PHYSICS APPLIED LETTERS 92, 023110 2008
Enhanced thermal conductivity by aggregation in heat transfer nanofluids containing metal oxide nanoparticles and carbon nanotubes
Jesse Wensel, Brian Wright, Dustin Thomas, Wayne Douglas, Bert Mannhalter, William Cross, Haiping Hong,a and Jon Kellar
Department of Material and Metallurgical Engineering, South Dakota School of Mines and Technology, Rapid City, South Dakota 57701, USA
Pauline Smith and Walter Roy
Army Research Laboratary, Aberdeen Proving Ground, Maryland 21005, USA
Received 18 November 2007; accepted 19 December 2007; published online 17 January 2008 An approximately 10% increase in the thermal conductivity TC of heat transfer nanofluids containing metal oxide nanoparticles and carbon nanotubes has been determined with very low percentage loading around 0.02 wt % of these two nanomaterials. These fluids are very stable and the viscosity remains approximately the same as water. A possible explanation for these interesting results is the aggregation of metal oxide particles on the surface of nanotubes by electrostatic attraction and form the aggregation chain along the nanotube. Time dependant magnetic results demonstrate that, under the influence of a strong outside magnetic field, the TC value decreases. Also, the TC value decreases when the pH is shifted from 7 to 11.45. 2008 American Institute of Physics. DOI: 10.1063/1.2834370 The discovery of carbon nanotubes CNTs has instigated considerable research efforts in recent years due to their promising thermal, electrical, mechanical, and functional properties. For example, single wall carbon nanotubes exhibit thermal conductivity TC as high as 2000 6000 W / m K.1 under ideal circumstances. By contrast, typical heat transfer fluids, such as water and ethylene glycol, have TC values of only 0.6 and 0.34 W / m K, respectively. There is a great need to increase the thermal conductivity significantly, while maintaining a desirably low fluid viscosity. Fluids containing carbon nanotubes called "nanofluids" hereafter should exhibit substantially improved TC values28 and could be useful for a variety of heat transfer related applications including coolants and lubricants. However, nanofluids utilizing a simple composite structure do not enhance the TC effectively. For instance, nanofluids with low percentage nanotube loading showed no significant improvements in TC, while at loading of 1 vol % CNTs 1.4 wt % , resulted in about 10%20% TC increase.9,10 Unforunately, at such a high concentration, the fluid became mudlike, thus, making the fluid much less useful for coolant and lubricant applications. A possible explanation for this unexpectedly small TC increase may be attributed to a lack of nanotube-nanotube physical contacts in the fluids. The carbon nanotubes are irregularly positioned in the fluids leading to infrequent contacts between the CNTs, therefore, only very high concentrations of CNTs produce noticeable TC improvements. Theoretical studies have indicated that aggregation of nanoparticles metal oxide, nanotube, etc. into clusters could enhance the thermal conductivity significantly.1113 To achieve this enhancement in practice, a concept of incorporating nanotube and metal oxide nanoparticles together in the fluid was developed. With proper control of the solution conditions, positively charged metal oxide particles will aggrea
Electronic mail: [e-mail].
gate on the negatively charged nanotube surface due to the chemical surfactant and probably form the aggregation chain along the nanotube. In this paper, the TC of nanofluid systems containing 0.02 wt % of carbon nanotube and metal oxide in water was investigated. Understanding of thermal conductivity behavior of such nanofluid at various pH values and in a magnetic field would be valuable in analysis and synthesis of nanofluids. Single wall carbon nanotubes SWNTs were purchased from Carbon Nanotechnologies Incorporation CNI, Houston, Texas . Metal oxide nanoparticles Fe2O3 and MgO and sodium dodecylbenzene sulfonate NaDDBS were purchased from Sigma Aldrich. The SWNTs and metal oxides were placed into water together with the appropriate amount of sodium dodecylbenzene sulfonate. Sonication was performed using a Branson Digital Sonifier, model 450. When needed, magnetic fields were produced by a pair of spaced apart Ba-ferrite magnet plates 4 6 1 in.3 and placing the sample in the middle of the gap between the magnets. The pH of the fluid containing the SWNTs and metal oxides was measured using a Denver Instrument UP-10 pH / mV meter. The viscosity value was obtained using a Brookfield LV viscometer. The thermal conductivity data was obtained by the Hot DiskTM thermal constants analyzer,14 using the following parameters: measurement depth of 6 mm, room temperature, power of 0.012 W, measurement time of 15 s, sensor radius of 3.189 mm, temperature coefficient of resistence of 0.0471/ K, disk type Kapton, and temperature drift. rec. yes. The magnetic field intensity was recorded by F.W. Bell Gaussmeter Model 5060. The dispersion stability of nanofluids containing nanotube and metal oxide particles was determined by visual inspection. The nanofluids were placed in a see-through glass beaker and observed to determine if any precipitation at the edge and/or bottom of the glass beaker had occurred.
2008 American Institute of Physics
0003-6951/2008/92 2 /023110/3/$23.00
92, 023110-1
023110-2
Wensel et al.
Appl. Phys. Lett. 92, 023110 2008
FIG. 1. Color online The picture of water A , nanotube+ metal oxide in water without chemical surfactant B and nanotube+ metal oxide in water with chemical surfactant C .
FIG. 2. Color online Thermal conductivities vs time in different systems. A Carbon nanotube+ Fe2O3 under magnetic field. B Carbon nanotube + Fe2O3, no magnetic field. C No carbon nanotube, only Fe2O3, no magnetic field.
Figrue 1 shows water A , nanotube+ metal oxide in water without chemical surfactant B , and nanotube+ metal oxide in water with chemical surfactant C . Relatively large, dark suspended spots, and much precipitation can be seen in bottle B, but is barely observed in bottle C. Therefore, bottle C appears well dispersed. This indicates that chemical surfactant is critical to the stability and homogeneity of the nanofluids. Nanofluids containing 0.017 wt % SWNT, 0.017 wt % MgO, and 0.17 wt % NaDDBS at pH 7 were made. The TC was measured to be 0.69 W / m K. This is a 10% increase over the base line water TC 0.63 W / m K . The measured solution viscosity was 1 cp, the same as water. This low waterlike viscosity is critical for a heat transfer coolant application. Previously, at least 0.5 wt % nanotube loading in the fluid was required in order to get a 10% TC increase. However, nanofluids with 0.5 wt % nanotubes are very viscous, and the fluid is nearly useless for coolant applications. One possible explanation of this TC enhancement is local aggregation of the metal oxide nanoparticles on the surface of the nanotubes by electrostatic attraction and possibly forming chains along the nanotube. Chemical surfactant is NaDDBS added to the solution to disperse the SWNT. NaDDBS adsorbs to the surface of the nanotubes and makes these surfaces negatively charged.15,16 The point of zero charge of the metal oxide MgO is around 11.17 As the used nanofluid has a pH value of 7, the MgO is positively charged. Due to its high dielectric constants, water is a good medium for the electrostatic attraction to take place over relatively long distances. Due to the electrostatic attraction, the positively charged metal oxides and the negatively charged nanotube aggregate. The visual observation indicates that dispersion of the nanoparticles is retained and these aggregations remain in the colloidal domain. Evidence to support the above assumption that local aggregation plays a crucial role in the TC enhancement of the nanofluids comes from altering the pH solution. Altering the nanofluids pH from 7 to 11.45 caused the TC of the nanofluid to drop back to 0.63 W / m K. In addition, the precipitation of nanoparticles was observed in the nanofluids. The
results indicate that aggregation could be broken by pH adjustment. This occurs because, at higher pH, a negatively charged MgO surface will occur. Therefore, while attractive van der Waals forces still exist between the nanoparticles and nanotubes, the electrostatic force is now repulsive and the MgO-SWNT separation increases. The nanoparticles aggregate and precipitate as their surface charge is relatively small and the ionic strength is relatively large. Consequently, the TC value of nanofluids decreases. More interestingly, after adjusting the pH value back to 7 and sonicating for a few minutes, the nanofluids became well dispersed again and the TC returned to 0.69 W / m K. Further evidence to support the local aggregation hypothesis is found from the time dependent magnetic results. Figure 2 shows the thermal conductivity versus time curves for different material systems. For Fe2O3 based nanofluids 0.02 wt % without carbon nanotube and magnetic field , the TC value is around 0.62 0.63 W / m K and remains reasonably constant versus the time. The TC value is essentially the value for the de-ionized DI water itself. With the addition of 0.02 wt % carbon nanotubes at pH 7, the TC increases to around 0.70 W / m K due to local aggregation of the nanoparticles, as described previously. This value is independent with time within experimental error. In the presence of a magnetic field, the TC shows very interesting behavior. The thermal conductivity initially increases with time but eventually reaches a peak. With more time in the magnetic field, the particles gradually form clumps of Fe2O3 particles and clumps of CNTs, thus, decreasing the TC as these clumps precipitate from the solution. The maximum TC value is around of 0.92 W / m K, about 50% higher than DI water value, 35% higher than the nanofluid without magnetic field. It is interesting to note that with a much longer time in magnetic field around 100 120 min , the TC value decreases to around 0.63 0.64 W / m K, even lower than that of the fluid with nanotubes but without magnetic particles. This is attributed to the possible breaking of the aggregation under the long time strong external magnetic force. More interestingly, after the fluid was removed from the magnetic
023110-3
Wensel et al.
Appl. Phys. Lett. 92, 023110 2008
field and resonicated for a few minutes, the TC value returns to 0.70 W / m K. The above results also indicate that TC enhancement is mostly independent of nanoparticles type, as long as the solution conditions were chosen to give oppositely charged nanoparticles and nanotubes. Thus, the TC value is mainly determined by the ability of the nanoparticles and nanotubes to aggregate locally. More work needs to be done to understand how particle size, type, and charge affect the TC. In addition to NaDDBS, positively charged surfactant acetyltrimethylammonium bromide CTAB could disperse SWNT well. However, MgO nanoparticles do not remain in the colloidal domain in the nanofluids containing SWNT and CTAB. This provides another strong evidence that electrostatic attraction plays critical role to keep fluids in good stability and dispersion. In summary, a 10% increase in the TC of the heat transfer nanofluids containing metal oxide particles and carbon nanotubes at very low percentage loading around 0.02 wt % of these two nanomaterials has been determined. The fluids show good dispersion and stability, while maintaining waterlike viscosity. These properties are very important for the fluids to be used in coolant applications. A possible explanation for the TC enhancement is related to the aggregation of metal oxide on the surface of nanotube by electrostatic attraction to probably form aggregations along the nanotube. This mechanism is strongly supported by the experimental results utilizing pH adjustment, magnetic field, and surfactant type. The type of metal oxide does not affect the TC value as long as the solution conditions are maintained such that electrostatic attraction between the nanoparticles and nanotubes remains. The work may open a route to engineer the next generation of high-efficiency heat transfer nanofluids. This initial
work needs to be supplemented by optimizing the experimental parameters weight percentage, type and ratio of nanotube, metal oxide and surfactant, pH value, etc. to obtain enhanced TC values. H. Hong would like to thank Army Research Laboratory Cooperative Agreement DAAD19-02-2-0011 for financial support. Special thanks to Professor Pawel Keblinski of Rensselaer Polytechnic Institute in Troy, New York for his critical comments related to these results.
S. Berber, Y. K. Kwon, and D. Tomnek, Phys. Rev. Lett. 84, 4613 2000 . 2 X. Wang, X. Xu, and S. Choi, J. Thermophys. Heat Transfer 13, 474 1999 . 3 S. Choi, Z. Zhang, W. Yu, F. E. Lockwood, and E. A. Grulke, Appl. Phys. Lett. 79, 2252 2001 . 4 B. H. Kim and G. P. Peterson, J. Thermophys. Heat Transfer 21, 451 2007 . 5 H. Hong, J. Wensel, F. Liang, W. E. Billups, and W. Roy, J. Thermophys. Heat Transfer 21, 234 2007 . 6 P. Keblinski, J. A. Eastman, and D. G. Cahill, Mater. Today 8, 36 2005 . 7 H. Hong, B. Wright, J. Wensel, S. Jin, X. Ye, and W. Roy, Synth. Met. 157, 437 2007 . 8 B. Wright, D. Thomas, H. Hong, L. Groven, J. Puszynski, E. Duke, X. Ye, and S. Jin, Appl. Phys. Lett. 91, 173116 2007 . 9 D. S. Wen and Y. L. Ding, J. Thermophys. Heat Transfer 18, 481 2004 . 10 H. Hong, Y. Zheng, and W. Roy, J. Nanosci. Nanotechnol. 7, 3180 2007 . 11 R. Prasher, W. Evans, P. Meakin, J. Fish, P. Phelan, and P. Keblinski, Appl. Phys. Lett. 89, 143119 2006 . 12 R. Prasher, P. Phelan, and P. Bhattacharya, Nano Lett. 6, 1529 2006 . 13 W. Evans, R. Prasher, J. Fish, P. Meakin, P. Phelan, and P. Keblinski, Int. J. Heat Mass Transfer unpublished . 14 Detail information see www.hotdisk.se . 15 O. Matarredona, H. Rhoads, Z. Li, J. Harewell, L. Balzano, and D. Resasco, J. Phys. Chem. B 107, 13357 2003 . 16 J. Wensel, M.S. thesis, South Dakota School of Mines, 2006. 17 K. Bourikas, C. Kordulis, and A. Lycourghiotis, Environ. Sci. Technol. 39, 4100 2005 .
1