Jean
Comtet
*a,
Bavand
Keshavarz
a and
John W. M.
Bush
b
aDepartment of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02138, USA. E-mail: jean.comtet@gmail.com
bDepartment of Mathematics, Massachusetts Institute of Technology, Cambridge, MA 02138, USA
First published on 1st October 2015
When a drop impacts a thin fiber, a critical impact speed can be defined, below which the drop is entirely captured by the fiber, and above which the drop pinches-off and fractures. We discuss here the capture dynamics of both inviscid and viscous drops on flexible fibers free to deform following impact. We characterize the impact-induced elongation of the drop thread for both high and low viscosity drops, and show that the capture dynamics depends on the relative magnitudes of the bending time of the fiber and deformation time of the drop. In particular, when these two timescales are comparable, drop capture is less prevalent, since the fiber rebounds when the drop deformation is maximal. Conversely, larger elasticity and slower bending time favor drop capture, as fiber rebound happens only after the drop has started to recoil. Finally, in the limit of highly flexible fibers, drop capture depends solely on the relative speed between the drop and the fiber directly after impact, as is prescribed by the momentum transferred during impact. Because the fiber speed directly after impact decreases with increasing fiber length and fiber mass, our study identifies an optimal fiber length for maximizing the efficiency of droplet capture.
The problem of drop impact on thin elongated structures, namely rods or fibers, was first examined by Hung and Yao9 and Patel et al.10 Lorenceau et al.11 quantified the critical velocity threshold between capture and fragmentation for inviscid droplets impacting a fixed fiber (Fig. 1b). For inviscid drops, V* was shown to depend on the relative size of the drop and fiber, and to increase when the impact occurs on an inclined fiber.12 Numerous other studies have been devoted to the optimization of fog harvesting structures, via alteration of mesh geometry,13 surface chemistry14–16 and fiber microstructure,5,6,17,18 with little attention being given to the initial capture stage.
To the best of our knowledge, the influence of fiber flexibility on the efficiency of capture of impacting droplets has yet to be considered. It is well-known that droplets can significantly alter the equilibrium shape of thin fibers even in static situations.19,20 In many cases of man-made and natural structures, the deformability of the structure may also play a critical role in droplet capture. Indeed, flexible substrates21 and membranes22,23 have been shown to delay or reduce splashing and jetting during drop impact, and can also significantly alter the drop fragmentation dynamics.24 Flexible beams have been used to measure forces of impacting drops25 and the surface chemistry of the beams has a significant impact on energy transfer during impact.26 In most of these studies, structural flexibility seems to act as a damper; however, it may also act to feed energy back into the system, as may arise during drop impact and fragmentation on plant leaves.24
We first describe the relevant timescales and lengthscales at play during impact, and how they are expected to interact. We then consider the impact of a drop on a rigid fiber and study the resulting dynamics for both high and low viscosity drops. Finally, we consider the role of fiber flexibility on droplet capture.
We consider the limit where the fiber radius is small relative to drop radius, and restrict ourselves to the case for which the center of mass of the drop is aligned with the center of the fiber before impact (Fig. 1b). This configuration allows us to define a critical capture speed,11 which we denote by V*. For impact speeds V < V*, the drops will be caught on the fiber (i), and for V > V*, the drops will fracture and pinch-off (ii). Fig. 2 shows typical impact sequences for the case of (i) capture and (ii) fracture. The fiber can substantially deform during impact, thus potentially playing an important role in the capture process in general and the critical impact speed V* in particular. Changing the fiber length l allows us to systematically change the elastic response of the fiber. In the case of droplet capture, the drop hangs below the fiber, at an equilibrium offset length Leq ∼ ρgR4/σa set by a balance between surface tension and drop weight. In the case of droplet fracture, some fraction of the drop remains on the fiber.
We first consider the case of a rigid fiber. As the drop hits the fiber, one expects inertia and gravity to favor drop fracture and escape; viscosity and surface tension to favor capture. The relative magnitudes of the two capture forces is prescribed by the Ohnesorge number, which we define here as (where the factor 3 is Trouton's ratio and characterizes the extensional viscosity of a thread), while the relative magnitude of the forces favoring escape and fracture is characterized by the Froude number Fr = V2/gR ≳ 1. Following impact, the drop elongates to a length L(t) over a characteristic elongation timescale τe (Fig. 1d and 2). For Oh ≪ 1, one expects τe to correspond to a typical inertia-capillary timescale ; when Oh > 1, one expects τe to correspond to the timescale of viscous momentum diffusion in the drop τv = ρR2/3η, and drop recoil to arise over the typical viscocapillary time τvc = 3ηR/σ. Relevant timescales in the impact process are summarized in Table 1.
We then consider drop impact on flexible fibers. Upon impact, we expect the fiber to deform by an amount prescribed by the transfer of momentum between the drop and the fiber during impact over a typical natural bending time with l the fiber length, ρf its density, E the Young modulus, a the fiber radius and I the area moment of inertia. Since fiber bending stores elastic energy, one expects that capture should in general be favored. However, as we shall see, elasticity may also feed energy back into the system at a critical time, thereby encouraging droplet fracture.
Fig. 3 Impact on a rigid fiber (inset). Variation of the critical capture speed V* (m s−1) with drop viscosity η (lower axis) and Ohnersorge number (upper axis). The dashed line is the expected capture speed for Oh ≪ 1 using the expression from Lorenceau et al.11 For Oh ≥ 1, V* increases linearly with viscosity. Drop radius is R = 850 μm and fiber radius a = 127 μm. Error bars characterize the uncertainty in the impact speed prevalent at low speeds. |
To understand the dynamics of impact, we show in Fig. 4 the typical elongation dynamics of drops impacting fibers in the two limits Oh ≪ 1 and Oh > 1. Following impact, the drop is stretched and elongates over a typical elongation timescale τe, before either (i) recoiling over a time τr or (ii) fracturing. For Oh ≪ 1, the elongation dynamics is symmetric, as both elongation time τe and recoil time τr scale as the typical inertia-capillary or Rayleigh timescale . When Oh > 1, the drop speed is first reduced over a viscous timescale τv = ρR2/3η; the thread then recoils with a characteristic visco-capillary time τvc = 3η/σR. In this case, the elongation dynamics is highly asymmetric, as the ratio between the recoil and elongation times, τr/τe scales as Oh2 > 1.
Fig. 4 Kymograph (spatiotemporal diagrams) of impact on a rigid fiber. Vertical slices of video images of 1 pixel width passing through the drop centerline are placed side by side with time increasing from left to right. (a) η = 1 cPs (Oh ≪ 1) and (b) η = 190 cPs (Oh > 1). The drop elongates for a time τe, and then either (i) recoils over a time τr to its equilibrium offset length Leq, or (ii) fractures. For Oh ≪ 1, τe ∼ τr is , while for Oh > 1, τe ∼ ρR2/3η and τr ∼ 3ηR/σ (cf.Table 1). |
Fig. 5 Impact on a flexible fiber. (a) Kymograph for the impact of a viscous drop (η = 100 cst) of speed V on a fiber of radius a = 127 μm and length l = 5 cm. The fiber starts oscillating with a typical timescale τb, amplitude δ0 and initial velocity Vfiber. (b) Variation of the typical bending timescale τb with fiber length L. We find τb ∼ l2, as expected on the basis of eqn (1). Black points: impacts for Oh ≪ 1, with a = 63.5 μm and a fiber tip of 250 μm. Red points: impacts for Oh > 1, with a = 127 μm. |
(1) |
(2) |
Fig. 6 Momentum transfer during impact. Variation of post-impact initial fiber speed Vfiber with drop impact speed V. (a and b) Low viscosities drops (η = 1 cPs), fiber radius is a = 63.5 μm and radius of the fiber tip is 215 μm. (a) Fiber length l is 8 cm. For impact speeds leading to drop fracture (black points), Vfiber is independent of drop impact speed V. (b) Vfiber decreases with increasing fiber length and fiber mass, according to eqn (4). (c and d) High viscosities drops (η > 100 cPs), fiber radius is a = 127 μm. (c) Vfiber is proportional to V and independent of liquid viscosity. Fiber length is 60 μm. (d) The coefficient of proportionality r = Vfiber/V decreases with increasing fiber length according to eqn (5). In (b) and (d), error bars correspond to one typical standard deviation. The dashed lines are fits to the data points (see text for details). |
To rationalize this behaviour, let us consider the processes accompanying impact. Because convection time R/V is typically small relative to the fiber response time τb, fiber elasticity can be neglected when considering the impact dynamics,25 and we can express the transfer of momentum between the drop and the fiber during impact as follows:
MeffVfiber ∼ FΔt | (3) |
For low viscosities, when the impact speed is too large for the drop to be caught on the fiber, the drop crosses the fiber with a contact time Δt ∼ 2R/V inversely proportional to the impact speed. Rewriting eqn (3), this condition predicts an initial fiber speed independent of the drop impact speed, as is apparent in the black points of Fig. 6a:
(4) |
As expected from eqn (4), Vfiber decreases with increasing fiber length, that is, larger effective fiber mass Meff (Fig. 6b). The dashed line in Fig. 6b is the best fit to the experimental points, using the expression Vfiber ≈ α·8πη(2R)2V/(ρfπa2l/3), with α ≈ 3 a fitting parameter. For the lowest impact speeds, where the drop is captured by the fiber (Fig. 6a, red points), eqn (4) breaks-down due to an increase in the interaction time between the fiber and the drop, which leads to enhanced momentum transfer. In this case, for which Δt ≈ τic, fiber speed is expected to grow linearly with drop impact speed: (Fig. 6a, red points).
For large viscosities, whether impact leads to capture or fracture, the drop initially “sticks” to the fiber, transferring momentum to the fiber for a time Δt ∼ ρR2/3η corresponding to the characteristic time of viscous penetration of the drop on the fiber. Rewriting eqn (3), we obtain:
(5) |
To understand the phase diagrams of Fig. 7 in term of the interaction of the drop and fiber, we present in Fig. 8 typical kymographs of capture events for Oh > 1. For short fibers (Fig. 8a), small fiber oscillations arise over a characteristic time τb short relative to the characteristic elongation time τe of the drop. In this limit we expect to recover the static critical capture condition (Fig. 8a, τb < τe, V* ≈ V0*). As fiber flexibility increases, we observe a decrease in the critical capture speed (Fig. 7, zone 1), which reaches a minimum when the elongation and bending timescales are of the same order. In this critical case, the deformed fiber begins to rebound just as the drop is reaching its maximal length, thus precipitating fracture at a speed lower than in the static fiber case (Fig. 8b, τb ∼ τe, V* < V0*). Thereafter, V* then increases progressively with flexibility, and exceeds that on a stiff fiber V0* (Fig. 7, zone 2). In this regime, some of the kinetic energy of impact is stored as elastic energy by the fiber, and is restored as the drop recoils (τb > τe, V* > V0*).
Fig. 8 Experimental kymographs for drop impact on a flexible fiber in the Oh > 1 case. (a) Short fiber. Fiber oscillations are small in amplitude and arise over a timescale short relative to the typical drop deformation time τe. (b) Intermediate flexibility. The fiber rebounds as the drop achieves its maximum deformation (Fig. 7, transition 1–2). (c) Long fibers. The fiber follows the drop, and rebounds only once the drop has recoiled, thus increasing the capture speed relative to the rigid-fiber case (Fig. 7, zone 3). |
Finally, for long fibers, or large bending period (Fig. 8c, τb ≳ 10·τe, zone 3) the critical capture speed saturates. In this limit, fiber oscillations are decorrelated from the drop temporal dynamics, and the fiber is carried with the drop at its initial velocity Vfiber, independent of fiber flexibility (see Fig. 6). This long-fiber limit leads to a maximal increase in the capture speed; since the fiber simply follows the drop, the initial elongation rate of the drop (0) = V − Vfiber can be greatly reduced relative to that arising in the static fiber case. For Oh ≪ 1, the initial fiber speed Vfiber is independent of drop impact speed (Fig. 6a), and we expect the capture speed V* to be increased by an amount Vfiber relative to that on a stiff fiber V0*. For Oh > 1, the ratio r = Vfiber/Vdrop is constant and we thus expect V* = V0*/(1 − r). These two predictions are consistent with our experimental data, and show the critical importance of momentum transfer between the drop and the fiber in optimizing capture efficiency.
When the fiber is free to bend, the impacting drop may excite oscillations of the fiber at its natural period τb, with an amplitude set by momentum transfer during impact (Fig. 5). For large viscosities, the impact is inelastic, and the initial fiber speed Vfiber is proportional to the drop speed V. For low viscosities, the contact time varies inversely with the impact speed, leading to an initial fiber speed Vfiber that is independent of drop speed V (Fig. 6). In both cases, momentum transfer is independent of fiber elasticity, but depends on the fiber length.
The ratio of the drop elongation timescale τe and the fiber bending time τb plays a critical role when considering capture criteria. For both viscous and inviscid drops, the critical capture speed varies non-monotically with fiber flexibility. In particular, the capture speed V* reaches a minimum when the fiber's bending time is comparable to the drop's elongation time, as the fiber begins to rebound just as the drop is reaching its maximal length, thus precipitating fracture. Here, the flexible structure does not act strictly as a damper, but instead promotes fragmentation.
For larger flexibility, the fiber begins to rebound only after the drop has started recoiling, leading to an increase in the critical capture speed. Beyond a critical flexibility, the fiber temporal dynamics occurs over a time much larger than the fiber elongation time, and the critical capture speed plateaus. In this limit, the fiber follows the drop with a constant speed, prescribed by the momentum transferred during impact, that decreases with increasing fiber length and fiber mass. We can thus define an optimal fiber length for capture. Specifically, we require that the bending time be large enough for fiber rebound to occur after drop recoil, and that the fiber mass Meff ∼ ρfa2l be small enough to allow maximal momentum transfer between the drop and the fiber, thereby reducing their relative speed. The fiber radius a and Young modulus E are thus critical parameters in attaining the regime τb ≳ 10·τe while maintaining a low fiber mass (τe ≈ 5 ms for a 1 mm water drop). We note that locally increasing the radius at the point of impact allows for large momentum transfer during impact, without reducing the capture efficiency associated with large drop-to-fiber aspect ratio.11
Droplet capture can thus be significantly enhanced by large fiber flexibility. This finding informs applications in which flexible structures are used to recover aerosols, as it provides a straightforward way to boost droplet recovery rates. Finally, we note that fiber surface chemistry and roughness will also affect criteria for droplet capture on flexible fibers, as will the detailed geometry of impact. Such effects are left for future consideration.
This journal is © The Royal Society of Chemistry 2016 |