Astonishingly distinct lubricity difference between the ionic liquid modified carbon nanoparticles grafted by anion and cation moieties

The astonishingly distinct lubricity difference between the ionic liquid modified carbon nanoparticles grafted by anion and cation moieties (A-g-CNPs and C-g-CNPs) was well established as additives of polyethylene glycol (PEG200). The peripheral anion moieties and positively charged inner parts of C-g-CNPs could successively absorb onto the friction interfaces by electrostatic interactions to form the organic—inorganic electric double layer structures, tremendously boosting the lubricity of PEG200. Contrarily, the preferentially electrostatic adsorption of negatively charged inner parts but repulsion of the peripheral cation moieties determined the weak embedded stability of A-g-CNPs between the friction interfaces, even impairing the lubricity of PEG200. This work can offer solidly experimental and theoretical guidance for designing and developing the high-performance nanoadditives modified by ionic molecules.


Introduction
It is reported that the friction and wear consumed about 23% of the global energy consumption [1]. The lubrication realized by loading lubricants between friction interfaces occupies an indispensable role in lessening the friction and wear. Compared with the lubrication achieved by solids, the fluid lubrication possesses many merits, including the prolonged service life, particularly low mechanical noise, huge enhancement of thermal diffusivity, and extremely light friction in elastohydrodynamic lubrication regime [2]. The fluid lubricants are usually formulated as base oils containing a series of additives including the friction modifiers, antioxidants, thickening agents, extreme pressure agents, anti-wear agents, and detergents. Although the amounts of additives in lubricants are commonly low, they have brought in qualitative leaps for the lubricity of base oils under the boundary lubrication regime [3]. Up to now, the most famous base oils are the mineral oils, which have been successfully applied on a large scale since 1800s, particularly for the vast machines invented during the period of Industrial Revolution. Meanwhile, the most impressive additives for mineral oils up to now must be zinc dialkyldithiophosphates (ZDDPs) because they have been in continuous use since late 1930s [4]. The major drawbacks of mineral oils including the poor high-temperature oxidation resistance and low-temperature liquidity became increasingly apparent since 1950s due to the discovery of machines worked under harsh conditions, and hence the synthetic lube oils including the silicone oils, perfluoropolyethers, polyalphaolefins (PAO), and polyesters were developed to satisfy the growing demands. Compared with base oils, the development of special additives as alternatives of ZDDPs is seriously lagged due to the particularly low cost and multiple functions of ZDDPs [5]. Considering the unfavorable comparability of ZDDPs with synthetic oils and rigorous limitation of phosphorus emission by modern environmental regulations, the development of eco-friendly lubricant additives with high-performance for synthetic oils is becoming a promising research topic.
During the past two decades, the development of lubricant additives for polyethylene glycol (PEG), the synthetic oil with biodegradability, has experienced two stages of ionic liquids and nanomaterials. The ionic liquids are naturally ideal additives for PEG because of their excellent comparability with base oil and unique electrostatic adsorption behaviors on the rubbing surfaces of metal friction pairs. The dense and ordered absorption layers constructed by ionic liquid molecules on the tribo-interfaces can directly act as the physically protecting films or even trigger the formation of tribo-films, laying the solid foundation for obviously boosting the lubricating functions of PEG [6][7][8]. The improvement and perfection of ionic liquid-based additives has been continually achieved by elaborately molecular design, displaying as the promotion of tribological performance and oxidation resistance [9][10][11], the alleviation of corrosion [12][13][14], and the reduction of pollution [15][16][17][18][19]. Various nanoadditives, including metal [20], oxides [21][22][23], graphite [24], grapheme [25], polymer [26], carbon nanoparticles (CNPs) [27], and composite [28], have exhibited better extreme pressure and anti-wear properties than ionic liquids once the aggregation issues in PEG were solved. The excellent performance of nanoadditives has been attributed to the formation of robust lubrication films on the tribo-interfaces by deposition or sintering. The CNP-based additives gained from the "bottom-up" methods are attracting more and more attentions due to their synthesismodification integration, dual designability from the peripheral surface groups to inner carbon cores, outstanding film-building ability on tribo-interfaces, favorable environmental friendliness, and unique self-lubricating effects [29][30][31][32][33][34][35][36][37][38][39]. Especially, the ionic liquid modified CNP-based additives, integrating the advantages of ionic liquids and CNPs, have been creatively designed and developed with the wide applied range and especially enhanced tribologcial performance [40][41][42]. It is well accepted that the ionic liquids on CNPs could not only improve their comparability with PEG and embedded stability on friction interfaces by the electrostatic interactions, but also directly take part in the construction of lubrication films, thereby playing the significant role in boosting the tribological performance of CNPs. However, it is worth noting that all the ionic liquid modified CNP-based additives reported so far were grafted by cation moieties, and their performance have been conveniently tuned by replacing the anion species [43][44][45]. Now that the synergistic effect between ionic liquid groups and carbon cores of CNPs has been well confirmed, then how about the performance of ionic liquid modified CNP-based additives grafted by anion moieties? Is the surface group grafting mean also a critical factor determining the tribological performance of ionic molecule modified nanoaddtives?
To shed light on the above issues, two ionic liquid modified CNPs grafted by anion and cation moieties (A-g-CNPs and C-g-CNPs) were designed and synthesized according to Fig. 1. The morphology and structures of A-g-CNPs and C-g-CNPs were revealed by many characterization technologies. The tribological   www.Springer.com/journal/40544 | Friction performance of A-g-CNPs and C-g-CNPs as the PEG-based additives were investigated in detail by varying the test conditions of additive, concentration (c), load, and duration. The influences of surface group grafting mean on the tribological properties of CNP-based additives modified by ionic liquid groups were clarified deeply. After the friction tests, the possible lubrication mechanisms of A-g-CNPs and C-g-CNPs were presented on account of wear track surface analysis results. The results presented here are hopeful to establish substantially experimental and theoretical guidance for designing and developing the high-performance ionic molecule modified nanoadditives from the point of grafting mean.

Synthesis of A-g-CNPs and C-g-CNPs
As illustrated in Fig. 1, both A-g-CNPs and C-g-CNPs were synthesized by the pyrolysis method followed by an ion exchange process. Specifically, to synthesize the A-g-CNPs, β-alanine (2.67 g, 0.03 mol) was first loaded into a round-bottom flask (100 mL) and melted by heating to 220 °C under N 2 atmosphere. Then, CA (1.92 g, 0.01 mol) was rapidly plunged into the above flask and pyrolyzed for 2 h under magnetic stirring. When the flask was naturally cooled to the room temperature, 0.1 M NaOH solution (50 mL) was injected to form a reddish-brown dispersion. The dispersion was loaded into a dialysis bag (1000 Da) to eliminate the water-soluble impurities by dialysis.
The CNPs capped by ionic molecules of sodium carboxylate (CNPs-Na) were collected as brown-black powders after the lyophilization treatment. The CNPs-Na (0.2 g) were dispersed into ultrapure water (5 mL) again by ultrasonic treatment. Subsequently, the [C 16 -MIm]Cl solution (0.1 M) in water was gradually trickled into the CNPs-Na dispersion until the mixture became colorless, and the black-brown viscous precipitate was totally separated out from the water phase, meaning the achievement of cation exchange process between Na + and [C 16 -MIm] + . The precipitate was rinsed with water (20 mL) for 6 times and dried in vacuum oven. Finally, the A-g-CNPs were gained as waxy powders (yield = 0.3266 g). If the above β-alanine (2.67 g, 0.03 mol), 0.1 M NaOH solution (50 mL), and ion exchange process between Na + and [C 16 -MIm] + were replaced by [AMIm]Cl (4.85 g, 0.03 mol), 0.1 M HCl solution (50 mL), and ion exchange between Cl − and [C 16 -HA] − , respectively, the C-g-CNPs were also obtained from CNPs-Cl (0.2 g) with a yield of 0.3115 g.

Characterization
The morphologies of CNPs-Na and CNPs-Cl were observed by the transmission electron microscopy (TEM; Tecnai G20) operating at 200 kV. The X-ray diffraction (XRD) patterns of various CNPs were acquired by an X-ray diffractometer (D8 Advance X Pert Pro, Bruker) at a wavelength (λ) of 0.15418 nm. The Fourier transform infrared (FTIR; Beifen Rayleigh WQF-520) spectra of various CNPs were obtained using a spectrophotometer by the KBr pellet technique. The X-ray photoelectron spectra (XPS) were gained using a photoelectron spectrometer (ESCALAB 250) equipped with a monochromatized Al Kα X-ray source (1,486.71 eV). The thermogravimetric analysis (TGA) curves were obtained on a thermogravimetric analyzer (Netzsch STA449F3) under N 2 atmosphere at a heating rate of 10 °C/min.

Tribological property tests and wear track surface analyses
The tribological performance evaluations of A-g-CNPs and C-g-CNPs as PEG200-based additives were systematacially performed on the universal mechanical tester (UMT) platform (Bruker UMT-TriboLab) equipped with a ball-on-disk linear reciprocating sliding module. The specific test conditions and processes in this work were similar to our previous work except that the loads used here were 50-250 N [45]. The average sliding speed was about 10 mm/s. The duration was 20 or 200 min. The amplitude was 5 mm. All the friction tests were conducted at room temperature (15-20 °C) and under the atmospheric environment with the relative humidity of 60%-70%. The loading concentration range of A-g-CNPs and C-g-CNPs was 0.1-1.5 wt%. After the friction test, the lower disk was cleaned in acetone by ultrasonic treatment for 10 min. The two-dimensional (2D) and three-dimensional (3D) profiles of wear track on the lower disk were recorded by a white light interferometer. The wear volume (WV) of the lower disk was calculated based on the integral area of the 2D profile and the amplitude. The morphology and elemental composition of the wear track surface were analyzed by a scanning electron microscope (SEM; JSM-7500F) equipped with an energy-dispersive X-ray spectroscope (EDS) and the XPS spectrometer. Figure 2 gives the TEM images of CNPs-Na and CNPs-Cl. The CNPs-Na and CNPs-Cl both appear as the spherical nanoparticles with the average diameters of about 20.2 and 22.7 nm (the insets of Fig. 2), respectively, confirming the successful synthesis of CNPs. The lattice fringes can hardly be found in CNPs-Na and CNPs-Cl marked out by the white circles, reflecting the relatively poor crystallinity of CNPs. Considering that the A-g-CNPs and C-g-CNPs are derived from the CNPs-Na and CNPs-Cl via the ion exchange processes, their TEM images are not obtained and shown because the negligible morphology and crystallinity vibrations triggered by ion exchanges can hardly be revealed by the TEM characterization. The crystallinity of CNPs-Na, CNPs-Cl, A-g-CNPs, and C-g-CNPs were also uncovered by the XRD analyses. In the XRD patterns of CNPs-Na and CNPs-Cl (Figs. 3(a) and 3(b)), only diffused and short peaks at 23.5° and 21.6° are observed, respectively, also confirming the poor crystallinity of CNPs-Na and CNPs-Cl, which is a commonplace for CNPs prepared from the "bottom-up" methods [46][47][48]. When the CNPs-Na and CNPs-Cl were transformed into A-g-CNPs and C-g-CNPs by ion exchanges, the above peaks only slightly shift to 24.3° and 20.1° without apparent shape changes (Figs. 3(c) and 3(d), respectively), which can be attributed to the displacements of small ions (Na + and Cl − ) by big ions with long carbon chains ([C 16 -MIm] + and [C 16 -HA] − ) on the surfaces of CNPs.

Characterization
The surface groups of various CNPs and the transformation processes between CNPs-Na/CNPs-Cl and A-g-CNPs/C-g-CNPs were monitored by the FTIR spectra (Fig. 4). The absorbance band at 3,426 cm −1 (Figs. 4(a) and 4(b)) is caused by the stretching vibrations of N-H. The intensity of this band is attenuated in the spectra of A-g-CNPs and C-g-CNPs (Figs. 4(c) and 4(d)), which can be attributed to the shielding effects of outer big ions containing long   carbon chains. The absorbance band at 2,926 cm −1 triggered by the stretching vibrations of C-H in methylene appears in all spectra. In addition, the band at 2,850 cm −1 referred to the stretching vibrations of C-H in -CH 3 newly emerges in Figs. 4(c) and 4(d), verifying the ion exchange processes between Na + /Cl − and [C 16 -MIm] + /[C 16 -HA] − . The bands at 1,654, 1,565, and 1,402 cm −1 , resulted from the stretching vibrations of C=O in amide bonds, C=C stretching, and C-H deformation vibrations, respectively, can be found in Fig. 4(a), indicating that the β-alanine molecules are grafted onto CNPs-Na via the amide bonds. Apart from the above bands, the bands at 1,654, 1,565, and 1,402 cm −1 derived from imidazole rings in Fig. 4 The XPS spectra were employed to reveal the compositions and existence states of the elements on the surfaces of various CNPs. The survey XPS spectrum presented in Fig. 5(a) shows that the main elemental compositions of CNPs-Na are C, O, N, and Na. It is worth noting that a peak at about 500 eV assigned to Sn element is also appeared. The Sn element derived from the precursors of CA and β-alanine can be survived during the synthesis process of CNPs-Na. According to the XPS spectrum in Fig. 5(b), the main elements in CNPs-Cl include C, O, N, and Cl. The elemental compositions of CNPs-Na and CNPs-Cl suggest that their surfaces should be modified by the sodium alanine and [AMIm]Cl groups. Compared with the CNPs-Na and CNPs-Cl, both the A-g-CNPs and C-g-CNPs are absolutely made up of C, O, and N along with the disappearances of respective Na and Cl, suggesting that the Na + /Cl − on the surfaces of CNPs-Na/CNPs-Cl should be replaced by the [C 16 -MIm] + /[C 16 -HA] − . The high-resolution XPS elemental spectra of various CNPs are given in Fig. S1 in the Electronic Supplementary Material (ESM), displaying  | https://mc03.manuscriptcentral.com/friction that all of the elemental existence states of CNPs, i.e., elemental species, are well in line with those of surface groups on CNPs shown in Fig. 1 and implying the smooth running of whole synthetic processes of A-g-CNPs and C-g-CNPs. In addition, the specific elemental compositions of various CNPs calculated from the XPS analyses are listed and compared in Table 1.
The thermal stability and over-all structures of various CNPs were disclosed by the TGA analyses. All the TGA curves (Fig. 6) demonstrate two thermal weight loss stages, which can be derived from the volatilization of moisture/adsorbates and the gradual decomposition of surface groups on CNPs. The thermal decomposition temperatures for CNPs-Na and A-g-CNPs are similar (about 265 °C) as the same anion moieties have been grafted onto their surfaces (Figs. 6(a) and 6(c)). Besides the decomposition temperatures (about 220 °C), the shapes of the TGA curves for CNPs-Cl and C-g-CNPs (Figs. 6(b) and 6(d)) are also analogous because their surfaces are modified by the same imidazolyl cation but different anions. As a whole, the relatively high thermal stability for A-g-CNPs and C-g-CNPs is a merit for their  applications as lubricant additives. The remaining weight fractions for CNPs-Na and CNPs-Cl are 60.4% and 47.9% (Figs. 6(a) and 6(b), respectively), indicating that the modification rates of their surface groups are relatively high. The remarkably lower remaining weight fractions for A-g-CNPs (6.9%, Fig. 6(c)) and C-g-CNPs (2.6%, Fig. 6(d)) are induced by the replacements of vast small ions (Na + /Cl − ) by much bigger ions ([C 16 -MIm] + /[C 16 -HA] − ). In view of the similar remaining weight fractions and the same composition of ionic liquid groups, the difference of modification rates of surface groups between A-g-CNPs and C-g-CNPs is moderate. Judging by the characterization results described above, the A-g-CNPs and C-g-CNPs with similar morphology and crystallinity, high modification rate, the same composition of surface ionic liquid group, but different surface group grafting means, were obtained according to the elaborately designed synthetic procedures (Fig. 1). To deep clarify the effects of grafting mean of surface groups on the tribological properties of CNPs modified by ionic liquid groups as additives, the PEG200 was chosen as base oil because the ionic liquids close to the surface groups on A-g-CNPs and C-g-CNPs have exhibited the superior comparability with the PEG200 [43][44][45].

Tribological performance of A-g-CNPs and C-g-CNPs as additives of PEG200
The favorable dispersion stability of nano-based additives is a critical issue to take full use of their lubricating functions [49,50]. The stability of A-g-CNPs and C-g-CNPs in PEG200 was evaluated by visual inspection. Figure 7 shows the photos of A-g-CNPs and C-g-CNPs dispersions in PEG200 (0. A-g-CNPs and C-g-CNPs in PEG200 can be attributed to their abundant ionic liquid surface groups. These ionic liquid groups not only possess excellent comparability with PEG200 but also provide high steric hindrance and strong electrostatic repulsion for A-g-CNPs and C-g-CNPs. As shown in Fig. 8, the tribological performance of A-g-CNPs and C-g-CNPs as PEG200-based additives were first investigated by altering c. An obvious running-in period manifesting as the sharp rising and falling of coefficient of friction (COF) is found in the COF curve of PEG200 ( Fig. 8(a)), indicating the occurrence of severe scraping between the friction pairs owing to the relatively poor lubricity of neat PEG200. If low amount (0.1 wt%) of C-g-CNPs is added into PEG200, the running-in period (sharp rising of COF between 0-100 s) is lessened but still existent, reflecting that the low amounts of C-g-CNPs only play slight friction-reducing function but can hardly protect the friction pair from direct contact. In other words, there are no enough C-g-CNPs to form the effective lubricating films. When the c of C-g-CNPs is ≥ 0.3 wt%, the running-in period is sharply shortened and presents as dramatically falling of COF, implying the effective boundary lubricating films gradually form on the rubbing surfaces. The lowest and smoothest COF curve is acquired for the friction pair lubricated by the C-g-CNPs dispersion with c of 0.7 wt%. That is, an optimum c of 0.7 wt% is existent. When the c of C-g-CNPs is > 0.7 wt%, the excess C-g-CNPs will aggregate on the friction interfaces. The aggregates are adverse to the COF reduction because they will bring the unnecessary  | https://mc03.manuscriptcentral.com/friction abrasive friction and damage the homogeneity of lubricating films. Thus, the C-g-CNPs at higher c will lead to the increase of COF. The average COF values shown in Fig. 8(a) again confirm that the C-g-CNPs have the best friction reduction property when the c is 0.7 wt% and make the COF of PEG200 reduce by about 51.0%. The variation tendency of WV with the increasing c is well consistence with that of COF. When the c is 0.7 wt%, the C-g-CNPs also display the best wear resistance ability, resulting in the biggest WV reduction of 76.2%.
Unexpectedly, the A-g-CNPs display the totally different tribological behaviors compared with the C-g-CNPs ( Fig. 8(b)). First, all the running-in periods of COF curves of A-g-CNPs dispersions in PEG200 with various concentrations are similar to that of PEG200, meaning the lack of boundary lubrication films on friction interfaces. Moreover, the whole COF curves are all tangled with each other even when they are amplified on Y axis (the inset of Fig. 8(b)), determining the negligible friction-reducing effect of A-g-CNPs. The average COF values given in Fig. 8(b) further prove the above point of view. Specifically, the average COF of base oil only reduces by 2.5% when 0.7 wt% A-g-CNPs is added. The slight friction reduction caused by A-g-CNPs can be attributed to their poor embedding stability between the asperities of rubbing surfaces. The A-g-CNPs are apt to aggregate or be squeezed out from the friction interfaces, thereby playing negligible friction-reducing effects. In other words, the A-g-CNPs always play slight friction-reducing function no matter the c is low or high. Therefore, the c change of A-g-CNPs exhibits little effect on the COF values. Correspondingly, the WV reductions of PEG200 induced by 0.7 wt% of A-g-CNPs is also mild (28.8%), further denying the existence of effective lubricating films on tribo-interfaces. In addition, when the c is higher, the more A-g-CNPs are retained on the rubbing surfaces and hence play more obvious anti-wear effect. Thus, the WV reduces with the increasing c. The astonishingly distinct lubricity difference between C-g-CNPs and A-g-CNPs can be attributed to the different grating means of surface groups, which greatly influence the action behaviors of C-g-CNPs and A-g-CNPs on sliding contact interfaces.
To visually sense the huge difference of wear resistance ability between A-g-CNPs and C-g-CNPs, the 3D and corresponding 2D profiles of wear tracks of the disks lubricated with PEG200, and A-g-CNPs and C-g-CNPs dispersions in PEG200 (0.7 wt%) were obtained and shown in Fig. 9. Both the wear track www.Springer.com/journal/40544 | Friction width and depth of PEG200 are conspicuous, resulting in a large 2D integral area (S) of ~2,816.3 μm 2 ( Fig. 9(a)). The S of cross profile of wear track of A-g-CNPs dispersion (0.7 wt%) is 2,004.8 μm 2 , indicating that the wear resistance ability of A-g-CNPs is moderate ( Fig. 9(b)). Strikingly, the S of wear track cross profile for C-g-CNPs dispersion (0.7 wt%) is tremendously reduced to 669.0 μm 2 , which is only one third of 2,004.8 μm 2 and even less than one fourth of 2,816.3 μm 2 ( Fig. 9(c)), directly reflecting the far superior wear resistance function of C-g-CNPs over A-g-CNPs.
At the fixed c (0.7 wt%) and duration (20 min), the effects of load on the tribological behaviors of A-g-CNPs and C-g-CNPs PEG200-based additives were studied. Figures 10(a) and 10(b) illustrate that all of the COF curves lubricated by PEG200 and A-g-CNPs dispersion (0.7 wt%) are fluctuated at high level and possess the distinct running-in periods, indicating that the dry friction happens at the beginning stage because of the absence of boundary lubricating films on the friction interfaces. In addition, the COF curves of A-g-CNPs dispersion (0.7 wt%) present the climbing trend at the end of friction tests under the high loads of 150-250 N (marked out by the red oval in Fig. 10(b)), reflecting that the A-g-CNPs even injures the friction-reducing function of PEG200. In sharp contrast, all the COF curves lubricated with C-g-CNPs dispersion (0.7 wt%) are smooth at low level with undetectable running-in periods despite a negligible climbing phenomenon of COF under 250 N (marked out by the red oval in Fig. 10(b)) is observed, indicating that the boundary lubrication films formed on friction interfaces are robust.
The variation tendencies of average COF and mean WV of disks lubricated by PEG200, and A-g-CNPs and C-g-CNPs dispersions (0.7 wt%) in PEG200 with the increasing load are given in Fig. 11. No frictionreducing function is found for A-g-CNPs under the load ranging from 50 N to 250 N, whereas the prominent friction-reducing function is always existent for C-g-CNPs despite it is attenuated from 51.8% at 100 N to 32.5% at 250 N ( Fig. 11(a)). The mild wear resistance ability of A-g-CNPs is almost disappeared when the load rises up to 200-250 N ( Fig. 11(b)), while the resistance ability of C-g-CNPs is gradually weakened but still noticeable with the increasing   The tribological test results gained under the loads ranging from 50 to 250 N further confirm the huge lubricity difference between A-g-CNPs and C-g-CNPs.
The long-time friction tests were carried out under the prolonged duration of 200 min to disclose the durability of A-g-CNPs and C-g-CNPs as PEG200based additives under the fixed load = 100 N and c = 0.7 wt%. Strikingly, the COF curve lubricated by A-g-CNPs dispersion is higher and waves more drastic than that lubricated with PEG200, reflecting that the A-g-CNPs will impose apparent damage on the friction-reducing ability of PEG200 once the duration is too long (Fig. 12(a)). Despite the slight upward tendency is found in the COF curve lubricated by C-g-CNPs dispersion, this curve is the smoothest and lowest generally, reflecting that the formed boundary lubrication film on friction interfaces is durable, and hence the service life of C-g-CNPs is particularly long. Figure 12(b) demonstrates that the A-g-CNPs deteriorate both the friction and wear reduction performance of PEG200 when the duration is long. The average COF and mean WV lubricated with A-g-CNPs dispersion are 11.6% and 26.8% higher than that lubricated by PEG200, respectively. On the contrary, the friction and wear reduction effects of C-g-CNPs are always remarkable under the long duration (200 min), resulting in significant COF and WV reductions for PEG200 (40.3% and 59.0%, respectively). The distinguished lubricity of C-g-CNPs can be attributed to their good embedding stability and strong film-building ability on the rubbing surfaces. The boundary lubrication films formed by C-g-CNPs are robust and durable enough to avoid the direct contact of friction interfaces for a long period of friction process. The overwhelming wear resistance gap between A-g-CNPs and C-g-CNPs is intuitively revealed by the 3D and 2D profiles shown in Fig. 13. The 2D integral areas of wear tracks lubricated by PEG200, and A-g-CNPs and C-g-CNPs dispersions (0.7 wt%) are 3,660.4, 4,640.0, and 1,507.3 μm 2 , respectively. The huge lubricity difference between A-g-CNPs and C-g-CNPs is further enlarged when the duration is increased.
To uncover the relationship between the surface groups and inner parts of A-g-CNPs/C-g-CNPs, the tribological properties of [C 16 -MIm]Cl/[C 16 -HA]Na as PEG200-based additives were inspected and compared with the A-g-CNPs/C-g-CNPs. Figure 14 shows that the [C 16 -MIm]Cl even exhibits better friction and wear reduction effects than A-g-CNPs because the [C 16 -MIm]Cl molecules can form well-organized absorption layers on the friction interfaces [8], whereas the lubricating function of [C 16 -HA]Na is far worse than that of C-g-CNPs as the boundary lubricating films induced by C-g-CNPs are more durable and robust than the ordered molecular absorption layers built by [C 16 -HA]Na [51]. In other words, the peripheral ion moieties ([C 16 -MIm] + / [C 16 -HA] − ) and inner parts (carbon cores) of A-g-CNPs/ C-g-CNPs play the inhibiting and collaborative lubricating effects, respectively, reflecting the great effects of surface group grafting mean on the lubricity of ionic molecule capped CNP-based lubricant additives.

Wear track surface analyses and lubrication mechanisms of A-g-CNPs and C-g-CNPs
After the friction tests, the micro-topographies and elemental compositions of wear track surfaces of disks  Fig. 15(a), the wear track surface of PEG200 is full of bumps and hollows with some abrasive dusts (marked out by the green arrows), indicating that the serious adhesive, abrasive, and fatigue wears are all existent on the worn surfaces because of the ordinary lubricating ability of PEG200 under the boundary lubrication regime. Moreover, the high C and O contents (15.81% and 5.07%, respectively) are detected out on the wear track surface, hinting that the complex tribochemical reactions, including severe oxidation of steel surface and thermal decomposition of PEG200, happen on the wear track surface. However, the above tribochemical products such as metal oxides and amorphous carbon are unable to form the effective lubrication film due to their poor mechanical properties and film-building abilities, and hence the heavy oxidation and wear are happened. The hole (marked out by the green arrow) on the wear track surface lubricated with A-g-CNPs dispersion ( Fig. 15(b)) seems even deeper than that lubricated with PEG200, implying that the A-g-CNPs   as additives not only cannot improve but also even damage the anti-wear function of PEG200. In addition, the local area with higher C content (marked out by the red circle) on the wear track surface is assigned to the aggregates of A-g-CNPs. These aggregates are not only unfavorable to the oil-film lubrication of PEG200 but also may bring in the crucial abrasive wear, hence resulting in more significant friction and wear. The micro-topography of wear track surface lubricated with C-g-CNPs dispersion (Fig. 15(c)) is significantly different from those shown in Figs. 15(a) and 15(b). Only some scuffing can be observed on the wear track surface, reflecting the just presence of mild abrasive wear and the nearly absence of adhesive and fatigue wears. The markedly low C and O contents (10.44% and 0.82%, respectively) on the worn surface imply that the oxidation of steel surface and thermal decomposition of PEG200 are greatly alleviated. At the same time, the appearance of N content (0.14%) on wear track surface suggests the existence of tribochemical reactions between the surface groups of C-g-CNPs and steel contact surfaces. On the basis of the lubrication mechanisms of CNPs described in previous works [35,39,43], the C-g-CNPs induce the generation of effective boundary lubricating film on wear track surface, prominently boosting the lubricity of PEG200.
To deep reveal the composition of tribo-film induced by C-g-CNPs, the fine XPS Fe 2p, C 1s, O 1s, and N 1s spectra of the wear track surfaces lubricated with PEG200, and A-g-CNPs and C-g-CNPs dispersions were given in Fig. 16. Generally, the wear track surface of C-g-CNPs dispersion displays the most abundant elemental species because of the intimate contact between the C-g-CNPs and wear track surface. Figure 16(a) show that all the wear track surfaces contain Fe 2 O 3 , FeO, and Fe 2 (CO 3 ) 3 . According to the fitting peaks given in Fig. 16(b), the wear track surface lubricated with C-g-CNPs dispersion has an obviously higher content of organic carboxylates (RCOO − ). According to the results given in Fig. 16(d), the N-H species on the wear track surface lubricated with A-g-CNPs dispersion is referred to the physical absorption of trace amount of [C 16 -MIm] + moieties, whereas the emerged Fe-N/Cr-N species on the wear track surface lubricated by C-g-CNPs dispersion verify the presence of tribochemical reactions between the surface groups of C-g-CNPs and wear track surface. The above results confirm that the tribo-film induced by C-g-CNPs is in fact composed of lots of compounds such as Fe 2 O 3 , FeO, Fe 2 (CO 3 ) 3 , organic carboxylates, and metal nitrides. The higher content of C-Fe species given in Fig. 16(b) and the emergence of C-N/C=N species given in Fig. 16(d) corporately imply that the inner parts of C-g-CNPs may also play the role of the third particles and directly take part in the formation of boundary lubricating film by embedding into the above tribo-film [52].
On account of the above convincing experimental findings, the reasonable lubrication mechanisms of A-g-CNPs and C-g-CNPs as additives of PEG200 (0.7 wt%) under the conditions of 100 N and 200 min www.Springer.com/journal/40544 | Friction were proposed and given in Fig. 17. Once the friction is started, the steel rubbing interfaces is positively charged due to the effusion of electrons [9]. Then the peripheral anions ([C 16 -HA] − ) of C-g-CNPs can primarily absorb onto the friction interfaces by electrostatic attraction to generate the dense and well-organized absorption layers. Later on, the inner parts of C-g-CNPs are fast absorbed onto the above anion layers to form the organic-inorganic electric double layer structures, which effectively avoid the intimate contact of asperities on friction interfaces and hence result in sudden falling of COF. But, when A-g-CNPs are used as additives, the above electric double layer structures cannot form on the rubbing interfaces because of the preferentially electrostatic adsorption of inner parts but the repulsion of peripheral cation moieties ([C 16 -MIm] + ) of A-g-CNPs, and thereby the original rubbing surfaces are completely contacted, exhibiting the distinguished running-in period (fast rising of COF). After the running-in period, the A-g-CNPs still cannot form the valid protecting films on the rubbing surfaces due to   | https://mc03.manuscriptcentral.com/friction their poor embedded stability. Although the A-g-CNPs can enter into the gaps of the rubbing surfaces, the direct contact of asperities is inevitable. Moreover, the A-g-CNPs will aggregate when the duration is long, which not only causes the abrasive wear but also damages the oil-film lubrication of PEG200, and hence aggravates the friction and wear (Fig. 12). By contrast, the carbon cores of C-g-CNPs with favorable embedded stability play the effective lubricating effects of solid particles (rolling, mending, and polishing). During the friction process, the C-g-CNPs with spherical morphology are prone to roll between the rubbing surfaces just like rolling bearings [53]. The C-g-CNPs with small and uniform particle size will mend the bumps and hollows of friction interfaces. Moreover, the C-g-CNPs as the abrasive particles can play the polishing effect to reduce the roughness of rubbing surfaces [54]. Meanwhile, the surface groups of C-g-CNPs react with the steel surfaces to generate the tribo-films. The synergistic effects between the inner parts and surface groups of C-g-CNPs induce the generation of robust lubrication films composed of tribochemical products and carbon cores of C-g-CNPs, finally accounting for the distinguished friction and wear reduction performance of C-g-CNPs.

Conclusions
A-g-CNPs and C-g-CNPs, with similar morphology and crystallinity, high modification rate, same composition of surface group, but different surface group grafting means, were successfully obtained. The tribological performance of A-g-CNPs and C-g-CNPs as PEG-based additives were investigated. By varying the test conditions of additive, c (0.1-1.5 wt%), load (50-250 N), and duration (20 and 200 min), the unexpected distinct lubricity difference between A-g-CNPs and C-g-CNPs were established, revealing the significant effects of surface group grafting mean on the tribological performance of CNPs modified by ionic molecules. The A-g-CNPs as additives not only could not visibly improve but also even damaged the lubricating ability of PEG200, whereas the C-g-CNPs remarkably boost the lubricity of PEG200. Especially, when the c, load, and duration were 0.7 wt%, 100 N, and 200 min, respectively, the A-g-CNPs made the mean COF and WV of PEG200 increase by 11.4% and 26.8%, respectively, while the C-g-CNPs made these values reduce by 40.3% and 59.0%, respectively. The results of wear track surface analyses illustrated that the C-g-CNPs could primarily form the organicinorganic electric double layer structures on friction interfaces by electrostatic interaction, guaranteeing the favorable embedded stability of C-g-CNPs. Subsequently, the surface groups and inner parts of C-g-CNPs corporately induced the building of durable and robust boundary lubricating films, which were in fact the tribo-films composed of tribo-chemical products and carbon cores, well explaining the excellent friction and wear reduction performance of C-g-CNPs. By contrast, the A-g-CNPs not only could not form the effective protecting films on friction interfaces due to their bad embedded stability, but also might aggregate and damage the oil-film lubrication of PEG200 when the duration was long, thereby aggravating the friction and wear. This work reported here is hopeful to provide the substantially experimental and theoretical guidance for designing and developing high-performance nanoadditives modified by ionic molecule groups from the point of grafting mean.
www.Springer.com/journal/40544 | Friction Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.