WorksPhysics and Astronomy
Aerosol generation by the splashing of low viscosity drops impacting liquid layers
Abstract
Using theory and numerical simulations, here we describe the early stages of the impact with a velocity of a drop of radius of a low viscosity liquid such as water against a layer of generic thickness of the same liquid. Our predictions for the initial velocity and the diameter of the toroidal rim bordering the edge of the thin lamella which is ejected radially outwards after the impact, are in fair agreement with both the numerical results and also with the experimental measurements reported by Zhang et al [J. Fluid Mech, 690, 5, 2012]. Consequently, the present findings can be employed, for instance, to predict the initial diameters and velocities of the fastest tiny droplets which are ejected right after a rain drop falls on a liquid pool, an ubiquitous phenomenon with implications in the dispersal of contaminants and in the generation of aerosols and of ice condensation nuclei
One of the main physical processes involved in the dispersal of contaminants and in the generation of sea spray aerosol is bubble bursting, a phenomenon which can be reliably quantified using very recent results in the literature [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]. While the production of aerosol particles by raindrop impact on dry soil is also driven by bubble bursting [13], the falling of raindrops on wet forests, puddles, ponds, lakes [14, 15] or the sea [16, 17, 18] induce the dispersion of contaminants [15] and the production of bioaerosols and of ice condensation nuclei [19] through a different physical mechanism namely, the ejection and subsequent breakup of the very thin and fast lamella which is issued shortly after the contact between the two liquid masses, see Figure 1, a physical process known as drop splashing. However, in spite of the number of experimental, theoretical, and numerical studies devoted to analyze the impact of a drop on a liquid layer [1, 16, 17, 18, 20, 21, 22, 23, 24, 25, 26, 27], much less is known about the crucial early stages taking place when raindrops fall on a liquid pool because none of the numerous and recent studies enumerated above addresses the discrepancies between the very precise experimental measurements reported by [1] over a decade ago using high speed X-ray images and the different quantitative analyses reported so far. Indeed, while the classical theoretical result used to quantify the initial velocity of the lamella ejected after the impact depicted in Figure 1 predicts that [21] , with indicating the liquid kinematic viscosity and the impact velocity of the drop, the experiments in [1] do not support the prediction in [21] but, instead, suggest a best fit to the data of the type . Motivated by this fact and with the purpose of describing the very early stages of drop impact on a liquid layer we develop a framework similar to the one used to quantify the splashing of drops impacting solid surfaces [30]. The results obtained will then be used to predict the initial velocity and the initial radius of curvature of the toroidal rim bordering the ejected liquid sheet, which breaks due to the growth of
Figure 1. Numerical simulation using Basilisk [28, 29] showing the impact with a velocity of a drop of radius , density and viscosity onto a liquid film of depth of the same liquid for , , , and .
azimuthal disturbances [1, 20, 31, 32, 33, 34], provoking an early splash characterized by the ejection of fast, micron-sized droplets [1, 20], with velocities and diameters [33], see Figure 1. Our predictions will be validated using the experimental measurements reported by [1, 20] and using the results of numerical simulations carried out using Basilisk [28, 29], see the Supplementary Material for details.
We first analyze the case of a liquid pool of thickness in terms of the dimensionless variables and parameters defined using , , and as the characteristic values of length, velocity, time and pressure, with , and indicating, respectively, density, viscosity and the interfacial tension coefficient. In the following, lower-case letters will denote dimensionless variables to differentiate them from their dimensional counterparts, written in capitals (e.g. , ) and the subscript will be used to denote gas quantities: indeed, the numerical results also depend on the ratios and , which are kept constant and equal to and unless otherwise specified.
The type of splashing analyzed here occurs within the two-jets regime reported in [1], which produces the spray reported in [20] for liquids with viscosities equal or similar to that of water. The spray is composed of micron-sized secondary droplets emitted from the toroidal rim bordering the lamella, which destabilizes due to the growth of azimuthal disturbances [1, 20, 31, 32, 33, 34]. Since splashing and the disintegration of the rim bordering the ejected lamella only occurs when the values of both the Weber and Reynolds numbers are such that [1], , with , for the cases of interest here, vorticity is confined within very narrow regions at the free interfaces, a fact implying that the dimensionless velocity field is irrotational outside the very thin boundary layers namely, with indicating the dimensionless velocity potential. Consequently, by virtue of the continuity equation , satisfies the Laplace equation . The potential flow will be described next for in a frame of reference moving with the vertical velocity ( in dimensionless terms), with the origin of times set at the instant when the bottom of the drop would contact the liquid layer if the gas was not present. In the moving frame of reference, the free interfaces of both the bottom layer and of the impacting drop—which in the moving frame of reference and for is a parabola of equation —are located, neglecting higher order terms [22, 25], at , with indicating the linearized free interface of the drop and the corresponding linearized position of the initially flat film. The solution of the Laplace equation subjected to the far field conditions , to the impenetrability condition and to the boundary conditions at the free interfaces —deduced using the Euler-Bernoulli equation in the limit [10, 22, 25, 30, 35], is symmetric with respect to —see the sketch in Figure 2—and yields the following analytical expression for the vertical velocity at the linearized free interfaces [30, 36]:
Figure 2. (a) Sketch showing the impact of a drop on a liquid layer in a frame of reference moving vertically with the velocity ; the dashed line indicates the parabola along which the two liquid masses collide. (b) Sketch of the linearized boundary conditions to be satisfied by , whose solution provides the vertical velocities given in Eq. (1), the key ingredient behind the Wagner condition [35] in Eq. (2).
In Eq. (1) indicates the dimensionless radius of the circular region, the wet region in what follows, where the two liquid masses in Figure 2 collide, entrapping a bubble of dimensionless radius , see Figure 1. The radius is calculated through the so-called Wagner condition [35], which is nothing but the time integral of the kinematic boundary condition at the free interfaces and permits to calculate the time taken by a material point located initially at , i.e., a point on the drop interface at , to reach the bottom liquid layer. Due to the fact that the vertical velocities at the linearized free interfaces are symmetric with respect to , the two liquid masses will collide at the midpoint between their initial positions namely, at —see Figure 2—and, therefore, using the expression for the vertical velocity at the upper free interface given in Eq. (1), the Wagner condition reads [35]:
which is the equation for deduced and solved in [30], where it was found that . Consequently, due to the fact that both the local velocity field and the thickness of the lamella depend on through , the results found in [30] are also applicable here and, hence, the radial velocity of the wet region is also ; the liquid velocity at the root of the lamella, see Figure 3, which can be deduced using the Euler-Bernoulli equation along a streamline at the free interface in a frame of reference moving with the radial velocity , is also in the laboratory frame of reference. Note that the solution for the velocity field at the free interface given in Eq. (1) is not uniformly valid because it diverges for where , with , and, hence, the term needs to be retained in the Euler-Bernoulli equation in an inner region namely, a lamella of thickness , where the blow-up of the outer velocity field in the limit of Eq. (1) is regularized. The thickness is calculated imposing that the flow rate through the lamella, in the inner region, is the one posed by the outer solution at the root of the lamella namely, , a classical result due to Wagner [22, 25, 35]. We then follow the same steps as in [30] and, similarly to the case of impact on a solid substrate, we also note from Figure 3 that the edge of the lamella is ejected at a radial velocity that coincides with the wetting velocity at the ejection time , i.e. , with deduced from the condition with denoting the dimensionless acceleration of the edge of the lamella. The deceleration is caused by the combined effect of both viscous and capillary stresses and is calculated by means of the momentum equation projected in the radial direction, [30], with the diameter of the edge of the lamella depicted in Figure 1 and Figure 3. Consequently, the equation for the ejection time is identical to the one deduced in [30]:
Figure 3. (a) In blue, numerical values obtained using Basilisk for the radial, , and vertical, , positions of the edge of the lamella defined in Figure 1, see also the Supplementary Material for further details. The thick black line is the theoretical prediction whereas the yellow lines, tangent to and , permit to determine the radial and vertical velocities of the edge of the lamella, and respectively, at , an instant of time at which the lamella follows a ballistic trajectory follows a straight line- and, hence, retains its velocity at the ejection instant predicted by the solution of Eq. (3), see also panel (b). (b) Numerical shape of the lamella at corresponding to the thin vertical black line in figure (a). The yellow circle of diameter , see also Figure 1, is the best fit to the local shape of the rim bordering the lamella. The dashed line corresponds to the parabola in Figure 2a but, in this case, it is represented in the laboratory frame of reference i.e., . In (a)-(b), , , .
with a constant coming from the factor in that was set to in [30] for the case of drops impacting solid substrates; here, and, for simplicity, here
we set in Eq. (3) and will absorb the dependence of on by adjusting a factor of order unity, as will be discussed below; notice that, for the reasons in [37, 38], the entrapped bubble does not appreciably influence the value of calculated through Eq. (3) which, in the limit of interest here yields . All the numerical values of , and in Figure 4-Figure 5 are calculated at , see Figure 3 because, at this instant of time, the lamella follows a ballistic trajectory [30, 33] that retains its initial velocity at when , see also [39].
Figure 4. Main: with the experimental values in Ref. [1] [ZD] and Ref. [20] [Th] for wide ranges of values ; with the ejection time calculated solving Eq. (3)- for the experimental data points in [20] [Th] because the data in this case represent drop velocities: indeed, using the experiments in [40], it was shown in [33] that the secondary drops emitted directly from the root of the lamella are ejected with the velocity . The values of represented using +, indicate numerical results; see the Supplementary Material for additional numerical data. Inset: numerical values of for and as a function of : notice that for over two decades in . Due to the fact that splash does not occur for [1, 26], in the figure.
Figure 5. Numerical values of (a) and (b), with and given by Eq. (3). The numerical results in (b) corresponding to , and to , have been calculated for and . The insets show the dependence on of: (a) , (b) .
Figure 4 compares the experiments of [1, 20] and our numerical results with the prediction , with given by the solution of Eq. (3), for wide ranges of values of , and finding that, for the conditions under which the rim disintegrates generating a spray namely, when [1], with relative errors . The results in Figure 4 also reveal that the dependence on of the initial radial velocity of the ejecta sheet is very small, and the reason for this lies in the fact that, except for in Eq. (3), the expressions for and in the limit are identical to those found for the case of drops impacting a rigid substrate considered in [30], a case that can be approached in the limit . The comparison between the numerical values of with in Figure 5a reveals that , with relative errors for over two decades in when [1]; note that the small deviations with respect to our prediction in Figure 5a are a consequence of the thickening of the rim caused by the fact that relative volume flux into the rim is [33] during the time interval , with and .
The initial value of the vertical velocity in Figure 3 can also be predicted using Wagner theory [35] in the limit , see also [25]. Indeed, the lamella is ejected at the midpoint between the initial position of the drop and the bottom liquid layer namely, at the parabola of equation -see Figure 2- and consequently, in the moving frame of reference, which means that, in the laboratory frame of reference, the vertical velocity at the ejection instant is , a value which is much smaller than the tangential velocity, with
because, in the limit , see Eq. (3). Figure 5b shows that and, in order to check if the deviations of with respect to are associated with gas effects [17, 31, 41], we have performed simulations at reduced values of and , finding only slight differences in the values of but not on , see Figure 5b as well as the Supplementary Material. Then, the reason why in Figure 5b is because the prediction refers to the initial value at , whereas the values depicted in Figure 5b are measured at . Indeed, for instants of time , the radial velocity at the root of the lamella is and, consequently, the vertical velocity at in the moving frame of reference is , which means that in the laboratory frame of reference. Then, the results in Figure 5b are a consequence of the relative flux of vertical momentum entering into the rim, which is accelerated vertically from to .
To conclude, here we have made use of Wagner’s theory [35] and its extension accounting for capillary and viscosity effects [30], with the purpose of quantifying the early stages of a drop impacting a liquid layer of arbitrary thickness of a low viscosity liquid such as water. Our main finding here is that, since the diameters of the droplets ejected from the rim bordering the thin ejecta sheet are proportional to the diameter [32, 33, 42] and the radial and vertical initial velocities of such droplets, and respectively, are the ones of the toroidal rim [33], we find that, in the usual limit , , , , with the dimensionless ejection time given by Eq. (3), which depends very weakly on the film thickness and recovers the analogous equation deduced in [30] in the limit . Our quantitative description remains valid within the range of dimensionless parameters characterizing the fall of raindrops on a liquid layer of generic depth: for instance, for a rain drop of radius mm, m s, where we assumed that the drag coefficient of the falling drop is , a spray composed of droplets with diameters m is ejected initially into the atmosphere with a vertical velocity m s and a tangential velocity m s (see Figure 4 and [20]), generating an aerosol containing particles with diameters m. At later times, , particles with diameters m will also be sprayed: indeed, since the radius of the wetted area increases in time as , the rim thickness grows in time as and the liquid velocity flowing into the lamella decays in time as , the flux of particles of diameter which will be issued at the instant into the surrounding atmosphere is approximately given by with defined in [43]; all the micron-sized solid particles emitted by drop splashing will be later on dispersed by wind. Hence, our results are applicable to describe the aerosol generated by the early splash of rain drops falling over liquid layers of arbitrary depth, an ubiquitous process with implications in climate [19] and in the dispersal of contaminants [14, 15].
Acknowledgements
This work has been supported by the Grant PID2024-156545NB-I00, financed by the Spanish MICIU/AEI/10.13039/501100011033 and by ERDF/EU.
References
- [1]L. V. Zhang, J. Toole, K. Fezzaa, and R. D. Deegan, Evolution of the ejecta sheet from the impact of a drop with a deep pool, Journal of Fluid Mechanics 690, 5–15 (2012).DOI
- [2]X. Wang, G. B. Deane, K. A. Moore, O. S. Ryder, M. D. Stokes, C. M. Beall, D. B. Collins, M. V. Santander, S. M. Burrows, C. M. Sultana, and K. A. Prather, The role of jet and film drops in controlling the mixing state of submicron sea spray aerosol particles, Proceedings of the National Academy of Sciences 114, 6978 (2017).DOI
- [3]A. Gañán Calvo, Revision of bubble bursting: Universal scaling laws of top jet drop size and speed, Phys. Rev. Lett. 119, 204502 (2017).DOI
- [4]J. M. Gordillo and J. Rodríguez-Rodríguez, Capillary waves control the ejection of bubble bursting jets, Journal of Fluid Mechanics 867, 557 (2019).DOI
- [5]F. J. Blanco-Rodríguez and J. M. Gordillo, On the sea spray aerosol originated from bubble bursting jets, Journal of Fluid Mechanics 886, R2 (2020).
- [6]L. Dubitsky, O. McRae, and J. C. Bird, Enrichment of scavenged particles in jet drops determined by bubble size and particle position, Phys. Rev. Lett. 130, 054001 (2023).DOI
- [7]Z. Yang, B. Ji, J. T. Ault, and J. Feng, Enhanced singular jet formation in oil-coated bubble bursting, Nat. Phys. (2023).DOI
- [8]L. Deike, Mass transfer at the ocean–atmosphere interface: The role of wave breaking, droplets, and bubbles, Annual Review of Fluid Mechanics 54, 191 (2022).DOI
- [9]X. Jiang, L. Rotily, E. Villermaux, and X. Wang, Submicron drops from flapping bursting bubbles, Proceedings of the National Academy of Sciences 119, e211924119 (2022).DOI
- [10]J. M. Gordillo and F. J. Blanco-Rodríguez, Theory of the jets ejected after the inertial collapse of cavities with applications to bubble bursting jets, Phys. Rev. Fluids 8, 073606 (2023).DOI
- [11]X. Jiang, L. Rotily, E. Villermaux, and X. Wang, Abyss aerosols: Drop production from underwater bubble collisions, Phys. Rev. Lett. 133, 024001 (2024).DOI
- [12]W. Chu, R. Zhang, X. Jiang, E. Villermaux, and X. Wang, Critical role of sea spray aerosol production pathways in particulate transfer across the sea-air interface, Journal of Geophysical Research: Atmospheres 130, e2025JD044168 (2025).
- [13]Y. Joung and C. Buie, Aerosol generation by raindrop impact on soil, Nat. Comm. 6, 6083 (2015).DOI
- [14]B. D. L. Fith, A. A. McCartney, and P. J. Walklate, The role of rain in dispersal of pathogen inoculum, Annual Review of Phytopathology 27, 241 (1989).
- [15]P. A. Paul, S. M. El-Allaf, P. E. Lipps, and L. V. Madden, Rain splash dispersal of gibberella zeae within wheat canopies in ohio, Phytopathology 94, 1342 (2004).DOI
- [16]D. W. Murphy, C. Li, V. d’Albignac, D. Morra, and J. Katz, Splash behavior and oily marine aerosol production by raindrops impacting oil slicks, Journal of Fluid Mechanics 780, 536–577 (2015).DOI
- [17]Y. S. Tian, A. B. Aldejani, T. Alghamdi, and S. T. Thoroddsen, Dancing ejecta, Journal of Fluid Mechanics 981, A4 (2024).DOI
- [18]W. Tran, G. Lubimov, J. Belden, N. Speirs, and J. Marston, Effects of particulate properties on splash activity and particulate destination for particle-laden drop impacts on liquid pools, Experiments in Fluids 67, 36 (2026).DOI
- [19]J. A. Huffman, A. J. Prenni, P. J. DeMott, C. Pöhlker, R. H. Mason, N. H. Robinson, J. Fröhlich-Nowoisky, Y. Tobo, V. R. Després, E. Garcia, D. J. Gochis, E. Harris, I. Müller-Germann, C. Ruzene, B. Schmer, B. Sinha, D. A. Day, M. O. Andreae, J. L. Jimenez, M. Gallagher, S. M. Kreidenweis, A. K. Bertram, and U. Pöschl, High concentrations of biological aerosol particles and ice nuclei during and after rain, Atmospheric Chemistry and Physics 13, 6151 (2013).DOI
- [20]S. T. Thoroddsen, The ejecta sheet generated by the impact of a drop, Journal of Fluid Mechanics 451, 373–381 (2002).DOI
- [21]C. Josserand and S. Zaleski, Droplet splashing on a thin liquid film, Physics of Fluids 15, 1650 (2003).DOI
- [22]S. D. Howison, J. R. Ockendon, J. M. Oliver, R. Purvis, and F. T. Smith, Droplet impact on a thin fluid layer, Journal of Fluid Mechanics 542, 1–23 (2005).DOI
- [23]C. Josserand and S. T. Thoroddsen, Drop impact on a solid surface, Annual Review of Fluid Mechanics 48, 365 (2016).DOI
- [24]C. Josserand, P. Ray, and S. Zaleski, Droplet impact on a thin liquid film: anatomy of the splash, Journal of Fluid Mechanics 802, 775–805 (2016).DOI
- [25]R. Cimpeanu and M. Moore, Early-time jet formation in liquid–liquid impact problems: theory and simulations, Journal of Fluid Mechanics 856, 764–796 (2018).DOI
- [26]T. C. Sykes, R. Cimpeanu, B. D. Fudge, J. R. Castrejón-Pita, and A. A. Castrejón-Pita, Droplet impact dynamics on shallow pools, Journal of Fluid Mechanics 970, A34 (2023).DOI
- [27]H. Wang, S. Liu, A.-C. Bayel-Lainé, D. Murphy, J. Katz, and O. Coutier-Delgosha, Analysis of high-speed drop impact onto deep liquid pool, Journal of Fluid Mechanics 972, A31 (2023).DOI
- [28]S. Popinet, An accurate adaptive solver for surface-tension-driven interfacial flows, Journal of Computational Physics 228, 5838 (2009).DOI
- [29]V. Sanjay, Code repository, https://github.com/VatsalSy/Impact-forces-of-water-drops-falling-on-superhydrophobic-surfaces (2022).
- [30]G. Riboux and J. M. Gordillo, Experiments of drops impacting a smooth solid surface: A model of a critical impact speed for drop splashing, Phys. Rev. Lett. 113, 024507 (2014).pubmed.ncbi.nlm.nih.gov/25062193
- [31]L. V. Zhang, J. Toole, K. Fezzaa, and R. D. Deegan, Splashing from drop impact into a deep pool: multiplicity of jets and the failure of conventional scaling, Journal of Fluid Mechanics 703, 402–413 (2012).DOI
- [32]G. Agbaglah and R. Deegan, Growth and instability of the liquid rim in the crown splash regime, Journal of Fluid Mechanics 752, 485–496 (2014).DOI
- [33]G. Riboux and J. Gordillo, The diameters and velocities of the droplets ejected after splashing, Journal of Fluid Mechanics 772, 630–648 (2015).DOI
- [34]E. Q. Li, M.-J. Thoraval, J. O. Marston, and S. T. Thoroddsen, Early azimuthal instability during drop impact, Journal of Fluid Mechanics 848, 821–835 (2018).DOI
- [35]H. Wagner, Über stoß-und gleitvorgänge an der oberfläche von flüssigkeiten, Z. Angew. Math. Mech. 12, 193 (1932).DOI
- [36]H. Lamb, Hydrodynamics (Cambridge University Press, 1932).
- [37]P. Hicks and R. Purvis, Air cushioning and bubble entrapment in three-dimensional droplet impacts, Journal of Fluid Mechanics 649, 135–163 (2010).DOI
- [38]For Oh ≪ 1, tₑ ≃ We⁻²/³; moreover, i) the radius from which the lamella is first ejected is a(tₑ) = √3 tₑ ≃ √3We⁻¹/³, ii) the radius of the entrapped bubble is r_b ≈ 6 (√We Oh⁻¹ μ/μ_g)⁻¹/³ [37]. Hence, the larger is the value of the ratio K = r_b/a(tₑ) ≃ 4 We¹/⁶ Oh¹/³ (μ_g/μ)¹/³, the larger is the influence of the entrapped bubble on the ejection of the lamella; in all cases considered here, K = r_b/a(tₑ) < 1.
- [39]For 0.1 ≤ h ≤ 0.5 and We ≥ 630, the value of t₀ > tₑ is just the instant before the rim collides with the falling droplet, t₀ ≈ 0.05.
- [40]S. T. Thoroddsen, K. Takehara, and T. G. Etoh, Micro-splashing by drop impacts, Journal of Fluid Mechanics 706, 560–570 (2012).DOI
- [41]S. T. Thoroddsen, M.-J. Thoraval, K. Takehara, and T. G. Etoh, Droplet splashing by slingshot mechanism, Phys. Rev. Lett. 106, 034501 (2011).pubmed.ncbi.nlm.nih.gov/21405277
- [42]G. Agbaglah, C. Josserand, and S. Zaleski, Longitudinal instability of a liquid rim, Physics of Fluids 25, 022103 (2013).DOI
- [43]C̄₊(Dₚ ≳ R_d t_d³/²) = 0 and C̄₊(Dₚ ≲ R_d t_d³/²) refers to the averaged value of the volume concentration of particles of diameter Dₚ existing at t = 0 at the interfaces of both the drop (+) and the liquid layer (−): C̄₊ = 1/(2πR_d t_d³/²) ∫₀²π ∫₀R_d t_d³/² C₊(θ,z,R = R_d√3t) dθ dz, with C₊ the concentration field at t = 0: indeed, R = R_d√3t is the radial position where the liquid feeding the lamella comes from [30]. C̄₋ is defined analogously.