Engineered silica nanoparticles as additives in lubricant oils

Silica nanoparticles (SiO2 NPs) synthesized by the sol–gel approach were engineered for size and surface properties by grafting hydrophobic chains to prevent their aggregation and facilitate their contact with the phase boundary, thus improving their dispersibility in lubricant base oils. The surface modification was performed by covalent binding of long chain alkyl functionalities using lauric acid and decanoyl chloride to the SiO2 NP surface. The hybrid SiO2 NPs were characterized by scanning electron microscopy, transmission electron microscopy, Fourier transform infrared spectroscopy, simultaneous differential thermal analysis, nuclear magnetic resonance and dynamic light scattering, while their dispersion in two base oils was studied by static multiple light scattering at low (0.01% w/v) and high (0.50%w/v) concentrations. The nature of the functional layer and the functionalization degree seemed to be directly involved in the stability of the suspensions. The potential use of the functional SiO2 NPs as lubricant additives in base oils, specially designed for being used in hydraulic circuits, has been outlined by analyzing the tribological properties of the dispersions. The dendritic structure of the external layer played a key role in the tribological characteristics of the material by reducing the friction coefficient and wear. These nanoparticles reduce drastically the waste of energy in friction processes and are more environmentally friendly than other additives.


Introduction
Nanoparticles (NPs), particles with at least one dimension between 1 and 100 nm, exhibit specific properties which are not present in the material at a macroscopic scale. These special properties have been widely studied and exploited in several scientific and technological fields, such as, medicine, electronics, optics, magnetic data storage, biotechnology, etc. Dispersing nanometer sized particles in a conventional base fluid results in nanoscale colloidal suspensions, commonly termed 'nano-fluids' [1,2]. When the base fluid is a lubricant oil the term 'nanolubricant' is commonly used [3].
The improved tribological properties of nano-lubricants, compared to the base lubricant, are the result of the combination of some features of the NPs. Firstly, due to their high surface area/volume ratio NPs provide a large interfacial surface area to enhance interaction with the surface material Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. of friction pairs forming a protective film, thus reducing friction and wear [4]. As a result of their small size, NPs also provide a mending effect by filling the surface roughness points, forming a thin physical tribofilm during the frictional process that not only bears the load but also separates the rubbing faces [5]. NPs can act as spacers, rolling between the friction surfaces, changing the pure sliding friction to mixed sliding-rolling friction. The abrasiveness of hard NPs may have a polishing effect by reducing the roughness of the rubbing surfaces [6]. Also, NPs may reduce compression stress concentrations associated with high contact pressure by bearing the compressive force depressively. There are a number of excellent reviews in which the synthesis, characterization and applications of nanofluids and nanolubrication are overviewed [7][8][9][10].
One of the main challenges to develop a stable nanolubricant formulation is to disperse the NPs in the liquid base oil. Due to their high surface energy, NPs tend to aggregate, cluster or coagulate. Besides, due to the weight of the aggregates formed, they cannot be maintained in suspension by Brownian agitation and tend to sediment. Many studies have been made to prepare stable nanolubricants, so preventing/minimizing those drawbacks. For example, intensive magnetic agitation, ultrasonic agitation and/or any other mechanical stirring are widely used physical techniques for breaking large agglomerates and dispersing particles in the base lubricant. Some other works deal with the electrostatic stabilization (charge repulsion) by which the NPs acquire surface charges through one or more of the following mechanisms: (i) adsorption of ions, (ii) dissociation of surface charged species, (iii) isomorphic substitution of ions, (iv) accumulation or depletion of electrons at the surface, and (v) physical adsorption of charged species onto the surface [11]. Addition of surfactants is also a widely used technique; for example, some metal oxide NPs do not present affinity with the base oil and their dispersion is not easy without the use of surfactants [12]. Ultrasonic dispersion with surfactants is also successful [13].
Long-term stability of nanolubricants can be achieved by use of surfactant-free functionalization by grafting hydrophobic functional groups on the surface of the NPs by chemical routes. In fact, synthesis of hybrid NPs in which the inorganic NPs are surface-modified with organic layers are promising as lubricant additives [14], although their use in the improvement and/or development of new nanolubricants is still in its infancy.
Many types of nanoparticles have been used as lubricant additives, such as organic [15][16][17] (graphenes, fullerenes, Teflon, graphite, diamond) and inorganic [15,[18][19][20][21][22] (metals, sulfur, fluorides, borates, metal oxides and hydroxides) ones. Among them, the use of surface-modified SiO 2 NPs as lubricant additives is scarcely reported [15,[23][24][25] in spite of the fact that SiO 2 NPs possess excellent mechanical properties in terms of hardness, thermal stability and large surface area. Besides, SiO 2 NPs are cheap, available on the market, and high quality SiO 2 NPs can even be prepared by sol-gel technology in the lab. In 2006, Li et al prepared dispersible SiO 2 NPs by an in situ surface modification approach in which the hydrolysis product of sodium metasilicate was used as the monomer and a silane coupling agent was the long chain terminator (iso-octanol, octylic acid) [26]. According to the authors, these NPs have superior dispersibility and stability in many types of organic solvents and tribological studies also revealed potential in lubrication. More recently, Peng et al investigated the tribological properties of liquid paraffin with oleic acid-modified SiO 2 NPs and found that optimal concentrations of SiO 2 NPs were associated with better tribological performance than pure paraffin oil [27]. They also reported that the SiO 2 NP tribological properties depended on the NP size, with those of 58 nm diameter providing better friction-reduction, anti-wear and load-carrying capacity than the pure paraffin oil. In a recent paper, Kim and Archer densely grafted SiO 2 NPs with amphiphilic organic chains to create a covalently tethered corona of sulfonate-functional alkylaryl chains around the NPs [14]. These NPs dispersed homogeneously in poly-α-olefin base oils and the resulting nanolubricant exhibited superior tribological properties in comparison with the pure base oil.
In line with the previous research work as reviewed above, in the present paper, the surface surfactant-free functionalization of SiO 2 NPs using organic long-chains was addressed and their performance as additives in base lubricant oils was studied. SiO 2 NPs were synthesized by the sol-gel approach and the surface was engineered by grafting long chain alkyl functionalities using lauric acid and decanoyl chloride with the aim of improving their dispersibility in lubricant base oils. The engineered hybrid SiO 2 NPs were characterized by scanning electron microscopy (SEM), Fourier transform infrared spectroscopy (FTIR), simultaneous differential thermal analysis (SDTA), nuclear magnetic resonance (NMR) and dynamic light scattering (DLS) experiments, while their dispersion in two base oils was studied by static multiple light scattering (MLS) in a Turbiscan Lab ® Expert apparatus at low (0.01% w/v) and high (0.50%w/v) concentrations. The potential use of the engineered SiO 2 NPs as additives in two lubricant oils specially designed for use in hydraulic circuits has been evaluated by checking the tribological properties of the dispersions.  [28]. Briefly, 22.5 mL of 25% ammonium hydroxide and 9 mL TEOS were added to 312.5 mL ethanol in a reaction flask under constant stirring (750 rpm). The reaction mixture was kept under reflux at 60°C for 24 h. Then, the solvent was removed in a rotary evaporator at 45°C until the sample was almost dry. Finally, the reaction product was heated overnight in an oven at 70°C to obtain a fine white powder of SiO 2 NPs. In order to activate the surface of the NPs, a thermal activation was carried out by heating the sample at 130°C for 21 h and then at 250°C for 3 h. These final activated SiO 2 NPs were stored in a desiccator until further use.

Functionalization of SiO 2 NPs.
(a) Synthesis of amino-functional SiO 2 NPs. Exposed amine groups on the SiO 2 surface were performed by functionalization with two different silane coupling agents, APTEOS and TMSDETA. An adaptation of the protocol by Kim and Archer was used [14]. One gram of SiO 2 NPs was suspended in 70 mL water using an ultrasound bath for a few minutes after which 0.1M NaOH solution was added dropwise under constant stirring until pH 9 was reached. The dispersion was warmed up to 95°C and 1.43 mL TMSDETA or 1.30 mL APTEOS was then added. The mixture was kept at 95°C until its volume was reduced to 1/ 3. This process was repeated until a total amount of 4.29 mL TMSDETA or 3.90 mL APTEOS was added. Finally, the suspension was left at 95°C until more than 50% solvent was removed. Once the suspension was cooled down, it was further dialyzed using a MWCO 12 000 dialysis sack versus deionized water for 2 days in order to remove the reagents excess. The final products were precipitated three times using THF to remove completely the remaining water and centrifuged at 7500 rpm for 20 min.
(b) Synthesis of amino-SiO 2 NPs with lauryl pendant groups. 1 g of the solid obtained in (a) was suspended in 180 mL of toluene:2-propanol mixture (5:1, v/v) and stirred overnight to ensure a homogeneous suspension. 3.30 g of LA solubilized in 50 mL of toluene:2-propanol mixture (5:1, v/v) was then added, allowing it to react for 72 h under constant stirring at 1000 rpm at room temperature. The NPs were then separated by centrifugation at 7500 rpm for 20 min and washed three times with toluene:2-propanol mixture (c) Functionalization of amino-SiO 2 NPs with decanoyl chloride. A proper amount of SiO 2 NP APTEOS obtained in (a) was heated at 45°C for two hours in an oven in order to completely remove any traces of remaining THF. One gram of solvent-free SiO 2 NP APTEOS was dispersed into 50 mL dichloromethane using an ultrasonic bath and 1.66 mL DCl, 1.67 mL TEA and about 100 milligrams of DMAP as catalyst were then added. The mixture was allowed to react at 45°C continuously stirring under inert N 2 atmosphere for 24 h. Then, the solid product was separated by centrifugation at 7500 rpm for 30 min, washed twice with dichloromethane and centrifuged. Finally, the material was washed with ethanol until the supernatant lost its orange color. The functional material SiO NP 2 APTEOS DCl was obtained after centrifugation and dried overnight at 40±2°C. It was stored in a desiccator until further use. A molar 1:1:3 ratio NPs:APTEOS:DCl was strictly kept throughout the syntheses. Morphology and dimensional analysis. The morphology and size of the NPs were checked using a JEOL-6610LV field emission scanning electron microscope (JEOL, Japan) equipped with an integrated INCA Energy 350-Xmax 50 microanalysis system. The microscope was operated at 20 kV. SEM samples were prepared as follows. A few milligrams of solid were suspended in 1 mL ethanol and sonicated for 3 min. Then, three aliquots of 0.1 mL of the homogeneous suspension were dropped on a sample holder and left to evaporate. Samples were coated with a thin gold layer using a high vacuum evaporator Bal-Tec SCD 005 with cold cathode vaporization due to their low electrical conductivity. The mean particle size was analyzed from the digitized images using ImageJ Tool software where at least 250 observations were performed for each sample. The size and morphology of the synthesized nanoparticles were also evaluated by a JEOL JEM-2100F field emission transmission electron microscope, operated at 200 kV. The microscope was equipped with an ultrahigh-resolution pole-piece, which provided a point-resolution better than 0.19 nm, and an energy dispersive x-ray (EDX) detector for elemental microanalysis. Size distribution by number was also .025%S). The assay consisted in a rotating steel ball under load against three stationary steel balls, held in a cradle shape and immersed in the tested nano-dispersion. After finishing the assay, the wear scars in the stationary balls were evaluated with a Nikon PFX optical microscope coupled to a Nikon F-301 CCD. Repeatability and reproducibility were verified following ASTM D2783 recommendations. All test-section components were ultrasonically cleaned in heptane for 3 min, then rinsed with ethanol and dried under hot air before and after tests.

Results and discussion
As a general rule, metal oxide NP suspensions in hydrophobic media tend to have a poor stability, especially when the concentration rises. This is due to aggregation and sedimentation phenomena which are related to the presence of hydroxyl groups on the SiO 2 NPs surface, their large specific area and specific surface energy. The higher the amount of surface hydroxyl groups, the lower the hydrophobicity of the NP and, therefore, the lower stability in hydrophobic media. A straightforward method to improve the compatibility of SiO 2 NPs with oily matrices, thus improving their dispersibility, is grafting hydrophobic chains to SiO 2 NPs surface.
In the present research, surface modification of SiO 2 NPs was addressed by using two silane coupling agents which provide amine groups for further reaction (figure 1). These silane coupling agents, TMSDETA and APTEOS (see figure  S1), bound to the SiO 2 NP surface forming 1-3 Si-O-Si links through a condensation reaction with the surface silanol groups. The presence of the amine groups allowed for facile linker chemistry with LA and DCl (see figure S1), which should provide the hydrophobic shell (hybrid SiO 2 NPs) to improve the NP stability in hydrophobic media.
SEM, high-resolution transmission electron microscopy (HR-TEM) and DLS techniques were used in order to study the shape, size and morphology of raw SiO 2 NPs, as well as to check possible changes after the surface modification steps. SEM and HR-TEM images revealed monodisperse spherical SiO 2 NPs with a mean diameter 102±33 nm (figures 2, S2 and S3(a)). This value was slightly smaller than that obtained by DLS, 115±36 nm. This may be due to the fact that DLS  provided the diameter of the solvated NPs [32]. Also, SEM and HR-TEM images of the hybrid SiO  DCl must be related to the synthesis procedure. In the latter, the solvents, temperature and presence of a catalyst should be responsible for the fluffy morphology and NP agglomeration ( figure 2). Also, the morphology of the hybrid NPs, SiO NP is not clearly evident from HR-TEM images (figure S2) because the nanoparticles are strongly aggregated (thanks to the hydrophobic surface shell) during the drying on the support film of the HR-TEM specimen grid. However, as will be shown below, the dispersibility of SiO NP 2 TMSDETA LA in the base oil was high. Table 2 summarizes the C, N and H elemental composition of the synthesized NPs. As can be observed, hybrid APTEOS-based NPs had a higher nitrogen content than hybrid-TMSDETA-based ones. These results may be surprising, taking into account that TMSDETA has three nitrogen atoms whereas APTEOS contains only one. However, the presence of three nitrogen atoms in TMSDETA may allow further functionalization with LA or DCl. In fact, a comparison of the carbon:nitrogen (C:N) experimental atomic ratio with the expected one, considering a reaction yield of 100%, the efficiency of the LA and DCl functionalization, was about 40% in the case of APTEOS-based SiO2 NPs while for SiO NP 2 TMSDETA

LA
the efficiency was about 80% if one LA was considered, 48% for two LA molecules and 35% for three LA groups. In light of these results, it is suggested that a more hydrophobic branched shell has been created around the SiO NP ) appeared due to the presence of C=O bonds [33]. These results indicated the successful grafting of the hydrophobic tails on the SiO 2 NPs. 13 C and 29 Si solid-state NMR spectra were also evaluated in order to obtain information regarding cross-linking, polymerization and functionalization of the different materials. The 13 C SS-NMR spectra of bare SiO 2 NPs (figure S5), no signals around 19 ppm or 60 ppm (ascribed to surface etoxygroups, Si-O-CH 2 -CH 3 ), could be detected as should be expected for NPs synthesized using TEOS, which confirmed that the precursor was fully hydrolyzed during the synthesis of the NPs [35]. On the other hand, 13 C SS-NMR of the hybrid NPs revealed the presence of signals characteristic of Si-CH 2 -CH 2 -CH 2 -N-(between 32.3 and 9.8 ppm) and CH 3 -(CH 2 ) n -CO-(n=8, 10, between 47.7 and 12.0 ppm) structures, due to the amine functional group and the amide formation with the long chain LA (or DCl), respectively (see figure S5). 29 Si SS-NMR spectra of raw SiO 2 NPs also showed typical peaks assignable to a Si-O bond: signals at −91.3 ppm, −101.5 ppm and −110.0 ppm that correspond to Q n shifts ( figure 3). The intense signal located at −101.5 ppm , is typical for Si atoms bond to a hydroxyl group and three siloxane bridges in a highly cross-linked matrix [36,37]. After functionalization, new signals appeared in the T n zone [34] at −57.3 ppm and −66.7 ppm, which confirmed the incorporation of amino groups provided by the silane coupling agents to Si atoms in the surface structure of the xero-gel. TGA and SDTA showed two different regions for the synthesized materials ( figure S6). In the first one, between 100°C and 350°C, all the materials exhibited a similar loss of mass (3.7%-5.2%), which can be attributed to a loss of physically adsorbed water (maximum around 100°C) and residual organic solvents. For unmodified SiO 2 , in the temperature range from 100 to 350°C, the weight loss was low (<2%). In the second region, between 300°C and 800°C, bare SiO 2 NPs remained more or less non affected (loss ∼6% mass) and the TGA curve was practically flat. According to literature, the temperature range from 200 to 400°C corresponds to surface silanol condensation whereas isolated hydroxyl groups (inside the mesoporous structure) are condensed at higher temperatures (up to 500°C) [38,39]. On the other hand, the hybrid NPs suffered the highest loss of mass (figure S6) between 350 and 800°C. This loss may be attributed to the thermal decomposition of chemically bonded LA and DCl groups. The weight loss for the SiO NP where the amount of organic composition introduced by LA and DCl functionalization was calculated from the TGA weight loss data obtained between 300 and 800°C. Functional SiO 2 was the weight retention by TGA corresponding to the bare amino-functional silica at 800°C. Results are summarized in table 3.
These results demonstrate that the grafting degree of SiO NP 2 TMSDETA LA was higher than that of amine-NPs based on APTEOS and are supported by the elemental analysis (vide supra) from which the efficiency of the grafting was found to be higher for the SiO 2 NPs based on the TMSDETA aminecoupling agent.
As stated before, the main objective of the functionalization process was to improve the dispersibility of the NPs used as lubricant additives. With this aim, the two base oils BO 32 and BO 68 (table 1), with high stability and specially designed for being used in hydraulic circuits, were selected as dispersion media in order to check the dispersibility of the synthesized NPs. Suspensions of 0.01%, 0.04%, 0.15% and 0.50% (w/v) were prepared according to the protocol described in the experimental section. Changes in the backscattered light (ΔBS) radiation as a function of time and cell height were measured 8 days after the corresponding suspension was prepared. These changes were evaluated by subtracting the initial value of the transmitted (TS) or backscattered (BS) light at t=0 (TS t=0 or BS t=0 ) from the values obtained at a time t=m, (TS t=m or BS t=m ) at a given cell height. The value of the backscattered radiation is related to the photon transport mean free path l * , which represents the mean free path run by the photons before losing the direction of the incident light. Considering a narrow area with a differential thickness, dh, like that in the measuring system, the BS can be expressed as [41]: The ΔBS parameter was evaluated in three different height positions of the cell. In the middle part, BS variations involve changes in particle size, whereas ΔBS and ΔTS in the upper and lower part of the cell provide information about migration, sedimentation and creaming phenomena [42]. Increase or decrease of BS with time is related to the ratio particle size/wavelength of incident light, so that when a light source of 880 nm is used, BS grows as the size of the NPs aggregate up to 0.6 μm. If the nanoparticle 'flakes' overtake this threshold, BS starts to diminish [43].
According to the experimental protocol used, there was a gap time between finishing the suspension of the NPs using ultrasound and starting the BS measurements. Thus, those unstable suspensions settled before being able to start the measurement (t=0). In these cases, data of ΔBS did not reflect accurately the real processes that occurred in the cell. This fact was particularly evident for SiO 2 NPs suspensions, where a white layer of settled NPs could be observed in the lower part of the cell by the naked eye before starting measurements. In contrast, this was not observed for hybrid NPs. For this reason, we missed out the discussion of SiO 2 NP results as they were not comparable to the data obtained for the functional materials. Figure 4 summarizes the BS profiles versus the cell height at three different times of the suspension preparation.  SiO NP 2 TMSDETA LA coating favored the self-assembling (aggregation) of the NPs via hydrophobic interactions among the alkane chains. Finally, figure 4(c) shows that an increase in the suspension concentration of SiO NP 2 APTEOS LA in BO 32 resulted not only in a BS increase due to aggregation, but also in a sedimentation process as observed for the BS decrease at the bottom of the cell (<5 mm height).
In figure 5, the BS results obtained for the different hybrid NPs in the two base oils are represented as a function of elapsed time after dispersions preparation and their concentration. In general, it was observed an increase of BS with time along the cell height whatever the base oil used. In most cases, after the second day, BS reached a plateau value that remained constant for at least 8 days. Also, as more diluted the suspensions, the lesser change in BS with time and cell height. The kinetic BS profiles were obtained plotting incremental values of BS (ΔBS) versus time for 8 days. Table 4 summarizes the maximum ΔBS (ΔBS max ) values obtained by applying equation ( From these data, it was clear that not only does the amine silane coupling agent for further functionalization of SiO 2 NPs have a decisive effect on the stability of the suspensions, but so does the synthesis procedure and physical properties of the base oil. So, when comparing the data ( LA the former exhibited a better dispersibility in both base oils than the latter. As stated before, a possible explanation for these results may take into account the dendrimer-like structure of the SiO NP 2 TMSDETA LA layer. In order to shed light on the migration of NPs from the bulk dispersion towards the bottom of the cell, the concentrated SiO NP 2 APTEOS LA 0.5% (w/v) dispersion was used. In figure 6(a) the % BS vs cell height (2-12 mm), after different days of dispersion preparation, was plotted for both base oils. As can be seen, the behavior of these systems does not follow the Mie theory [41], which explains the classical sedimentation phenomenon: BS increase in the lower part of the cell as a consequence of the increment in the concentration of the dispersed phase (sediment) and a simultaneous decrease in the upper part of the cell as a consequence of the clarification phenomenon.
Our system showed atypical behavior, since BS decreased in the lower part of the cell while it increased in the upper part of the cell. After 1 day of preparation, BS decreased in the height range 2-8 mm and that decrease shifted to about 5 mm and 7 mm height after 5 days of dispersion aging in BO 32 and BO 68 , respectively. At the bottom of the cell, when the concentration is increased and a 'cake' of NPs is formed, it is possible to have dependent diffusion leading to a decrease of BS, which is a characteristic phenomenon of particle packing. This trend was also observed in figure 4(c).
According to data in figure 6(a), we can assume that after a consolidation period of about 5 days, three zones may be considered: a zone of NPs with similar size free in solution and a sediment layer, separated by a diffusion interface (1 mm). figure 6(b) shows the rate of NP settling by observing the change in height of the sediment layer with time in both base oils. It can be seen that the rate of sedimentation decrease in the more viscous oil BS 68 as expected due to increased interactions (friction) of NPs with the oil which, in turn, increase the oil dispersion viscosity [44].

Outlook of engineered hybrid NPs use as lubricant additives
Once the objective of the stability was achieved, the tribological properties of the hybrid NPs were evaluated. The friction coefficient and the average diameter on the wear scar were calculated in a standard 4-ball ASTM 4172-94 assay using hybrid NPs suspended in both BO 32 and BO 68 base oils at 0.5% w/v as lubricant. The suspensions were prepared just before starting the tribological experiment so that the aggregation or precipitation effect could be neglected (ΔBS<3% within the first six hours for cell-heights 10-30 mm).
The results demonstrated that the diameter of the wear scar was larger when NPs were incorporated either the base oil used, except in the case of SiO NP , 2 TMSDETA LA for which the wear scar diameter was similar to that observed for the neat base oil ( figure 7). These results could be explained by taking into account the functional surface of the NPs (figure 1). The dendritic-like structure of the cover shell of SiO NP   nanoparticles. On the other hand, only those LA-functional NPs improved the friction coefficient at a similar level than the BO 32 ( figure 7).
These data suggested that the alkane functional coating of the SiO 2 NPs may act as a protective buffer between the contact area, smoothing the rough areas and reducing the friction under shear forces through a rolling mechanism.

Conclusions
SiO 2 NPs have been engineered with hydrophobic layers in order to prepare stable suspensions in base oil matrices for their potential use as anti-wear and friction reducing additives. Long alkane chain-hybrid SiO 2 NPs have been demonstrated to reduce the friction coefficient and wear. This reduction implies a strong reduction in the spend of energy as a consequence of the friction processes, thus also reducing the cost of lubrication. The nature of the functional layer and the functionalization degree seemed to be directly involved in the stability of the suspensions. The dendritic structure of the external layer played a key role in the tribological characteristics of the material. Furthermore, the physical characteristics of the base oil also have an important influence on the anti-wear properties of the NPs. Future work will include the tribological performance of these materials in a formulated oil.