Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Thermally activated vapor bubble nucleation: the Landau-Lifshitz/Van der Waals approach

Thermally activated vapor bubble nucleation: the Landau-Lifshitz/Van der Waals approach Thermally activated vapor bubble nucleation: the Landau–Lifshitz/Van der Waals approach Mirko Gallo, Francesco Magaletti, and Carlo Massimo Casciola Department of Mechanical and Aerospace Engineering, Sapienza Università di Roma, Rome, Italy (Dated: September 17, 2018) Vapor bubbles are formed in liquids by two mechanisms: evaporation (temperature above the boiling threshold) and cavitation (pressure below the vapor pressure). The liquid resists in these metastable (overheating and tensile, respectively) states for a long time since bubble nucleation is an activated process that needs to surmount the free energy barrier separating the liquid and the vapor states. The bubble nucleation rate is difficult to assess and, typically, only for extremely small systems treated at atomistic level of detail. In this work a powerful approach, based on a continuum diffuse interface modeling of the two-phase fluid embedded with thermal fluctuations (Fluctuating Hydrodynamics) is exploited to study the nucleation process in homogeneous conditions, evaluating the bubble nucleation rates and following the long term dynamics of the metastable system, up to the bubble coalescence and expansion stages. In comparison with more classical approaches, this methodology allows on the one hand to deal with much larger systems observed for a much longer times than possible with even the most advanced atomistic models. On the other it extends contin- uum formulations to thermally activated processes, impossible to deal with in a purely determinist setting. Thermal fluctuations play a dominant role in the dy- phase vapor–liquid system [25] similar to the one recently namics of fluid systems below the micrometer scale. exploited by Chaudhri et al. [26] to address the spin- Their effects are significant in, e.g., the smallest micro- odal decomposition. The thermodynamic range of ap- fluidic devices [1, 2] or in biological systems such as lipid plicability of this approach is subjected to some restric- membranes [3], for Brownian engines and in artificial tions: i) at the very first stage of nucleation the vapor molecular motors [4]. They are crucial for thermally ac- nucleii, smaller than the critical size, need to be numer- tivated processes such as nucleation, the precursor of the ically resolved; analogously, ii) the thin liquid-vapor in- phase change in metastable systems. Nucleation is di- terface needs to be captured for the correct evaluation of rectly connected to the phenomenon of bubble cavitation the capillary stresses; iii) fluctuating hydrodynamics pre- [5] and of freezing rain [6], to cite a few. There, ther- dicts that the fluctuation intensity grows with the inverse mal fluctuations allow to overcome the energy barriers cell volume, V , leading to intense fluctuations, con- for phase transitions [7–9]. Depending on the thermody- trary to the assumption of weak noise needed to derive namic conditions, the nucleation time may be exceedingly the model ( hf i=hfi  1). Notwithstanding these long, the so-called “rare-event” issue. Classical nucleation restrictions, where it can be applied, this mesoscale ap- theory (CNT) [10] provides the basic understanding of proach offers a good level of accuracy (as will be shown the phenomenon which is nowadays addressed through when discussing the results) at a very cheap computa- more sophisticated models like density functional theory tional cost compared to other techniques. The typical (DFT) [11, 12] or by means of molecular dynamics (MD) size of the system we simulate on a small computational simulations [13]. These approaches need to be coupled cluster (200 200 200 nm , corresponding to a system to specialized techniques for rare events, like the string of order 10 atomistic particles) is comparable with one method [14], the forward flux sampling [15] and the tran- of the largest MD simulations [27] on a tier-0 machine. sition path sampling [16], to reliably evaluate the nucle- Moreover the simulated time is here T  s to be com- max ation barrier and determine the transition path [17]. For pared with the MD T  ns. The enormous difference max many real systems they are often computationally too between the two time extensions allows us to address the expensive and therefore limited to very small domains. simultaneous nucleation of several vapor bubbles, their expansion, coalescence and, at variance with most of the Here we adopt a mesoscopic continuum approach, em- available methods dealing with quasi-static conditions, bedding stochastic fluctuations, for the numerical simula- the resulting excitation of the macroscopic velocity field. tion of thermally activated bubble nucleation. Since the pioneering work of Landau and Lifshitz (1958, 1959) [18] The diffuse interface modeling adopted here has a several works contributed to the growing field of “Fluc- strict relationship with more fundamental atomistic ap- tuating Hydrodynamics” (FH) [19]. More recently the proaches, since it is based on a suitable approximation theoretical effort has been followed by a flourishing of of the free energy functional derived in DFT [12]. It highly specialized numerical methods for the treatment of dates back to the famous Van der Waals square gradi- the stochastic contributions [20–23]. The present model ent approximation of the Helmholtz free energy func- is based on a diffuse interface [24] description of the two– tional F [; ] = dV (f (; ) + 1=2r  r), where arXiv:1801.06817v1 [cond-mat.soft] 21 Jan 2018 2 f is the classical bulk free energy, expressed as a func- specific heat at constant volume and k =  c as ther- 0 r r vr tion of density  and temperature .  is the cap- mal conductivity. The only dimensionless control group 2 2 illarity coefficient that controls the (equilibrium) sur- is the capillary number C =  = U , fixed here to face tension and interface thickness, e.g. () = C = 5:244 in order to reproduce the LJ surface tension R sat p l [32]. The symbol will be omitted in the following to sat 2 [! () ! ( )]d with ! (; ) the bulk sat 0 0 0 simplify notation. grand potential per unit volume and the superscript sat The system volume V = (600) has been discretized on denoting the (temperature dependent) saturation condi- an equi-spaced grid with 50 cells per direction. Critical tions, see Supplemental Material [28]. The mesoscopic nucleus size, interface thickness and transition time are fields obey mass, momentum and energy conservation, crucial to prepare a well resolved simulation. The prelim- with the addition of stochastic contributions (Lifshitz- inary set-up was based on the expected transition time Landau-Navier-Stokes equations with capillarity): derived from the free energy barriers and on the size R of the critical nucleus at the different thermodynamic con- +r (u) = 0 ; (1) @t ditions, as estimated from the string method [14] applied @u to the present system. Technical details are available in +r (u u) = rp +r  +r  ; @t the SM [28]. The comparison with the predictions of @E CNT are summarized in Tab. I. It turns out that, for the +r (uE) = r (pu + u  q) + present relatively large system, there is no quantitative @t +r (u  q) ; difference between the barriers estimated from CNT in the different ensembles, and we may use the grand canon- CNT where u is the fluid velocity, p =  @(f =)=@ is 0 2 ical expression = 4=3 R . After a convergence the pressure, E is the total energy density, E = U + analysis we found that a grid size x = 12 is sufficient for 2 2 1=2juj + 1=2jrj , with U the internal energy den- a reliable simulation in these thermodynamic conditions, sity. In the momentum and energy equations,  and q see SM [28]. Thanks to the extension of the simulated are the classical deterministic stress tensor and energy domain, ten runs for each condition provide a reasonably flux, respectively, and the terms with the prefix  are the well converged statistics. stochastic parts, required to satisfy a suitable fluctuation- Among the different conditions we have investigated, dissipation balance (FDB). The deterministic stress ten- we mainly focused on the initial temperature  = 1:25 sor  and the energy flux q follow by standard non- at changing bulk density to explore the corresponding equilibrium thermodynamic considerations [29] as:  = metastable range  2 [ ;  ] = [0:44; 0:51], where 2 T L spin sat (=2jrj +r (r))Ir r+[(ru +ru ) and  are the saturation and spinodal densities, sat spin 2=3r  uI ], q = rr  u kr, with  and k the respectively. A few snapshots of the evolution for two dynamic viscosity and the thermal conductivity, respec- different initial conditions are shown in the top panels of tively. Enforcing the FDB, the covariance of the stochas- y Fig.s 1 and 2. Starting from a homogeneous metastable ^ ~ tic fluxes follows as, h(x; ^ t) (x; ~ t)i = Q (x ^ y liquid phase, the fluctuations lead the system to spon- ^ ~ ^ ~ ^ ~ x ~)(tt), and hq(x; ^ t) q (x; ~ t)i = Q (x ^x ~)(tt), taneously nucleate vapor bubbles. The nucleii start out where Q = 2k  (  +   2=3  ) with a complex, far from spherical, shape, see, e.g., [13]. q 2 and Q = 2k  k , with k the Boltzmann con- B B Roughly, when they happen to reach a size larger than stant (see the Supplemental Material [28] for details). critical they typically expand. Eventually, after a long Thanks to the Curie-Prigogine principle [30], the cross- and complex dynamics where bubbles expand and co- correlation between different tensor rank fluxes vanishes, ~ ^ i.e. ( q (x; ~ t) (x; ^ t) = 0). Bubble nucleation is investigated in a metastable liq- CNT CNT g g R R 0 L c c 0 0 uid enclosed in a cubic box with periodic boundary con- 1:25 0:45 12:8 8:07 1:68 12:89 ditions, with fixed volume, total mass and energy (NVE). The equation of state (EoS) we use, which can be cho- 1:25 0:46 10:2 8:42 5:90 14:05 sen freely among available models, e.g. van der Waals 1:25 0:47 9:5 9:17 13:90 16:67 or IAPWS [31] EoS for water, corresponds to a Lennard- 1:25 0:48 10:1 10:64 29:22 22:41 Jones (LJ) fluid [32] to allow direct comparison with MD 1:20 0:51 7:1 6:35 10:55 18:13 simulations. Quantities are made dimensionless accord- 1:20 0:52 7:0 6:93 20:60 21:60 ing to  = = ,  = = , u = u=U , by introducing r r r as reference quantities the parameters of the LJ poten- Table I. Comparison between CNT and the string method ap- 10 21 tial,  = 3:4  10 m as length,  = 1:65  10 J plied to the Diffuse Interface model. Critical radii and (Lan- 26 1=2 as energy, m = 6:63 10 kg as mass, U = (=m) dau) free energy barriers for bubble nucleation from the as velocity, T = =U as time,  = =k as temper- liquid. The discrepancy close to the spinodal is well known r r r B from the literature. ature,  = m= as shear viscosity, c = mk as r vr B 3 Figure 1. Top panel: bubble configurations along nucleation Figure 2. Same as Figure 1 at  = 0:48;  = 1:25. Snapshots ( = 0:46;  = 1:25), from left to right t = 400, t = 2000, taken at t = 2000, t = 11000, t = 230000. The bubble number t = 25000. Animation available in [28]. Bottom panel: bubble vs time in the bottom panel is fitted by the dotted red line number evolution (red symbols) and number of coalescence for better readability. events (blue line). competing-growth mechanism [33] due to the constraint of constant mass, explaining the high number of super- alesce, stable equilibrium conditions are reached. This critical bubble collapses. The coalescence events start new configuration is characterized by several stable va- being less and less probable during the slowly-expanding por bubbles in equilibrium with the surrounding liquid. regime that characterizes the long-time dynamics of the The case at  = 0:46, the closest one to the spinodal we multi-bubble system. The inset of the Fig. 1 zooms into considered here, is the more populated, Fig. 1 in compar- this regime showing that isolated collision events are still ison with Fig. 2. This system has a lower barrier, hence occurring, contributing to important acceleration toward it nucleates faster. The initial (metastable) thermody- the final equilibrium condition. namic condition also influences the number and typical The volume history of the distinct bubbles (in par- dimension of the bubbles in the final stage, bottom pan- ticular those survived up to the last time investigated) els of Fig.s 1 and 2 providing the bubble number N as a have been plotted in Fig. 3. Among the different bub- function of time. A tracking procedure has been put for- ble evolutions, we highlighted in red the volume histo- ward to follow the evolution of the distinct bubbles. By ries of those bubbles that experienced intense coalescence monitoring volume, mass, center of mass and its velocity, events, characterized by a sudden increase in volume. It the tracking algorithm allows to detect coalescence events is apparent that the larger bubbles gained substantial (see [28] for the procedure details). The actual number part of their volume by coalescence. To substantiate this of collisions N evaluated at each time step is char- coll impression, for each bubble in the last configuration, the acterized by a highly discontinuous fingerprint, hence we sum of the volumes acquired by coalescence throughout smoothed it out with a Gaussian kernel with standard de- viation on the order of 50 time units to gain more robust the whole evolution, V , was calculated, inset of Fig. 3. coll indications. The smoothed number of collisions N , The present mesoscale approach allows to access the coll plotted with the blue line in the bottom panel of the statistics of bubble dimensions. The probability distribu- Fig.s 1 and 2, shows a strong correlation with the num- tion function of bubble volumes f (V ) is plotted in Fig. 4. ber of bubbles throughout the initial nucleating stage – During both the nucleating and collapsing stages the pdf when N grows linearly with time (i.e. at a constant is sharply peaked at small volumes, of the order of 2–4 nucleation rate) – and the first part of the expansion V . The successive bubble expansion phase is substan- phase – characterized by a rapidly decreasing number of tially slower and calls for a much longer observation time bubbles mainly due to collapse – that we will call col- to detect a significant growth (green dash-dotted curve lapsing stage. These two stages are characterized by a at t = 34000). The intense coalescence events explain 4 Figure 3. Volume history of the bubbles surviving the entire simulation ( = 0:46 and  = 1:25). Intense coalescence events, characterized by a sudden volume jump, are identified in the red curves. The corresponding volumes are shown by the red dots in the inset providing the V V scatter plot, coll where V is the volume acquired by coalescence. The largest coll Figure 5. Nucleation rates: FH simulations (red squares at bubbles experienced intense coalescence events. = 1:25 and blue squares at  = 1:20) vs CNT (green 0 0 circles at  = 1:25 and purple circles at  = 1:20). The 0 0 inset shows the comparison with other authors. one in Fig. 5 which also provides the comparison with some MD simulations [13, 34]. Our results are in reason- able agreement with molecular dynamic simulations in the NVE ensemble [13], as shown in the inset of Fig. 5, and consistent with the order of magnitude predicted by CNT. The apparent discrepancy that, contrary to expec- tation, the nucleation rate is smaller where the barriers are smaller than CNT, may be explained by the compres- sion induced by bubble nucleation in the NVE ensemble. In conclusion, the FH approach together with a diffuse interface modeling of the multiphase system have been exploited to study homogeneous nucleation of vapor bub- bles in metastable liquids. We evaluated the nucleation rate and compared it favorably with state of the art simu- lations and theories. The present technique has revealed Figure 4. Probability distribution function f (V ) of the bubble extremely cheaper with respect to MD simulations, allow- volumes during the nucleation, at different times (  = 1:25, ing the analysis of the very long bubble expansion stage = 0:46, critical volume V = 4445). where bubble-bubble interaction/coalescence events turn out to determine the eventual bubble size distribution. The accurate results and the efficiency of the modeling the presence of the second peak in the pdf at very large encourage the exploitation to more complex conditions, volume (black curve in the inset of Fig. 4 at t = 163760). like e.g. heterogeneous nucleation and multi-species sys- The initial nucleating stage, where the bubble num- tems, and could pave the way for the development of ber increases linearly, gives access to the nucleation rate innovative continuum formulation to address thermally J in terms of bubbles formed per unit time and vol- activated processes. ume. It is here calculated as the slope of the linear fit to the curves of Fig.s 1 and 2 near the origin. The The research leading to these results has received fund- CNT expression for the (dimensional) nucleation rate is ing from the European Research Council under the Eu- J = n 2 =m exp( =k ), where n is the ropean Union’s Seventh Framework Programme (FP7/ CNT L B L liquid number density. It is compared with the measured 2007-2013)/ERC Grant agreement no. [339446]. 5 ciola, Langmuir 28, 10764 (2012). [18] L. Landau and E. Lifshitz, Course of theoretical physics, Volume 5, part I, 3rd Edition (1980). carlomassimo.casciola@uniroma1.it; [19] R. F. Fox and G. E. Uhlenbeck, The Physics of Fluids http://flumacs.site.uniroma1.it 13, 1893 (1970). [1] F. Detcheverry and L. Bocquet, Physical review letters [20] A. Donev, E. Vanden-Eijnden, A. Garcia, and J. Bell, 109, 024501 (2012). Communications in Applied Mathematics and Computa- [2] L. Bocquet and E. Charlaix, Chemical Society Reviews tional Science 5, 149 (2010). 39, 1073 (2010). [21] S. Delong, B. E. Griffith, E. Vanden-Eijnden, and [3] A. Naji, P. J. Atzberger, and F. L. Brown, Physical A. Donev, Physical Review E 87, 033302 (2013). review letters 102, 138102 (2009). [22] F. Balboa, J. B. Bell, R. Delgado-Buscalioni, A. Donev, [4] C. S. Peskin, G. M. Odell, and G. F. Oster, Biophysical T. G. Fai, B. E. Griffith, and C. S. Peskin, Multiscale journal 65, 316 (1993). Modeling & Simulation 10, 1369 (2012). [5] C. E. Brennen, Cavitation and bubble dynamics (Cam- [23] A. Donev, A. Nonaka, Y. Sun, T. Fai, A. Garcia, and bridge University Press, 2013). J. Bell, Communications in Applied Mathematics and [6] L. Cao, A. K. Jones, V. K. Sikka, J. Wu, and D. Gao, Computational Science 9, 47 (2014). Langmuir 25, 12444 (2009). [24] J. Van der Waals, Journal of Statistical Physics 20, 200 [7] S. Jones, G. Evans, and K. Galvin, Advances in colloid (1979). and interface science 80, 27 (1999). [25] F. Magaletti, L. Marino, and C. Casciola, Physical Re- [8] D. Kashchiev and G. Van Rosmalen, Crystal Research view Letters 114, 064501 (2015). and Technology 38, 555 (2003). [26] A. Chaudhri, J. B. Bell, A. L. Garcia, and A. Donev, [9] D. Lohse and A. Prosperetti, Proceedings of the National Physical Review E 90, 033014 (2014). Academy of Sciences 113, 13549 (2016). [27] R. Angélil, J. Diemand, K. K. Tanaka, and H. Tanaka, [10] M. Blander and J. L. Katz, AIChE Journal 21, 833 Physical review E 90, 063301 (2014). (1975). [28] see Supplemental Material at http://correct.site. [11] D. W. Oxtoby and R. Evans, The Journal of chemical url for details, . physics 89, 7521 (1988). [29] F. Magaletti, M. Gallo, L. Marino, and C. M. Casciola, [12] J. F. Lutsko, The Journal of chemical physics 134, International Journal of Multiphase Flow 84, 34 (2016). 164501 (2011). [30] S. R. De Groot and P. Mazur, Non-equilibrium thermo- [13] J. Diemand, R. Angélil, K. K. Tanaka, and H. Tanaka, dynamics (Courier Dover Publications, 2013). Physical review E 90, 052407 (2014). [31] J. Kestin, J. Sengers, B. Kamgar-Parsi, and J. L. Sen- [14] E. Weinan, W. Ren, and E. Vanden-Eijnden, Physical gers, Journal of Physical and Chemical Reference Data Review B 66, 052301 (2002). 13, 175 (1984). [15] R. J. Allen, C. Valeriani, and P. R. ten Wolde, Journal [32] J. K. Johnson, J. A. Zollweg, and K. E. Gubbins, Molec- of physics: Condensed matter 21, 463102 (2009). ular Physics 78, 591 (1993). [16] P. G. Bolhuis, D. Chandler, C. Dellago, and P. L. [33] S.-i. Tsuda, S. Takagi, and Y. Matsumoto, Fluid dynam- Geissler, Annual review of physical chemistry 53, 291 ics research 40, 606 (2008). (2002). [34] B. R. Novak, E. J. Maginn, and M. J. McCready, Phys. [17] A. Giacomello, S. Meloni, M. Chinappi, and C. M. Cas- Rev. B 75, 085413 (2007). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Condensed Matter arXiv (Cornell University)

Thermally activated vapor bubble nucleation: the Landau-Lifshitz/Van der Waals approach

Download PDF
5 pages

Loading next page...
 
/lp/arxiv-cornell-university/thermally-activated-vapor-bubble-nucleation-the-landau-lifshitz-van-LO7EgbC3L5

References (58)

ISSN
2469-990X
eISSN
ARCH-3331
DOI
10.1103/PhysRevFluids.3.053604
Publisher site
See Article on Publisher Site

Abstract

Thermally activated vapor bubble nucleation: the Landau–Lifshitz/Van der Waals approach Mirko Gallo, Francesco Magaletti, and Carlo Massimo Casciola Department of Mechanical and Aerospace Engineering, Sapienza Università di Roma, Rome, Italy (Dated: September 17, 2018) Vapor bubbles are formed in liquids by two mechanisms: evaporation (temperature above the boiling threshold) and cavitation (pressure below the vapor pressure). The liquid resists in these metastable (overheating and tensile, respectively) states for a long time since bubble nucleation is an activated process that needs to surmount the free energy barrier separating the liquid and the vapor states. The bubble nucleation rate is difficult to assess and, typically, only for extremely small systems treated at atomistic level of detail. In this work a powerful approach, based on a continuum diffuse interface modeling of the two-phase fluid embedded with thermal fluctuations (Fluctuating Hydrodynamics) is exploited to study the nucleation process in homogeneous conditions, evaluating the bubble nucleation rates and following the long term dynamics of the metastable system, up to the bubble coalescence and expansion stages. In comparison with more classical approaches, this methodology allows on the one hand to deal with much larger systems observed for a much longer times than possible with even the most advanced atomistic models. On the other it extends contin- uum formulations to thermally activated processes, impossible to deal with in a purely determinist setting. Thermal fluctuations play a dominant role in the dy- phase vapor–liquid system [25] similar to the one recently namics of fluid systems below the micrometer scale. exploited by Chaudhri et al. [26] to address the spin- Their effects are significant in, e.g., the smallest micro- odal decomposition. The thermodynamic range of ap- fluidic devices [1, 2] or in biological systems such as lipid plicability of this approach is subjected to some restric- membranes [3], for Brownian engines and in artificial tions: i) at the very first stage of nucleation the vapor molecular motors [4]. They are crucial for thermally ac- nucleii, smaller than the critical size, need to be numer- tivated processes such as nucleation, the precursor of the ically resolved; analogously, ii) the thin liquid-vapor in- phase change in metastable systems. Nucleation is di- terface needs to be captured for the correct evaluation of rectly connected to the phenomenon of bubble cavitation the capillary stresses; iii) fluctuating hydrodynamics pre- [5] and of freezing rain [6], to cite a few. There, ther- dicts that the fluctuation intensity grows with the inverse mal fluctuations allow to overcome the energy barriers cell volume, V , leading to intense fluctuations, con- for phase transitions [7–9]. Depending on the thermody- trary to the assumption of weak noise needed to derive namic conditions, the nucleation time may be exceedingly the model ( hf i=hfi  1). Notwithstanding these long, the so-called “rare-event” issue. Classical nucleation restrictions, where it can be applied, this mesoscale ap- theory (CNT) [10] provides the basic understanding of proach offers a good level of accuracy (as will be shown the phenomenon which is nowadays addressed through when discussing the results) at a very cheap computa- more sophisticated models like density functional theory tional cost compared to other techniques. The typical (DFT) [11, 12] or by means of molecular dynamics (MD) size of the system we simulate on a small computational simulations [13]. These approaches need to be coupled cluster (200 200 200 nm , corresponding to a system to specialized techniques for rare events, like the string of order 10 atomistic particles) is comparable with one method [14], the forward flux sampling [15] and the tran- of the largest MD simulations [27] on a tier-0 machine. sition path sampling [16], to reliably evaluate the nucle- Moreover the simulated time is here T  s to be com- max ation barrier and determine the transition path [17]. For pared with the MD T  ns. The enormous difference max many real systems they are often computationally too between the two time extensions allows us to address the expensive and therefore limited to very small domains. simultaneous nucleation of several vapor bubbles, their expansion, coalescence and, at variance with most of the Here we adopt a mesoscopic continuum approach, em- available methods dealing with quasi-static conditions, bedding stochastic fluctuations, for the numerical simula- the resulting excitation of the macroscopic velocity field. tion of thermally activated bubble nucleation. Since the pioneering work of Landau and Lifshitz (1958, 1959) [18] The diffuse interface modeling adopted here has a several works contributed to the growing field of “Fluc- strict relationship with more fundamental atomistic ap- tuating Hydrodynamics” (FH) [19]. More recently the proaches, since it is based on a suitable approximation theoretical effort has been followed by a flourishing of of the free energy functional derived in DFT [12]. It highly specialized numerical methods for the treatment of dates back to the famous Van der Waals square gradi- the stochastic contributions [20–23]. The present model ent approximation of the Helmholtz free energy func- is based on a diffuse interface [24] description of the two– tional F [; ] = dV (f (; ) + 1=2r  r), where arXiv:1801.06817v1 [cond-mat.soft] 21 Jan 2018 2 f is the classical bulk free energy, expressed as a func- specific heat at constant volume and k =  c as ther- 0 r r vr tion of density  and temperature .  is the cap- mal conductivity. The only dimensionless control group 2 2 illarity coefficient that controls the (equilibrium) sur- is the capillary number C =  = U , fixed here to face tension and interface thickness, e.g. () = C = 5:244 in order to reproduce the LJ surface tension R sat p l [32]. The symbol will be omitted in the following to sat 2 [! () ! ( )]d with ! (; ) the bulk sat 0 0 0 simplify notation. grand potential per unit volume and the superscript sat The system volume V = (600) has been discretized on denoting the (temperature dependent) saturation condi- an equi-spaced grid with 50 cells per direction. Critical tions, see Supplemental Material [28]. The mesoscopic nucleus size, interface thickness and transition time are fields obey mass, momentum and energy conservation, crucial to prepare a well resolved simulation. The prelim- with the addition of stochastic contributions (Lifshitz- inary set-up was based on the expected transition time Landau-Navier-Stokes equations with capillarity): derived from the free energy barriers and on the size R of the critical nucleus at the different thermodynamic con- +r (u) = 0 ; (1) @t ditions, as estimated from the string method [14] applied @u to the present system. Technical details are available in +r (u u) = rp +r  +r  ; @t the SM [28]. The comparison with the predictions of @E CNT are summarized in Tab. I. It turns out that, for the +r (uE) = r (pu + u  q) + present relatively large system, there is no quantitative @t +r (u  q) ; difference between the barriers estimated from CNT in the different ensembles, and we may use the grand canon- CNT where u is the fluid velocity, p =  @(f =)=@ is 0 2 ical expression = 4=3 R . After a convergence the pressure, E is the total energy density, E = U + analysis we found that a grid size x = 12 is sufficient for 2 2 1=2juj + 1=2jrj , with U the internal energy den- a reliable simulation in these thermodynamic conditions, sity. In the momentum and energy equations,  and q see SM [28]. Thanks to the extension of the simulated are the classical deterministic stress tensor and energy domain, ten runs for each condition provide a reasonably flux, respectively, and the terms with the prefix  are the well converged statistics. stochastic parts, required to satisfy a suitable fluctuation- Among the different conditions we have investigated, dissipation balance (FDB). The deterministic stress ten- we mainly focused on the initial temperature  = 1:25 sor  and the energy flux q follow by standard non- at changing bulk density to explore the corresponding equilibrium thermodynamic considerations [29] as:  = metastable range  2 [ ;  ] = [0:44; 0:51], where 2 T L spin sat (=2jrj +r (r))Ir r+[(ru +ru ) and  are the saturation and spinodal densities, sat spin 2=3r  uI ], q = rr  u kr, with  and k the respectively. A few snapshots of the evolution for two dynamic viscosity and the thermal conductivity, respec- different initial conditions are shown in the top panels of tively. Enforcing the FDB, the covariance of the stochas- y Fig.s 1 and 2. Starting from a homogeneous metastable ^ ~ tic fluxes follows as, h(x; ^ t) (x; ~ t)i = Q (x ^ y liquid phase, the fluctuations lead the system to spon- ^ ~ ^ ~ ^ ~ x ~)(tt), and hq(x; ^ t) q (x; ~ t)i = Q (x ^x ~)(tt), taneously nucleate vapor bubbles. The nucleii start out where Q = 2k  (  +   2=3  ) with a complex, far from spherical, shape, see, e.g., [13]. q 2 and Q = 2k  k , with k the Boltzmann con- B B Roughly, when they happen to reach a size larger than stant (see the Supplemental Material [28] for details). critical they typically expand. Eventually, after a long Thanks to the Curie-Prigogine principle [30], the cross- and complex dynamics where bubbles expand and co- correlation between different tensor rank fluxes vanishes, ~ ^ i.e. ( q (x; ~ t) (x; ^ t) = 0). Bubble nucleation is investigated in a metastable liq- CNT CNT g g R R 0 L c c 0 0 uid enclosed in a cubic box with periodic boundary con- 1:25 0:45 12:8 8:07 1:68 12:89 ditions, with fixed volume, total mass and energy (NVE). The equation of state (EoS) we use, which can be cho- 1:25 0:46 10:2 8:42 5:90 14:05 sen freely among available models, e.g. van der Waals 1:25 0:47 9:5 9:17 13:90 16:67 or IAPWS [31] EoS for water, corresponds to a Lennard- 1:25 0:48 10:1 10:64 29:22 22:41 Jones (LJ) fluid [32] to allow direct comparison with MD 1:20 0:51 7:1 6:35 10:55 18:13 simulations. Quantities are made dimensionless accord- 1:20 0:52 7:0 6:93 20:60 21:60 ing to  = = ,  = = , u = u=U , by introducing r r r as reference quantities the parameters of the LJ poten- Table I. Comparison between CNT and the string method ap- 10 21 tial,  = 3:4  10 m as length,  = 1:65  10 J plied to the Diffuse Interface model. Critical radii and (Lan- 26 1=2 as energy, m = 6:63 10 kg as mass, U = (=m) dau) free energy barriers for bubble nucleation from the as velocity, T = =U as time,  = =k as temper- liquid. The discrepancy close to the spinodal is well known r r r B from the literature. ature,  = m= as shear viscosity, c = mk as r vr B 3 Figure 1. Top panel: bubble configurations along nucleation Figure 2. Same as Figure 1 at  = 0:48;  = 1:25. Snapshots ( = 0:46;  = 1:25), from left to right t = 400, t = 2000, taken at t = 2000, t = 11000, t = 230000. The bubble number t = 25000. Animation available in [28]. Bottom panel: bubble vs time in the bottom panel is fitted by the dotted red line number evolution (red symbols) and number of coalescence for better readability. events (blue line). competing-growth mechanism [33] due to the constraint of constant mass, explaining the high number of super- alesce, stable equilibrium conditions are reached. This critical bubble collapses. The coalescence events start new configuration is characterized by several stable va- being less and less probable during the slowly-expanding por bubbles in equilibrium with the surrounding liquid. regime that characterizes the long-time dynamics of the The case at  = 0:46, the closest one to the spinodal we multi-bubble system. The inset of the Fig. 1 zooms into considered here, is the more populated, Fig. 1 in compar- this regime showing that isolated collision events are still ison with Fig. 2. This system has a lower barrier, hence occurring, contributing to important acceleration toward it nucleates faster. The initial (metastable) thermody- the final equilibrium condition. namic condition also influences the number and typical The volume history of the distinct bubbles (in par- dimension of the bubbles in the final stage, bottom pan- ticular those survived up to the last time investigated) els of Fig.s 1 and 2 providing the bubble number N as a have been plotted in Fig. 3. Among the different bub- function of time. A tracking procedure has been put for- ble evolutions, we highlighted in red the volume histo- ward to follow the evolution of the distinct bubbles. By ries of those bubbles that experienced intense coalescence monitoring volume, mass, center of mass and its velocity, events, characterized by a sudden increase in volume. It the tracking algorithm allows to detect coalescence events is apparent that the larger bubbles gained substantial (see [28] for the procedure details). The actual number part of their volume by coalescence. To substantiate this of collisions N evaluated at each time step is char- coll impression, for each bubble in the last configuration, the acterized by a highly discontinuous fingerprint, hence we sum of the volumes acquired by coalescence throughout smoothed it out with a Gaussian kernel with standard de- viation on the order of 50 time units to gain more robust the whole evolution, V , was calculated, inset of Fig. 3. coll indications. The smoothed number of collisions N , The present mesoscale approach allows to access the coll plotted with the blue line in the bottom panel of the statistics of bubble dimensions. The probability distribu- Fig.s 1 and 2, shows a strong correlation with the num- tion function of bubble volumes f (V ) is plotted in Fig. 4. ber of bubbles throughout the initial nucleating stage – During both the nucleating and collapsing stages the pdf when N grows linearly with time (i.e. at a constant is sharply peaked at small volumes, of the order of 2–4 nucleation rate) – and the first part of the expansion V . The successive bubble expansion phase is substan- phase – characterized by a rapidly decreasing number of tially slower and calls for a much longer observation time bubbles mainly due to collapse – that we will call col- to detect a significant growth (green dash-dotted curve lapsing stage. These two stages are characterized by a at t = 34000). The intense coalescence events explain 4 Figure 3. Volume history of the bubbles surviving the entire simulation ( = 0:46 and  = 1:25). Intense coalescence events, characterized by a sudden volume jump, are identified in the red curves. The corresponding volumes are shown by the red dots in the inset providing the V V scatter plot, coll where V is the volume acquired by coalescence. The largest coll Figure 5. Nucleation rates: FH simulations (red squares at bubbles experienced intense coalescence events. = 1:25 and blue squares at  = 1:20) vs CNT (green 0 0 circles at  = 1:25 and purple circles at  = 1:20). The 0 0 inset shows the comparison with other authors. one in Fig. 5 which also provides the comparison with some MD simulations [13, 34]. Our results are in reason- able agreement with molecular dynamic simulations in the NVE ensemble [13], as shown in the inset of Fig. 5, and consistent with the order of magnitude predicted by CNT. The apparent discrepancy that, contrary to expec- tation, the nucleation rate is smaller where the barriers are smaller than CNT, may be explained by the compres- sion induced by bubble nucleation in the NVE ensemble. In conclusion, the FH approach together with a diffuse interface modeling of the multiphase system have been exploited to study homogeneous nucleation of vapor bub- bles in metastable liquids. We evaluated the nucleation rate and compared it favorably with state of the art simu- lations and theories. The present technique has revealed Figure 4. Probability distribution function f (V ) of the bubble extremely cheaper with respect to MD simulations, allow- volumes during the nucleation, at different times (  = 1:25, ing the analysis of the very long bubble expansion stage = 0:46, critical volume V = 4445). where bubble-bubble interaction/coalescence events turn out to determine the eventual bubble size distribution. The accurate results and the efficiency of the modeling the presence of the second peak in the pdf at very large encourage the exploitation to more complex conditions, volume (black curve in the inset of Fig. 4 at t = 163760). like e.g. heterogeneous nucleation and multi-species sys- The initial nucleating stage, where the bubble num- tems, and could pave the way for the development of ber increases linearly, gives access to the nucleation rate innovative continuum formulation to address thermally J in terms of bubbles formed per unit time and vol- activated processes. ume. It is here calculated as the slope of the linear fit to the curves of Fig.s 1 and 2 near the origin. The The research leading to these results has received fund- CNT expression for the (dimensional) nucleation rate is ing from the European Research Council under the Eu- J = n 2 =m exp( =k ), where n is the ropean Union’s Seventh Framework Programme (FP7/ CNT L B L liquid number density. It is compared with the measured 2007-2013)/ERC Grant agreement no. [339446]. 5 ciola, Langmuir 28, 10764 (2012). [18] L. Landau and E. Lifshitz, Course of theoretical physics, Volume 5, part I, 3rd Edition (1980). carlomassimo.casciola@uniroma1.it; [19] R. F. Fox and G. E. Uhlenbeck, The Physics of Fluids http://flumacs.site.uniroma1.it 13, 1893 (1970). [1] F. Detcheverry and L. Bocquet, Physical review letters [20] A. Donev, E. Vanden-Eijnden, A. Garcia, and J. Bell, 109, 024501 (2012). Communications in Applied Mathematics and Computa- [2] L. Bocquet and E. Charlaix, Chemical Society Reviews tional Science 5, 149 (2010). 39, 1073 (2010). [21] S. Delong, B. E. Griffith, E. Vanden-Eijnden, and [3] A. Naji, P. J. Atzberger, and F. L. Brown, Physical A. Donev, Physical Review E 87, 033302 (2013). review letters 102, 138102 (2009). [22] F. Balboa, J. B. Bell, R. Delgado-Buscalioni, A. Donev, [4] C. S. Peskin, G. M. Odell, and G. F. Oster, Biophysical T. G. Fai, B. E. Griffith, and C. S. Peskin, Multiscale journal 65, 316 (1993). Modeling & Simulation 10, 1369 (2012). [5] C. E. Brennen, Cavitation and bubble dynamics (Cam- [23] A. Donev, A. Nonaka, Y. Sun, T. Fai, A. Garcia, and bridge University Press, 2013). J. Bell, Communications in Applied Mathematics and [6] L. Cao, A. K. Jones, V. K. Sikka, J. Wu, and D. Gao, Computational Science 9, 47 (2014). Langmuir 25, 12444 (2009). [24] J. Van der Waals, Journal of Statistical Physics 20, 200 [7] S. Jones, G. Evans, and K. Galvin, Advances in colloid (1979). and interface science 80, 27 (1999). [25] F. Magaletti, L. Marino, and C. Casciola, Physical Re- [8] D. Kashchiev and G. Van Rosmalen, Crystal Research view Letters 114, 064501 (2015). and Technology 38, 555 (2003). [26] A. Chaudhri, J. B. Bell, A. L. Garcia, and A. Donev, [9] D. Lohse and A. Prosperetti, Proceedings of the National Physical Review E 90, 033014 (2014). Academy of Sciences 113, 13549 (2016). [27] R. Angélil, J. Diemand, K. K. Tanaka, and H. Tanaka, [10] M. Blander and J. L. Katz, AIChE Journal 21, 833 Physical review E 90, 063301 (2014). (1975). [28] see Supplemental Material at http://correct.site. [11] D. W. Oxtoby and R. Evans, The Journal of chemical url for details, . physics 89, 7521 (1988). [29] F. Magaletti, M. Gallo, L. Marino, and C. M. Casciola, [12] J. F. Lutsko, The Journal of chemical physics 134, International Journal of Multiphase Flow 84, 34 (2016). 164501 (2011). [30] S. R. De Groot and P. Mazur, Non-equilibrium thermo- [13] J. Diemand, R. Angélil, K. K. Tanaka, and H. Tanaka, dynamics (Courier Dover Publications, 2013). Physical review E 90, 052407 (2014). [31] J. Kestin, J. Sengers, B. Kamgar-Parsi, and J. L. Sen- [14] E. Weinan, W. Ren, and E. Vanden-Eijnden, Physical gers, Journal of Physical and Chemical Reference Data Review B 66, 052301 (2002). 13, 175 (1984). [15] R. J. Allen, C. Valeriani, and P. R. ten Wolde, Journal [32] J. K. Johnson, J. A. Zollweg, and K. E. Gubbins, Molec- of physics: Condensed matter 21, 463102 (2009). ular Physics 78, 591 (1993). [16] P. G. Bolhuis, D. Chandler, C. Dellago, and P. L. [33] S.-i. Tsuda, S. Takagi, and Y. Matsumoto, Fluid dynam- Geissler, Annual review of physical chemistry 53, 291 ics research 40, 606 (2008). (2002). [34] B. R. Novak, E. J. Maginn, and M. J. McCready, Phys. [17] A. Giacomello, S. Meloni, M. Chinappi, and C. M. Cas- Rev. B 75, 085413 (2007).

Journal

Condensed Matter – arXiv (Cornell University)

Published: Jan 21, 2018

There are no references for this article.