On the specific heat capacity enhancement in nanofluids
- 1.3k Downloads
Molten salts are used as heat transfer fluids and for short-term heat energy storage in solar power plants. Experiments show that the specific heat capacity of the base salt may be significantly enhanced by adding small amounts of certain nanoparticles. This effect, which is technically interesting and economically important, is not yet understood. This paper presents a critical discussion of the existing attendant experimental literature and the phenomenological models put forward thus far. A common assumption, the existence of nanolayers surrounding the nanoparticles, which are thought to be the source of, in some cases, the large increase of a nanofluid’s specific heat capacity is criticized and a different model is proposed. The model assumes that the influence of the nanoparticles in the surrounding liquid is of long range. The attendant long-range interfacial layers may interact with each other upon increase of nanoparticle concentration. This can explain the specific heat maximum observed by different groups, for which no other theoretical explanation appears to exist.
KeywordsEnergy storage Solar power plants Heat transfer nanofluids Specific heat capacity
transmission electron micrograph
A standard problem in courses on Statistical Mechanics, when the topic is Stefan’s law, is the calculation of the size of a “solar panel” large enough to collect the energy currently consumed by the earth’s population. Even though this is an academic problem, the resulting area is surprisingly small, providing a feeling for the vast amount of solar energy received by the surface of the earth. One of the technologies, being developed in this context, is that of thermal solar power plants, especially of the parabolic trough and power tower types. Inside these plants, heat transfer fluids, e.g., molten salts, are used for energy transport as well as storage . Heat transfer fluids must meet certain key criteria - aside from being inexpensive. They should be thermally stable at high temperatures. Their melting point should be as low as possible. The same is true for their vapor pressure at high temperatures (several hundred degree Celsius). Other aspects include corrosivity, viscosity, thermal conductivity, toxicity, etc.
The thermodynamic property of interest in this study is the specific heat capacity, i.e., the heat capacity in units of joules per gram Kelvin, of the heat transfer fluid, which should be as high as possible. The aforementioned molten salts or rather salt mixtures do possess a wide temperature range of useful application - an advantage outweighing their comparatively low specific heat capacity of about 0.75 to 2 J/(gK). However, a number of experiments (cf. the recent review ) do show that the specific heat capacity of the base salt mixture, here and in the following we talk about c P , the specific heat capacity at constant pressure, is substantially enhanced by addition of nanoparticles in small amounts (≈ 1 wt.%), i.e., particles whose size distribution is peaked around 10 to 100 nm. Salt mixtures doped with nanoparticles belong to a larger and more general class of fluids, so-called nanofluids.
Nanofluids doped with suspended nanoparticles do show enhanced thermal conductivities [3, 4, 5, 6, 7, 8]. The effect of nanoparticle addition on the specific heat capacity of the fluids, however, does not yield a consistent picture (e.g., [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37]). Das and co-workers [9, 10, 11] find reduced specific heat capacities of nanofluids consisting of silicon dioxide, zinc oxide, and alumina nanoparticles, respectively, dispersed in a mixture of water and ethylene glycol compared to the base fluid. The specific heat capacity of the nanofluid is found to decrease with increasing nanoparticle concentration. Zhou and Ni  also find a reduced specific heat capacity of the water-based alumina nanofluid and a similar decrease of specific heat capacity with increasing particle concentration. An extensive study by O’Hanley et al.  confirms this detrimental effect of various types of nanoparticles on the heat capacity of water. Recent work by Sekhar and Sharma  comes to similar conclusions. In contrast, Nelson et al.  find that c P of polyalphaolefin is enhanced by 50 % when mixed with graphite nanoparticles at 0.6 % mass fraction. The c P of Li 2CO3/K2CO3 eutectic salt is enhanced by 19 % when mixed with carbon nanotubes at 1 % mass fraction as reported by Shin et al. . Refs. [17, 18] by Jo and Banerjee are continuations of this work using the same base fluid. In , the effect of nanoparticle dispersion is studied based on graphite nanoparticles. Greater enhancement in the specific heat capacity was observed from the nanomaterial samples with more homogeneous dispersion of the nanoparticles. In , multiwalled CNTs (carbon nanotubes) were dispersed using a surfactant (SDS) to obtain homogeneous dispersion. Four different concentrations (0.1, 0.5, 1, and 5 wt.%) of CNT were employed. It was observed that the specific heat capacity is enhanced by doping with the nanotubes in both solid and liquid phase. In addition, the enhancements of the specific heat capacity are increased with increase of the CNT concentration. Zhou et al.  find a maximum enhancement of about 6 % of the specific heat capacity of their ethylene glycol-based CuO nanofluid. In addition, Shin and Banerjee [20, 21, 22] obtain specific heat capacity enhancements of about 14 and 19 % to 24 % in different nanofluids, i.e., Li 2CO3/K2CO3 eutectic and chloride eutectic, doped with 1 wt.% SiO 2 nanoparticles. Most recently, they report specific heat capacities in the Li 2CO3/K2CO3 (62:38) eutectic doped with 1.5 wt.% silica nanoparticles (size between 2 and 20 nm), which are about 120 % above the base salt specific heat capacity . These researchers have also used alumina nanoparticles in again the same base salt mixture . They report an enhancement of around 30 % at 1 % mass concentration. Lu and Huang  study the NaNO 3/KNO 3 (60:40) liquid base salt mixture doped with alumina nanoparticles of two distinct sizes. Their key observation is that c P is larger for the nanofluid containing the larger nanoparticles under otherwise identical conditions. This is also found by Dudda and Shin  in NaNO 3/KNO 3 (60:40) doped (1 wt.%) with SiO 2 nanoparticles. In contrast, Tiznobaik and Shin  do find no influence of particle size in the Li 2CO3/K2CO3 eutectic doped with silicon dioxide nanoparticles (5, 10, 30, and 60 nm in diameter - same sizes as in ). In another publication, Tiznobaik and Shin  tie the enhancement of the specific heat capacity observed in the previous system to the formation of nanostructures surrounding the particles. In subsequent work Shin, Tiznobaik, and Banerjee  discuss the possible role of fractal flocs formed by the nanoparticles on the enhancement of the specific heat capacity in nanofluid. The two investigations by Chieruzzi  and Lasfargues et al. [31, 32] study the eutectic mixture of NaNO 3/KNO 3 doped with different nanoparticles at different concentrations. These researchers find that in general the specific heat capacity is enhanced by a few percent (up to about 10 % using copper oxide in ). Most notably, however, a maximum enhancement is observed in the range between 0.1 and 1 wt.% nanoparticle. An analogous maximum was also observed by Heilmann  using the same base salt doped with Al 2O3 nanoparticles. Another confirmation of the maximum comes from Andreu-Cabedo . They study the eutectic mixture of NaNO 3/KNO 3 doped with silica nanoparticles in the concentration range from 0.5 to 2 wt.%. Their maximum heat capacity increase is 25 %.
It is important to stress that the observation of enhanced heat capacity is not limited to nanofluids whose base is a binary mixture. Recently, Chieruzzi et al.  have studied a single component base fluid, KNO 3, doped with different size nanoparticles of silica, alumina, and a mix thereof. In both the solid and the liquid phase, the authors do observe an increase of c P . At 1 wt.% added nanoparticles c P in the solid phase is increased between 5 and 10 %. In the liquid phase, the increase is around 6 %. Ho and Pan  have looked for the optimal concentration of alumina nanoparticles in a molten ternary salt mixture. They obtain a specific heat capacity increase of about 20 % at a very small nanoparticle concentration, i.e., 0.063 wt.%. A still larger increase of c P has been found by Paul , who has studied different nanoparticle-enhanced ionic liquids. Ionic liquids are salts whose cations are large organic species and anions are organic or inorganic species. Interestingly, the considerable increases of c P reported in this work appear to be monotonous functions of nanoparticle volume fraction in the entire range of volume fractions studied (up to 2.5 wt.%).
In summary, it is probably safe to say that the specific heat capacity of salt mixtures is enhanced by every type of nanoparticle used. The particle mass concentration should not exceed 1 wt.%, however. Details of the methods of preparation may vary (e.g., the temperatures during drying of the samples). Quite independent of these variations, c P enhancements are between 10 and 30 % on average. Doping water with nanoparticles thus far yields a decrease of the heat capacity. A molecular theory explaining the above effects of nanoparticles on the specific heat capacity of heat transfer fluids does not yet exist.
Results and Discussion
The first part of this section, phenomenological heat capacity of nanofluids - the nanolayer concept, is an exposition of the existing phenomenological theory and where it fails to describe the experimental findings. The second part, phenomenological heat capacity of nanofluids - the mesolayer concept, presents arguments in support of the idea that the effect of the nanoparticles on the base fluid should be of much longer range than assumed previously. In the third part, the interacting mesolayer model, it is shown how this concept can be developed into a model explaining the specific heat maximum observed in several experimental studies. The section concludes with a suggestion on how to tie this phenomenological theory to molecular interaction.
where ρ liq and ρ np are the mass densities of the base liquid and the nanoparticles, respectively. In , the authors investigate water-based silica, alumina, and copper oxide nanofluids. In some cases, Eq. 1, e.g., the open symbols in the top panel of Fig. 1, provides a good description of the experimental data. In other cases, e.g., the closed symbols in the top panel of Fig. 1, Eq. 1 is off.
It is worth noting that the specific heat capacity of the nanoparticles, c P,np, is significantly below the specific heat capacity of the base fluid, c P,liq - and not only in the case of water. Theoretical calculations of c P,np indicate that the nanoparticles possess somewhat larger specific heat capacities (at high temperatures) than their bulk materials [38, 39]. However, the difference does not alter this statement. In particular, the concentration of nanoparticles in the experiments is so small that the difference between the nanoparticle’s c P,np and the corresponding bulk-specific heat capacities is negligible in the present context.
The quantity x np=1−x liq=M np/(M np+M liq) is the mass fraction nanoparticles (notice: x np=y/(1+y)). In the limit λ=1 (or κ=1), this equation is identical to Eq. 1. However, if momentarily we treat λ as an adjustable parameter, we can improve the approximation to the data in Fig. 1 (solid lines in the top panel), which suggests that the concept of an interfacial layer indeed possesses significance.
The bottom panel of Fig. 1 shows experimental results obtained by Lu and Huang . These researchers study the NaNO 3/KNO 3 (60:40) liquid base salt mixture doped with alumina nanoparticles of two distinct sizes. Their key observation is that c P is larger for the nanofluid containing the larger nanoparticles under otherwise identical conditions. Using again Eq. 5, we find reasonable agreement with the experimental data when λ is around 2.
What does this mean for the λ values used in the bottom panel of Fig. 1? Suppose we use A≈150 m 2/g for the small particles and A≈30 m 2/g for the large particles. In addition, we set ρ liq,i ≈ρ liq≈1.8 g/cm 3 (cf. ). Of course, the particles induce structuring of the liquid close to their surface  and thus this may or may not be a good approximation. But it is fully sufficient if we want to explore the basic consequences of the interface. If we assume Δ=10 nm, then we need κ≈0.55 for the small particles and κ≈−0.30 for the large ones. Negative values for κ are in violation of thermal stability, however. We can avoid this by increasing the other quantities, i.e., A, ρ liq or Δ. For instance, assuming Δ=100 nm, we obtain κ≈0.96 in case of the small particles and κ≈0.87 in the case of the large particles. Now the κ values are reasonable, but the width of the interface is unexpectedly large.
Phenomenological heat capacity of nanofluids - the mesolayer concept: The above discussion suggests that the nanoparticles induce a far reaching effect in the surrounding liquid. Of course, there is also the possibility that κ instead of Δ may become large, i.e. c P,i ≫c P,liq. Even though large or diverging specific heat capacities may occur at a phase transition, i.e., in particular a second-order transition, these conditions are rather special and strongly temperature dependent . Therefore, it is worth noting that most authors do observe only a slight change, usually a slight linear increase, of c P in a fairly large temperature interval in the liquid state of the nanofluid (examples include Figs. 2 and 3 in , Fig. 3 in , Fig. 2 in , Figs. 2, 3, and 4 in , or Fig. 1 in ). All in all, a mechanism which increases c P,i by a factor greater than two compared to the base fluid and at the same time is weakly dependent on temperature is difficult to imagine.
Next, we focus on recent work by Shin and Banerjee . These authors find increases of the specific heat capacity of over 120 % relative to the base fluid (Li 2CO3/K 2CO3) (62:38). The nanoparticles are silica particles. The authors assume a κ of around 4. However, they do not provide a mechanism explaining this large increase of c P,i relative to c P,liq . They also try to avoid large interface widths by proposing that the main effect is due to small nanoparticles with diameters of around 2 nm. This however makes it difficult to fix the attendant mass fraction (the overall x np is around 1.5 wt.%). Small particles, as is illustrated in Fig. 2, do possess a large surface area. Thus, one may obtain a sufficiently large value for λ, according to Eq. 7, with a moderate Δ.
There is another subtlety worth mentioning. Apparently, the authors do observe segregation of their nanofluid into two “phases,” called types A and B, during preparation. The large increases of c P reported in the paper are solely observed in the type A nanofluid, apparently containing well-dispersed nanoparticles <20 nm. However, it is not clear how large the mass content nanoparticle really is in type A. If we calculate the average center of mass separation for 2-nm particles at 1.5 wt.%, we obtain around 7 nm (on a cubic lattice). However, the transmission electron micrograph of a type A sample printed in the paper shows particles separated much farther than this. The picture is consistent with 20-nm particles as the dominant species, which yield an average separation of around 70 nm. For these particles, we obtain Δ≈50 nm if κ=4 is assumed. Because large κ values, and 4 already is large, are difficult to rationalize, this again suggest that Δ is in the 100-nm range. It is also worth mentioning that in a previous publication, the same authors , based on the same system (with a different silica supplier), do find a mere 20 % enhancement of c P . The reason for this substantial discrepancy is not clear.
The interacting mesolayer model: There is an explanation for the observed c P maximum if we accept the above idea that the “interfacial layer” induced by a nanoparticle in the surrounding fluid is significantly wider than currently assumed. Notice that Eq. 5 does not account for the overlap between “interfacial layers” of nanoparticles. There is of course no need for this within the nanolayer concept, because at the nanoparticle concentrations of interest, no significant overlap of nanolayers occurs. If the layers are much wider, then overlap occurs even at very small nanoparticle concentrations, and this, as we shall see, can produce the c P vs. x np maximum.
Notice that κ=κ max when Y=1, which means that little or no overlap between shells occurs on average. κ=κ min when Y=0. This means that the overlap is at a maximum. We use κ max=1.2(1.14), κ min=0.87(1.02), and Δ/R=3.5(5.5) for the Heilmann (Lasfargues et al.) data. The theory now is in better agreement with both data sets. Notice that κ is reduced with increasing overlap.
But why should κ be reduced (below one) with increasing overlap? After all, our phenomenological model matches the experiments showing enhanced heat capacity only if κ max is larger than one. The presence of a nanoparticle may apparently enhance the specific heat capacity in a surrounding liquid shell of width Δ by a factor κ max>1. So why should the presence of a second nanaoparticle in the proximity of the first have the opposite effect? The interpretation is that increasing overlap means that nanoparticles may be getting close. This imposes significant positional ordering on the liquid molecules, which may form a “solid-like” layer structure between the two “walls.” This type of surface-induced ordering can for instance be measured using the surface force apparatus (e.g., ). Experiments show that c P,solid of the base salt in its solid state is significantly lower than the corresponding c P,liq. An example is , where the authors provide both c P,solid and c P,liq. Outside the immediate transition region, both specific heat capacities are well described by a linear dependence on temperature. If we linearly extrapolate c P,solid into the temperature regime where c P,liq is measured, then c P,solid is up to 10 % less than c P,liq. Thus, if the c p of the base liquid in this overlap area between nanoparticles becomes more like c P,solid, we should expect κ to decrease - even below one. Due to the complexity of the systems, we believe that this point should be studied using molecular simulation techniques rather than analytical approaches. The foremost question however remains - why is κ max larger than one in the first place? We shall return to this question below.
We note that we did not alter the ratio Δ/R according to the two very different mean particle sizes used by Lu and Huang . The ratio Δ/R determines the position of the c P maximum in terms of nanoparticle concentration. We do not think that Δ and R are independent. The effect of nanoparticles on viscosity or, in the case of elastomers, the effect of nanoparticles on the elastic modulus is derived from hydrodynamic [45, 46, 47] or elastic equations . The only length scale in these theories is the particle radius, R. The effect of the presence of a particle in the surrounding medium at a distance r from the particle center decays algebraically in terms of certain powers of R/r. We consider it very likely that the effect of the particles on c P will follow from an analogous mathematical description. This means that Δ has no precise meaning. It should rather be thought of as some r at which R/r has decayed below a certain value. In this sense, we view Δ/R as one single parameter - independent of the actual particle radius.
However, we still need to tackle the question what possible mechanism leads to κ max>1. Before addressing this point, we mention two observations, based on experimental evidence, which may provide helpful hints. First, it is worth noting that experimental studies find that the qualitative effect of nanoparticles on c P appears to be the same in the solid and in the liquid state of the respective nanofluids (e.g., [21, 25, 26, 27, 35]). The solid state being of course not so much of technical importance. In the aforementioned studies, the solid state c P rises close to linearly with increasing temperature in a rather wide range of temperatures. In most of these studies (exceptions are , where c P is close to constant in the studied temperature range, and , where the authors find a decrease of c P with rising temperature), this holds true also in the liquid state. These similarities between solid and liquid phase are interesting, because the solid state theory of specific heat capacity is much better developed than its counterpart for liquids. Secondly, we note that base salt mixtures like NaNO 3/KNO 3 (60:40) or the Li 2CO3/K 2CO3 eutectic (as well as water) possess a c P , which is rather close to c V calculated using the Dulong-Petit law.
One recent approach to the heat capacity in liquids, which explicitly exploits similarities between liquids and solids, is the phonon theory by Bolmatov et al. . Bolmatov et al. show that their theory yields the isochoric heat capacity, C V , vs. T for a great number of fluids in quite good quantitative agreement with experimental measurements. The key idea, based on an old proposition of Frenkel, is that a liquid becomes solid-like at frequencies larger than τ −1∼η −1, when it may support shear waves. Here, τ is a certain relaxation time and η is the liquid’s viscosity. The authors derive an expression for the internal energy of the liquid consisting of contributions due to longitudinal phonons as well as transverse (shear) phonons beyond τ −1 and a diffusion contribution below τ −1. We emphasize this theory also because of one particular aspect. Since Einstein [45, 46], it is well known how the addition of dispersed particles to a liquid increases its viscosity, i.e., η=η ∗(1+2.5ϕ) for ϕ≪1 (cf. ). Here, η ∗ is the viscosity of the pure liquid and ϕ is the volume fraction of the added particles. In the present case, this means that τ −1 should decrease, i.e., the nanoliquid is able to support more shear modes at a particular temperature in comparison to the base fluid. This in turn increases C V , and thus addition of nanoparticles, at first glance, leads to an increase of the isochoric heat capacity. However, the full relation between τ −1 and η −1 in  is τ −1=G ∞ /η, where G ∞ is the infinite-frequency shear modulus. Whether or not G ∞ possesses the same dependence on the concentration of added particles as η (cf. ) is a subtle question. If it does, then there is no effect of the nanoparticles on τ −1. In the case of filled rubbers, which have much in common with liquids, it is well known that the glass transition is indeed not shifted by addition of nanoparticles (even at large amounts), i.e., the attendant relaxation times are not dependent on filler.
Notice that anharmonic molecular interactions are the source of thermal expansion. Both heat capacity and thermal expansion are closely related - at least in the case of solids (e.g., § 67 in ; To the best of the author‘s knowledge, this particular relation has not been explored thus far in experiments on nanofluids in the present context. Thermal expansion of water-based nanofluids has been obtained in several studies primarily aimed at quantities other than heat capacity (e.g., ). But in these water-based systems, excess effects due to the presence of nanoparticles are difficult to discern.) Because the expression Bolmatov et al. derive for C V , based on their phonon theory, also contains the thermal expansion coefficient (in fact, their C V increases with increasing thermal expansion), the calculation or modeling of the latter may offer a theoretical approach to c P . At least from the perspective of molecular computer simulation, thermal expansion appears to be the simpler quantity.
We remark in closing that recently, Tiznobaik, Shin, and Banerjee [28, 29] have carried out an interesting experiment, which shows that addition of NaOH in minuscule amounts prevents the formation of structure and eliminates the previously observed c P enhancement. In the present picture, we would interpret the effect of NaOH as one that relaxes anharmonicity effects.
Specific heat enhancements of the type observed thus far requires one or both of the following to be true: (i) Every particle is surrounded by a nanolayer with a c P exceeding the bulk fluid’s c P by a large factor. This is necessary to explain the large increases observed in some experiments in which c P is enhanced through addition of minute amounts of nanoparticle to the base fluid. (ii) The effect of the particles on the base fluid has a long range (100 nm or more), which only requires a moderate change of c P in the attendant mesolayer. The new phenomenological theory, i.e., the interacting mesolayer model, developed on the assumption of (ii) can explain the occurrence of the c P vs x np maximum. It contains three parameters: (1) Δ/R, i.e., the width of the mesolayer in units of R, the particle radius; (2) κ max, the ratio of the average specific heat capacity inside the mesolayer at vanishing overlap to the specific heat capacity of the base fluid at the same temperature; (3) κ min, the ratio of the average specific heat capacity inside the mesolayer at maximum overlap to the specific heat capacity of the base fluid at the same temperature. Thus far, no microscopic theory exists, from which these parameters can be computed. Nevertheless, we have proposed a physical picture according to which κ max>1 is possible, based on induced enhanced anharmonicity of the molecular interactions, as well as a mechanism for the reduction of κ to κ min, via ‘solidification’ of the base liquid due to confinement.
This work was supported through DFG-grant HE 1545/18-1.
- 4.Choi SUS (1995) Enhancing thermal conductivity of fluids with nanoparticles In: Developments and Applications of Non-Newtonian Flows vol. ASME Publications FED-Vol. 231/MD-Vol. 66, 99–105.. The American Society of Mechanical Engineers, New York.Google Scholar
- 5.Xuan Y, Li Q (2000) Heat transfer enhancement of nanofluids. Int Commun Heat Mass 21: 58–64.Google Scholar
- 16.Shin D, Jo B, Kwak H, Banerjee D (2010) Investigation of high temperature nanofluids for solar thermal power conversion and storage applications In: 14th International Heat Transfer Conference, Washington, DC, 583–591.. ASME, New York.Google Scholar
- 21.Shin D, Banerjee D (2010) Effects of silica nanoparticles on enhancing the specific heat capacity of carbonate salt eutectic (work in progress). Int J Struct Changes Solids - Mech Appl 2: 25–31.Google Scholar
- 31.Lasfargues M (2014) Nitrate based high temperature nano-heat-transfer-fluids: formulation & characterisation. PhD thesis. University of Leeds.Google Scholar
- 33.Heilmann P (2013) Evaluation, neuentwicklung und optimierung des eigenschaftsprofils von salzschmelzen für die verwendung als wärmeträgerfluide. PhD thesis. Bergische Universität, Wuppertal.Google Scholar
- 37.Paul TC (2014) Investigation of thermal performance of nanoparticle enhanced ionic liquids (NEILs) for solar collector applications. PhD thesis. University of South Carolina.Google Scholar
- 41.Evans DF, Wennerström H (1994) The colloidal domain. VCH, New York.Google Scholar
- 42.Stanley HE (1971) Introduction to phase transitions and critical phenomena. Oxford University Press, New York.Google Scholar
- 44.Israelachvili J (1991) Intermolecular and surface forces. Academic Press, London.Google Scholar
- 47.Felderhof BU (1990) Hydrodynamic of suspensions. In: van Beijren H (ed)Fundamental Problems in Statistical Mechanics VII.. North-Holland, Amsterdam.Google Scholar
- 50.Landau LD, Lifshitz EM (1980) Statistical physics (Vol. 5). Elsevier, Oxford.Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License(http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.