Improved tribological and thermal properties of lubricants by graphene based nano-additives

Abstract

Enhancing the tribological performance of lubricants with nanoparticle additives is a recent challenge. The purpose of this study was to investigate the potential advantage of nanolubricants, a new class of advanced lubricants integrating nano-sized materials, to reduce friction and wear processes, in view of applications in compressors for refrigeration. An investigation of tribological and thermal properties of nanolubricants for vane-on-roller systems was carried out through ball-on-disk wear tests and thermal diffusivity photo-acoustic measurements from room temperature to 70 °C. Nanofluids containing different concentrations of graphene based nanostructures in oil were tested. Poly-alkylene glycol was selected as the oil, being a lubricant suitable for compressors operating with CO₂ refrigerants. The suspensions resulted stable with time. The dispersed nanostructures played an important role in protecting surfaces against wear phenomena and in improving the friction properties and load carrying capability of raw oil. A maximum decrease of 18% for friction coefficient and over 70% for worn volume were estimated in boundary lubrication conditions, the most severe for tribological couplings. Finally, durability experiments and Stribeck tests confirmed the benefits produced by nano-additives in different lubrication regimes. Thermal investigation proved also the advantage of using nanolubricants in heat exchange properties in laminar flow conditions, e.g. during compressor power-up phases.

---

Introduction

Energy saving through increased efficiency of all devices is among the cornerstones of energy and environmental commitments of society as a whole. A prime example is the energy consumption by refrigerators, which is growing in the last years, accounting for almost 14% of the total electricity consumption from the residential sector.¹ Therefore, an increase in their efficiency plays a crucial role in reducing energetic costs globally. At present, the refrigeration industry is working to achieve an equilibrium between environmental safeguards and energy saving: in terms of environmental protection, the former use of chloro-fluorocarbons (CFCs) and then of hydro-fluorocarbons (HFCs) as working fluids in refrigerator compressors is being replaced by more ecological solutions.² Among low global warming potential gases, CO₂ seems to be the most promising option,^3,4 having low toxicity and low flammability,⁵ and it is going to become the most common substitute refrigerant in refrigeration units of domestic refrigerators, chillers, air conditioners, and so on. Indeed, carbon dioxide is non-corrosive, has high volumetric capacity, low pressure ratio, superior heat transfer properties and complete compatibility with conventional lubricants, ready availability, low price and no recycling issues.^6,7

The introduction of alternative refrigerants and lubricants modifies the severity of the tribological contacts in devices, increasing operational failure probability for traditional designs. In this context, the tribological investigation on critical couplings, the development of new materials and surface solutions, as well as the understanding of lubrication mechanisms especially in the boundary/mixed regime are under intense research investigation.⁸ A critical component in all refrigeration and air-conditioning systems is the compressor, which is typically rolling piston or vane type. Rotary type compressors are widespread since they are simple and compact, have good dynamic equilibrium characteristics and high reliability.⁹ Nowadays, in these devices there is a need for more robust and wear-resistant components, as well as for a more efficient lubricating action in order to reduce frictional losses and achieve reliability in slithering components. Rapid wear on the sliding parts is a critical issue, especially in the rotary compressor^10,11 and, typically, the wear phenomena between vane and roller are the most harmful,⁹ since the increase of wear and friction on that tribological coupling induce great power consumption and shorten the operating life of whole device.

Considering the lubricant action in terms of heat exchange capacity, the working fluid is a mixture of air and oil vapors, so the unit typically behaves like an open system, characterized by input and output enthalpy flows.¹² In general, the convection heat transfer coefficient depends on the thermal conductivity of the fluid, the difference between the bulk fluid temperature and the wall temperature and, finally, the temperature gradient.¹³ Thus, the thermal conductivity of the working mixture becomes decisive. Recently scientists used nanoparticles in refrigeration systems because of their considerable improvement in thermo-physical and heat transfer capabilities, to enhance the efficiency and reliability of refrigeration and air conditioning systems.^14,15

Lately, nanolubricants have been widely studied as an alternative solution to conventional lubricant oils, since they allow obtaining a significant friction reduction and interesting improvements in load-carrying capability.¹⁶ The nano-additives reduce friction and wear on surfaces that operate in sliding contact with each other.^17,18 Graphene based materials are of particular interest, showing an improvement in thermal and tribological properties of mineral oil¹⁹ or grease.²⁰ However, in most cases to obtain stable suspensions the functionalization of graphene is necessary²¹ or the use of particular morphologies as crumpled graphene balls, that show superior performances in lubricants as polyalphaolefins.²² Good suspension stabilities and performances can be achieved with carbon based nanomaterials such as carbon nano-onions,²³ carbon nanotubes (CNTs)²⁴ and graphene “nanostars” known as carbon nanohorns (CNHs). CNHs with respect to CNTs have low manufacturing cost and weak toxicity,²⁵ due to both the lack of fibril-like structure and the absence of metal nanoparticle catalysts. CNHs have been the subject of numerous studies due to their unique morphology and wide ranging properties of graphene, including chemical stability, low surface energy and high thermal and electrical conductivity.^26–28 Tribological tests performed on the macroscale with these graphene-based nanostructures contained in polyimide coatings in dry conditions resulted in reduced friction and wear effects.²⁹ Nanolubricants containing such nano-additives were also studied in an internal combustion engine:³⁰ benefits were observed due to the presence of nanostructures, as reduction of both friction coefficient and mean wear rate, compared to raw oil.

In this work, an extensive study on the tribological and thermal properties of nanolubricants for vane-on-roller systems is carried out, by testing suspensions with different concentrations of CNHs in oil. Poly-Alkylene Glycol (PAG) was selected as the base fluid to prepare nanolubricants, being suitable to operate with CO₂ as refrigerant. PAG lubricants offer some interesting advantages over other mineral oils, since they have low solubility in the CO₂ refrigerant and provide high chemical stability and very good lubricity at high pressures and temperatures.³¹ Modifications in tribological and thermal behaviors of produced nanolubricants are evaluated through wear tests and photo-acoustic measurements at different temperatures. In term of lubrication properties, aim of the current article was to measure friction (boundary and mixed) and wear under pure sliding conditions, recognize differences among prepared nanolubricants in their friction behavior, and thereby develop a better understanding of action mechanisms played by dispersed nanoparticles in a conventional lubricating oil.

Experimental

A Poly-Alkylene Glycol (PAG) for CO₂ refrigeration systems was used as base fluid for nanolubricants. The general chemical structure is reported in Fig. 1.

Fig. 1 General chemical structure for synthetic PAG.

The chemical–physical features of the commercial oil are summarized in Table 1.

Table 1 Characteristics of as supplied BREOX PAG. The viscosity-pressure coefficient α was calculated according to So et al.³²

Features
Viscosity – at 40 °C	46 cSt
Viscosity – at 100 °C	10.7 cSt
Density at 20 °C	0.998 g cm^−3
Pour point	−49 °C
Flash point	>200 °C
Viscosity-pressure coefficient α at 25 °C	17.2 GPa^−1
Viscosity-pressure coefficient α at 70 °C	15.2 GPa^−1

---

Carbon nanohorns are tiny graphene sheets, wrapped to form horn-shaped cones with a half-fullerene cap, having 30–50 nm length and 2–5 nm diameter. They generally group together and form aggregates (spherical clusters or bundles) like nanostars.²⁷ A schematic reproduction of their structure can be found in ESI, Fig. S1.† These nanostructures were provided by Carbonium S.r.l. and were produced by a process based on rapid condensation of carbon atoms without any catalyst and with highly reduced costs of production.³³ Fig. 2 shows a SEM picture of their morphology. The aggregation is due to the drying process in preparing the specimen to perform SEM observation. The mean dimension of carbon nanohorns was evaluated through image analysis carried out on SEM micrographs by using ImageJ 1.46r software.³⁴ The estimated mean diameter was (80 ± 6) nm.

Fig. 2 SEM picture of as supplied CNHs.

Nanolubricants were prepared by a two-step method. Suspensions containing 0.04%, 0.1%, 0.2%, 0.5% and 1% weight fractions of nano-additives were prepared through ultrasonic irradiation, with a Sonics&Materials VCX130 ultrasonic processor, operating at 20 kHz and 130 W maximum power, and equipped with a 6 mm diameter Ti–6Al–4V alloy tip. Ultrasonic emission was applied for 30 min of actual irradiation, divided in cycles of 5 s of sonication and 5 s of pause. Heating produced by ultrasonic irradiation was offset by a thermostatic bath, in order to avoid any possible structural modification and/or degradation phenomenon induced by possible increase of temperature due to cavitation. Nanofluids viscosity was measured at 25 and 70 °C by a rotational rheometer, AR G2, TA Instruments, in a range of shear rate between 80 and 1200 s⁻¹, using a cone-plate geometry.

In suspensions, the particle size and the possible presence of aggregates were examined by Dynamic Light Scattering (DLS) (Zetasizer Nano, Malvern). It is important to outline that, since DLS is reliable only when particles are strictly subjected to Brownian motion, data collected in oils are only indicative of the agglomeration trend and cannot be used as accurate or absolute measurements for actual hydrodynamic diameters even if measured viscosity values where used to calculate mean size values. Moreover, the presence of additives in the pristine fluid can affect the real size of nanoparticles and prevent any measure of zeta potential. Thus, DLS data were used in this work for evaluating the aggregate size evolution with time and then the nanolubricants stability.

The thermal diffusivity (α) evaluations were carried out using a homemade device exploiting the photoacoustic effect. In such instrumental setup, a laser with wavelength of 808 nm was modulated at the frequency of 2 Hz. The laser beam impinges on a Si slab immersed in the liquid and generates thermal waves that travel through the sample and reach a photoacoustic chamber able to detect them. The distance between the Si slab and the photoacoustic chamber was changed during the measurement, and the instrument provided two independent estimations of thermal diffusivity in the temperature range 25–65 °C, one from the signal amplitude and one from the phase shift between the laser and the signal recorded by the photoacoustic chamber. All values reported in this paper are the average of the two estimations. More details about the technique and instrument are reported elsewhere.³⁵

The tribological wear tests were carried out with a Bruker UMT-2 tribotester, set for pure sliding contact geometry. Since in a rotary type compressor severe wear occurs on the contact surfaces between the vane and the roller, a ball-on-disk geometry was chosen in sliding tests. A non-conformal contact type was realized, which is similar to that normally achieved between the vane and the roller during the common operation of the compressor. In pure sliding experiments, ball was held stationary, while disk was allowed to rotate, so SRR (i.e. the slide-roll-ratio, defined as the ratio of sliding speed to mean speed) equaled to 2. Fig. 3 reports a picture and a sketch of the measurement set-up. The disks were made of AISI 304 stainless steel and the balls were standard AISI 52100 polished ϕ 5 mm bearing balls with a certified hardness of 65 HRc. The surface of the roller disks was ground with SiC papers to obtain a mean roughness of R_a 0.1 μm that is in the typical range for conventional rollers.

Fig. 3 (a) Picture of the set-up, (b) sketch of the ball-on-disk contact geometry.

The non-conformal contact between the steel ball and the substrate was investigated in the boundary lubrication regime, which is the most severe and generally rules the service life of the coupled mechanical components.³⁶ Tests lasted for 120 min and an average sliding speed of 75 mm s⁻¹ was set, via rotary motion mode. Thus, the overall friction path was 550 m. The test temperature was kept constant at respectively 25 and 70 °C, controlled through a thermocouple under the disk holder. The lubricant oil was filled to cover the disk: the geometry of the tribometer and the rotating speed guaranteed the continuous lubrication of tribological coupling and the absence of starvation mechanism. After temperature stabilization, the normal load was kept constant in order to obtain a 1 GPa of initial Hertzian contact pressure. The sliding speed was selected to guarantee the development of boundary lubrication conditions.

The topography of worn surfaces was examined by optical microscopy, while detailed characterization was carried out by FE-SEM (Sigma, Zeiss). For each sample, wear track depth, width and mean roughness at the bottom were measured with a stylus profiler (Dektak XT, Bruker). In addition, the most promising lubricating system was tested in different lubrication regimes by collecting its Stribeck curve, according to which the coefficient of friction was plotted against a factor including operative parameters such as sliding speed, applied load and lubricant viscosity. Stribeck tests were carried out at 25 and 70 °C. It was not possible to reach the hydrodynamic lubrication condition, because of the limitation of the tribometer, given that the applied Hertzian contact pressure was kept at 1 GPa, to reproduce the settings of wear tests. The sliding speed was varied in the range 0.002–0.2 m s⁻¹. Moreover, the Electric Contact Resistance (ECR) was also collected to evaluate the evolution of the direct metal–metal contact. Lastly some long lasting wear tests were carried out by applying 5 MPa of contact pressure in pin-on-disk configuration, for 10⁶ cycles at 200 mm s⁻¹ sliding speed. These conditions were achieved with the aim of operating for a sufficiently long time by avoiding a too rapid and catastrophic failure of coupled materials, in order to highlight possible differences between oil and nanolubricant.

Results and discussion

The sonication procedure developed for the preparation of nanoparticle suspensions in PAG oil and the peculiar morphology of CNHs resulted very effective in achieving fine suspensions of self-dispersed nanoparticles stable with time without the need of surfactants or CNH functionalization procedures. In fact, Dynamic Light Scattering (DLS) measurements performed for 2 weeks confirmed the stability of nanolubricants in static conditions. DLS data showed stable size trends for nanolubricants containing 0.1–0.5%_wt (1%_wt sample was not tested due to its high viscosity), since measured average hydrodynamic diameter of nanoparticles did not change significantly during measuring time, leading to assume that there were no important aggregation phenomena. Fig. S2† reports an example of the mean aggregate size measured for 2 weeks in oil containing 0.1%_wt CNHs. The percentage deviation of mean size was below 10% for all suspensions and moreover no sedimentation was observed, thus confirming a good stability with time.

Viscosity tests were carried out to verify the influence of dispersed nanostructures on the physical properties of the raw oil at both 25 and 70 °C. As expected, a decrease of dynamic viscosity with temperature was observed. Results are reported in Table 2 and Fig. S3.† Results confirmed a weak increase of mean viscosity value for CNHs concentration up to 0.2%_wt, with deviations within the rheometer uncertainty. For nanohorns concentration over 0.2%_wt the viscosity underwent a significant enhancement, up to a maximum increase of 35% at 70 °C, that could significantly influence the lubricant properties and become unfavorable for the proper operation of the tribological system. The viscosity trend with shear rate (see Fig. S3†) shows the Newtonian behavior of the fluids at both tested temperatures: viscosity was constant with shear rate, and the shear stress raised proportionally to shear rate for nanolubricants containing CNHs up to 0.2%_wt. Instead, the most concentrated nanolubricants showed a shear thinning behavior, i.e. the viscosity decreases when the shear rate increases. This behavior has been frequently noticed in highly concentrated nanofluids.³⁷

Table 2 Values of viscosity measured for raw oil and nanolubricants. Dev.% indicates the viscosity percentage deviation of nanolubricants compared to raw oil

Sample	Dynamic viscosity (Pa s)
	T = 25 °C	Dev.%	T = 70 °C	Dev.%
PAG	0.0819 ± 0.0003	—	0.0172 ± 0.0001	—
PAG + CNH 0.1%wt	0.0832 ± 0.0004	+1.6	0.0174 ± 0.0001	+1.2
PAG + CNH 0.2%wt	0.0847 ± 0.0008	+3.4	0.0176 ± 0.0001	+2.6
PAG + CNH 0.5%wt	0.0897 ± 0.0008	+9.6	0.0199 ± 0.0001	+15.7
PAG + CNH 1%wt	0.0952 ± 0.0002	+16.2	0.0232 ± 0.0001	+35.3

---

In agreement with what supposed by Ettefaghi et al.,³⁸ it was assumed that, as nano-additives fraction increased, aggregation and formation of larger and asymmetric particles might occur, which could prevent the movement of oil layers on each other, thereby increasing mean viscosity. In hydrodynamic lubrication conditions, the increase in lubricant viscosity could be advantageous in terms of load carrying capability, while in boundary lubrication condition the phenomenon would result in a reduction of the friction power loss.³⁹ Since the viscosity modification is a critical aspect, an exceeding viscosity modification could be detrimental for anti-friction properties and behavior of tested lubricants.

As to thermal performance, Fig. 4(a) reports the thermal diffusivity of raw oil and nanolubricants in the temperature range 25–65 °C, as evaluated by photoacoustic measurements. The thermal diffusivity was nearly constant with temperature for all fluids and Fig. 4(b) shows a non-linear increase of average thermal diffusivity with nanoparticles concentration. Interestingly, the thermal diffusivity enhancements were well above those predicted by the Maxwell theory extended to the thermal diffusivity of nanofluids⁴⁰ (6.2% enhancement at 0.5%_wt CNHs versus 0.75% enhancement as predicted by Maxwell model). It is worth remembering that the Maxwell model is valid for diluted colloid where nanoparticles are static, non-interacting and equally spaced.

Fig. 4 (a) Thermal diffusivity of raw oil and nanolubricants at different temperatures. (b) Average thermal diffusivity enhancement as a function of CNH concentration.

The achieved enhancements are however within the Hashin and Shtrikman mean-field bounds for non-homogeneous materials.⁴¹ These bounds take into account the fact that nanoparticles in colloidal suspensions can take several configurations ranging from the perfectly dispersed state assumed by the Maxwell model to chain like structures or fractal clustering proposed by other authors.⁴² Several authors experimentally showed the effect of nanoparticles aggregation on thermal diffusivity.^43–45 Therefore, it is safe to presume that partial aggregation of nanoparticles could be the cause of thermal diffusivity enhancements above the Maxwell model that was observed.

Anyway, the heat transfer rate, which tells the ability of removing heat from the system, is dependent on thermal conductivity and on other thermo-physical properties of the fluid and on flow conditions. In case of laminar flow, that typically occurs in power-up conditions and when the working temperatures are low in compressors, Prasher et al.⁴⁶ developed two expressions, eqn (1) and (2), for fully developed laminar flow conditions (R_e < 2300):

where subscripts nf and f indicate nanofluid and base fluid respectively, ϕ is the volume fraction and C_k and C_η are the thermal conductivity and viscosity enhancement coefficients. As for Prasher et al., the nanofluid is beneficial if C_η/C_k < 4. In this work, the coefficients were calculated for dispersions containing 0.1 and 0.2%_wt nano-additives, which showed an almost constant viscosity. The ratio was about 1.1 and 0.9 for the dispersion with 0.1%_wt CNHs at 25 °C and 70 °C respectively, while it was 1.7 and 0.7 for the dispersion with 0.2%_wt. Considering that obtained values were much lower than 4, the benefit in terms of heat transfer in laminar flow conditions was considerable. Conversely, in case of internal turbulent flow, which is realized in many operating conditions, the Mouromtseff number (Mo) well represents the balance among fluid properties and is given by eqn (3):which takes into account the density (ρ), the thermal conductivity (k), the specific heat (C_p) and the viscosity (η) of the fluid.⁴⁷ Therefore, a sort of figure of merit of heat transfer rates could be estimated by calculating the ratio between the Mouromtseff numbers of nanolubricant with respect to that of raw oil. A Mo value higher than 1 indicates a benefit in term of heat transfer. Calculating the values for dispersions with 0.1%_wt and 0.2%_wt of nanohorns, the Mo number was around 1 for 0.1%_wt and around 1.01 for 0.2%_wt dispersions, thus indicating that in turbulent flow conditions the heat transfer properties were essentially unaffected by the presence of dispersed CNHs.

The friction and wear behaviors of all nanofluids were investigated by using a rotational ball-on-disk device at both 25 and 70 °C, in environmental pressure and humidity conditions. The operational parameters were chosen in order to develop boundary lubrication, for which the coefficient of friction (μ) is in the range 0.1 < μ < 0.3.⁴⁸ The coefficient of friction may give an indication of the lubrication, but for highly loaded contacts the film parameter can provide a better indication of the lubrication mode.

From Hamrock and Dowson⁴⁹ the central-lubrication film-thickness was estimated in two test conditions. It was found to be about 14 nm at 25 °C and 42 nm at 70 °C. The lambda ratio λ, i.e. ratio of calculated elasto-hydrodynamic (EHD) lubricant film thickness to composite surface roughness, corresponding to the mean sliding speed of 75 mm s⁻¹ and to a contact pressure of 1 GPa, was also calculated. These calculations were based on initial roughness and lubricant properties listed in Table 1. The typical definition of lubrication regimes based on lambda ratio is: λ > 3, 0.5 < λ < 3 and λ < 0.5,⁵⁰ corresponding, respectively, to thin film/EHD, mixed and boundary lubrication. Calculated values were found to be respectively 0.38 at 25 °C and 0.12 at 70 °C, both below 0.5, thus indicating that boundary lubrication conditions were achieved during wear tests. All experiments were repeated three times and showed very low variation in friction, so only responses of one representative experiment were presented here.

Fig. 5(a) and (b) show the frictional performances recorded during rotational ball-on-disk wear tests. Mean calculated values for coefficient of friction (μ) was over 0.1 in all tests, stating that the operating conditions were suitable for developing the desired boundary lubrication regime.^51,52 The graphs evidence the friction reduction due to nanohorns addition. Since the friction curves were noisy, a dot-plot was selected for a clearer presentation of data. Thus, each point in Fig. 5 was calculated as the average of friction coefficient values recorded over ten minutes of sliding test. The details of the whole test procedure are reported in the experimental section.

Fig. 5 Coefficient of friction calculated during tribological tests with raw oil and nanolubricants: at (a) 25 °C and (b) 70 °C.

The data showed the typical trend of a wear process: first a short running-in period, during which the highest asperities and the contact surfaces were plastically deformed and worn. This is followed by a steady-state period in which the wear depth is directly proportional to sliding distance. The coefficient of friction was significantly lower for all nanofluids if compared with values recorded for the raw oil, in particular at low temperature. The improvement of μ was evident at both temperatures and the best results were obtained for nanohorns concentration up to 0.2%_wt, which seemed a threshold value for nanoparticles concentration, as already observed by Liu et al.²⁴ that found about 0.025%_wt as the best concentration for the suspensions of functionalized multi walled CNTs in paraffin oil. At high temperature the coefficient of friction of raw oil tended to increase with time, as breakage of the substrate took place. Instead, all nanolubricants at 70 °C showed a stable and constant trend, index of less damage.

Table 3 reports the improvement mean values of coefficient of friction. The PAG/CNH 0.1%_wt system exhibited superior performance, with a minimum coefficient of friction of about 0.134 at 25 °C, for which a significant reduction of 18% was obtained with respect to raw oil, while at 70 °C a reduction up to 7% was detected. As a general remark, carbon nanostructures seemed to better behave at room temperature.

Table 3 Calculated variation of friction coefficient mean value during tribological tests

Lubricant	Coefficient of friction	Dev.%	Coefficient of friction	Dev.%
	T = 25 °C		T = 70 °C
PAG	0.164 ± 0.002	—	0.152 ± 0.005	—
PAG + 0.04%wt	0.145 ± 0.007	−12%	0.144 ± 0.001	−5%
PAG + 0.1%wt	0.134 ± 0.003	−18%	0.142 ± 0.003	−7%
PAG + 0.2%wt	0.136 ± 0.005	−17%	0.142 ± 0.002	−7%
PAG + 0.5%wt	0.140 ± 0.005	−15%	0.146 ± 0.002	−4%
PAG + 1.0%wt	0.156 ± 0.003	−5%	0.149 ± 0.003	−2%

---

The general reduction of coefficient of friction can have two origins: the nanolubricant fills in the gap between friction surfaces while forming an oil film coating on them, the so-called tribofilm, which contains CNHs and thus prevents direct metal–metal contact. In this case the separation between sliding surfaces is enhanced by the presence of dispersed nanoparticles, which partially support the applied normal load. On the other hand, the nanohorns inside the lubricant oil could improve the lubricating yield through the low contact area with surfaces provided by the peculiar horns at their surface. The maximum increase of anti-friction properties was achieved with nano-additive contents that cause negligible variation of viscosity of the base oil. The worsening at higher concentration could be attributed to aggregation and/or interaction among free nanoparticles during operation that affect the real contact area, shape and volume.

Generally, the friction-reduction and anti-wear mechanisms of nanoparticles in lubricants have been reported in several studies as due to colloidal effect, rolling effect, ball bearing, protective film formation, and third body.^53,54 Since these graphene-based nanostructures are chemically inert, extremely hard and show remarkable mechanical characteristics,⁵⁵ the formation of a protective film adherent to coupled surfaces was not reputed plausible. In fact, it generally originates from the chemical interaction of nanoparticles with metallic surfaces to form a self-healing, wear-protection tribofilm.⁵⁶ Therefore, in tested conditions the friction reduction was attributed to the sliding process over nanohorns and nanoparticles dragging with the counter-body.⁵⁷ Moreover, when the counter-ball run over nanoparticles aggregates, it was possible that undeformed nanohorns rolled between surfaces, like a sort of ball-bearings.⁵⁴

For a prospective lubricant to be considered effective, it is also necessary safeguard the underlying surface from wear and damage. Fig. 6(a) and b show the results of wear scars measurements. According to ASTM G99-95a,⁵⁸ referring to the pin-on-disk tribotester, the volumetric wear loss was computed using the geometrical relations given in the standard, by measuring appropriate linear dimensions of the wear tracks after the test. Considering that there was no significant ball wear, since no worn area could be detected on balls surfaces by optical microscopy, the Disk Volume Loss (DVL) was given by eqn (4).

where R was the wear track radius, d was the wear track width and r was the ball radius. According to ASTM standard, the disk volume loss thus estimated was affected by certain not estimable errors due to variations around the wear track, accumulations of debris and plastic deformation. Thus, no error bars could be estimated for these values. The worn volume was reduced in all tests performed with nanolubricants (see Fig. 6(a)). Furthermore, mean roughness measured at the bottom of the wear scars was affected by the concentration of dispersed nano-additives inside the lubricant (Fig. 6(b)). All measured parameters first decreased up to CNH concentration of 0.2%_wt, and then raised and in some cases exceed the value obtained with raw oil. The existence of an optimal concentration, already observed in the coefficient of friction measurements, was interpreted also by Liu et al.²⁴ for carbon nanotubes as follows. When the concentration is less than this optimal value, the amount of nanoparticles is insufficient to protect from wear. On the other hand, when the concentration is too high, nanoparticles tend to agglomerate, thus reducing suspension stability and causing large size aggregate to scratch the surface under loading. Similar results were obtained also by Mosleh et al.⁵⁹ and Hernádez et al.⁶⁰ for hard ceramic nanoparticles: they found a critical value for concentration of dispersed nanoparticles to obtain best improvements of lubricity. In this work, the maximum wear protection occurred with 0.1%_wt system at 70 °C, for which a reduction of 75% of worn volume was measured. A similar decrease for disk volume loss was measured for 0.2%_wt system at 25 °C.

Fig. 6 (a) Estimated disk volume loss after wear tests; (b) mean roughness measured at the bottom of the wear tracks.

Therefore, the presence of CNHs resulted in a surprising improvement of tribological behavior of the nanolubricant compared to raw oil. Comparable behavior was observed by Dou et al. at room temperature with crumpled graphene balls in polyalphaolefin base oil.²² Lahouij et al.⁶¹ suggested rolling as a valid operation for reduction of friction and wear. Moreover, depending on the operating conditions, rolling of nanoparticles is the main action mechanism under quite low contact pressure, while at higher pressure sliding is more likely to take place. Tevet et al.⁵⁷ also obtained similar results. Therefore, CNHs were probably partially dragged along with the counter body as wear scar was formed. As more nanoparticles were able to support the applied load, the contact pressure was reduced and this allowed the possibility of rolling of other nanohorns, which were not dragged, as active part of the friction reduction mechanism. Thus, the overall mechanism could be described as a combination of the different actions previously described, enhanced by the horns presence that reduce the contact area of nanoparticles with metal surfaces.

Wear coefficients were also evaluated for each nanolubricant, to investigate the realized lubrication regime and the wear mode. The simplest classification of the types of wears is based on type I (mild), type II (severe) and type III (catastrophic). According to Ashby and Lim,⁶² at sliding velocities close to about 0.1 m s⁻¹ for metals there is no significant heating of any sort. Therefore, wear advances according a “cold” mechanism: in a non-corrosive environment, like in these experiments, wear of ductile materials is dominated by plasticity-mechanism, while brittle materials damage through brittle fracture and edge spalling. Both cases, empirically, are approximately described by Archard's law, reported in eqn (5). It is well known that the Archard's wear model correlates the wear volume with the hardness of the sliding bodies, the applied normal load and the sliding distance:⁶³

where K is the dimensionless generalized wear coefficient, V is the worn volume [m³], s is the sliding distance [m], F_N is the normal contact force [N] and H is the material penetration hardness [N m⁻²] of the surface which is worn away. The hardness of the disks was measured by means of nanoindentation and was found to be (4.8 ± 0.4) GPa.

The Archard's wear equation was used to estimate the generalized dimensionless wear coefficient K and results are reported in Table S4.† For boundary lubricated contact surfaces running under mild wear conditions, the value of the dimensional wear coefficient k (i.e. the ratio K/H in Pa⁻¹) may be of the order of 10⁻¹⁸ Pa⁻¹. If the k value is higher than 10⁻¹⁶ Pa⁻¹, sliding conditions are considered as severe wear. In realized wear tests the calculated k values for various nanolubricants were between 2.5 × 10⁻¹⁴ and 2.7 × 10⁻¹⁵ Pa⁻¹, thus revealing that severe sliding conditions were achieved in all tests.⁶⁴ In fact, significant scoring was found in wear tracks and the plastic deformation in the contact was easily observed on disks after tests. From data in Table S4† it clearly appeared that nanolubricants containing low concentrations of CNHs exhibited the lowest values of wear coefficient K, which corresponded to less severe wear phenomena. The tracks observed by optical microscopy (not reported here) evidently showed that a significant part of the track surface was scored, seized or strongly plasticized, although the lower damage in case of nanolubricants was clear. Moreover, experiments carried out at high temperature showed the presence of oxide inside the scars, which was much more visible in wear tracks obtained with raw oil.

Since nanoparticles are graphene-based, and therefore chemically stable,⁶⁵ no chemical reactions are expected between surface materials and moisture that can be absorbed by the raw oil during sliding process. Thus, no chemical deterioration of nanoparticles was foreseen during tribological tests. Scanning electron microscopy was carried out on wear scars to assess the condition of the worn surface and verify the action mechanism of dispersed additives. Fig. 7 shows the surface of wear track obtained at room temperature with a nanolubricant containing 0.2%_wt CNHs. It can be noticed that small aggregates of CNHs randomly appeared at the wear scar bottom. X-EDS analyses carried out on aggregates distributed on wear track surface confirmed that visible particles having irregular shape and enhanced size were mainly constituted of iron and chromium (see Fig. S5†). Thus, they were considered as debris, while particles having round shape were identified as unbroken and undeformed nanoparticles. Such observations were similar to what observed by Dou et al.²² and suggested that the strain-hardening property of these nanostructures prevents deformation and carbon film formation. To confirm the CNHs influence on friction coefficient, Stribeck tests were carried out for the PAG/CNH 0.1%_wt system (Fig. 8). The Stribeck curve describes the coefficient of friction variation with lubrication regimes. It clearly appeared that the nanolubricant brought important benefits to the tribological behavior of raw oil in all lubrication regimes. The coefficient of friction was lower for nanolubricant while the ECR value appeared higher. It further confirmed the advantage of using nanohorns, because the electric insulation was enhanced in tests performed in presence of CNHs, thus restricting the direct metal–metal contact, which was responsible of wear and failure.

Fig. 7 SEM picture of nanohorns deposited on wear scar surface (circle highlights some CNHs).

Fig. 8 Stribeck tests performed on PAG/CNH 0.1%_wt system at (a) 25 °C and (b) 70 °C.

Long lasting wear tests allowed reaching the catastrophic failure point of disk, revealed as a sudden and abrupt increase of μ value and acoustic emission signal. Friction coefficient mean value appeared to be similar between the nanolubricant and the base fluid as reported in Fig. 9(a).

Fig. 9 (a) Coefficient of friction and (b) acoustic emission signals of long-lasting wear tests for PAG and on PAG + CNH 0.1%_wt.

The breakage point for the disk material was reached after about 68 400 cycles (about 1500 m) with the raw oil, while the test carried out in the presence of nanolubricant lasted for over the initially planned 100 000 cycles, without a catastrophic damage of the sliding materials. Failure occurred at almost 125 000 cycles (about 2700 m). Acoustic emission signal confirmed the failure of the tribological contact in the test performed with raw oil (see Fig. 9(b)). These tests further confirmed the ability of dispersed nanoparticles in preventing and limiting detrimental wear phenomena.

Conclusions

Graphene based nanostructures were homogeneously dispersed in PAG oil for applications in refrigeration/air conditioning field. Thermal, rheological and tribological properties of nanolubricants were investigated at 25 and 70 °C. Tribological properties of nanolubricants were evaluated in boundary lubrication regime and in severe plastically dominated wear conditions. Very promising results were obtained for nanolubricants containing 0.1%_wt and 0.2%_wt CNHs, with significant reduction in coefficient of friction at both temperatures (about 18% at 25 °C and 7% at 70 °C for lubricant containing 0.1%_wt of CNHs). The friction reduction was attributed to the reduced contact area provided by the peculiar morphology of nanohorns and by rolling/sliding mechanisms. A threshold value of 0.2%_wt for nano-additives concentration was identified, over which tribological behavior showed less improvements. The wear track analyses (a remarkable 75% of reduction in volume of wear scar) and Stribeck curves confirmed such important benefits of carbon nanostructures.

Nanolubricants containing 0.1%_wt and 0.2%_wt CNHs showed a negligible change in viscosity. A slight increase in thermal conductivity was observed, thus suggesting a benefit in heat transfer properties in laminar flow conditions. A decrease in the coefficient of friction could correspond to an increase in compressor efficiency and a reduction of power required for the same number of revolutions. At the same time, the increased anti-wear properties could ensure a longer durability of devices.

Acknowledgements

This work has been funded by the Italian National Research Council – Italian Ministry of Economic Development agreement “Ricerca di Sistema Elettrico Nazionale”.

Notes and references

1. C. Barthel, L. Tholen, T. Götz and A. Durand, Halving worldwide electricity demand for residential cold appliances through appropriate policy packages, ed. T. Lindström, European Council for an Energy Efficient Economy, Stockholm, 2013, pp. 1841–1850.
2. T. Jia, R. Wang and R. Xu, Int. J. Refrig., 2014, 45, 120.
3. S. B. Riffat, C. F. Afonso, A. C. Oliveira and D. A. Reay, Appl. Therm. Eng., 1997, 17(1), 33.
4. European Parliament and Council, Directive 2006/40/EC relating to emissions from air conditioning systems in motor vehicles and amending Council Directive 70/156/EEC; Official J. European Union L161, 2006, 12–18.
5. I. Sarbu, Int. J. Refrig., 2014, 46, 123.
6. P. Bansal, Appl. Therm. Eng., 2012, 41, 18.
7. Y. Ma, Z. Liu and H. Tian, Energy, 2013, 55, 156.
8. Y. Z. Lee and S. D. Oh, Wear, 2003, 255, 1168.
9. H. J. Kim, Int. J. Refrig., 2005, 28(4), 498.
10. H. C. Sung, Wear, 1998, 221, 77.
11. N. P. Garland and M. Hadfield, Mater. Des., 2005, 26(7), 578.
12. G. Bianchi and R. Cipollone, Appl. Energy, 2015, 142, 95.
13. K. M. Tan and K. T. Ooi, Appl. Therm. Eng., 2011, 31(8–9), 1519.
14. M. S. Liu, M. C. Lin, I. T. Huang and C. C. Wang, Chem. Eng. Technol., 2006, 29(1), 72.
15. A. Celen, A. Çebi, M. Aktas, O. Mahian, A. S. Dalkilic and S. Wongwises, Int. J. Refrig., 2014, 44, 125.
16. B. Li, X. Wang, W. Liu and Q. Xue, Tribol. Lett., 2006, 22, 79.
17. D. Maharaj and B. Bhushan, Tribol. Lett., 2013, 49(2), 323.
18. L. Joly-Pottuz and N. Ohmae, Nanolubricants, ed. J. M. Martin and N. Ohmae, John Wiley & Sons Ltd., Chichester, England, 2008, pp. 93–147.
19. J. Taha-Tijerina, L. Peña-Paras, T. N. Narayanan, L. Garza, C. Lapray, J. Gonzalez, E. Palacios, D. Molina, A. García, D. Maldonado and P. M. Ajayan, Wear, 2013, 302, 1241.
20. J. Ota, S. K. Hait, M. I. S. Sastry and S. S. V. Ramakumar, RSC Adv., 2015, 5, 53326.
21. X. Fan, L. Wang and W. Li, Tribol. Lett., 2015, 58, 12.
22. X. Dou, A. R. Koltonow, X. He, H. Dong Jang, Q. Wang, Y. Chung and J. Huang, Proc. Natl. Acad. Sci. U. S. A., 2016, 113, 1528.
23. L. Joly-Pottuz, E. E. Bucholz, N. Matsumoto, S. R. Phillpot, S. B. Sinnott, N. Ohmae and J. M. Martin, Tribol. Lett., 2010, 37, 75.
24. L. Liu, Z. Fang, A. Gu and Z. Guo, Tribol. Lett., 2011, 42, 59.
25. J. Miyawaki, M. Yudasaka, T. Azami, Y. Kubo and S. Iijima, ACS Nano, 2008, 2, 213.
26. E. Sani, S. Barison, C. Pagura, L. Mercatelli, P. Sansoni, D. Fontani, D. Jafrancesco and F. Francini, Opt. Express, 2010, 18(5), 5179.
27. S. Iijima, M. Yudasaka, R. Yamada, S. Bandow, K. Suenaga, F. Kokai and K. Takahashi, Chem. Phys. Lett., 1999, 309(3–4), 165.
28. N. Karousis, I. Suarez-artinez, C. P. Ewels and N. Tagmatarchis, Chem. Rev., 2016, 116, 4850.
29. A. Tanaka, K. Umeda, M. Yudasaka, M. Suzuki, T. Ohana, M. Yumura and S. Iijima, Tribol. Lett., 2005, 19(2), 135.
30. V. Zin, F. Agresti, S. Barison, L. Colla, E. Mercadelli, M. Fabrizio and C. Pagura, Tribol. Lett., 2014, 55, 45.
31. C. Dang, K. Hoshika, E. Hihara and M. Kaneko, in Proceedings of the International Refrigeration and Air Conditioning Conference, Purdue, IN, USA, 2010, Paper 1143.
32. B. Y. C. So and E. E. Klaus, ASLE Trans., 1980, 23, 409.
33. M. Schiavon, Patent EP1428794 A3, 2006.
34. W. Rasband, ImageJ 1.46r software, National Institute of Health, U.S.A.
35. F. Agresti, A. Ferrario, S. Boldrini, A. Miozzo, F. Montagner, S. Barison, C. Pagura and M. Fabrizio, Thermochim. Acta, 2015, 619, 48.
36. S. M. Hsu and R. S. Gates, Tribol. Int., 2005, 38(3), 305.
37. H. Chen, Y. Ding, A. Lapkin and X. Fan, J. Nanopart. Res., 2009, 11, 1513.
38. E. I. Ettefaghi, A. Rashidi, H. Ahmadi, S. S. Mohtasebi and M. Pourkhalil, Int. Commun. Heat Mass Transfer, 2013, 48, 178.
39. R. K. Sabareesh, N. Gobinath, V. Sajith, S. Das and C. B. Sobhan, Int. J. Refrig., 2012, 35(7), 1989.
40. J. C. Maxwell, A treatise on electricity and magnetism, ed. William Garnett, Clarendon Press, Oxford, 1881, vol. 1.
41. Z. Hashin and S. Shtrikman, J. Mech. Phys. Solids, 1963, 11(2), 127.
42. R. Prasher, W. Evans, P. Meakin, J. Fish, P. Phelan and P. Keblinski, Appl. Phys. Lett., 2006, 89(14), 143119.
43. J. Philip, P. Shima and B. Raj, Appl. Phys. Lett., 2007, 91, 203108.
44. N. Karthikeyan, J. Philip and B. Raj, Mater. Chem. Phys., 2008, 109(1), 50.
45. F. Agresti, S. Barison, S. Battiston, C. Pagura, L. Colla, L. Fedele and M. Fabrizio, Nanotechnology, 2013, 24, 365601.
46. R. Prasher, D. Song, J. Wang and P. Phelan, Appl. Phys. Lett., 2006, 89, 133108.
47. R. E. Simons, Electronics Cooling, 2006, 12, 2.
48. B. Bhushan, Tribology: Friction, Wear and Lubrication, in The Engineering Handbook, CRC Press, Florida, USA, 2000.
49. B. J. Hamrock and D. Dowson, in Fluid-Film Lubrication, NASA-TM-81700, 1983.
50. H. A. Spikes, Lubr. Sci., 1997, 9(3), 221.
51. B. J. Hamrock, S. R. Schmid and B. O. Jacobson, Fundamentals of Fluid Film Lubrication, CRC Press, 2004.
52. K. C. Ludema, Friction, Wear, Lubrication: A Textbook in Tribology, CRC Press, 1996.
53. Y. Y. Wu, W. C. Tsui and T. C. Liu, Wear, 2007, 262(7–8), 819.
54. Y. Hwang, C. Lee, Y. Choi, S. Cheong, D. Kim, K. Lee, J. Lee and S. Kim, J. Mech. Sci. Technol., 2011, 25, 2853.
55. D. Kumar, V. Verma, H. S. Bhatti and K. Dharamvir, J. Nanomater., 2011, 1.
56. J. Qu, H. M. Meyer III, Z. B. Cai, C. Ma and H. Luo, Wear, 2015, 332–333, 1273.
57. O. Tevet, P. Von-Huth, R. Popovitz-Biro, R. Rosentsveig, H. D. Wagner and R. Tenne, Proc. Natl. Acad. Sci. U. S. A., 2011, 108, 19901.
58. ASTM G99–95a(2000)e1, Standard Test Method for Wear Testing with a Pin-on-Disk Apparatus, ASTM International, West Conshohocken, PA, 2000.
59. M. Mosleh, N. D. Atnafu, J. H. Belk and O. M. Nobles, Wear, 2009, 267(5–8), 1220.
60. A. Hernández Battez, R. González, J. L. Viesca, J. E. Fernández, J. M. Díaz Fernández, A. Machado, R. Chou and J. Riba, Wear, 2008, 265, 422.
61. I. Lahouij, F. Dassenoy, L. de Knoop, J. M. Martin and B. Vacher, Tribol. Lett., 2011, 42(2), 133.
62. M. F. Ashby and S. C. Lim, Acta Metall., 1987, 35, 1.
63. J. F. Archard, J. Appl. Phys., 1953, 24, 981.
64. S. Andersson, Wear Simulation, Advanced Knowledge Application in Practice, ed. Igor Fuerstner, 2010.
65. K. S. Kim, H. J. Lee, C. Lee, S. K. Lee, H. Jang, J. H. Ahn, J. H. Kim and H. J. Lee, ACS Nano, 2011, 5(6), 5107.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra12029f

---