Particle size effects in the thermal conductivity enhancement of copper-based nanofluids
© Saterlie et al; licensee Springer. 2011
Received: 11 November 2010
Accepted: 14 March 2011
Published: 14 March 2011
Skip to main content
© Saterlie et al; licensee Springer. 2011
Received: 11 November 2010
Accepted: 14 March 2011
Published: 14 March 2011
We present an analysis of the dispersion characteristics and thermal conductivity performance of copper-based nanofluids. The copper nanoparticles were prepared using a chemical reduction methodology in the presence of a stabilizing surfactant, oleic acid or cetyl trimethylammonium bromide (CTAB). Nanofluids were prepared using water as the base fluid with copper nanoparticle concentrations of 0.55 and 1.0 vol.%. A dispersing agent, sodium dodecylbenzene sulfonate (SDBS), and subsequent ultrasonication was used to ensure homogenous dispersion of the copper nanopowders in water. Particle size distribution of the copper nanoparticles in the base fluid was determined by dynamic light scattering. We found that the 0.55 vol.% Cu nanofluids exhibited excellent dispersion in the presence of SDBS. In addition, a dynamic thermal conductivity setup was developed and used to measure the thermal conductivity performance of the nanofluids. The 0.55 vol.% Cu nanofluids exhibited a thermal conductivity enhancement of approximately 22%. In the case of the nanofluids prepared from the powders synthesized in the presence of CTAB, the enhancement was approximately 48% over the base fluid for the 1.0 vol.% Cu nanofluids, which is higher than the enhancement values found in the literature. These results can be directly related to the particle/agglomerate size of the copper nanoparticles in water, as determined from dynamic light scattering.
in which K, K1, and K2 are the thermal conductivities of the nanofluid, the dispersed phase, and the base fluid, respectively, n is an empirical shape factor dependent on the sphericity of the particles, and V2 is the volume fraction of the particles. This model includes the effect of particle shape and size and has been applied to a copper/water nanofluid system with three different sphericity factors, showing that the thermal conductivity of a nanofluid depends on both the particle volume fraction and the shape .
A comparison of particle shape in SiC nanofluids of water and ethylene glycol has been completed with powders of spherical morphology of 26-nm average crystallite size and cylindrical morphology of 600-nm average crystallite size along the fiber axis direction . It was concluded that because heat transfer occurs between the surface of the particles and the fluid, heat transfer is more efficient for a system with a larger interfacial area. Therefore, smaller crystallites will exhibit the greatest amount of thermal conductivity enhancement. However, this conclusion does not take into consideration the effect of agglomeration of the powders in the fluid. If the powders are heavily agglomerated, then heat transfer is possibly hindered, since transfer will only occur on the surfaces of the agglomerates, even if the powder crystallite sizes are in the nanometer range.
In nanofluids of Al2O3/water, an increase of 12% in thermal conductivity enhancement was observed with the addition of Al2O3 powders with crystallites of 28-nm average size at a particle volume fraction of 3% . Other studies of this same system, with powders of 13 and 38 nm Al2O3 average crystallite size exhibited a 20% and 8% thermal conductivity enhancement [5, 6]. The effect can be attributed to the phonon mean free path in the nanofluid  such that in a nanofluid containing nanoparticles of a crystallite size much different from the phonon mean free path, the thermal conductivity will increase with decreasing crystallite size, while in a nanofluid containing nanoparticles less than or equal to the mean free path, the thermal conductivity will be reduced with crystallite size reduction due to the scattering of phonons. None of these studies reported the level of agglomeration (i.e., the size of the particle/agglomerates) in the nanofluid. Since powders tend to heavily agglomerate in polar fluids, a comparison connected to the agglomeration of the particles and not just to crystallite size, cannot be ascertained without proper particle/agglomerate size distribution analysis [8–14].
Crystallite and particle size results in copper-based nanofluids
Crystallite size (nm)
Particle size (nm)
Cu prepared using oleic acid
Cu prepared using CTAB
Sinha et al. 
Wang et al. 
Liu et al. 
Eastman et al. 
Xuan and Li 
Velasco et al. 
Li et al. 
Yu et al. 
In this study, we prepared copper nanopowders and then incorporated these powders into water, with SDBS as a dispersant, for the formation of well-dispersed nanofluids. The copper nanoparticles were produced through the reduction of copper (II) chloride with sodium borohydride in the presence of a surfactant [i.e., oleic acid or cetyl trimethylammonium bromide (CTAB)]. After preparation of powders using various surfactant concentrations, the optimal samples were chosen based on phase and particle size criteria. The small agglomerate sizes for the 0.55 and 1.0 vol.% copper nanofluids exhibited thermal conductivity enhancements of up to 48% over the base fluid with a mean thermal conductivity of 0.89 W/m·K, higher than the enhancement values found in other studies, as will be discussed later. We conclude that minimization of agglomerate size in nanofluids is important in order to take full advantage of the presence of nanopowders in the base fluid.
The main experimental result and contribution of this work to the nanofluid field is that excessive aggregation of the dispersed phase reduces the effectiveness of the produced nanofluid through the settling of particles and disruption of nanofluid flow during thermal conductivity testing. While this has been hypothesized extensively, our work shows definite experimental results to support this. When testing the thermal conductivity of the nanofluids, the increase in particle loading for the oleic acid powders from 0.55 to 1.0 vol.%, results in rapid settling due to increased agglomeration, as carefully determined by dynamic light scattering measurements.
Due to the rapid release of hydrogen during the reaction and the presence of bubbling nitrogen gas, significant foam was formed upon reduction of the copper salt/surfactant solution. To counteract this foaming, octyl aldehyde (98%, Sigma Aldrich, St. Louis, MO, USA), a known de-foaming agent, was added as needed throughout the process.
Once the reaction was complete, the fluid was removed from the reactor, emptied into 50-mL centrifuge tubes, and centrifuged using an Eppendorf Centrifuge 5810 (Eppendorf, Hamburg, Germany 22339) for 10 min at 11,000 rpm. The clear supernatant liquid was discarded and a solution of 50/50 semiconductor grade methanol (99.9%, Alfa Aesar, Ward Hill, MA 01835, USA) and de-ionized water was added to the centrifuge tubes in order to remove sodium chloride and the surfactant from the surfaces of the particles. These tubes were shaken vigorously for 5 min. A second methanol/water wash and a final methanol wash were applied on the powders while decanting the supernatant after each centrifugation. The copper powders were then placed in a vacuum dessicator for 2 to 3 days. Once dry, the powders were ground by hand using a mortar and pestle.
In later experiments, the batch sizes were increased. This required one solution to contain 67.08 g of CuCl2 in 200 mL of de-ionized water and a separate solution of 148.86 g of sodium borohydride in 300 mL of de-ionized water. The same molar ratios for the surfactants were used for the production of the larger batch sizes, resulting in amounts of 6.94 g of oleic acid and 1.29 g of CTAB.
Two different nanofluids, of 0.55 and 1.0 vol.% copper concentrations in water, were prepared. For preparation of the 0.55 vol.% copper nanofluid with 15 wt.% dispersant, 2.61 g of SDBS (Sigma Aldrich, St. Louis, MO, USA) was added to 296 mL of de-ionized water and allowed to stir at a slow speed, so that a high shearing force did not result in bubbles on the surface. A mass of 14.78 g of copper powder was then slowly added to the dispersant solution under slow stirring and allowed to mix for 1 h. The 1.0 vol.% copper nanofluid contained 4.74 g of SDBS and 26.88 g of copper powder in 292 mL of de-ionized water. The fluid was transferred to a jacketed reaction vessel and then ultrasonicated using an Ultrasonic Processor (Ace Glass, Vineland, NJ, USA) for 50 min with amplitude of 70%, pulsed on for 3 s, then off for 3 s.
Particle size measurements were performed on a Microtrac Nanotrac Ultra dynamic light scattering system (Microtrac Inc., Montgomeryville, PA, USA). The Microtrac Nanotrac Ultra instrument measures the particle size distribution in solution, with measurement capability from 0.8 nm to 6.5 μm. Multiple measurements were done on each sample, using the appropriate parameters determined by the estimated particle size range and fluid viscosity. With these parameters, the measurements were taken at a run time of 30 s. At least five measurements were taken for each sample and then averaged to produce accurate particle size distribution analysis for each sample as is recommended by the instrument manufacturer and in conjunction with ASTM standard E2490-09.
There are several accepted measurement methodologies for thermal conductivity of a nanofluid. In this study, a flow cell experimental setup (Figure 1) was used for the evaluation of dynamic thermal conductivity of the nanofluids. This experimental setup allows for the continuous flow of the nanofluid between an insulated heating element and a cooling fan. Two thermoelectric modules were affixed to both walls of the heating channel and connected to a power supply. The power supply created an electric current which increased the temperature of the walls in the heating channel. A dc pump supplied the circulation of the nanofluid within the closed loop under laminar flow conditions, Re < 2,300. When the nanofluid was passed through the heating channel, it absorbed some of the heat, increasing the temperature of the nanofluid, and then was subsequently cooled back to room temperature by the cooling fan. The temperature of the nanofluid and the walls were monitored by four thermocouples located at the walls and at the inlet and outlet channels of the heating channel. By comparing the steady-state channel wall temperature, the effectiveness of a nanofluid was evaluated.
where, ΔQ is the heat input, A is the heat transfer surface area, and ΔT is the temperature difference between the solid surface and the surrounding fluid. The Nusselt number (Nu) is the ratio of convective to conductive heat transfer across the wall within the channel and is directly proportional to the convective heat transfer coefficient and the hydraulic diameter of the tube, while indirectly proportional to the thermal conductivity of the fluid. Calibration of the experimental setup was validated by testing the system with pure de-ionized water using a Nu of 5.9 for the calculations. A Nusselt number of 5.9 provides a reasonable estimate of k for pure water, but it may not be as accurate for nanofluids . Thus, there is likely some error in the determination of the nanofluid thermal conductivities. The thermal conductivity values of de-ionized water for three independent measurements were 0.61, 0.58, and 0.63 W/m·K, with errors of 1.67%, -3.33%, and 5.00%, respectively. For the case of the nanofluid thermal conductivities, the experimental errors are likely slightly greater than 5%, although the exact magnitude of these errors is not known.
More importantly, we observe that our nanofluid of 1.0 vol.% CTAB-prepared copper powders exhibits comparable or greater enhancement than some nanofluids with much greater particle loadings. For example, the water-based nanofluids prepared by Xuan and Li  have particle loadings greater than 30 wt.%, but exhibit almost the same (in fact, slightly lower) thermal conductivity enhancement as our own nanofluids, which have modest particle loadings of 1 vol.% (8.3 wt.%). We attribute this directly to our smaller particle/agglomerate size and possibly our lower viscosity, which plays an important role in dynamic heat transfer applications. An efficient nanofluid involves the greatest thermal conductivity enhancement, with the least amount of particle loading, and therefore, a low viscosity. Thus, we propose that determination of particle/agglomerate size in nanofluids is an important variable that should be determined in order to obtain a complete picture of the characteristics and behavior of nanofluids. An appropriate technique for these measurements is dynamic light scattering.
Copper nanopowders were successfully synthesized using a chemical precipitation method in the presence of two surfactants, oleic acid and CTAB. This study has explored the importance of particle size distribution analysis in the reporting of nanofluid results. The particle sizes of the powders produced in this study were successfully reduced from approximately 1 μm to 120 and 80 nm for the oleic acid- and CTAB-prepared powders, respectively, upon addition of a dispersant in water. Two particle loadings were used for producing the copper-based nanofluids, 0.55 and 1.0 vol.%. A dynamic thermal conductivity test setup was devised and thermal conductivity measurements were performed on both nanofluids. A thermal conductivity enhancement of 22% over water was observed for the 0.55 vol.% Cu nanofluids. When the particle loading was increased to 1.0 vol.%, the nanofluids of oleic acid-prepared Cu powders settled and clogged the test setup. The nanofluids of CTAB-prepared copper powders remained well dispersed, allowing for a successful thermal conductivity measurement. A maximum increase of 48% was observed for the 1.0 vol.% copper nanofluid from the CTAB-prepared Cu powders. The thermal conductivity enhancement in these latter fluids is directly attributable to the excellent dispersion of the nanoparticles in the fluid.
This project was funded by the National Science Foundation under contract no. IIP 0823112.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.