Unlike the almost perfectly parallel orientation of the imidazolium ring plane on the HOPG surface, the [BMIM] + cations of both BB (Fig. 8(d)) and BP ( Fig. S6(d) in the ESM) exhibit preferential distributions for maximizing hydrogen bonding interactions between the cations and the dangling O atoms on the SiO 2 surface [69,75,76]. We observed that angles θ of approximately 15° and 170° were favored by the imidazolium ring distance from the SiO 2 surface but with a broad rough angle probability distribution in between, indicating different distributions for the configurations and orientations of the cations on SiO 2 from those on the graphene surface. The broad rough peaks in the imidazolium ring probability distribution profiles indicate that a few rings exhibit tilted orientations from the SiO 2 surface in varying slit pores. The observed angles of 90° in the butyl chains of both BB and BP to the surface normal correspond to the chains "sleeping" on the SiO 2 surface (Fig. 8(e) and Fig. S6(e) in the ESM). In contrast to the HOPG surface, the majority of [BF 4 ] − anions exhibit a tilted orientation of approximately 65° from the SiO 2 surface, and some of them are tilted 90°-150° away from the surface (Fig. 8(f)). The [PF 6 ] − anions indicate maximum values at approximately 53° and 125° on the SiO 2 surface, which is less probable on the HOPG surface, corresponding to the weaker adsorption of this conformation on SiO 2 (Fig. S6(f) in the ESM).

Ion orientation and number density profiles
The effect of charging on the orientation of [BMIM] + cations confined in 2 nm graphene-SiO 2 slit pores is shown in Fig. 9(a). The parallel orientation of the imidazolium ring (red, structure (II), θ = 0° or 180°) dominates at the uncharged surfaces of the slit pore, thereby facilitating the formation of compact and ordered ion layers. In comparison, the parallel "sleeping" structure (II) is more pronounced during the slit pore charging of either sign (8 V, −8 V) in both BB (the left panel in Fig. 9(a)) and BP (the right panel in Fig. 9(a)). This implies more ordered ions at the charged surfaces, which agreed well with the stiffer AFM-detected ion layers at the surfaces with either positive-or negative-bias voltages, as shown in Fig. 6. Except for the [BMIM] + cation, the [BF 4 ] − and/or [PF 6 ] − anions dominated the layer composition, as shown in the number density profiles ( Fig. 9(b)), and primarily adhered to the vicinity of the slit pore surface. The number density profiles in Fig. 9(b) show well-defined interfacial structures of ions at the charged slit pores (width = 2 nm) under biased voltages of 8 and −8 V. The cations accumulate near the graphene surface at −8 V, whereas more anions accumulate near the SiO 2 surface. By contrast, at a biased voltage of 8 V, cations and anions accumulate near the SiO 2 and graphene surfaces, respectively. This indicates the ordering of ILs into alternating charge layers at the charged surfaces. The compact and ordered IL layers formed on the charged surfaces result in a greater friction force [25] than those on the uncharged surfaces. In comparison with the ion accumulation at 0 V, the accumulation of ions near the charged graphene and SiO 2 surfaces under biased voltages of 8 and −8 V is not distinct, which is likely due to the steric hindrance that the slit pore can exert on the internal IL structure [77].

Angular probability distributions
It is noteworthy that the differences in the contour plots of the imidazolium ring distribution under biased voltages can not be observed clearly with the naked eye. Hence, we provide detailed angular probability distributions of the imidazolium ring, butyl chains, and anions on HOPG and SiO 2 surfaces (slit width = 2 nm) under biased voltages of 8 and −8 V, as shown in Fig. 10 and Fig. S7 in the ESM, which show different accumulations of cations and anions.
As shown in Fig. 10(a) and Fig. S7(a) in the ESM, on the HOPG surface at 8 V, the angles between the imidazolium ring are normal for BB and BP, as well as the surface normal are 8°-25° and 160°-170°, On the SiO 2 surface, the orientational angles of the imidazolium rings for the BB (Fig. 10(d)) show two peaks at approximately 14° and 170° at 8 V, whereas those at −8 V show two peaks at approximately 23° and 160°. For BP, as shown in Fig. S7(d) in the ESM, the imidazolium rings tilt by 10° and 170° from the

Conclusions
Colloid probe AFM was employed to investigate in situ changes in the nanofriction and microstructures of ILs (BB and BP) on uncharged and charged surfaces. A higher nanoscale friction force and a higher torsional resonance frequency were observed when the normal load on the uncharged surfaces increased. A higher torsional resonance frequency at a higher normal load corresponded to a higher contact stiffness, indicating a more compact and ordered IL layer that provided a higher friction. By contrast, on the charged surfaces, the nanofriction of the ILs increased with the positive-bias voltage, and a similar variation tendency was observed under the negative-bias voltages. The dependence of friction force on the applied bias voltage was similar at various normal loads. Additionally, the recorded torsional resonance frequency in the ILs was higher when a higher surface bias of either sign was applied, indicating a stiffer, more compact, and ordered IL layer, which contributed to the higher friction on the charged surfaces. MD simulations showed that the parallel orientation of the imidazolium ring in the [BMIM] + cation dominated on the uncharged surfaces of the adopted slit pore model, thereby facilitating the formation of compact and ordered ion layers. Additionally, the parallel "sleeping" structure was more pronounced on the charged surfaces, which implied more ordered ions; this confirmed the presence of stiffer ion layers, as detected via AFM on the charged surfaces. These findings are expected to contribute significantly to the use of ILs as lubricants in electric vehicles, electrolytes in energy storage devices, and other applications where charged surfaces are used.