The influence of covalent and non-covalent functionalization of GNP based nanofluids on its thermophysical, rheological and suspension stability properties

Covalent functionalization (CF-GNPs) and non-covalent functionalization (NCF-GNPs) approaches were applied to prepare graphene nanoplatelets (GNPs). The impact of using four surfactants (SDS, CTAB, Tween-80, and Triton X-100) was studied with four test times (15, 30, 60, and 90 min) and four weight concentrations. The stable thermal conductivity and viscosity were measured as a function of temperature. Fourier transform infrared spectroscopy (FTIR), thermo-gravimetric analysis (TGA), X-ray diffraction (XRD) and Raman spectroscopy verified the fundamental efficient and stable CF. Several techniques, such as dispersion of particle size, FESEM, FETEM, EDX, zeta potential, and UV-vis spectrophotometry, were employed to characterize both the dispersion stability and morphology of functionalized materials. At ultrasonic test time, the highest stability of nanofluids was achieved at 60 min. As a result, the thermal conductivity displayed by CF-GNPs was higher than NCF-GNPs and distilled water. In conclusion, the improvement in thermal conductivity and stability displayed by CF-GNPs was higher than those of NCF-GNPs, while the lowest viscosity was 8% higher than distilled water, and the best thermal conductivity improvement was recorded at 29.2%.


Introduction
Nanouid refers to the stable and homogenous suspension exhibited by nanoparticles (NPs) within conventional working uids (e.g., ethylene glycol, oil, and water). 1,2 Prior studies reported that the addition of a small weight concentration of NPs could improve the efficiency of heat transfer and thermal conductivity in the base uid, and hence the suitability for wide range applications, such as in solar collectors, cooling appliances, and heat exchangers. 3,4 Nanoparticles (NPs) are made of either metal oxide (e.g., Al 2 O 3 , SiO 2 , CuO, and TiO 2 ) 5-7 or carbonbased particles (e.g., carbon nanotube (CNT)), 8,9 graphene oxide (GO), 10 or graphene nanoplatelets (GNPs). 11,12 Numerous types of NPs have been proposed in the last decades to generate nanouids, wherein CuO, Al 2 O 3 , ZnO, and SiO 2 are more commonly employed than nanomaterials based on metal oxide. [13][14][15] Both graphene and carbon-based NPs (e.g., CNT) 16 have been assessed empirically. 12,[17][18][19][20] Simply put, nanouids have garnered much attention due to their chemical and physical properties, as well as their nanometre size. More importantly, nanouids have improved thermal conductivity, hence their suitability to serve as heat-exchanging uids. Upon inclusion of a NP base uid, some empirical studies have also improved the convective heat transfer, in comparison to the pure base uid.
Choi 1 discovered an improvement in thermal conductivity aer including copper NPs into the base uids. Meanwhile, Zeinali Heris et al., 21 assessed the convective heat transfer coefficient of CuO NPs, wherein the outputs were compared with Al 2 O 3 -water nanouid. The reported outcomes displayed a higher convective heat transfer coefficient for Al 2 O 3 -water nanouids than CuO-water nanouids. Here, the integral role of NPs in thermal conductivity is highlighted for nanouid heat transfer. A maximum decrease of 24% was noted for thermal resistance in water-based titanium dioxide and gold nano-uids. 22 Shanbedi et al., 23 determined the impact of multiwalled carbon nanotubes (MWCNT) on the effectiveness of two-phase closed thermosyphon (TPCT). As a result, 11% improvement was recorded for thermal efficiency at 90 kW in TPCT with functionalized MWCNT. Thermo-physical properties, such as thermal conductivity of NPs, have a signicant function in heat transfer appliances. 24,25 The thermal conductivity of GNPs exceeded that displayed by other carbon allotropes, such as diamond, MWCNTs, and SWCNT. 26 The GNPs have vast applications due to its exceptional properties, particularly in the scientic domain, for instance in making batteries and sensors. 27 Nevertheless, most applications are from being realized due to the weak interaction GNPs has with other materials. Hence, to enhance this interaction, covalent (amino acids) and non-covalent (NC) functionalization have been proposed as the solution in other studies. 28 Covalent and non-covalent functionalization can improve the aspect of the dispersibility of GNPs in organic/aqueous solvents. The NCF of carbon nanostructures may be executed by using a range of surfactants. [28][29][30] Four surfactants can be applied to enhance carbon nanostructure dispersibility in an aqueous medium, namely Tween-80 (Tw-80), sodium dodecyl sulphonate (SDS), Triton X-100 (Tx-100), and Cetyltrimethylammonium bromide (CTAB). The benzene function in Tx-100 generates powerful p-p interaction with the surface of carbon nanostructures. Besides, Tw-80 possesses better dispersibility than Tx-100, mainly due to the hindrance of steric in tip chains of Tx-100, thus the low concentration in Triton on the surface of carbon nanostructures. 31 Despite the exceptional dispersion of carbon nanostructures in Tx-100, when compared to CTAB and SDS, numerous issues emerge due to increased mixture viscosity, such as drop in pressure in thermal appliances. 23 Hence, Tw-80 was chosen in this study to synthesise NCF nanouid. The CF is composed of hydrophilic or hydrophobic groups in light of high-energy characteristics, for example, GNPs edges.
In this study, GNPs were functionalized using two approaches, namely covalent and non-covalent reactions, for comparison purposes. The experiment included rapid NCF of GNPs-surfactant and CF with carboxyl groups. Next, characterization of samples and performance of thermal were assessed. The effects of CF and NCF on thermal conductivity were investigated, while the aspect of viscosity was assessed by considering a range of concentrations and temperatures. The signicant increment of the viscosity of suspension vitiates in the NCF groups improved the feature of heat transfer in GNPs nanouids. The outputs revealed the more improved thermophysical property in covalent nanouids (CF-GNPs) than those of water and non-covalent nanouids (GNPs-surfactant).

Nanoparticles and chemicals
GNPs with the purity of 98%, the maximum particle diameter of 2 mm and specic surface area 750 m 2 g À1 purchased from, XG Sciences, Lansing, MI, USA, were used in this study. Sulfuric acid 95-97% (H 2 SO 4 ) and nitric acid 65% analytical reagent grade (HNO 3 ) were purchased from Sigma-Aldrich Co., Selangor, Malaysia, were used as the functionalization media. All chemicals were used as received without further purication. Surfactants used were SDS, CTAB, Tween-80 and Tx-100 which are purchased from Sigma-Aldrich Co., Selangor, Malaysia.

Covalent functionalization method
Meanwhile, GNPs are obviously hydrophobic and it cannot be dispersed in any solvent which is polar like distilled water. Proper ways to make GNPs hydrophilic do functionalization by acid treatment. This functionalization procedure helps to present functional groups such as carboxyl and hydroxyl groups on the surface of GNPs. The pristine GNPs was transferred into H 2 SO 4 and the solution was shaking, next, the container was moved to an ice bath and the nitric acid drops were added to the solution mixture. The ratio of sulfuric to nitric (3 : 1) was used due to its preference as an acid treatment 32,33 then the solution was stirred for 30 min at room temperature and then was under bath-ultrasonication for 3 h. Aer 3 h, of the probe sonication process, the sample mixture was reuxed for 30 min at room temperature with constant stirring. Aer that GNPs were washed a number of times with distilled water thoroughly then centrifuged at 6000 rpm for 15 min to remove excess acid and then dried under drying oven for 24 h at 80 C. Rich samples of (0.02, 0.05, 0.08, and, 0.1 wt%) were prepared with the addition of a known amount of functionalized GNPs into distilled water. Fig. 1 presented the schematic diagram of the functionalization and CF-GNPs preparation process of the nanouids.

Non-covalent functionalization method
Ultrasonication probe sonication was used to dispersed GNPs in distilled water (Sonics Vibra-Cell, VC 750, Sonics & Materials, Inc., USA) having an output power of (750 W) and a power source of (20 kHz) frequency. The lack of a surfactant or functionalization cannot be stably dispersed carbon-based nanoparticles in water, ever since they have a hydrophobic surface. 34 Accordingly, NCF-GNPs were achieved using different additives surfactants; SDS, CTAB, Tween-80 and Triton X-100. Four different ultrasonication times of 15, 30, 60 and 90 min were used. Weight concentrations of GNPs were in the range of 0.02, 0.05, 0.08, and, 0.1 wt% for the preparation of samples. Samples compositions are presented in Table 1 and Fig. 2. The NCF-GNPs weight concentration ratio (1 : 1) and 60 min GNPs were reserved constant during the preparation of samples. During the probe sonication process caused a considerable increase in heat generated in the temperature and evaporation in water of the sample with the subsequent alteration in the weight concentration of the sample. Therefore, ice bath used to regularly the temperature at a suitable degree. For long periods and stable suspension without sedimentation the prepared nano-uid would be an agglomerate-free. The weight percentage of GNPs was calculated based on eqn (1). 35 Weight percentage of GNPs ¼ weight of GNPs weight of base fluid Â 100% (1)

Measurement devices
2.4.1. Evaluation of stability. Zeta potential and ultravioletvisible spectrophotometry (UV-vis) analysis were conducted to assess the stability of GNPs dispersions. The measurement of light absorbance of a suspension by UV-vis spectroscopy can be used to make available a quantiable characterization of the stability. The use of UV-vis is Lambda operating in the range of 200-800 nm wavelengths (UV-800/900, Lambda Company, U.S.A.). Light absorbance was measured by special quartz cuvettes suitable for the UV region since all samples were at certain time intervals for an extra 30 days. A 1 : 20 ratio diluted in distilled water to allow proper light transmission for all samples. Zetasizer Nano ZS (Anton Paar, GmbH Ltd, Malvern, UK) measured zeta potential using the prepared nanouids principle of Electrophoretic Light Scattering (ELS). The degree of revulsion in the measurement of zeta potential between close particles shows the same load in nanouid dispersal. 36 2.4.2. Morphology and elemental analysis. Using Fourier transform infrared (FTIR) spectroscopy, Raman spectroscopy, Field Emission Transmission Electron Microscope (FETEM, HT 7700, Hitachi) and Field Emission Scanning Electron Microscopy (FESEM, SU8000, Hitachi), the main structure of CF-GNPs nanouids was analyzed. The evaluation of the FTIR spectra within a wavenumber range of 400-4000 cm À1 for FTIR spectroscopy. Phase compositions were determined by using an Xray diffractometer (XRD, EMPYREAN, PANALYTICAL) with Cu-Ka radiation over a 2q range from 20 to 80 . The "PANalytical X'Pert HighScore" soware was employed to compare the XRD proles with the standards compiled by the Joint Committee on Powder Diffraction and Standards (JCPDS).

Measurement of thermophysical properties.
To measure the steady-shear rheological properties of water and water-based GNPs nanouids, Anton Paar rotational rheometer (Model Physica MCR 301, Anton Paar GmbH Ltd, Malvern, UK) was used. The temperatures in the ranges of (20-60 C) at a shear rates of 200 s À1 were used to achieve the tests for all samples. Thermal conductivity was determined by the thermal property's analyzer device KD2 Pro (Decagon Devices, Inc., USA), with an accuracy of approximately 5%. A 1.4 kW water bath WNB22 Memmert (Germany) was used to keep the samples at the preferred temperature during measurements and 0.1 C accuracy. The rotational rheometer consists of a moving cylindrical plate and a stationary cylindrical surface which are parallel with a small gap between them.

Characterization of CF-GNPs nanoparticles
The FTIR spectra of pristine-GNPs and CF-GNPs are illustrated in Fig. 3(a). Table 2 summarises the assigned bonds and IR peaks derived from the CF-GNPs FTIR spectra.   Paper wavenumbers between 3000 and 3500 cm À1 ; attributable to -OH chains from acid treatment. 37,38 The CF-GNPs peaks indicate the success of the CF procedure. The Raman analysis had been performed to determine the success of the covalent approach. In this analysis, G-band displays the existence of sp 2 carbon, D-band shows the defects of chemical construction. 39 Fig. 3(b) illustrates the Raman spectra of P-GNPs and CF-GNPs. Ratios I D /I G of P-GNPs and CF-GNPs are 0.843 and 0.855, respectively (see Table 3). Ratio I D /I G for 3 h was higher for CF-GNPs than P-GNPs because of the existence of covalent new bonds on carbon sheets deriving from CF-GNPs. Fig. 3(c) illustrates the TGA curve of P-GNPs, as well as 1 and 3 h CF-GNPs. The initial loss of weight reected in pristine GNPs and CF-GNPs were 7% and 10%, respectively, at 0-100 C, attributable to adsorbed moisture. Slight loss of weight was noted at 60-800 C range for P-GNPs; attributable to the pyrolysis of primary carbon-based structures. 33 On the contrary, the second slight weight loss by 16% was recorded for 100-500 C range for CF-GNPs, mainly because of the formation of hydroxyl and carboxyl groups in oxidation step as unstable organic moieties, as well as degradation in the functionality of acid treatment. 38 The third slight weight loss by 24% was recorded for 500-800 C range for FC-GNPs, mainly because of the degradation of the graphitic structures in air (see Table 4).
The energy dispersive X-ray (XRD) test was conducted to nd the purity and degree of oxidization of the material. Fig. 3(d) shows the XRD patterns of functionalized GNPs. Two clear distinct peaks of plane (002) at around 25 and the plane (004) at 44.4 represent the structure of GNPs. The results show that the CF-GNPs did not affect the crystalline structure of GNPs which conforms with the results obtained by Yarmand et al. 40 The morphologies of pristine GNPs and CF-GNPs via FETEM and FESEM are portrayed in Fig. 4. Based on Fig. 4(a), pristine GNPs was composed of dual sheets with intact edges, transparent structures, and smooth surfaces. Aer functionalization, acid treatment, as well as carboxyl group on the edges and surfaces of GNPs, gave slightly blur effect on the sheets with wrinkles and crumples (see Fig. 4(b) and (c)). Changes in morphology and functional groups analysis point out the successful reaction between acid molecules and GNP-COOH. Defective folded akes and rough edges (see Fig. 4(d)) signify the success of the CF method. FESEM images also conrm that the functionalization of GNPs results in the broken sheets of GNPs (see Fig. 4(e) and (g)). The EDX measurements portray two components present in CF-GNPs; carbon and oxygen. Despite the best dispersion stability offered by the small sheets, the thermal conductivity may be reduced slightly. 41 On the contrary, the functionalization treatment of nanouids explains the upper surface roughness found in P-GNPs. Damage noted partially in graphitic carbon implies higher roughness, in effect of basic carboxylation. Lines observed in FETEM images refer to wrinkles on GNPs surface as a result of inherent stability in 2D structures. The functionalization method increases in these lines during sonication, which could be due to higher than prior wrinkling or waviness. Fig. 5(a)-(c) illustrates the UV-vis spectra of water-based CF-GNPs nanouids prepared at varied weight concentrations (0.02, 0.05, 0.08, and 0.1) wt%. A single peak was noted within wavelength ranging between 270 and 290 nm. The sharp absorption peak observed at $275 nm is associated with p / p * transition of the C]C bond of GNPs. 42 Increased peak intensity due to increment in particle concentration adheres to the Beer-Lambert law. 43 The sediment generated in the nanouids was insignicant for particle concentrations aer 30 days, which resulted in 91, 92, 92.8, and 94% with 0.02, 0.05, 0.08, and 0.1 wt% of CF-GNPs, respectively. The most suitable factors that generate high stability aqueous GNPs dispersion were determined, in terms of surfactant types (SDS, CTAB, Tween-80, and Triton X-100), and various ultrasonication probe times (15,30,60, and 90 min) with varied weight concentrations utilised to disperse GNPs in distilled water. 43 The P-GNPs and varied NCF-GNPs dispersed in distilled water by UV-vis spectrum are displayed in Fig. 5(d)-(i). All the samples were placed in the wavelength that ranged from 270 to 290 nm, the peak absorption was clearly due to the presence of GNPs. A decrease in absorbance was noted for all samples, as shown in Fig. 5. 42 Based on Fig. 5(a), the P-GNPs dispersion cannot continue in height stability for the range of ultrasonication times assessed, wherein NCF-GNPs and CF-GNPs functionalization are vital to obtaining high stability GNPs nanouid dispersions. Hence, a 60 min ultrasonication probe time emerged as the most effective time to prepare nanouids in this study. This 60 min ultrasonication probe  time exhibited higher stability for Tw-80-GNPs sample than the other three samples for a similar amount of days, as illustrated in Fig. 5(f)-(i). Fig. 5(e) displays that SDS-GNPs at the ratio of (1 : 1) gave better stability, when compared to other tested ratios. Fig. 5(j) shows the long-term stability of CF-GNPs via UVvis spectroscopy and the quantitative analysis of the dispersion Paper state for varied weight concentrations. Over time, the CF-GNPs decreased insignicantly with relative to concentration. As a result, the maximum weight concentration of 0.1 wt% and the maximum sediment of approximately 6% gave the appropriate dispersibility of CF-GNPs. In contrast, the CF-GNPs show remarkable colloidal stability in distilled water aer 24 h, as shown in Fig. 5(k). Another evaluation, known as zeta potential, was applied to assess sample stability. 43 Zeta potential can be related in a direct manner with the dispersed nanouid stability. 44,45 The rising repulsion between particles with similar charges determined dispersion stability. Particles with high-surface charges can lead to the low occurrence of agglomeration. 46 Values of zeta potential for the nanouids ranging between <À30 mV and >+30 mV were linked with physically-stable colloids. 47 Fig. 6 and  7 illustrate the values of zeta potential and particle size distribution, for both pristine GNPs and CF-GNPs samples aer one and twenty days. Tables 5 and 6 tabulate the retrieved outcomes. Aer 20 days, CF-GNPs had a high value (À37.1 mV) and slightly minimized value of zeta potential (À35.1 mV). The pristine GNPs portrayed shallow values of zeta potential at À21.7 mV and À18.4 mV for one and twenty days, respectively. Hence, it is clear that CF-GNPs gave exceptional stability outcomes, in comparison to pristine GNPs.
The average particle size for pristine GNPs was more prominent than that of CF-GNPs, while the dispersion of particle size for pristine GNPs was 228.4 nm and 292.5 nm aer one and twenty days, respectively. The results were higher than the CF-GNPs values, whereby a dispersion of particle size had been 184.5 nm and 280.7 nm aer one and twenty days, respectively. The stability of nanouids via dispersal of CF-GNP generated high stability suspension, in comparison to pristine GNPs in isolation. 48,49 3.2. Foaming test Fig. 8 illustrates an image of six nanouid samples prepared via shaking for 16 s to assess the generation of foam. The foam was generated above the nanouid with the inclusion of surfactants (CTAB, SDS, Tw-80, and Tx-100), except for CF-GNPs, which did not form foam. This reects an exceptional setting for heat transfer and uid ow applications, as foams deteriorate the efficiency of heat transfer and interrupt uid ow. 50 3.3. Thermo-physical properties analysis Fig. 9(a)-(c) presents the thermal conductivity of nanouids at varying temperatures with CF-GNPs and NCF-GNPs for the following weight concentrations: (0.02, 0.05, 0.08, and 0.1) wt%. To validate the reliability of the thermal conductivity measurements for distilled water, KD2 Pro was used in the temperature range of 20-60 C. The obtained data are in good agreement with the National Institute of Standards and Technology (NIST) 51 as seen in Fig. 9(a). The average error in the thermal conductivity was found to be AE1.172%, indicating that the KD2 Pro thermal property analyzer is reliable to be used in the thermal conductivity measurements of the samples. Based on Fig. 9(b) and (c), the thermal conductivity of CF-GNPs nano-uids was higher than that of NCF-GNPs nanouids and distilled water. Increment in weight concentrations of GNPs for water-based nanouids increased its thermal conductivity. Improving thermal conductivity is a dominant function of temperature, which is attributable to Brownian motion of suspended particles. 52 Carbon nanostructures-based nanouid that comprises of GNPs, along with Brownian motion of GNPs and chemical treatment functionalization; the surface nanolayers tend to dominate the energy heat transfer in the nanouids. With the presence of covalent functional groups, a higher effective heat transfer area is attributable to the higher thermal conductivity of CF-GNPs. Based on Fig. 9(b) and (c), thermal conductivity increased with increment in temperature. The agglomeration between the NPs could easily break down at higher temperatures, which may occur due to the uniform    dispersion of GNPs in water. This reects the best conduction of heat in the uid ow, which generated layers around the GNPs of liquid molecules, thus the interface area increases the local ordering of liquid layer. It is reasonable to record higher thermal conductivity in liquid layer than bulk liquid at the interface. 53 Increased thermal conductivity of GNPs nanouids appears to be an essential aspect in nanolayer. The formation of layers surrounding the nanostructures on the surface of GNPs  can be increased when effective heat transfer region is decreased, mainly because the non-covalent groups warp around GNPs akes and reduce the local liquid layer by absorbing non-covalent groups at the interface area of GNPs. Oxygen-containing functional groups (COOH) on the surface of GNPs, as well as the formation of more hydrophilic phase, could lead to new homogeneous dispersion in base uids. 53 Hence, to enhance the aspect of dispersibility upon utilizing surfactant of carbon nanostructures, the surfactants tend to wrap around them, thus lowering micro convection that hinders the nanolayer from improvement. 54 More recent works that compared the carbonbased nanouids are listed in Table 7. Based on Table 7, improve thermal conductivity offers signicant evidence that CF-GNPs nanouids, in comparison to other samples, had higher concentrations of additives or NPs. It is highlighted here that medium-temperature applications can achieve suitable thermal conductivity by using low-weight GNPs concentration. The viscosity in characterizing the suitability of nanouids is a signicant parameter for heat transfer applications. The viscosity measurements obtained using the rotational rheometer have shown good agreement with NIST standards as seen in Fig. 10(a). 51 The average error was found to be AE3.25% indicating the reliability of the rotational rheometer to be used for measuring the viscosity of the samples. Fig. 10(b) and (c),  displays the CF-GNPs nanouids for the viscosity of distilled water as a function of weight concentration and at 20-60 C range with xed shear rate of 200 s À1 . The results were obtained with increased weight concentration of nanouids, primarily because of the increased concentration for the viscosity of nanouids demands a direct inuence on the uid internal shear stress. 55 Increment in temperature declined the viscosity, due to the waning of inter-molecular and inter-particle bond forces. 55 At 0.1 w% at 60 C, the viscosity increased by approximately 27%, when compared to the viscosity of base uid at the same condition. As shown in Fig. 10(c), at the concentration of 0.1 wt%, the NCF-GNPs nanouid had a higher viscosity than that of CF-GNPs nanouid at similar concentrations, mainly due to the inuence of surfactant-GNPs on the viscosity of the nanouids. This is the main reason for the increase of viscosity in the surfactants (SDS, CTAB, Tw-80, and Tx-100) of nanouids.

Conclusion
The current study focused on the stability behavior of functionalized GNPs water-based nanouids for different types of surfactants (SDS, CTAB, Tw-80, and Tx-100) and different weight concentrations of (0.02, 0.05, 0.08, and, 0.1) wt%. The prepared samples were characterized using various measurement tools such as FESEM, FETEM, FTIR, Raman, TGA, EDX, XRD, UV-vis, zeta potential, and particle size distribution. The thermophysical properties were tested at a different temperature span to examine the heat transfer enhancements. The following ndings would be drawn: (1) The functionalization was conrmed by the appearance of peaks at 3410 (3000-2800) and 1650 cm À1 for the -OH, C-H stretching vibrations and C]O bending vibration, respectively.
(2) An increase in the I D /I G ratio reveals that the number of sp 2 hybridized carbons changed to sp 3 hybridization carbons because of the covalent functionalization.
(3) The mass change of the functionalized samples with the shi in temperature observed at 24% while for pristine was only 14% for 500-800 C.
(4) The morphology and surface deterioration of the functionalized samples were compared to the pristine to study the structure of GNPs.
(5) The sonication time showed an impact on the absorption 0.1 wt% of CF-GNPs provided higher absorption compared to the other concentrations. The ration of (1 : 1) in the noncovalent functionalization presented a higher absorption rate, among different ratios the maximum sediment of about 6%, which conrmed the appropriate dispersibility of CF-GNPs.
(6) Particle size distributions and zeta potential were the standard procedures for characterization of dispersion stability of the colloids by measuring the magnitude of electrostatic interaction between colloidal particles. (7) To discuss the foaming test, six samples were tested and CF-GNPs were with no foam which is required for the heat transfer application to prevent the blockage inside the piping lines.
(8) The present study showed a 29.2% enhancement of the thermal conductivity at 0.1 wt%. At 0.1% weight concentration with 60 C of nanouid the viscosity increases of about 27% compared to the viscosity of the base uid.