Introduction

Energy transport is continuously facing challenges in various industries for cooling and heating applications. Energy transport with conventional heat transfer fluids are widely used in many industries and different thermal systems. However, the thermal conductivity of conventional heat transfer fluids is limited resulting in poor heat transfer performance. The fluids having high thermal conductivity are needed to enhance the energy transfer rate of compact cooling system. Hence, there is a need for new and innovative heat transfer fluids to achieve high-performance cooling.

Over the past few decades, different nanomaterials are seeded with conventional heat transfer fluids in order to increase the thermal conductivity and energy transfer rate of such fluids. The nanomaterials are available with different shapes such as spherical, cylindrical and platelets. The spherical-shaped metallic and metallic oxide nanomaterials were used for the preparation of nanofluids with different basefluids, and enhancement of thermal conductivity has been summarized in the literature [1–4]. There is no promising enhancement in thermal conductivity with spherical nanoinclusions due to the high interfacial thermal resistance between the nanoparticles and the surrounding fluids [5, 6].

To avoid this issue, recently carbon-based cylindrical and platelet-shaped nanostructures are used to prepare high-conductivity nanofluids [7–11]. Experimental study on thermal conductivity with carbon-based nanostructures such as carbon nanotubes (CNTs) and graphene nanoplatelets (GnP) have been performed by several researchers [12–23]. Sabiha et al. [17] performed the thermal conductivity of SWCNT-H2O nanofluids experimentally, and highest enhancement of thermal conductivity is reported to be ~36% for 0.25 vol%. The graphene nanoplatelets (GnP) could play a significant role in the enhancement of thermal conductivity due to its high thermal conductivity in the order of 3000–5000 W m−1 K−1. Amiri et al. [20] studied the thermal conductivity of GnP/H2O–EG (H2O:EG = 40:60) nanofluids experimentally and reported a 65% enhancement at ~0. 1 vol%. From the above literature, it is possible to conclude that 2D nanoplatelets play a significant role in the enhancement of thermal conductivity compared to other dimensionalities. The GnPs (2D nanostructure) are highly beneficial due to their high thermal conductivity and high aspect ratio as compared to the 1D and spherical nanomaterials [24, 25]. Recent effective medium theory predictions and experimental results show that the GnP-based nanocomposites show a better thermal conductivity enhancement due to their low thermal boundary resistance as compared to the carbon nanotubes [26]. Enhancements in heat transfer coefficient using different nanofluids with spherical nanoparticles have been performed experimentally by several researchers [27–33]. The higher enhancement in heat transfer coefficient was found at higher volume concentration of the spherical nanoparticles in basefluid. This will increase viscosity of nanofluids significantly resulting in higher pressure drop. Many researchers have reported the enhancement of heat transfer coefficient by more than 50–100% at lower concentration (<1% itself) of 2D nanostructure with limited penalty in the pressure drop [23, 34–38]. It is clearly seen that use of 2D nanostructure (GnP) with low density and high aspect ratio in basefluid will increase the thermal conductivity and heat transfer coefficient significantly.

The experimental studies on graphene nanoplatelets with water–ethylene glycol mixture are limited in the published literature. In order to understand the variation in thermophysical properties and thermal boundary resistance between graphene nanoplatelets and surrounding basefluids, further experimental works are required. In the current study, we report (1) thermophysical properties such as thermal conductivity and specific heat capacity of GnP/H2O–EG nanofluids, (2) the enhancement in thermal conductivity of nanofluids and comparison of experimental data with predicted values based on effective medium theory predictions and (3) the interfacial thermal resistance between GnP and water–ethylene glycol mixture.

Materials and methods

In this work, graphene nanoplatelets (GnP) and water–ethylene glycol mixture (H2O–EG) have been used as nanomaterial and basefluid, respectively. Generally, the water–ethylene glycol mixture with 30% volume of ethylene glycol is recommended to be used as a radiator coolant in winter [39]. Viscosity of water–ethylene glycol mixture will be increased when more quantity of ethylene glycol is added to the water. Hence, in the present study, we use water–ethylene glycol mixture (H2O:EG = 70:30) as a base fluid which has lower freezing point and higher boiling point than pure water with minimum enhancement in viscosity value. Multilayered graphene nanoplatelets (GnP) were obtained from the XG Sciences, USA (Grade M, diameter: 15 µm, average thickness: 5–10 nm and density: 2.2 g cm−3). In the material used, the number of graphene layers varies from 10 to 30. Scanning electron microscope visualization (SEM) and transmission electron microscope visualization (TEM model JEOL, JEM-2000EX. Acceleration voltage 200 kV) of GnP are shown in Fig. 1a, b. The thermophysical properties and purity of materials and basefluids are listed in Table 1.

Fig. 1
figure 1

a SEM and b TEM visualization of GnP

Full size image
Table 1 Thermophysical properties and purity of materials and basefluids at 30 °C
Full size table

Due to the hydrophobic nature of GnP, we used of non-covalent functionalization to prepare stable dispersions. Covalent or non-covalent treatment is widely used to prepare stable nanofluid dispersions with carbon nanostructures. Covalent treatment using acids often damages the nanostructure and introduces structural defects in the planar structure of GnP. The intrinsic thermal conductivity of GnP will be reduced due to enhanced phonon scattering. To avoid this problem, non-covalent preparation method was used in this work.

In this work, sodium deoxycholate (SDC) surfactant (C24H39NaO4, molecular weight: 414.55 g mol−1) was used to prepare stable nanofluid dispersions. The molecular structure of SDC is shown in Fig. 2. Harish et al. [41] prepared SWCNT-H2O nanofluids with SDC surfactant. Based on optical absorption spectroscopy and photoluminescence spectroscopy studies, they reported that SDC is an effective surfactant to disperse hydrophobic nature materials in basefluids. Pure ethylene glycol was added to pure water in the volume (%) ratio 30:70 and stirred well for 30 min using magnetic stirrer. Sodium deoxycholate (SDC) surfactant with 0.75 vol% (0.82% by mass) was added to the water–ethylene glycol (H2O–EG) mixture and stirred in the same way for 30 min. In this study, the basefluid (H2O–EG) with 0.75 vol% SDC considered as 0 vol% nanofluid. The GnPs were added directly to the basefluid (H2O–EG + 0.75 vol% SDC) by intensive ultrasonic vibration (QSonica, USA) for 2 h. The nanofluids were prepared with 0.001, 0.01, 0.05, 0.1, 0.15, 0.3 and 0.45% volume fractions of GnP (mass %: 0.0021, 0.021, 0.21, 0.42, 0.63, 0.83 and 1.04%). There was no visual sedimentation observed for the nanofluid suspensions kept undisturbed for more than 15 days. Furthermore, we also performed optical absorption (PG instruments, UK) and zeta potential (Horiba Nanopartica SZ-100, USA) measurements to investigate the stability of nanofluid dispersions.

Fig. 2
figure 2

Chemical structure of SDC (C24H39NaO4)

Full size image

Measurement of thermophysical properties

In this work, the thermal conductivity of nanofluids was measured using KD2 Pro thermal properties analyzer which works based on the transient hot wire method (Decagon Devices, USA). The KD2 Pro thermal properties analyzer is designed as per IEEE Standard 442–1981 and ASTM Standard D5334-08. In this instrument, the thermal conductivity of fluids can be measured with ks−1 sensor which is having an uncertainty of ±5.0% (for thermal conductivity ranges from 0.2 to 2 W m−1 K−1). The thermal conductivity of nanofluid was measured at a temperature of 30 °C (±0.15 °C accuracy) and normal atmosphere (±1% accuracy).

Specific heat capacity of nanofluid was measured by differential scanning calorimeter (TA instruments, USA). The measurements were taken for the temperature ranging from 30 to 50 °C with a scanning rate of 0.015 °C min−1. The combined uncertainty of specific heat measurements lies within ±3% deviation.

Results and discussion

Characterization of nanofluids

The prepared nanofluids were characterized by UV–Vis absorption spectrometer to predict the stability of dispersion. The stability test was performed for 15 days from the day of sample preparation for 0.01 vol% and 0.1 vol% of GnP loadings. Figure 3a illustrates the absorption spectra of GnP/H2O–EG nanofluids for lower volume fraction GnP loadings which are performed on same day after the sample preparation. The absorbance is found to increase with respect to volume fraction of GnP. The maximum peak of the absorbance spectrum is found to be around 255 nm. Long-term stability test was carried out quantitatively using UV–Vis absorption spectrometer. The predicted relative concentrations of GnP/H2O–EG nanofluids with respect to time are shown in Fig. 3b. The predicted relative concentration by absorbance at a wavelength of 255 nm is taken for time of 15 days with an interval of 5 days. It is found that the relative concentration of GnP/H2O–EG nanofluids marginally decreases over a period of 15 days.

Fig. 3
figure 3

a UV–Vis spectrum of GnP/H2O–EG nanofluids, b the stability of GnP/H2O–EG nanofluids with respect to time, c the reversibility characterization of GnP/H2O–EG nanofluids with respect to time

Full size image

The reversibility characterization of prepared nanofluids was performed using UV–Vis absorption spectrometer. The prepared nanofluids are sonicated again for 30 min after 20 days in order to predict the reversible/irreversible characterization of dispersions. The comparison of UV–Vis spectrum of GnP/H2O–EG nanofluids before sonication and after sonication is illustrated in Fig. 3c. The maximum sedimentation is found to be within 2% after sonication which indicates that the particle aggregation is reversible. From the UV–Vis results, it is observed that the prepared nanofluids are stable for the experimentation. However, the particle agglomeration will be reversible during the circulation process in the heat transfer applications.

Zeta potential distribution

In order to confirm the stability of prepared nanofluid at higher volume fraction of GnP loadings, the zeta potential distribution analysis was carried out. Zeta potential measurement is taken on the 20th day of sample preparation with 30 min sonication and without sonication. Figure 4 illustrates the zeta potential distribution of the nanofluids before sonication at 0.45 vol% GnP as a function of intensity. The measured value of the electric potential at the slip plane between the bound layer of diluent molecules surrounding the particle and the bulk solution is known as zeta potential. Zeta potential distribution values are expected to be more positive than +30 mV or more negative than −30 mV in order to confirm the stability of solid–liquid dispersion [42]. The average zeta potential distribution value of nanofluid (0.45 vol% GnP) is found to be −43.8 mV. It is observed that the particles are highly electronegative which indicates a good stability of the prepared nanofluid dispersion. The average zeta potential distribution of the nanofluid with 0.45 vol% of GnP sonicated for 30 min was found to be −53.2 mV. It is clearly seen that the particle aggregation is reversible, and zeta potential distribution is more negative than −43.8 mV.

Fig. 4
figure 4

Zeta potential distribution of (0.45 vol % GnP) nanofluids

Full size image

Calibration of measurement techniques

Initially, the measurements of thermal conductivity and specific heat capacity of water–ethylene glycol mixture (H2O:EG = 70:30) were taken for the temperatures ranging from 30 to 50 °C. In order to calibrate the measurement techniques, the measured values were compared with the reference (ASHRAE standard) [43] values as listed in Table 2. It is found that the measured thermal conductivity and specific heat capacity values are in good agreement with the reference values within ±2.5%.

Table 2 Comparison of measured data with reference data
Full size table

Thermal conductivity

The measurement of thermal conductivity was taken at 30 °C on the same day after the sample preparation. At-least 10 measurements were taken, and the average value was reported as the resultant thermal conductivity. Figure 5 illustrates the variation in thermal conductivity of nanofluids as a function of GnP loadings (vol%) at 30 °C. The thermal conductivity of nanofluids is found to increase significantly with respect to GnP loadings. Thermal conductivity is slightly enhanced (2%) even when the surfactant (SDC) added to the pure basefluid. Leong et al. [44] reported that the thermal conductivity of basefluid is enhanced up to 6.6% when gum Arabic surfactant was added to the basefluid. The highest enhancement of thermal conductivity is found to be ~18% at 0.45 vol%. Mechanism behind the thermal conductivity enhancement of nanofluids is still under debate. However, recent experimental works on freezing of nanofluid suspensions [45, 46] reveal that clustering of nanoparticles play a significant role in enhancing the thermal conductivity of suspensions. The works of Zheng et al. [45] and Harish et al. [46] show that the enhancement in thermal conductivity is significantly higher when clustering of nanoparticle is carefully controlled to form chain-like structures which enable efficient thermal transport thereby enhancing the thermal conductivity. Hence, it is possible that GnP with its 2D structure can form long chains which could be the reason for the significant enhancement in thermal conductivity even at very low loadings. Thermal conductivity measurement of nanofluid at higher concentration (0.45%) was taken after 20 days without sonication in order to check the repeatability. The thermal conductivity of 0.45 vol% GnP nanofluid is found to be 0.543 ± 0.0055 W m−1 K−1. The result shows the 2.5% deviation as compared to the previous measurement due to error in the instrument. However, the deviation is negligible with respect to uncertainty in the measurement.

Fig. 5
figure 5

Thermal conductivity of GnP/H2O–EG as a function of GnP loadings at 30 °C

Full size image

Furthermore, the measured thermal conductivity ratio (k nf/k bf) of GnP/H2O–EG nanofluid is compared with those obtained from Maxwell–Garnett-type effective medium theory (EMT) model. Many parameters such as the nanoparticle shape, size, aspect ratio, volume concentration and the interfacial thermal resistance between the nanomaterial and basefluid directly influence the enhancement in thermal conductivity. Nan et al. [47] proposed EMT model considering these parameters to predict the thermal conductivity enhancement. The thermal conductivity ratio (k nf/k bf) of nanofluid was determined using the following Eqs. (1–6).

$$\frac{{k_{\text{nf}} }}{{k_{\text{bf}} }} = \frac{{3 + \phi \left[ {2\beta_{11} (1 - L_{11} ) + \beta_{33} (1 - L_{33} )} \right]}}{{3 - \phi \left[ {2\beta_{11} L_{11} + \beta_{33} L_{33} } \right]}}$$
(1)

where the dimensionless parameters β11, β33, L11 and L33 are defined as follows in Eqs. (2–4)

$$\beta_{11} = \frac{{k_{11}^{c} - k_{\text{bf}} }}{{k_{\text{bf}} + L_{11} (k_{11}^{c} - k_{\text{bf}} )}},^{{}} \beta_{33} = \frac{{k_{33}^{c} - k_{\text{bf}} }}{{k_{\text{bf}} + L_{33} (k_{33}^{c} - k_{\text{bf}} )}}$$
(2)
$$L_{11} = \frac{{a^{2} }}{{2(a^{2} - 1)}} - \frac{{a^{2} }}{{2(1 - a^{2} )^{3/2} }}\text{Cos} \,h^{ - 1} a$$
(3)
$$L_{33} = 1 - 2L_{11}$$
(4)

where k nf and k bf refer to thermal conductivity of nanofluid and basefluid, respectively. φ is the volume fraction of graphene nanoplatelets inclusions. L 11 and L 33 are geometric shape factors of the spheroid particle. a (a = l/t) is the aspect ratio of ellipsoid (l and t are length and thickness of graphene nanoplatelets inclusions, respectively). k c11 and k c33 are refer to the equivalent thermal conductivity of the composite along the longitudinal and transverse directions.

The equivalent thermal conductivities of the composite considering the role of thermal interface resistance can be estimated as follows:

$$k_{\text{ii}}^{c} = \frac{{k_{\text{P}} }}{{1 + \frac{{\gamma L_{\text{ii}} k_{\text{P}} }}{{k_{\text{bf}} }}}},\quad\,i = 1,^{{}} 3$$
(5)

With

$$^{{}} \gamma = (1 + 2a)\frac{{Rk_{\text{bf}} }}{t}$$
(6)

where kP and R refer to the thermal conductivity of GnP and the interfacial thermal resistance between the GnP and basefluid, respectively.

For model calculations, the length and thickness of GnP are considered to be 1 μm and 10 nm, respectively. The thermal conductivity of the GnP is considered to be 3000 W m−1 K−1. The interfacial thermal resistance between GnP and basefluid is determined using the EMT model.

Figure 6 illustrates the comparison of measured thermal conductivity enhancement with those obtained from the EMT model predictions. The interfacial thermal resistance is extracted from the EMT model for the present experimental conditions and parameters. The measured thermal conductivity data are in good agreement with the EMT model predictions. The interfacial thermal resistance between graphene nanoplatelets and water–ethylene glycol mixture is found to be 1.7 × 10−8 m2KW−1. The thermal conductivity enhancement is significantly increased, when the interfacial thermal resistance is considered as 0 m2 KW−1 is shown in Fig. 6. Hence, it is found that the interfacial thermal resistance is limiting the enhancement of thermal conductivity. The thermal boundary conductance (inverse of interfacial thermal resistance) of GnP/H2O–EG nanofluid is found to be ~60 MW m2 K−1 from EMT prediction. Konatham et al. [48] studied the thermal conductivity enhancement of graphene–organic interfaces nanofluids. The maximum thermal boundary conductance of nanofluids is reported to be in the range of 50–270 MW m2 K−1. The thermal boundary conductance (TBC) of present nanofluid is in good agreement within the range reported in the published literature [48]. The GnP/H2O–EG nanofluids show better thermal performance than the carbon nanotube-based nanofluids due to its high TBC. Carbon nanotube-based nanofluids have a very low TBC in the range of 2.4–12 MW m2 K−1 [49] which indicates the graphene nanoplatelets are highly beneficial as compared to CNTs.

Fig. 6
figure 6

Comparison of thermal conductivity enhancement with EMT model predictions

Full size image

Specific heat capacity

Figure 7 illustrates the variation of specific heat capacity with respect to volume fraction of GnP loadings at 30 °C. The specific heat of nanofluid is found to decrease with increase in GnP loadings. The specific heat capacity of the nanofluid reduces with addition of GnP due to lower specific heat capacity of GnP as compared to the basefluid. The highest decrement in specific heat is found to be ~8% for 0.45 vol% as compared to 0 vol%. Further, the measured specific heat capacity of GnP nanofluid is compared with those obtained from physical principle of mixture rule as given in Eq. (7).

Fig. 7
figure 7

Specific heat capacity of GnP/H2O–EG as a function of GnP loadings at 30 °C

Full size image
$$(C_{\text {p}})_{\text{nf}} = \frac{{\left( {1 - \phi_{\text{P}} } \right)\left( {\rho C_{\text {p}}} \right)_{\text{bf}} + \phi_{\text{P}} \left( {\rho C_{\text {p}}} \right)_{\text{P}} }}{{(1 - \phi_{\text{P}} )\rho_{\text{bf}} + \phi_{\text{P}} \rho_{\text{P}} }}$$
(7)

For model calculation, the specific heat capacity of GnP and basefluid (0 vol%) is considered to be 0.643 [50] and 3.43 kJ kg−1 K−1, respectively.

The comparison of measured specific heat capacity with mixture rule is illustrated in Fig. 7. It is clearly seen that the measured specific heat capacity of nanofluids are lower than those of the predicted values using mixture rule. It is observed that the specific heat capacity marginally reduces with respect to GnP loadings which is obtained from mixture rule. The variation of specific heat capacity lies within the error limits up to 0.15 vol% of GnP loadings, while beyond 0.15 vol% of GnP the deviation exceeds the error bandwidth. Rule of mixtures was developed based on thermal equilibrium between the nanoparticles and basefluid. The dispersion of nanoparticles in basefluid leads to the faster propagation of heat into and out of the medium due to the enhancement in thermal diffusivity. It is inferred that nanoparticles and basefluid are not in thermal equilibrium due to enhancement in thermal diffusivity [51]. Hence, the deviation is higher between the measured values and predicted values using mixture rule.

Conclusions

The GnP/H2O–EG nanofluids were prepared by non-covalent method, and their thermal properties such as thermal conductivity and specific heat capacity were measured experimentally. The UV–Vis absorption spectrum and zeta potential analyses were performed in order to confirm the stability of the prepared nanofluid dispersions. From the results, it is found that the prepared nanofluid was stable. Thermal conductivity of nanofluid significantly increases with increasing volume concentration of graphene nanoplatelets loading. The highest enhancement of thermal conductivity was found to be ~18% at 0.45 vol% of GnP, respectively. The measured thermal conductivities of GnP/H2O–EG nanofluids were compared with those predicted using Maxwell–Garnett-type effective medium theory. The interfacial thermal resistance between graphene nanoplatelets and water–ethylene glycol mixture was estimated to be 1.7 × 10−8 m2 K W−1. A study on reducing the interfacial thermal resistance between graphene nanoplatelets and base fluid to further enhance the thermal conductivity of nanofluid could be a topic for further research.