How to improve superlubricity performance of diketone at steel interface: Effects of oxygen gas

Achievement of steady and reliable super-low friction at the steel/steel contact interface, one of the most tribological systems applied for mechanical moving parts, is of importance for prolonging machine lifetime and reducing energy consumption. Here we reported that the superlubricity performance of the steel/steel sliding interface lubricated with tiny amounts of diketone solution strongly depends on the oxygen content in surrounding environment. The increase of oxygen not only significantly shortens the initial running-in time but also further reduces the stable coefficient of friction in superlubricity stage due to the enhancement of tribochemical reactions. On the one hand, more severe oxidation wear occurring at higher oxygen content facilitates material removal of the contact interface, lowering the contact pressure and the corresponding initial friction. On the other hand, the growth of iron ions during the shear process in high oxygen environment promotes the formation of chelate which acted as an effective lubricated film chemisorbed at the steel/steel friction interface to further lower the interfacial friction. The results provide a new opportunity to further optimize the tribological performance of diketone superlubricity system, especially towards the lubrication of mechanical engineering materials.


Introduction
In order to further prolong using life of machines and reduce energy consumption, a large number of liquid superlubricity lubricants have been developed [1][2][3][4][5][6]. Basically, liquid superlubricity is defined as water-based superlubricity and oil-based superlubricity based on the different base liquids. Due to its origination of electrostatic double-layer force or hydration lubrication, water-based superlubricity is normally available for very limited tribological interfaces, such as ceramic and glass [7][8][9]. In comparison, oil-based superlubricity has attracted more and more attention because of its wide applications in lubricating mechanical engineering materials (especially steel) [10][11][12][13][14].
The early studies demonstrated that some kinds of oil adding the matched nanoparticles enable extreme low friction and even superlubricity for steel tribological pairs. For instance, through mixing fullerene nanoparticles in pure mineral oil, the coefficient of friction (COF) of steel sliding interface can be reduced to below 0.01 (superlubricity) due to smoothing of the contact area [10]. Furthermore, Bogunovic et al. [11] found that the additional TiO 2 nanoparticles in a highly purified paraffin oil can facilitate tribochemical reaction and promote the formation of chemical adsorption film at the steel interface, resulting in ultra-low friction. Although the lubrication performance of oil may be significantly improved by nanoparticles, the problem of how to stably and durably disperse lubricating oil must be resolved [12].
In recent years, some studies show that the steel tribological interface is capable of achieving superlubricity under lubrication of specific oil without any additive. Zeng and Dong [13] reported the superlubricity behaviors of Nitinol 60 alloy/steel interface with the lubrication of castor oil. They indicated that the castor oil dissociated at the tribological area to form OH-terminated surfaces, and the boundary wear-protective films is conducive to reducing friction largely. However, the unstable chemical properties of castor oil due to the poor oxidation resistance limits its further applications at high sliding speed [13]. Amann and Kailer [14] found that the lubrication of 1,3-diketone enables superlubricity at steel sliding interface even under relatively highspeed conditions. Later Li et al. [15][16][17][18][19] and Zhang et al. [20,21] indicated that the formed chemisorption film of ferric chelate in contact area plays a significant role in the superlubricity of steel tribological interface. The most severe problem of the 1,3-diketone lubrication system is the long running-in time with high friction before superlubricity (normally more than 6 h) [14,16]. Therefore, putting forward the method to shorten the running-in process and stabilize super low friction performance will provide a reference for the industrial applications of superlubricity lubrication.
In this study, we studied the lubrication performance of steel/steel pair lubricated with diketone (EPBD-0201 solution) in the atmospheres with different oxygen contents. It was found that the oxygen in the surrounding air not only largely shortens the running-in process but also produces a lower stable COF in the superlubricity stage. The contribution of oxygen in proving lubrication performance in this tribological system was further detected based on the characterizations of wear tracks using the white light interferometer (WLI) and scanning X-ray photoelectron spectrometer. The purpose of this study is to further improve the lubrication performance of diketone superlubricity system and provide theoretical support for its future applications.

Experimental details
Both the steel specimens and the steel balls used in the tribological tests are made of 440C, and the details are shown in Table 1. Based on the white light interferometry results, the roughnesses of the steel substrates and balls are measured as 10±3 and 6±1 nm, respectively. The 440C steel balls with a diameter of around 12.7 mm were purchased from Jiangsu Sak Bearing Co., Ltd. Before tribological tests, all friction pairs were ultrasonically cleaned with acetone and ethanol for 3 min each time. EPBD-0201 (1-(4-ethyl phenyl) butane-1,3-dione) as one kind of β-diketone molecule was applied as the lubricant in this study. More details about the synthesis of this reagent have been described in Ref. [16]. Most experiments in this study were repeated more than 3 times.
All tribological tests of the steel/steel sliding interfaces lubricated with EPBD-0201 were carried out by a high-precision universal tribotester (UMT3, Bruker, USA) under a rotary ball-on-disc contact form in a high vacuum chamber. During the sliding process, the steel ball remains stationary and applies a constant load of 2 N while the steel specimen rotates stably around the center point with a track radius of 6 mm and a sliding speed of 0.5 m/s. The maximum contact pressure before surface wear occurring was estimated to be around 483.3 MPa. The environmental temperature was set to 23±2 °C. When the tribological tests were operated in different oxygen content environments, the chamber was introducing pure nitrogen (N 2 ) to eliminate the potentially residual air at first. After that, pure oxygen (O 2 ) was introduced, and the oxygen content in the chamber was controlled according to the feedback information from an oxygen sensor. After the friction tests, the morphologies and chemical

Results and discussion
3.1 Superlubricity performance depending on surrounding oxygen content Figure 1 shows a typical friction curve of the steel/steel sliding interface lubricated with EPBD-0201 solution under a simulated air condition with ~20.9 wt% O 2 and ~79.1 wt% N 2 . In general, the friction that decreases largely along with the sliding cycles undergoes two running-in stages. At first, the COF decreases from ~0.14 to 0.05 as the sliding time increases to 4,500 s. After that, the COF reduces much more slowly to a final value of around 0.01 when the sliding time goes up to 22,000 s (around 6 h) (Fig. 1). Our results are consistent with the previous reports [14] where the superlubricity (COF ≤ 0.01) of the steel/steel tribological interface can be achieved with the lubrication of EPBD-0201 solution after a long running-in process.
The two running-in stages with different decrease rates of COF indicate the change of lubricated mechanism during this process, which will be discussed in detail later. Normally, oxygen plays a significant role in metal wear due to the oxidation reaction [22,23]. Here, we demonstrate that the oxygen content in the surrounding environment also has a strong influence on the superlubricity performance of the steel/steel interface lubricated with EPBD-0201 solution. Figure 2 compares the friction behaviors of this tribological system in the environments with a low O 2 content of ~0.3 wt% and a high O 2 content of ~50 wt%. Although superlubricity can be matched under two conditions, the friction behaviors especially during the running-in processes are significantly diverse. On the one hand, both the running-in Stages I and II need far less time as the surrounding O 2 goes up. In comparison, the running-in time decreases from ~15,000 to ~2,500 s (decreases by ~83%) in Stage I, and from ~30,000 to ~6,000 s (decreases by ~80%) in Stage II when the O 2 content increases from ~0.3 to ~50 wt%. On the other hand, more O 2 in the surrounding environment is conducive to producing a more stable superlubricity state with lower COF. The friction data in ~20.9 wt% O 2 environment ( Fig. 1) compared with those under ~0.3 and ~50 wt% O 2 condition (Fig. 2) is summarized in Fig. 3. Generally, higher surrounding O 2 content not only shortens the initial running-in period to reach superlubricity of the steel/steel sliding interface lubricated with EPBD-0201 solution but also further reduces the stable COF (which is defined based on the constant COF after running-in process) by 60%, i.e., 0.01 in ~0.3 and ~20.9 wt% O 2 environments dropping to 0.004 in ~50 wt% O 2 condition.

Evolution of contact pressure during running-in process
Morphologies of the worn surfaces on the steel balls and the steel substrates were imaged at the ends of Stages I and II, respectively. Due to much longer running-in time needed for lower O 2 environments with ~0.3 wt% content, more severe wear of the ball and the substrate was observed at both Stages I and II (Figs. 4(a) and 4(b)). The running-in time at ~0.3 wt% oxygen content is around 5 times longer than that at ~50 wt% oxygen content ( Fig. 3(a)), so the wear rates that mean wear volume per sliding time show the larger values at Stage I under higher O 2 content for the steel ball ( Fig. 4(a)) and the steel substrate ( Fig. 4(b)). Differently, the wear rates are close at Stage II for the two steel sliding surfaces. It has been reported that the lubrication performance can be improved with the decrease of contact pressure originating from surface wear [24][25][26]. Based on the mechanism of steel wear facilitated by oxygen [22,23], hereby we deduce that the severe oxidation wear of the steel ball in high O 2 condition plays a significant role in the sharp drop of the COF in Stage I. Table 2 summarizes the wear information of the steel balls and substrates under different conditions. The O 2 content in the surrounding environment not only changes the wear rate but also impacts the surface roughness of the worn region. In comparison with ~0.3 wt% O 2 condition, both the two stages in the running-in process end with slighter interfacial wear (i.e., the smaller radius of wear scar on the ball and narrower wear track on the substrate) under ~50 wt% O 2 environment. Besides that, the surface roughness inside the worn regions (estimated based on the topographies in Fig. 4) presents a dramatic increase in Stage I process followed with an obvious decrease in Stage II under the two environments. However,  Before wear occurs, the Hertz contact pressure of the steel/steel interface can be estimated as around 483.6 MPa with a single contact model due to the extreme low surface roughness of the original ball and substrate. The surface roughening along with the surface wear and the formation of the platform on the ball during the running-in process should result in a multi-asperity contact. Since the surface roughness of the worn ball is much lower than that of the worn substrate (Table 2), the multi-asperity contact can be simplified as a smooth plane in ball side contacting with a rough substrate surface [27]. Taking the crosssection profile of the wear track formed at ~50 wt% O 2 (Fig. 5(a)) as an example, each single asperity can be fitted circularly ( Fig. 5(b)). Through fitting all the asperity curves, the radii of the asperities on the worn substrate surfaces formed at the end of Stage II    (Fig. 5(c)). Then, the real contact pressure P can be estimated based on the applied load F n averaged to each asperity with an average radius r ave , P = F n /λS(r ave ), where λ is the number of asperities in contact region, and S(r ave ) is the average contact area of single asperity contact calculated with average radius r ave using the Hertz model. It is intriguing that the real contact pressures under both two O 2 conditions reduce sharply from ~483.6 MPa to 20-40 MPa at the end of the Stage I running-in process, and then slowly drop to around 10 MPa at the end of the Stage II process (Fig. 5(d)). The results confirm that the reduction of contact stress due to interfacial wear has the main contribution to the sharp drop of COF in Stage I, and the O 2 in the surrounding environment can facilitate this process [28,29].

Role of oxygen in the formation of superlubricity film
It has been reported that the superlubricity of the steel/steel sliding interface also originates from the formation of a chemisorption film due to the chelation reaction besides the reduction of contact pressure along with interface wear [15,[16][17][18][19][20][21]. The chelate with a proposed octahedral structure can provide an extreme low shear strength between two steel contact surfaces [16][17][18][19][20][21]. Different from that in Stage I, the real contact pressures decrease much more slowly corresponding to a final running-in process to superlubricity state (COF < 0.01) in Stage II under two O 2 conditions. It is reasonable to infer that the growth of chelate plays a more important role in Stage II running-in process.
The much shorter running-in time of Stage II at higher O 2 content (Fig. 2) indicates that O 2 should facilitate the chelation reaction between the steel/steel interfaces. The product of diketone molecules and Fe 3+ is firmly adsorbed on the outermost layer of the friction interface to constitute the chemisorbed layer [20].
To detect the contribution of oxygen to chelation reaction in deep, the wear tracks on the steel substrates formed at different running-in conditions are characterized using the X-ray photoelectron spectroscopy (XPS) in comparison with the pristine steel surface. Before sliding, the C 1s XPS spectrum of the pristine steel surface has four peaks in the range of bond energy from 282 to 292 eV, which are assigned to the C-C bond at 284.8 eV, C-O bond at 286.1 eV, and C=O bond at 288.6 eV ( Fig. 6(a)) [30][31][32][33][34]. In comparison, besides the peak at 283.5 eV assigned as the metal carbide, two extra peaks at 285.4 eV and 286 eV are observed for the C 1s XPS spectra inside all wear tracks (Fig. 6(b)). These two peaks also correspond to the C-C bonds coming from the diketone molecules in chemisorption film. As the inset schematic in Fig. 6(c), the binding energy of C-C bond connected with one carbonyl increases to 285.4 eV, and that connected with two carbonyls increases to 286 eV [20]. The increase in the area fraction of the C-C (285.4 eV) peak ( Fig. 6(c)) confirms that the chemisorbed chelate grows along with the running-in process at the steel/steel sliding interfaces. Figure 7 compares the O 1s XPS spectra of the worn regions and a pristine steel surface. The peak  locations at 529.9, 532.1, and 533.4 eV are assigned to the metal oxides, C=O bond, and C-O bond on the steel surface, respectively ( Fig. 7(a)) [35,36], which are much different from those inside the worn regions. Especially, the new peaks located at 530.9 eV are observed inside all worn tracks ( Fig. 7(b)). This bond energy should correspond to the C=O bonds from the chemisorbed chelate molecules where the influence from adjacent carbonyl group and benzene ring results in the red shift of binding energy [20]. Both under ~0.3 and ~50 wt% O 2 conditions, the area fraction δ of the C=O (530.9 eV) peak increases from Stage I to Stage II running-in process (Fig. 7(c)), indicating the growth of superlubricity film between the steel/steel sliding interface. Furthermore, the higher values of δ (530.9 eV) at each stage under ~50 wt% O 2 condition further confirm that oxygen promotes the formation of the chelate film. Figure 8 compares the XPS spectra of Fe 2p inside the wear tracks formed in ~0.3 and ~50 wt% oxygen environments. Three different peaks located at 706.7, 709.2, and 711.1 eV were observed, corresponding to the Fe 0 , Fe 2+ , and Fe 3+ , respectively [37]. The change of these peaks along with sliding at different oxygen conditions are summarized in Table 3. Both under two oxygen environments, the Fe 0 and Fe 2+ peaks decrease whereas the Fe 3+ peak increases as the running-in progress goes from Stage Ι to Stage II. It implies that the final chelation products should be Fe 3+ -diketone, which chemisorbs on the steel/steel interface to provide super-low shear strength [16][17][18][19][20][21]. Furthermore, the peaks at 711.1 eV are much stronger in ~50 wt% O 2 condition, indicating that the surrounding oxygen can facilitate the generation of Fe 3+ as well as the chemisorbed diketone molecules. Similar phenomena were reported in Refs. [22,23] where H 2 O 2 can promote the formation of Fe 3+ chelates.

Chelation reaction suppressed on the preoxidized steel surface
To further detect the chelation reaction mechanism for Fe 3+ -diketone formation, we conducted the tribological test on a pre-oxidized steel surface with lubrication of EPBD-0201 solution. After thermally treated for 1 h at 1,000 °C under air condition, both Fe 2+ oxide and Fe 3+ oxide were formed at the steel surface. Compared with the pre-oxidized 440C surface, the Fe 2+ and Fe 3+ oxides on the original 440C surface are less, and the Fe 0 can be detected on the pristine 440C surface ( Fig. 9(b)). Different from that on the pristine steel surface, the COF on the pre-oxidized steel surface roughly maintains a stable value of 0.13±0.01 during the whole 2 h sliding times ( Fig. 9(a)).
No remarkable difference in the XPS spectra measured at the pre-oxidized steel surface and inside the wear track (Figs. 9(b) and 9(c)) indicates that the chelation reaction on the iron oxide surface is very limited. It is reasonable to hypothesize that the Fe 3+ iron in the oxide layer is insoluble and then difficult to react with diketone molecules. In other word, the iron ions produced in the sliding process that chelate with diketone molecules to achieve low shear should be capable of dissolving in diketone solution [38][39][40].
Our results suggest that pre-removal of the native iron oxide layer should be of great significance for practical applications of the steel/steel sliding interface to achieve superlubricity with diketone lubrication.

Role of oxygen in superlubricity of steel/steel with diketone lubrication
Superlubricity can be achieved at the steel/steel interface lubricated with EPBD-0201 solution mainly due to the formation of chelate with low shear strength [15,[16][17][18][19][20][21]; however, the long running-in process largely restrains the applications of this technology.
Our results show that rich oxygen in the surrounding environment not only substantially shortens the running-in time but also further reduces the stable friction force (Figs. 1 and 2). On the one hand, oxidation wear of the steel ball is facilitated by oxygen, resulting in the sharp drop of contact pressure ( Fig. 10(a)) [22]. On the other hand, oxygen enhances the chelation reactions ( Figs. 6 and 7), improving the chemisorption layer forming at the steel/steel sliding interface. The roles of surrounding oxygen may be similar with the contributions of H 2 O 2 in the chelation reactions on steel surface, which has been pictured by molecular dynamic simulations [41]. First, oxidation reactions cause the evagination of iron from steel surface along with the dissociation of Fe-Fe bonds [41]. The raised Fe atoms are activated and react with diketone molecules to form chelate under mechanical shear stress ( Fig. 10(b)). Second, oxidation wear results in the growth of free irons in the diketone solution confined between the sliding interface, which is capable of promoting the chelation reaction with diketone molecule [42][43][44]. Furthermore, we also found that the sliding in rich oxygen environment produces the smoother worn steel ball and substrate surfaces (Fig. 4), which chemisorb more chelate (Figs. 6 and 7). It is reasonable to infer that lower contact pressure benefits the chemisorption of the chelate layer due to the weaker squeezing action. This may be the reason for the rapid growth of surface adsorption in Stage II of running-in process.

Conclusions
In the present study, we demonstrated that rich surrounding oxygen can significantly improve superlubricity performance of the steel/steel sliding interface lubricated with diketone solution, i.e., largely shortening the running-in time and further lowering the final stable COF. Compared to oxygen-free condition, a 50 wt% O 2 content environment can shorten the running-in time by ~80% and reduce the stable COF by 60% to a final value of 0.004. In general, the initial decrease of friction between the steel/steel interfaces lubricated with diketone undergoes two running-in processes. In the first running-in stage, the sharp drop of COF is mainly attributed to the decrease of contact pressure, which is facilitated by severe oxidation wear in rich O 2 environment. In the second running-in stage, oxygen promotes the growth of the chelate layer, which owns super-low shear strength and finally results in the stable superlubricity of the steel/steel interface. Here, we proposed a new and simple method to improve the stability and durability of superlubricity between steel/steel contacts, which would greatly promote the industrial applications of superlubricity technologies.