This journal is© the Owner Societies 2016 Phys. Chem. Chem. Phys., 2016, 18, 15363--15368 | 15363

Cite this:Phys.Chem.Chem.Phys.,

2016, 18, 15363

Temperature-dependent effect of percolationand Brownian motion on the thermal conductivityof TiO2–ethanol nanofluids†

Chien-Cheng Li,a Nga Yu Hau,a Yuechen Wang,a Ai Kah Soh*b andShien-Ping Feng*a

Ethanol-based nanofluids have attracted much attention due to the enhancement in heat transfer and

their potential applications in nanofluid-type fuels and thermal storage. Most research has been

conducted on ethanol-based nanofluids containing various nanoparticles in low mass fraction; however,

to-date such studies based on high weight fraction of nanoparticles are limited due to the poor stability

problem. In addition, very little existing work has considered the inevitable water content in ethanol for

the change of thermal conductivity. In this paper, the highly stable and well-dispersed TiO2–ethanol

nanofluids of high weight fraction of up to 3 wt% can be fabricated by stirred bead milling, which

enables the studies of thermal conductivity of TiO2–ethanol nanofluids over a wide range of operating

temperatures. Our results provide evidence that the enhanced thermal conductivity is mainly contributed

by the percolation network of nanoparticles at low temperatures, while it is in combination with both

Brownian motion and local percolation of nanoparticle clustering at high temperatures.

1. Introduction

Nanofluids, in which nano-sized particles are suspended inliquids, have emerged as a potential candidate in the design ofheat transfer fluids. The experimental observations of nanofluidshave shown much higher thermal conductivity than the predictionsof Maxwell effective medium theory. Potential mechanisms,such as local Brownian motion, percolation of nanoparticleclustering, and liquid layering, have been proposed to explainthis enhancement.1–3 Recently, ethanol-based nanofluids haveattracted increasing interest due to their potential applicationsin fuels, thermal convection and thermal storage.4–7 However,there are contradictory data regarding the thermal conductivityof ethanol-based nanofluids, such as temperature dependency,particle size, and mass fraction.1,3,8 In reviewing the previouslyreported experiments on ethanol-based nanofluids, we summarizefour points which are the possible causes of the inconsistencies inthermal conductivity data reported by different groups. Firstly,nanofluids are usually made by a two-step process, in whichthe nanoparticles are pre-synthesized and then added to the

base fluids. Since nanoparticles have high surface energy, theyreadily aggregate together when added to the solution, leadingto a large variation of particle size distribution.9–12 The aggregationand sedimentation of nanoparticles cause a stability problem inthe nanofluids, particularly in high mass fractions and at hightemperatures. This is the reason why most research work onnanofluids is focused on low mass fractions (0.01–1 wt%),5,13,14 butvery little work has been conducted on high mass fractions.2,15,16

Secondly, the surfactant or additives are usually used to preventnanoparticle aggregation, but their effects are not considered.Thirdly, the accuracy of thermal conductivity is highly dependenton the measured environment (temperature, vibration, and noise).Lastly, very little existing studies have considered the effect ofinevitable water content in ethanol, which influences the thermalconductivity of base fluids versus temperature.

In this paper, stirred bead milling was employed to fabricateTiO2–ethanol nanofluids without the need for adding surfactants/additives,17,18 which produced a uniform particle distribution, highmass fraction and excellent long-term and thermal stability. TiO2

nanoparticles were selected because of their low material cost, goodelectrocatalytic activity, and excellent thermal/chemical stability,enabling them to serve as potential candidates in nanofluids forcombustion systems and compact reactor-heat exchanger.19,20

A home-made thermal insulation cell was used to provide anon-disturbing environment for thermal conductivity measure-ments. By using this stable suspension, we investigated the heattransport behaviors of TiO2–ethanol nanofluids over a broad range

a Department of Mechanical Engineering, The University of Hong Kong, Hong Kong,

China. E-mail: hpfeng@hku.hk; Fax: +852 2858 5415; Tel: +852 2859 2639b School of Engineering, Monash University Malaysia, Malaysia.

E-mail: soh.ai.kah@monash.edu; Fax: +60 3 55146207; Tel: +60 3 55146138

† Electronic supplementary information (ESI) available: Additional details aboutthe home-made thermal insulation cell and experimental set-up for thermalconductivity measurements. See DOI: 10.1039/c6cp00500d

Received 23rd January 2016,Accepted 6th May 2016

DOI: 10.1039/c6cp00500d

www.rsc.org/pccp

PCCP

PAPER

15364 | Phys. Chem. Chem. Phys., 2016, 18, 15363--15368 This journal is© the Owner Societies 2016

of mass fractions (0.5–5 wt%) and temperatures (10–50 1C). Wealso considered the water content in ethanol, which causes thetemperature-dependent transition for the thermal conductivityof base fluids. And the same method has also been used for theinvestigation of graphene aqueous nanofluids by our group now.

2. Experiment2.1 The preparation of TiO2–ethanol nanofluids

The TiO2 nanoparticles used for dispersions were purchasedfrom Evonik Degussa (Aeroxides P90, Germany) with a BETsurface area of 70–100 m2 g�1 and anatase content greater than90%. The average diameter of TiO2 nanoparticles was about21 nm. Ethanol (96% BP, Guangdong Guanghua Chemical FactoryCo., Ltd, China) was used as a base fluid. Ethanol (99.5% GC,Merck, Germany) and ethanol/DI water mixtures were used for thecalibration of thermal conductivity. TiO2–ethanol nanofluids wereprepared by dispersing TiO2 nanoparticles in ethanol (96%)through stirred bead milling (JBM-B035, JUSTNANO, Taiwan) for6 h. The ZrO2 ball loading and stirrer speed were maintained at600 g and 2000 rpm, respectively. The particle size of the ZrO2 ballis about 0.05 mm. The nanoparticles were mechanically dispersedin ethanol at various mass fractions (0.5 wt% (0.092 vol%), 1 wt%(0.185 vol%), 3 wt% (0.566 vol%), 5 wt% (0.959 vol%)). Noadditional dispersants/additives were added into the solutions.

2.2 Material characterization

The morphology and microstructure of the TiO2 nanoparticleswere determined by HRTEM (Tecnai, G220S-Twin). The TEMsamples were prepared by dropping 2 ml of upper well-suspendedsolutions on the copper grid and then dried in the ambientenvironment for 1 min. The particle size distributions and zetapotentials were measured using a dynamic light scattering (DLS)analyzer (Microtrac, Nanotrac Wave, USA) at room temperature.Here the particle size determined by the scattered light intensityin DLS is the size of nanoparticle clusters in the suspension (notthe size of the individual TiO2 nanoparticle). Fourier transforminfrared (FTIR) spectra were obtained using a Bruker Tensor 27spectrometer.

2.3 Measurement of nanofluids

The electrical conductivity, viscosity, and thermal conductivityof the nanofluids were measured at the temperatures of 10 1C,20 1C, 30 1C, 40 1C, and 50 1C. The viscosity of the nanofluidswas measured using a rotating viscometer (NDJ-9S, ShanghaiPingxuan Instrument Co., Ltd, China) at the spinning rate of20 rpm. Thermal conductivity was measured by the transienthot-wire method using a thermal conductivity meter (KD2 pro,Decagon Devices, USA). The probe, 1.3 mm in diameter and60 mm in length, was vertically immersed in the center ofnanofluids. Calibration of the probe was done by measuring thethermal conductivity of DI water, ethanol (99.5%), and ethanol(96%). The dimensions of the home-made thermal insulation cellwere 30 mm diameter and 70 mm length (Fig. S1, ESI†). The probeis fixed to the cap of the cell. The entire unit was immersed in a

circulating water bath. Thermal conductivity was recordedautomatically every 15 min for 6 h. The reported values representthe average of at least 12 measured data. Electrochemical impedancespectroscopy was performed using an AutoLab-PGSTAT302Nworkstation.

3. Results and discussion

Ultrasonic dispersion is the most common method for preparingnanofluids.10,21,22 In this paper, we successfully produced a uniformand stable suspension of TiO2 nanoparticles in ethanol using stirredbead-milling methods,11,23 where the particle–particle interactionbetween the nanoparticles can be controlled appropriately duringthe process. Fig. 1 shows the TEM images of TiO2 nanoparticlesbefore and after stirred bead milling. As shown in Fig. 1a, thereare more aggregated nanoparticles in TiO2–ethanol nanofluidsby using ultrasonic dispersion as compared to other samplesusing stirred-bead milling (Fig. 1b–e). The diffraction patterns ofTiO2 nanoparticles in the inset images show that all the samplesare in anatase phase, indicating that the crystallinity of TiO2 NPsdoes not change after the stirred bead-milling process. A photo-graph of TiO2–ethanol nanofluids with mass fractions of 0.5 wt%,1 wt%, 3 wt%, and 5 wt% after 60 days of preparation is shown inFig. 1f, which shows excellent long-term stability without obvioussedimentation for 0.5 wt%, 1 wt%, and 3 wt% TiO2 nanofluids(Note: the change in thermal conductivity is maintained within�1% after 60 days). For 5 wt% TiO2 nanofluids, it can be seen thatsome sediment particles appeared at the bottom of the vial after60 days. Fig. 2 shows the particles size distribution of TiO2–ethanol nanofluids for the as-prepared sample (black curve) andthe sample after 60 days (red curve) at room temperature, as wellas for the sample at 50 1C (blue curve). After 60 days, the meandiameters of TiO2 nanoparticles in nanofluids for the massfractions of 0.5 wt%, 1 wt%, and 3 wt% can be still maintainedat 30 nm, which are similar to those of the as-prepared samples(Fig. 2a–c); while the mean diameter became smaller (B10 nm) inthe 5 wt% TiO2–ethanol nanofluids because only the small-sizednanoparticles remained in suspension after the sedimentation ofbig nanoparticle aggregates (Fig. 2d). As shown by the blue curvesin Fig. 2a–c, the particle size in the TiO2–ethanol nanofluidsslightly increased after heating up to 50 1C and becamewell-dispersed TiO2 nanoparticle clusters in ethanol withoutsedimentation. This indicates that local clustering of nano-particles occurred with the increase of temperature. Thesenanoparticle clusters would re-disperse in the base fluid andthus return back to their original individual particles aftercooling to room temperature. In comparison, 3 wt% TiO2–ethanol nanofluids prepared by the ultrasonic dispersionmethod have a wide range of nanoparticle distribution, asshown in the green curve of Fig. 2c. Table 1 shows the valuesof zeta potential for TiO2–ethanol nanofluids with 0.5, 1, and 3wt% at different time intervals of 1, 15, 30, and 60 days and at50 1C. All the measured samples show non-degradation of zetapotentials even if heating up to 50 1C or after 60 days. The highzeta potential comes from the stable repulsive forces between

Paper PCCP

This journal is© the Owner Societies 2016 Phys. Chem. Chem. Phys., 2016, 18, 15363--15368 | 15365

the TiO2 nanoparticles in ethanol. Fig. 3 presents the FTIRspectra of ethanol and TiO2–ethanol nanofluids, where the

characteristic peaks at 1055, 2981 and 3391 cm�1 correspondto the C–H (as shown in Fig. 3), C–H and O–H stretchingvibrations, respectively (three values corresponding to 3 terms).The additional peak in the spectra of TiO2 nanofluids in thetransmittance band at 1659 cm�1 corresponds to the stretchingvibration of Ti–OH. It is believed that the abundant and uniformTi–OH functional group can lead to the well dispersion of TiO2

nanoparticles in ethanol by electrostatic repulsive forces.24 Thus,the stirred-bead milling produces TiO2–ethanol nanofluids with-out the need for adding surfactants/additives, which have auniform particle distribution, high mass fraction and excellentlong-term and thermal stability that enable the investigation ofthermal conductivity over a broad range of mass fractions andtemperatures in TiO2–ethanol nanofluids.

Before measuring the thermal conductivity of TiO2–ethanolnanofluids, the thermal conductivity of ethanol (base fluid) wasmeasured in the temperature range of 10–50 1C. Fig. 4a showsthe thermal conductivity versus temperature for ethanol withdifferent volume concentrations of water. The measured value

Fig. 2 Particle size distribution of (a) 0.5 wt%, (b) 1 wt%, (c) 3 wt% and(d) 5 wt% mass fractions of TiO2 nanoparticles in ethanol-based nanofluidsfor the as-prepared sample (black curve), sample after 60 days (red curve)at room temperature, and sample at 50 1C (blue curve) and sampleprepared by an ultrasonic method (green curve).

Table 1 Zeta potentials for TiO2–ethanol nanofluids with 0.5, 1, and 3 wt%concentrations at different time intervals of 1, 15, 30, and 60 days and thesamples after heating up to 50 1C

TiO2 nanofluids 0 wt% 0.5 wt% 1 wt% 3 wt% 5 wt%

Zetapotential(mV)

As-prepared 0 11.85 11.77 11.47 11.42After 15 days 0 11.51 11.24 11.62 11.30After 30 days 0 11.42 11.08 11.25 11.63After 60 days 0 11.28 11.17 11.72 11.75Heated up to 50 1C 0 10.32 10.46 10.33 10.53

Fig. 1 TEM images of TiO2 nanoparticles before (a) and after bead-milling for 6 h at various concentrations of (b) 0.5 wt%, (c) 1 wt%, (d) 3 wt%, and(e) 5 wt% (insets: the corresponding SAED pattern of TiO2 nanoparticles). (f) Photograph of TiO2–ethanol nanofluids with mass fractions of 0.5, 1, 3, and5 wt% after 60 days of preparation.

Fig. 3 FT-IR spectra of ethanol and TiO2–ethanol nanofluids at theconcentration of 3 wt%.

PCCP Paper

15366 | Phys. Chem. Chem. Phys., 2016, 18, 15363--15368 This journal is© the Owner Societies 2016

of 99.5% ethanol at 30 1C is 1.823 � 0.036 W mK�1, comparedwith the theoretical data for 1.775 W mK�1 at 30 1C for pureethanol (100%). This difference should come from the contentof water in ethanol, which is inevitable due to azeotropy. Thereis a V-shaped transition of thermal conductivity as a function oftemperature because the thermal conductivity of ethanol decreaseswith increasing temperature while that of water increaseswith increasing temperature. With increasing water content,the temperature dependence of thermal conductivity becomessteeper. Fig. 4b and c shows the thermal conductivity of TiO2–ethanol nanofluids at various temperatures and mass fractionsof TiO2 nanoparticles (0.5–5 wt%). Basically, the V-shaped curveof thermal conductivity as a function of temperature follows theperformance of base fluids, and the incorporation of TiO2

nanoparticles enhances the thermal conductivity. The dashedlines in Fig. 4c are the thermal conductivities versus tempera-tures based on the predictions of the modified Maxwell’smodel.25 The discrepancy between the measured and calcu-lated thermal conductivities suggests a lack of explanation forthe temperature-dependent effect on thermal conductivitiesof TiO2–ethanol nanofluids. As seen in 3 wt% TiO2–ethanolsample, the enhancement in thermal conductivity is about 8%at 20 1C and about 11% at 50 1C. Over the past decade, themechanisms of thermal conductivity enhancement in nano-fluids were intensely debated, such as the Brownian motion,percolation, nanoparticle clustering, ballistic transport andinterfacial layering.2,3,26–28 According to the previously reportedpercolation threshold of 0.05–0.1 vol% for TiO2-based nanofluids,the particle concentrations of 0.092–0.96 vol% (0.5–5 wt%) in oursystems should be higher than the percolation threshold.19,29

In our experiment, when temperature was low (o20 1C), thethermal conductivities of TiO2–ethanol nanofluids increasedwith increasing TiO2 mass fraction but were not affected muchby temperature. Note that when Brownian motion exists, thethermal conductivity should increase with increasing temperature.Therefore, one may infer that the contribution of Brownian motionis only nominal at this low temperature range. Our previousresearch has found that the percolation network of nanoparticleswas the key contributor to this thermal conductivity enhance-ment.2,30 When the temperature was over 30 1C, the thermalconductivity of TiO2–ethanol nanofluids increased with increasingtemperature, which indicated that the enhanced thermal conduc-tivity was related to microconvection caused by Brownian motionof the nanoparticles.3,28 Although the role of Brownian motion isdebatable, it may be an important factor when the viscosity ofnanofluids significantly decreases with increasing temperature,which is the possible mechanism in our experimental observation(Fig. 4d). The shear thinning effect is negligible in the range ofviscosity measurement.31 It has been reported that Brownianmotion and nanoparticle clustering were related, and not com-pletely independent of each other.27,28,32 As mentioned above, therapid clustering of nanoparticles took place to form nanoparticleaggregates when the temperature was increased. This nanoparticleclustering would decrease the Brownian motion due to theincrease of the mass of the aggregates; whereas, the thermalconductivity would be increased due to the local percolationbehavior as the nanoparticles came in contact with each otherwithin the aggregates. The individual aggregates would have ahigher thermal conductivity, which can be considered as newparticles with larger effective radii. As mentioned in Fig. 2, thesize of nanoparticles in our system was slightly increased afterheating up to 50 1C, which indicates that the well-dispersed TiO2

nanoparticles would rapidly and locally aggregate as well-dispersednanoparticle clusters in ethanol while increasing the temperatureand thus leading to a local percolation effect. Meanwhile, thedecrease of fluid viscosity would compensate the mass effect oflarge-size nanoparticle clusters, leading to the microconvectioncaused by the enhancement of Brownian motion. Therefore, onemay infer that the combination of nanoparticle clustering (localpercolation behavior) and Brownian motion (microconvection)would cause the increase of thermal conductivity while increasingthe temperature. Note that the thermal conductivity enhancementdecreased with continuous agglomeration of clusters to make amuch bigger size, as in the case of 5 wt% TiO2–ethanol nanofluids.

Therefore, we propose a model that the enhancement ofthermal conductivity is dominated by the percolation networkformed by nanoparticle clustering at low temperatures; how-ever, at high temperatures it is governed by a combination ofthe local percolation behavior of nanoparticle clustering andthe microconvection caused by Brownian motion. By takingadvantage of the V-shape transition shown in Fig. 4, thethermal conductivities of ethanol were similar at 20 1C and50 1C, but those of TiO2–ethanol nanofluids were different at20 1C and 50 1C. In the present study, AC impedance isemployed to study the impact of structural transformation onthe transport properties of 96% ethanol (base fluid) and 3 wt%

Fig. 4 (a) Thermal conductivity of ethanol in concentrations of 99.5%,96%, 90% and 85% vs. temperature. (b) Thermal conductivity of 96%ethanol and 0.5, 1, 3 and 5 wt% TiO2–ethanol nanofluids vs. temperature.(c) Thermal conductivity ratio of 0.5, 1, 3 and 5 wt% TiO2–ethanol nanofluidscompared to 96% ethanol vs. temperature. (d) Viscosity of 96% ethanol and0.5, 1, 3 and 5 wt% TiO2–ethanol nanofluids vs. temperature. The corres-ponding volume fractions (vol%) are 0.092%, 0.185%, 0.566%, and 0.959%for 0.5, 1, 3 and 5 wt% TiO2–ethanol nanofluids.

Paper PCCP

This journal is© the Owner Societies 2016 Phys. Chem. Chem. Phys., 2016, 18, 15363--15368 | 15367

TiO2–ethanol nanofluids at 20 and 50 1C. Nyquist and Bodeplots, as presented in Fig. 5a and b respectively, show that theimpedances of ethanol at 20 and 50 1C are similar, but that of3 wt% TiO2–ethanol nanofluids at 50 1C is smaller than that at20 1C. This indicates different heat transport behaviors of thenanoparticles at low and high temperatures because of theconstant mass fractions of nanoparticles and similar thermalconductivities of base fluids. The large semicircle in theNyquist plot can be modeled as the intra- and interclusterimpedance responses by two RC parallel circuits in series.2,33,34

The RC unit with higher characteristic frequency represents theintracluster impedance response, while the RC unit with lowercharacteristic frequency represents the intercluster impedance.The results show that the intracluster resistance decreases to26% and the intercluster resistance decreases to 13% with theincrease of temperature from 20 1C to 50 1C. The real andimaginary parts of the complex capacitance in Fig. 5c and dshow an increase of capacitance at 50 1C as compared to that at20 1C. When increasing the temperature, the viscosity andpermittivity of ethanol decrease and its electrical conductivitydoes not change much, as shown in Fig. 4d and Fig. S2 (ESI†).The effective surface area would be reduced due to the effect ofnanoparticle clustering at 50 1C. Therefore, the explanation forthe increased capacitance at 50 1C should come from thereduction of EDL (electric double layer) effective thickness.35

This result led us to infer that the microconvection induced byBrownian motion possibly plays a role in the reduction of EDLeffective thickness when increasing the temperature.

4. Conclusions

This paper presents a stirred bead-milling method, which doesnot need the addition of additives, to prepare TiO2–ethanol

nanofluids having uniform particle distribution, high massfraction, and excellent stability. By using these nanofluids,the thermal conductivity behaviors of ethanol-based TiO2 nano-fluids were investigated over a broad range of concentrations(0.5–5 wt%) and temperatures (10–50 1C). At low temperatures,the nanoparticle clustering formed a percolation network,which dominated the enhanced thermal conductivity. Byincreasing the temperature, the well dispersed TiO2 nano-particles rapidly aggregated as well-dispersed nanoparticleclusters in ethanol, which gave rise to a local percolationbehavior. Meanwhile, the decrease of fluid viscosity enhancedthe Brownian motion of nanoparticles, which led to the micro-convection effect. A combination of nanoparticle clustering andBrownian motion caused the enhancement of thermal conduc-tivity at high temperatures. The impedance spectroscopy pro-vided evidence that both the intra-/inter-cluster resistancesdecreased with increasing temperature from 20 1C to 50 1C atconstant mass fractions of TiO2 nanoparticles and constantimpedances of base fluids.

Conflict of interest

The authors declare no competing financial interest.

Acknowledgements

This work was supported by the General Research Fund of theResearch Grants Council of Hong Kong Special AdministrativeRegion, China: Award Number: HKU 712213E (S. P. Feng). Itwas also partially supported by the FRGS Grant (Project no.FRGS/2/2013/SG06/MUSM/01/1) provided by the Ministry ofHigher Education (MOHE), Malaysia (A. K. Soh).

References

1 S. Cingarapu, D. Singh, E. V. Timofeeva and M. R. Moravek,Int. J. Energy Res., 2014, 38, 51–59.

2 R. Zheng, J. Gao, J. Wang, S.-P. Feng, H. Ohtani, J. Wang andG. Chen, Nano Lett., 2011, 12, 188–192.

3 R. Prasher, P. Bhattacharya and P. E. Phelan, J. Heat Transfer,2006, 128, 588.

4 A. K. Singh and V. S. Raykar, Colloid Polym. Sci., 2008, 286,1667–1673.

5 S. S. Park and N. J. Kim, J. Renewable Sustainable Energy,2014, 6, 042005.

6 L. Lu, L.-C. Lv and Z.-H. Liu, Thermochim. Acta, 2011, 512,98–104.

7 Y. Gan, Y. S. Lim and L. Qiao, Combust. Flame, 2012, 159,1732–1740.

8 R. Prasher, P. Bhattacharya and P. E. Phelan, Phys. Rev. Lett.,2005, 94, 025901.

9 B. T. Branson, P. S. Beauchamp, J. C. Beam, C. M. Lukehartand J. L. Davidson, ACS Nano, 2013, 7, 3183–3189.

10 J.-H. Lee, S.-H. Lee and S. Pil Jang, Appl. Phys. Lett., 2014,104, 161908.

Fig. 5 Electrochemical impedance spectroscopy (EIS) characterization of96% ethanol and 3 wt% TiO2–ethanol nanofluids at 20 1C and 50 1Crespectively. (a) Nyquist plots and (b) Bode plots. Bode plots of the (c) realand (d) imaginary parts of the complex capacitance.

PCCP Paper

15368 | Phys. Chem. Chem. Phys., 2016, 18, 15363--15368 This journal is© the Owner Societies 2016

11 M. Silambarasan, S. Manikandan and K. S. Rajan, Int.J. Heat Mass Transfer, 2012, 55, 7991–8002.

12 L. S. Sundar, M. K. Singh, E. V. Ramana, B. Singh, J. Gracioand A. C. M. Sousa, Sci. Rep., 2014, 4, 4039.

13 R. Prasher, P. E. Phelan and P. Bhattacharya, Nano Lett.,2006, 6, 1529–1534.

14 S. Srivastava, Appl. Nanosci., 2012, 2, 325–331.15 P. Keblinski, J. A. Eastman and D. G. Cahill, Mater. Today,

2005, 8, 36–44.16 S. Shenogin, A. Bodapati, L. Xue, R. Ozisik and P. Keblinski,

Appl. Phys. Lett., 2004, 85, 2229–2231.17 H. Chang and S.-C. Lin, Mater. Trans., JIM, 2007, 48,

836–841.18 M. Reddy, V. V. Rao, B. Reddy, S. N. Sarada and L. Ramesh,

Nanosci. Nanotechnol. Lett., 2012, 4, 105–109.19 S. M. S. Murshed, K. C. Leong and C. Yang, Int. J. Therm. Sci.,

2005, 44, 367–373.20 A. Turgut, I. Tavman, M. Chirtoc, H. P. Schuchmann,

C. Sauter and S. Tavman, Int. J. Thermophys., 2009, 30,1213–1226.

21 M. Hadadian, E. K. Goharshadi and A. Youssefi, J. Nanopart.Res., 2014, 16, 1–17.

22 T. T. Baby and S. Ramaprabhu, J. Appl. Phys., 2010,108, 124308.

23 Y. Wang and E. Forssberg, Int. J. Miner. Process., 2006, 81,1–14.

24 K. Kwak and C. Kim, Korea-aust rheol j, 2005, 17, 35–40.25 R. L. Hamilton and O. K. Crosser, Industrial & Engineering

Chemistry Fundamentals, 1962, 1, 187–191.26 P. Keblinski, S. Phillpot, S. Choi and J. Eastman, Int. J. Heat

Mass Transfer, 2002, 45, 855–863.27 D. Wen and Y. Ding, Int. J. Heat Mass Transfer, 2004, 47,

5181–5188.28 J. Gao, R. Zheng, H. Ohtani, D. Zhu and G. Chen, Nano Lett.,

2009, 9, 4128–4132.29 L. S. Semko, Y. I. Kruchek, Y. A. Shevlyakov, P. P. Gorbik and

E. I. Oranskaya, Inorg. Mater., 2007, 43, 358–363.30 C. Liu, H. Lee, Y.-H. Chang and S.-P. Feng, J. Colloid Interface

Sci., 2016, 469, 17–24.31 H. Chen, Y. Ding and C. Tan, New J. Phys., 2007, 9, 367.32 W. Evans, J. Fish and P. Keblinski, Appl. Phys. Lett., 2006,

88, 093116.33 N. Y. Hau, Y.-H. Chang, Y.-T. Huang, T.-C. Wei and S.-P.

Feng, Langmuir, 2013, 30, 132–139.34 N. Y. Hau, Y.-H. Chang and S.-P. Feng, Electrochim. Acta,

2015, 158, 121–128.35 S. Ganguly, S. Sikdar and S. Basu, Powder Technol., 2009,

196, 326–330.

Paper PCCP